{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999891664680519,
  "eval_steps": 500,
  "global_step": 55383,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.027943290071429e-05,
      "grad_norm": 2.874858856201172,
      "learning_rate": 1.9999999597786356e-05,
      "loss": 0.5785,
      "step": 5
    },
    {
      "epoch": 0.00018055886580142859,
      "grad_norm": 1.3120938539505005,
      "learning_rate": 1.999999839114545e-05,
      "loss": 0.5951,
      "step": 10
    },
    {
      "epoch": 0.00027083829870214285,
      "grad_norm": 1.8099818229675293,
      "learning_rate": 1.999999638007738e-05,
      "loss": 0.4052,
      "step": 15
    },
    {
      "epoch": 0.00036111773160285717,
      "grad_norm": 2.011772394180298,
      "learning_rate": 1.9999993564582305e-05,
      "loss": 0.4159,
      "step": 20
    },
    {
      "epoch": 0.00045139716450357144,
      "grad_norm": 1.8650308847427368,
      "learning_rate": 1.999998994466046e-05,
      "loss": 0.4756,
      "step": 25
    },
    {
      "epoch": 0.0005416765974042857,
      "grad_norm": 0.7753961682319641,
      "learning_rate": 1.9999985520312126e-05,
      "loss": 0.5705,
      "step": 30
    },
    {
      "epoch": 0.0006319560303050001,
      "grad_norm": 1.2066733837127686,
      "learning_rate": 1.9999980291537665e-05,
      "loss": 0.4,
      "step": 35
    },
    {
      "epoch": 0.0007222354632057143,
      "grad_norm": 1.1299852132797241,
      "learning_rate": 1.9999974258337497e-05,
      "loss": 0.5143,
      "step": 40
    },
    {
      "epoch": 0.0008125148961064286,
      "grad_norm": 1.0047327280044556,
      "learning_rate": 1.9999967420712108e-05,
      "loss": 0.4731,
      "step": 45
    },
    {
      "epoch": 0.0009027943290071429,
      "grad_norm": 1.0650899410247803,
      "learning_rate": 1.9999959778662046e-05,
      "loss": 0.4378,
      "step": 50
    },
    {
      "epoch": 0.0009930737619078573,
      "grad_norm": 1.0157448053359985,
      "learning_rate": 1.999995133218793e-05,
      "loss": 0.4402,
      "step": 55
    },
    {
      "epoch": 0.0010833531948085714,
      "grad_norm": 0.650635302066803,
      "learning_rate": 1.999994208129043e-05,
      "loss": 0.4935,
      "step": 60
    },
    {
      "epoch": 0.0011736326277092858,
      "grad_norm": 0.9630825519561768,
      "learning_rate": 1.99999320259703e-05,
      "loss": 0.3306,
      "step": 65
    },
    {
      "epoch": 0.0012639120606100002,
      "grad_norm": 0.9416863918304443,
      "learning_rate": 1.999992116622834e-05,
      "loss": 0.4445,
      "step": 70
    },
    {
      "epoch": 0.0013541914935107143,
      "grad_norm": 0.8302100896835327,
      "learning_rate": 1.9999909502065433e-05,
      "loss": 0.4469,
      "step": 75
    },
    {
      "epoch": 0.0014444709264114287,
      "grad_norm": 1.1551083326339722,
      "learning_rate": 1.999989703348251e-05,
      "loss": 0.3983,
      "step": 80
    },
    {
      "epoch": 0.0015347503593121428,
      "grad_norm": 1.1567633152008057,
      "learning_rate": 1.9999883760480585e-05,
      "loss": 0.4067,
      "step": 85
    },
    {
      "epoch": 0.0016250297922128572,
      "grad_norm": 1.1382306814193726,
      "learning_rate": 1.9999869683060708e-05,
      "loss": 0.4439,
      "step": 90
    },
    {
      "epoch": 0.0017153092251135716,
      "grad_norm": 0.9161384105682373,
      "learning_rate": 1.9999854801224025e-05,
      "loss": 0.4142,
      "step": 95
    },
    {
      "epoch": 0.0018055886580142858,
      "grad_norm": 1.4126842021942139,
      "learning_rate": 1.9999839114971727e-05,
      "loss": 0.4189,
      "step": 100
    },
    {
      "epoch": 0.0018958680909150001,
      "grad_norm": 0.44530755281448364,
      "learning_rate": 1.9999822624305083e-05,
      "loss": 0.3662,
      "step": 105
    },
    {
      "epoch": 0.0019861475238157145,
      "grad_norm": 1.3806456327438354,
      "learning_rate": 1.999980532922541e-05,
      "loss": 0.4043,
      "step": 110
    },
    {
      "epoch": 0.002076426956716429,
      "grad_norm": 0.7793863415718079,
      "learning_rate": 1.9999787229734108e-05,
      "loss": 0.4335,
      "step": 115
    },
    {
      "epoch": 0.002166706389617143,
      "grad_norm": 0.9610792398452759,
      "learning_rate": 1.9999768325832626e-05,
      "loss": 0.455,
      "step": 120
    },
    {
      "epoch": 0.002256985822517857,
      "grad_norm": 1.081809163093567,
      "learning_rate": 1.9999748617522488e-05,
      "loss": 0.3969,
      "step": 125
    },
    {
      "epoch": 0.0023472652554185716,
      "grad_norm": 1.3969955444335938,
      "learning_rate": 1.9999728104805278e-05,
      "loss": 0.4921,
      "step": 130
    },
    {
      "epoch": 0.002437544688319286,
      "grad_norm": 0.729268491268158,
      "learning_rate": 1.999970678768265e-05,
      "loss": 0.3292,
      "step": 135
    },
    {
      "epoch": 0.0025278241212200003,
      "grad_norm": 1.072990894317627,
      "learning_rate": 1.9999684666156313e-05,
      "loss": 0.4629,
      "step": 140
    },
    {
      "epoch": 0.0026181035541207143,
      "grad_norm": 1.3183773756027222,
      "learning_rate": 1.999966174022805e-05,
      "loss": 0.4235,
      "step": 145
    },
    {
      "epoch": 0.0027083829870214286,
      "grad_norm": 1.0797187089920044,
      "learning_rate": 1.9999638009899706e-05,
      "loss": 0.4071,
      "step": 150
    },
    {
      "epoch": 0.002798662419922143,
      "grad_norm": 0.5764801502227783,
      "learning_rate": 1.9999613475173184e-05,
      "loss": 0.4883,
      "step": 155
    },
    {
      "epoch": 0.0028889418528228574,
      "grad_norm": 1.0525012016296387,
      "learning_rate": 1.999958813605047e-05,
      "loss": 0.4519,
      "step": 160
    },
    {
      "epoch": 0.0029792212857235718,
      "grad_norm": 1.0718201398849487,
      "learning_rate": 1.9999561992533588e-05,
      "loss": 0.4336,
      "step": 165
    },
    {
      "epoch": 0.0030695007186242857,
      "grad_norm": 0.711060106754303,
      "learning_rate": 1.999953504462465e-05,
      "loss": 0.5076,
      "step": 170
    },
    {
      "epoch": 0.003159780151525,
      "grad_norm": 4.838633060455322,
      "learning_rate": 1.9999507292325822e-05,
      "loss": 0.4658,
      "step": 175
    },
    {
      "epoch": 0.0032500595844257144,
      "grad_norm": 1.0590788125991821,
      "learning_rate": 1.9999478735639333e-05,
      "loss": 0.3615,
      "step": 180
    },
    {
      "epoch": 0.003340339017326429,
      "grad_norm": 0.704115092754364,
      "learning_rate": 1.9999449374567484e-05,
      "loss": 0.4115,
      "step": 185
    },
    {
      "epoch": 0.003430618450227143,
      "grad_norm": 0.6360671520233154,
      "learning_rate": 1.9999419209112637e-05,
      "loss": 0.4122,
      "step": 190
    },
    {
      "epoch": 0.003520897883127857,
      "grad_norm": 0.7924306988716125,
      "learning_rate": 1.9999388239277217e-05,
      "loss": 0.4475,
      "step": 195
    },
    {
      "epoch": 0.0036111773160285715,
      "grad_norm": 1.2045217752456665,
      "learning_rate": 1.9999356465063714e-05,
      "loss": 0.4365,
      "step": 200
    },
    {
      "epoch": 0.003701456748929286,
      "grad_norm": 0.6561562418937683,
      "learning_rate": 1.9999323886474687e-05,
      "loss": 0.5271,
      "step": 205
    },
    {
      "epoch": 0.0037917361818300003,
      "grad_norm": 0.776882529258728,
      "learning_rate": 1.9999290503512755e-05,
      "loss": 0.4197,
      "step": 210
    },
    {
      "epoch": 0.0038820156147307146,
      "grad_norm": 0.7396628260612488,
      "learning_rate": 1.9999256316180604e-05,
      "loss": 0.425,
      "step": 215
    },
    {
      "epoch": 0.003972295047631429,
      "grad_norm": 0.872489333152771,
      "learning_rate": 1.9999221324480986e-05,
      "loss": 0.4407,
      "step": 220
    },
    {
      "epoch": 0.004062574480532143,
      "grad_norm": 0.542452335357666,
      "learning_rate": 1.9999185528416712e-05,
      "loss": 0.3691,
      "step": 225
    },
    {
      "epoch": 0.004152853913432858,
      "grad_norm": 0.538821816444397,
      "learning_rate": 1.9999148927990663e-05,
      "loss": 0.3504,
      "step": 230
    },
    {
      "epoch": 0.004243133346333572,
      "grad_norm": 0.6316237449645996,
      "learning_rate": 1.9999111523205787e-05,
      "loss": 0.4251,
      "step": 235
    },
    {
      "epoch": 0.004333412779234286,
      "grad_norm": 0.9031202793121338,
      "learning_rate": 1.9999073314065084e-05,
      "loss": 0.5062,
      "step": 240
    },
    {
      "epoch": 0.0044236922121350004,
      "grad_norm": 0.5457375645637512,
      "learning_rate": 1.9999034300571636e-05,
      "loss": 0.4341,
      "step": 245
    },
    {
      "epoch": 0.004513971645035714,
      "grad_norm": 1.1680082082748413,
      "learning_rate": 1.999899448272858e-05,
      "loss": 0.4187,
      "step": 250
    },
    {
      "epoch": 0.004604251077936429,
      "grad_norm": 1.056076169013977,
      "learning_rate": 1.9998953860539116e-05,
      "loss": 0.3579,
      "step": 255
    },
    {
      "epoch": 0.004694530510837143,
      "grad_norm": 0.7869812846183777,
      "learning_rate": 1.999891243400651e-05,
      "loss": 0.4457,
      "step": 260
    },
    {
      "epoch": 0.004784809943737857,
      "grad_norm": 0.7695295214653015,
      "learning_rate": 1.9998870203134104e-05,
      "loss": 0.3827,
      "step": 265
    },
    {
      "epoch": 0.004875089376638572,
      "grad_norm": 1.0620439052581787,
      "learning_rate": 1.9998827167925284e-05,
      "loss": 0.477,
      "step": 270
    },
    {
      "epoch": 0.004965368809539286,
      "grad_norm": 0.7931966185569763,
      "learning_rate": 1.999878332838352e-05,
      "loss": 0.2773,
      "step": 275
    },
    {
      "epoch": 0.005055648242440001,
      "grad_norm": 0.8393620252609253,
      "learning_rate": 1.9998738684512335e-05,
      "loss": 0.4401,
      "step": 280
    },
    {
      "epoch": 0.005145927675340715,
      "grad_norm": 0.7053926587104797,
      "learning_rate": 1.999869323631532e-05,
      "loss": 0.3441,
      "step": 285
    },
    {
      "epoch": 0.0052362071082414285,
      "grad_norm": 0.5869619846343994,
      "learning_rate": 1.999864698379613e-05,
      "loss": 0.402,
      "step": 290
    },
    {
      "epoch": 0.005326486541142143,
      "grad_norm": 0.5323655009269714,
      "learning_rate": 1.999859992695849e-05,
      "loss": 0.3474,
      "step": 295
    },
    {
      "epoch": 0.005416765974042857,
      "grad_norm": 0.8899840712547302,
      "learning_rate": 1.9998552065806183e-05,
      "loss": 0.5102,
      "step": 300
    },
    {
      "epoch": 0.005507045406943572,
      "grad_norm": 0.5503301620483398,
      "learning_rate": 1.9998503400343058e-05,
      "loss": 0.5319,
      "step": 305
    },
    {
      "epoch": 0.005597324839844286,
      "grad_norm": 0.8930386304855347,
      "learning_rate": 1.999845393057303e-05,
      "loss": 0.4279,
      "step": 310
    },
    {
      "epoch": 0.005687604272745,
      "grad_norm": 0.7680234313011169,
      "learning_rate": 1.9998403656500082e-05,
      "loss": 0.445,
      "step": 315
    },
    {
      "epoch": 0.005777883705645715,
      "grad_norm": 0.6237252354621887,
      "learning_rate": 1.999835257812825e-05,
      "loss": 0.3322,
      "step": 320
    },
    {
      "epoch": 0.005868163138546429,
      "grad_norm": 0.726040244102478,
      "learning_rate": 1.999830069546165e-05,
      "loss": 0.3773,
      "step": 325
    },
    {
      "epoch": 0.0059584425714471435,
      "grad_norm": 0.8470166921615601,
      "learning_rate": 1.9998248008504454e-05,
      "loss": 0.4471,
      "step": 330
    },
    {
      "epoch": 0.0060487220043478574,
      "grad_norm": 0.7286909222602844,
      "learning_rate": 1.99981945172609e-05,
      "loss": 0.3542,
      "step": 335
    },
    {
      "epoch": 0.006139001437248571,
      "grad_norm": 1.338836908340454,
      "learning_rate": 1.9998140221735292e-05,
      "loss": 0.2662,
      "step": 340
    },
    {
      "epoch": 0.006229280870149286,
      "grad_norm": 0.9748361706733704,
      "learning_rate": 1.9998085121931995e-05,
      "loss": 0.4134,
      "step": 345
    },
    {
      "epoch": 0.00631956030305,
      "grad_norm": 0.4502583146095276,
      "learning_rate": 1.9998029217855444e-05,
      "loss": 0.4096,
      "step": 350
    },
    {
      "epoch": 0.006409839735950715,
      "grad_norm": 0.6040946245193481,
      "learning_rate": 1.9997972509510136e-05,
      "loss": 0.3897,
      "step": 355
    },
    {
      "epoch": 0.006500119168851429,
      "grad_norm": 0.8161752223968506,
      "learning_rate": 1.999791499690063e-05,
      "loss": 0.3511,
      "step": 360
    },
    {
      "epoch": 0.006590398601752143,
      "grad_norm": 0.436606764793396,
      "learning_rate": 1.999785668003156e-05,
      "loss": 0.4545,
      "step": 365
    },
    {
      "epoch": 0.006680678034652858,
      "grad_norm": 0.7064688205718994,
      "learning_rate": 1.9997797558907604e-05,
      "loss": 0.3563,
      "step": 370
    },
    {
      "epoch": 0.006770957467553572,
      "grad_norm": 0.5320123434066772,
      "learning_rate": 1.999773763353353e-05,
      "loss": 0.4064,
      "step": 375
    },
    {
      "epoch": 0.006861236900454286,
      "grad_norm": 0.6215265393257141,
      "learning_rate": 1.999767690391415e-05,
      "loss": 0.3495,
      "step": 380
    },
    {
      "epoch": 0.006951516333355,
      "grad_norm": 0.8487299680709839,
      "learning_rate": 1.9997615370054357e-05,
      "loss": 0.3919,
      "step": 385
    },
    {
      "epoch": 0.007041795766255714,
      "grad_norm": 0.7624923586845398,
      "learning_rate": 1.9997553031959093e-05,
      "loss": 0.4063,
      "step": 390
    },
    {
      "epoch": 0.007132075199156429,
      "grad_norm": 0.5340708494186401,
      "learning_rate": 1.9997489889633385e-05,
      "loss": 0.3908,
      "step": 395
    },
    {
      "epoch": 0.007222354632057143,
      "grad_norm": 0.8394657373428345,
      "learning_rate": 1.9997425943082297e-05,
      "loss": 0.4193,
      "step": 400
    },
    {
      "epoch": 0.007312634064957858,
      "grad_norm": 4.428290367126465,
      "learning_rate": 1.999736119231098e-05,
      "loss": 0.5262,
      "step": 405
    },
    {
      "epoch": 0.007402913497858572,
      "grad_norm": 0.6705797910690308,
      "learning_rate": 1.9997295637324644e-05,
      "loss": 0.4652,
      "step": 410
    },
    {
      "epoch": 0.0074931929307592866,
      "grad_norm": 0.5221576690673828,
      "learning_rate": 1.9997229278128562e-05,
      "loss": 0.4423,
      "step": 415
    },
    {
      "epoch": 0.0075834723636600005,
      "grad_norm": 0.7182873487472534,
      "learning_rate": 1.9997162114728073e-05,
      "loss": 0.3028,
      "step": 420
    },
    {
      "epoch": 0.0076737517965607144,
      "grad_norm": 0.6148266196250916,
      "learning_rate": 1.999709414712858e-05,
      "loss": 0.3227,
      "step": 425
    },
    {
      "epoch": 0.007764031229461429,
      "grad_norm": 0.541401743888855,
      "learning_rate": 1.9997025375335542e-05,
      "loss": 0.4572,
      "step": 430
    },
    {
      "epoch": 0.007854310662362144,
      "grad_norm": 0.7312717437744141,
      "learning_rate": 1.99969557993545e-05,
      "loss": 0.4082,
      "step": 435
    },
    {
      "epoch": 0.007944590095262858,
      "grad_norm": 0.7365412712097168,
      "learning_rate": 1.9996885419191053e-05,
      "loss": 0.3968,
      "step": 440
    },
    {
      "epoch": 0.008034869528163572,
      "grad_norm": 0.6317394375801086,
      "learning_rate": 1.9996814234850855e-05,
      "loss": 0.3782,
      "step": 445
    },
    {
      "epoch": 0.008125148961064286,
      "grad_norm": 0.49483615159988403,
      "learning_rate": 1.9996742246339637e-05,
      "loss": 0.4659,
      "step": 450
    },
    {
      "epoch": 0.008215428393965,
      "grad_norm": 0.5639483332633972,
      "learning_rate": 1.9996669453663186e-05,
      "loss": 0.3857,
      "step": 455
    },
    {
      "epoch": 0.008305707826865716,
      "grad_norm": 0.5206676125526428,
      "learning_rate": 1.9996595856827366e-05,
      "loss": 0.305,
      "step": 460
    },
    {
      "epoch": 0.00839598725976643,
      "grad_norm": 0.6142938137054443,
      "learning_rate": 1.9996521455838087e-05,
      "loss": 0.3415,
      "step": 465
    },
    {
      "epoch": 0.008486266692667143,
      "grad_norm": 0.6466097235679626,
      "learning_rate": 1.999644625070134e-05,
      "loss": 0.4893,
      "step": 470
    },
    {
      "epoch": 0.008576546125567857,
      "grad_norm": 0.6824833154678345,
      "learning_rate": 1.9996370241423172e-05,
      "loss": 0.6019,
      "step": 475
    },
    {
      "epoch": 0.008666825558468571,
      "grad_norm": 0.42456185817718506,
      "learning_rate": 1.99962934280097e-05,
      "loss": 0.349,
      "step": 480
    },
    {
      "epoch": 0.008757104991369287,
      "grad_norm": 0.7232670187950134,
      "learning_rate": 1.9996215810467102e-05,
      "loss": 0.4968,
      "step": 485
    },
    {
      "epoch": 0.008847384424270001,
      "grad_norm": 0.7302543520927429,
      "learning_rate": 1.9996137388801622e-05,
      "loss": 0.447,
      "step": 490
    },
    {
      "epoch": 0.008937663857170715,
      "grad_norm": 0.4422917068004608,
      "learning_rate": 1.9996058163019566e-05,
      "loss": 0.4172,
      "step": 495
    },
    {
      "epoch": 0.009027943290071429,
      "grad_norm": 0.51226407289505,
      "learning_rate": 1.999597813312731e-05,
      "loss": 0.3597,
      "step": 500
    },
    {
      "epoch": 0.009118222722972143,
      "grad_norm": 0.7817170023918152,
      "learning_rate": 1.999589729913129e-05,
      "loss": 0.4316,
      "step": 505
    },
    {
      "epoch": 0.009208502155872858,
      "grad_norm": 0.5358217358589172,
      "learning_rate": 1.999581566103801e-05,
      "loss": 0.3517,
      "step": 510
    },
    {
      "epoch": 0.009298781588773572,
      "grad_norm": 0.7174142003059387,
      "learning_rate": 1.9995733218854038e-05,
      "loss": 0.4131,
      "step": 515
    },
    {
      "epoch": 0.009389061021674286,
      "grad_norm": 0.5455174446105957,
      "learning_rate": 1.9995649972586005e-05,
      "loss": 0.3305,
      "step": 520
    },
    {
      "epoch": 0.009479340454575,
      "grad_norm": 1.0940828323364258,
      "learning_rate": 1.9995565922240606e-05,
      "loss": 0.4694,
      "step": 525
    },
    {
      "epoch": 0.009569619887475714,
      "grad_norm": 0.7484090328216553,
      "learning_rate": 1.99954810678246e-05,
      "loss": 0.348,
      "step": 530
    },
    {
      "epoch": 0.00965989932037643,
      "grad_norm": 0.6891335248947144,
      "learning_rate": 1.999539540934482e-05,
      "loss": 0.4109,
      "step": 535
    },
    {
      "epoch": 0.009750178753277144,
      "grad_norm": 0.46111416816711426,
      "learning_rate": 1.9995308946808153e-05,
      "loss": 0.373,
      "step": 540
    },
    {
      "epoch": 0.009840458186177858,
      "grad_norm": 0.5738707780838013,
      "learning_rate": 1.9995221680221554e-05,
      "loss": 0.4119,
      "step": 545
    },
    {
      "epoch": 0.009930737619078572,
      "grad_norm": 2.263345241546631,
      "learning_rate": 1.9995133609592043e-05,
      "loss": 0.4038,
      "step": 550
    },
    {
      "epoch": 0.010021017051979286,
      "grad_norm": 0.5366809368133545,
      "learning_rate": 1.9995044734926705e-05,
      "loss": 0.5166,
      "step": 555
    },
    {
      "epoch": 0.010111296484880001,
      "grad_norm": 0.591261088848114,
      "learning_rate": 1.999495505623269e-05,
      "loss": 0.3459,
      "step": 560
    },
    {
      "epoch": 0.010201575917780715,
      "grad_norm": 0.513255774974823,
      "learning_rate": 1.999486457351721e-05,
      "loss": 0.3736,
      "step": 565
    },
    {
      "epoch": 0.01029185535068143,
      "grad_norm": 0.8982439041137695,
      "learning_rate": 1.9994773286787544e-05,
      "loss": 0.3046,
      "step": 570
    },
    {
      "epoch": 0.010382134783582143,
      "grad_norm": 0.5183768272399902,
      "learning_rate": 1.9994681196051036e-05,
      "loss": 0.2987,
      "step": 575
    },
    {
      "epoch": 0.010472414216482857,
      "grad_norm": 0.595223069190979,
      "learning_rate": 1.9994588301315092e-05,
      "loss": 0.4161,
      "step": 580
    },
    {
      "epoch": 0.010562693649383573,
      "grad_norm": 0.5426445603370667,
      "learning_rate": 1.999449460258719e-05,
      "loss": 0.3414,
      "step": 585
    },
    {
      "epoch": 0.010652973082284287,
      "grad_norm": 0.7920253872871399,
      "learning_rate": 1.9994400099874864e-05,
      "loss": 0.3525,
      "step": 590
    },
    {
      "epoch": 0.010743252515185,
      "grad_norm": 0.3906225264072418,
      "learning_rate": 1.9994304793185718e-05,
      "loss": 0.2805,
      "step": 595
    },
    {
      "epoch": 0.010833531948085715,
      "grad_norm": 0.7094693183898926,
      "learning_rate": 1.9994208682527413e-05,
      "loss": 0.373,
      "step": 600
    },
    {
      "epoch": 0.010923811380986428,
      "grad_norm": 0.583308219909668,
      "learning_rate": 1.9994111767907685e-05,
      "loss": 0.4022,
      "step": 605
    },
    {
      "epoch": 0.011014090813887144,
      "grad_norm": 0.7794875502586365,
      "learning_rate": 1.9994014049334333e-05,
      "loss": 0.3641,
      "step": 610
    },
    {
      "epoch": 0.011104370246787858,
      "grad_norm": 0.613642156124115,
      "learning_rate": 1.9993915526815214e-05,
      "loss": 0.3642,
      "step": 615
    },
    {
      "epoch": 0.011194649679688572,
      "grad_norm": 0.8832647204399109,
      "learning_rate": 1.999381620035825e-05,
      "loss": 0.4075,
      "step": 620
    },
    {
      "epoch": 0.011284929112589286,
      "grad_norm": 0.47653380036354065,
      "learning_rate": 1.9993716069971435e-05,
      "loss": 0.2557,
      "step": 625
    },
    {
      "epoch": 0.01137520854549,
      "grad_norm": 0.6913342475891113,
      "learning_rate": 1.9993615135662825e-05,
      "loss": 0.4108,
      "step": 630
    },
    {
      "epoch": 0.011465487978390716,
      "grad_norm": 0.6558578610420227,
      "learning_rate": 1.9993513397440537e-05,
      "loss": 0.3773,
      "step": 635
    },
    {
      "epoch": 0.01155576741129143,
      "grad_norm": 0.7672843337059021,
      "learning_rate": 1.9993410855312757e-05,
      "loss": 0.3645,
      "step": 640
    },
    {
      "epoch": 0.011646046844192143,
      "grad_norm": 0.7909203171730042,
      "learning_rate": 1.9993307509287734e-05,
      "loss": 0.4225,
      "step": 645
    },
    {
      "epoch": 0.011736326277092857,
      "grad_norm": 0.5562388300895691,
      "learning_rate": 1.999320335937378e-05,
      "loss": 0.3487,
      "step": 650
    },
    {
      "epoch": 0.011826605709993571,
      "grad_norm": 0.49056869745254517,
      "learning_rate": 1.999309840557927e-05,
      "loss": 0.3879,
      "step": 655
    },
    {
      "epoch": 0.011916885142894287,
      "grad_norm": 0.4736185371875763,
      "learning_rate": 1.9992992647912653e-05,
      "loss": 0.3523,
      "step": 660
    },
    {
      "epoch": 0.012007164575795001,
      "grad_norm": 0.5065646767616272,
      "learning_rate": 1.9992886086382434e-05,
      "loss": 0.5149,
      "step": 665
    },
    {
      "epoch": 0.012097444008695715,
      "grad_norm": 0.44907593727111816,
      "learning_rate": 1.9992778720997185e-05,
      "loss": 0.3551,
      "step": 670
    },
    {
      "epoch": 0.012187723441596429,
      "grad_norm": 0.9845952987670898,
      "learning_rate": 1.999267055176554e-05,
      "loss": 0.3669,
      "step": 675
    },
    {
      "epoch": 0.012278002874497143,
      "grad_norm": 0.5695990920066833,
      "learning_rate": 1.9992561578696205e-05,
      "loss": 0.336,
      "step": 680
    },
    {
      "epoch": 0.012368282307397858,
      "grad_norm": 0.6252821683883667,
      "learning_rate": 1.999245180179794e-05,
      "loss": 0.3034,
      "step": 685
    },
    {
      "epoch": 0.012458561740298572,
      "grad_norm": 1.414514183998108,
      "learning_rate": 1.9992341221079584e-05,
      "loss": 0.3144,
      "step": 690
    },
    {
      "epoch": 0.012548841173199286,
      "grad_norm": 0.5516579151153564,
      "learning_rate": 1.9992229836550024e-05,
      "loss": 0.3976,
      "step": 695
    },
    {
      "epoch": 0.0126391206061,
      "grad_norm": 0.6496583223342896,
      "learning_rate": 1.999211764821823e-05,
      "loss": 0.3827,
      "step": 700
    },
    {
      "epoch": 0.012729400039000714,
      "grad_norm": 0.408225953578949,
      "learning_rate": 1.9992004656093217e-05,
      "loss": 0.3794,
      "step": 705
    },
    {
      "epoch": 0.01281967947190143,
      "grad_norm": 0.5428416728973389,
      "learning_rate": 1.999189086018408e-05,
      "loss": 0.3612,
      "step": 710
    },
    {
      "epoch": 0.012909958904802144,
      "grad_norm": 0.6431609392166138,
      "learning_rate": 1.999177626049997e-05,
      "loss": 0.3706,
      "step": 715
    },
    {
      "epoch": 0.013000238337702858,
      "grad_norm": 0.5790135264396667,
      "learning_rate": 1.9991660857050106e-05,
      "loss": 0.361,
      "step": 720
    },
    {
      "epoch": 0.013090517770603572,
      "grad_norm": 0.45080527663230896,
      "learning_rate": 1.9991544649843776e-05,
      "loss": 0.2959,
      "step": 725
    },
    {
      "epoch": 0.013180797203504286,
      "grad_norm": 0.5176795721054077,
      "learning_rate": 1.999142763889032e-05,
      "loss": 0.407,
      "step": 730
    },
    {
      "epoch": 0.013271076636405001,
      "grad_norm": 0.7002350091934204,
      "learning_rate": 1.999130982419916e-05,
      "loss": 0.383,
      "step": 735
    },
    {
      "epoch": 0.013361356069305715,
      "grad_norm": 0.48800963163375854,
      "learning_rate": 1.9991191205779764e-05,
      "loss": 0.3827,
      "step": 740
    },
    {
      "epoch": 0.01345163550220643,
      "grad_norm": 0.6332141160964966,
      "learning_rate": 1.9991071783641682e-05,
      "loss": 0.3713,
      "step": 745
    },
    {
      "epoch": 0.013541914935107143,
      "grad_norm": 0.6171541213989258,
      "learning_rate": 1.9990951557794514e-05,
      "loss": 0.4427,
      "step": 750
    },
    {
      "epoch": 0.013632194368007857,
      "grad_norm": 0.5151376724243164,
      "learning_rate": 1.9990830528247937e-05,
      "loss": 0.3107,
      "step": 755
    },
    {
      "epoch": 0.013722473800908573,
      "grad_norm": 0.3900178074836731,
      "learning_rate": 1.9990708695011685e-05,
      "loss": 0.3266,
      "step": 760
    },
    {
      "epoch": 0.013812753233809287,
      "grad_norm": 0.502527117729187,
      "learning_rate": 1.9990586058095557e-05,
      "loss": 0.4022,
      "step": 765
    },
    {
      "epoch": 0.01390303266671,
      "grad_norm": 0.7116743326187134,
      "learning_rate": 1.9990462617509418e-05,
      "loss": 0.4046,
      "step": 770
    },
    {
      "epoch": 0.013993312099610715,
      "grad_norm": 0.6583302617073059,
      "learning_rate": 1.99903383732632e-05,
      "loss": 0.3114,
      "step": 775
    },
    {
      "epoch": 0.014083591532511429,
      "grad_norm": 1.1406691074371338,
      "learning_rate": 1.9990213325366893e-05,
      "loss": 0.3487,
      "step": 780
    },
    {
      "epoch": 0.014173870965412144,
      "grad_norm": 0.5100271105766296,
      "learning_rate": 1.9990087473830567e-05,
      "loss": 0.3689,
      "step": 785
    },
    {
      "epoch": 0.014264150398312858,
      "grad_norm": 0.6300409436225891,
      "learning_rate": 1.9989960818664334e-05,
      "loss": 0.365,
      "step": 790
    },
    {
      "epoch": 0.014354429831213572,
      "grad_norm": 0.5625520944595337,
      "learning_rate": 1.9989833359878387e-05,
      "loss": 0.4753,
      "step": 795
    },
    {
      "epoch": 0.014444709264114286,
      "grad_norm": 0.6765584349632263,
      "learning_rate": 1.9989705097482982e-05,
      "loss": 0.2894,
      "step": 800
    },
    {
      "epoch": 0.014534988697015002,
      "grad_norm": 0.6390443444252014,
      "learning_rate": 1.998957603148843e-05,
      "loss": 0.4293,
      "step": 805
    },
    {
      "epoch": 0.014625268129915716,
      "grad_norm": 1.139591932296753,
      "learning_rate": 1.998944616190512e-05,
      "loss": 0.3266,
      "step": 810
    },
    {
      "epoch": 0.01471554756281643,
      "grad_norm": 0.677528440952301,
      "learning_rate": 1.99893154887435e-05,
      "loss": 0.4459,
      "step": 815
    },
    {
      "epoch": 0.014805826995717144,
      "grad_norm": 1.0746852159500122,
      "learning_rate": 1.9989184012014073e-05,
      "loss": 0.3805,
      "step": 820
    },
    {
      "epoch": 0.014896106428617857,
      "grad_norm": 0.7234352231025696,
      "learning_rate": 1.9989051731727424e-05,
      "loss": 0.2925,
      "step": 825
    },
    {
      "epoch": 0.014986385861518573,
      "grad_norm": 0.4438038170337677,
      "learning_rate": 1.998891864789419e-05,
      "loss": 0.3536,
      "step": 830
    },
    {
      "epoch": 0.015076665294419287,
      "grad_norm": 0.573958158493042,
      "learning_rate": 1.9988784760525073e-05,
      "loss": 0.5212,
      "step": 835
    },
    {
      "epoch": 0.015166944727320001,
      "grad_norm": 0.6227238178253174,
      "learning_rate": 1.9988650069630852e-05,
      "loss": 0.4595,
      "step": 840
    },
    {
      "epoch": 0.015257224160220715,
      "grad_norm": 0.5828679203987122,
      "learning_rate": 1.9988514575222356e-05,
      "loss": 0.3545,
      "step": 845
    },
    {
      "epoch": 0.015347503593121429,
      "grad_norm": 0.4298834800720215,
      "learning_rate": 1.9988378277310486e-05,
      "loss": 0.4366,
      "step": 850
    },
    {
      "epoch": 0.015437783026022145,
      "grad_norm": 0.5449682474136353,
      "learning_rate": 1.9988241175906204e-05,
      "loss": 0.4333,
      "step": 855
    },
    {
      "epoch": 0.015528062458922859,
      "grad_norm": 1.4678033590316772,
      "learning_rate": 1.9988103271020545e-05,
      "loss": 0.3295,
      "step": 860
    },
    {
      "epoch": 0.015618341891823572,
      "grad_norm": 1.1345720291137695,
      "learning_rate": 1.9987964562664597e-05,
      "loss": 0.494,
      "step": 865
    },
    {
      "epoch": 0.015708621324724288,
      "grad_norm": 0.6344099640846252,
      "learning_rate": 1.998782505084952e-05,
      "loss": 0.4833,
      "step": 870
    },
    {
      "epoch": 0.015798900757625002,
      "grad_norm": 0.4467625916004181,
      "learning_rate": 1.9987684735586533e-05,
      "loss": 0.2867,
      "step": 875
    },
    {
      "epoch": 0.015889180190525716,
      "grad_norm": 0.821846604347229,
      "learning_rate": 1.998754361688693e-05,
      "loss": 0.2907,
      "step": 880
    },
    {
      "epoch": 0.01597945962342643,
      "grad_norm": 0.4069104492664337,
      "learning_rate": 1.998740169476206e-05,
      "loss": 0.3291,
      "step": 885
    },
    {
      "epoch": 0.016069739056327144,
      "grad_norm": 0.7173675298690796,
      "learning_rate": 1.9987258969223334e-05,
      "loss": 0.4522,
      "step": 890
    },
    {
      "epoch": 0.016160018489227858,
      "grad_norm": 0.5012965798377991,
      "learning_rate": 1.9987115440282242e-05,
      "loss": 0.3062,
      "step": 895
    },
    {
      "epoch": 0.016250297922128572,
      "grad_norm": 0.5497738122940063,
      "learning_rate": 1.9986971107950325e-05,
      "loss": 0.2864,
      "step": 900
    },
    {
      "epoch": 0.016340577355029286,
      "grad_norm": 0.6531277894973755,
      "learning_rate": 1.9986825972239197e-05,
      "loss": 0.388,
      "step": 905
    },
    {
      "epoch": 0.01643085678793,
      "grad_norm": 0.622791588306427,
      "learning_rate": 1.9986680033160528e-05,
      "loss": 0.4287,
      "step": 910
    },
    {
      "epoch": 0.016521136220830714,
      "grad_norm": 0.5567917227745056,
      "learning_rate": 1.998653329072606e-05,
      "loss": 0.3517,
      "step": 915
    },
    {
      "epoch": 0.01661141565373143,
      "grad_norm": 0.4722416400909424,
      "learning_rate": 1.99863857449476e-05,
      "loss": 0.3337,
      "step": 920
    },
    {
      "epoch": 0.016701695086632145,
      "grad_norm": 0.6758808493614197,
      "learning_rate": 1.9986237395837015e-05,
      "loss": 0.4052,
      "step": 925
    },
    {
      "epoch": 0.01679197451953286,
      "grad_norm": 0.5101285576820374,
      "learning_rate": 1.998608824340624e-05,
      "loss": 0.3068,
      "step": 930
    },
    {
      "epoch": 0.016882253952433573,
      "grad_norm": 0.6921379566192627,
      "learning_rate": 1.9985938287667268e-05,
      "loss": 0.3838,
      "step": 935
    },
    {
      "epoch": 0.016972533385334287,
      "grad_norm": 0.5071640014648438,
      "learning_rate": 1.9985787528632167e-05,
      "loss": 0.3373,
      "step": 940
    },
    {
      "epoch": 0.017062812818235,
      "grad_norm": 0.5234969854354858,
      "learning_rate": 1.9985635966313065e-05,
      "loss": 0.3708,
      "step": 945
    },
    {
      "epoch": 0.017153092251135715,
      "grad_norm": 0.3468939960002899,
      "learning_rate": 1.998548360072215e-05,
      "loss": 0.3431,
      "step": 950
    },
    {
      "epoch": 0.01724337168403643,
      "grad_norm": 0.5182576179504395,
      "learning_rate": 1.998533043187168e-05,
      "loss": 0.3471,
      "step": 955
    },
    {
      "epoch": 0.017333651116937143,
      "grad_norm": 0.4680151641368866,
      "learning_rate": 1.9985176459773977e-05,
      "loss": 0.4855,
      "step": 960
    },
    {
      "epoch": 0.017423930549837856,
      "grad_norm": 0.5592324733734131,
      "learning_rate": 1.998502168444143e-05,
      "loss": 0.3305,
      "step": 965
    },
    {
      "epoch": 0.017514209982738574,
      "grad_norm": 0.5374587178230286,
      "learning_rate": 1.9984866105886484e-05,
      "loss": 0.3335,
      "step": 970
    },
    {
      "epoch": 0.017604489415639288,
      "grad_norm": 0.4037306308746338,
      "learning_rate": 1.998470972412166e-05,
      "loss": 0.4198,
      "step": 975
    },
    {
      "epoch": 0.017694768848540002,
      "grad_norm": 0.7788845300674438,
      "learning_rate": 1.998455253915953e-05,
      "loss": 0.4744,
      "step": 980
    },
    {
      "epoch": 0.017785048281440716,
      "grad_norm": 0.43953633308410645,
      "learning_rate": 1.9984394551012745e-05,
      "loss": 0.4211,
      "step": 985
    },
    {
      "epoch": 0.01787532771434143,
      "grad_norm": 0.5609700083732605,
      "learning_rate": 1.998423575969401e-05,
      "loss": 0.3761,
      "step": 990
    },
    {
      "epoch": 0.017965607147242144,
      "grad_norm": 0.5562583804130554,
      "learning_rate": 1.9984076165216106e-05,
      "loss": 0.4629,
      "step": 995
    },
    {
      "epoch": 0.018055886580142858,
      "grad_norm": 0.7671899795532227,
      "learning_rate": 1.9983915767591863e-05,
      "loss": 0.401,
      "step": 1000
    },
    {
      "epoch": 0.01814616601304357,
      "grad_norm": 0.44110381603240967,
      "learning_rate": 1.9983754566834185e-05,
      "loss": 0.43,
      "step": 1005
    },
    {
      "epoch": 0.018236445445944285,
      "grad_norm": 1.3212000131607056,
      "learning_rate": 1.998359256295604e-05,
      "loss": 0.4265,
      "step": 1010
    },
    {
      "epoch": 0.018326724878845,
      "grad_norm": 0.6430184841156006,
      "learning_rate": 1.9983429755970466e-05,
      "loss": 0.3988,
      "step": 1015
    },
    {
      "epoch": 0.018417004311745717,
      "grad_norm": 0.921191394329071,
      "learning_rate": 1.998326614589055e-05,
      "loss": 0.3036,
      "step": 1020
    },
    {
      "epoch": 0.01850728374464643,
      "grad_norm": 0.5766294598579407,
      "learning_rate": 1.998310173272946e-05,
      "loss": 0.4316,
      "step": 1025
    },
    {
      "epoch": 0.018597563177547145,
      "grad_norm": 0.4915804862976074,
      "learning_rate": 1.998293651650042e-05,
      "loss": 0.2762,
      "step": 1030
    },
    {
      "epoch": 0.01868784261044786,
      "grad_norm": 0.4122258126735687,
      "learning_rate": 1.9982770497216718e-05,
      "loss": 0.3431,
      "step": 1035
    },
    {
      "epoch": 0.018778122043348573,
      "grad_norm": 0.539563775062561,
      "learning_rate": 1.9982603674891714e-05,
      "loss": 0.3714,
      "step": 1040
    },
    {
      "epoch": 0.018868401476249286,
      "grad_norm": 0.59839928150177,
      "learning_rate": 1.9982436049538822e-05,
      "loss": 0.3429,
      "step": 1045
    },
    {
      "epoch": 0.01895868090915,
      "grad_norm": 0.6879011392593384,
      "learning_rate": 1.998226762117153e-05,
      "loss": 0.3609,
      "step": 1050
    },
    {
      "epoch": 0.019048960342050714,
      "grad_norm": 0.7054517269134521,
      "learning_rate": 1.9982098389803388e-05,
      "loss": 0.3349,
      "step": 1055
    },
    {
      "epoch": 0.01913923977495143,
      "grad_norm": 0.37130406498908997,
      "learning_rate": 1.9981928355448007e-05,
      "loss": 0.3453,
      "step": 1060
    },
    {
      "epoch": 0.019229519207852146,
      "grad_norm": 0.405746728181839,
      "learning_rate": 1.9981757518119064e-05,
      "loss": 0.3002,
      "step": 1065
    },
    {
      "epoch": 0.01931979864075286,
      "grad_norm": 0.49931222200393677,
      "learning_rate": 1.9981585877830304e-05,
      "loss": 0.2688,
      "step": 1070
    },
    {
      "epoch": 0.019410078073653574,
      "grad_norm": 0.7177553176879883,
      "learning_rate": 1.9981413434595533e-05,
      "loss": 0.3903,
      "step": 1075
    },
    {
      "epoch": 0.019500357506554288,
      "grad_norm": 0.46105924248695374,
      "learning_rate": 1.9981240188428622e-05,
      "loss": 0.4226,
      "step": 1080
    },
    {
      "epoch": 0.019590636939455,
      "grad_norm": 0.5767070651054382,
      "learning_rate": 1.998106613934351e-05,
      "loss": 0.3401,
      "step": 1085
    },
    {
      "epoch": 0.019680916372355715,
      "grad_norm": 0.41604000329971313,
      "learning_rate": 1.9980891287354195e-05,
      "loss": 0.3416,
      "step": 1090
    },
    {
      "epoch": 0.01977119580525643,
      "grad_norm": 0.45730865001678467,
      "learning_rate": 1.9980715632474746e-05,
      "loss": 0.4165,
      "step": 1095
    },
    {
      "epoch": 0.019861475238157143,
      "grad_norm": 0.4547334611415863,
      "learning_rate": 1.998053917471929e-05,
      "loss": 0.4183,
      "step": 1100
    },
    {
      "epoch": 0.019951754671057857,
      "grad_norm": 0.5578950643539429,
      "learning_rate": 1.9980361914102025e-05,
      "loss": 0.3189,
      "step": 1105
    },
    {
      "epoch": 0.02004203410395857,
      "grad_norm": 0.6025101542472839,
      "learning_rate": 1.9980183850637204e-05,
      "loss": 0.3669,
      "step": 1110
    },
    {
      "epoch": 0.02013231353685929,
      "grad_norm": 0.7206414341926575,
      "learning_rate": 1.998000498433916e-05,
      "loss": 0.375,
      "step": 1115
    },
    {
      "epoch": 0.020222592969760003,
      "grad_norm": 0.4863177239894867,
      "learning_rate": 1.997982531522227e-05,
      "loss": 0.284,
      "step": 1120
    },
    {
      "epoch": 0.020312872402660716,
      "grad_norm": 0.40691885352134705,
      "learning_rate": 1.9979644843301e-05,
      "loss": 0.4479,
      "step": 1125
    },
    {
      "epoch": 0.02040315183556143,
      "grad_norm": 0.5481736660003662,
      "learning_rate": 1.997946356858986e-05,
      "loss": 0.31,
      "step": 1130
    },
    {
      "epoch": 0.020493431268462144,
      "grad_norm": 0.7137806415557861,
      "learning_rate": 1.9979281491103433e-05,
      "loss": 0.4124,
      "step": 1135
    },
    {
      "epoch": 0.02058371070136286,
      "grad_norm": 0.6022760272026062,
      "learning_rate": 1.9979098610856368e-05,
      "loss": 0.4443,
      "step": 1140
    },
    {
      "epoch": 0.020673990134263572,
      "grad_norm": 0.47027885913848877,
      "learning_rate": 1.9978914927863375e-05,
      "loss": 0.3898,
      "step": 1145
    },
    {
      "epoch": 0.020764269567164286,
      "grad_norm": 0.7193593978881836,
      "learning_rate": 1.9978730442139226e-05,
      "loss": 0.3634,
      "step": 1150
    },
    {
      "epoch": 0.020854549000065,
      "grad_norm": 0.4863266944885254,
      "learning_rate": 1.9978545153698765e-05,
      "loss": 0.2502,
      "step": 1155
    },
    {
      "epoch": 0.020944828432965714,
      "grad_norm": 0.5944661498069763,
      "learning_rate": 1.99783590625569e-05,
      "loss": 0.3268,
      "step": 1160
    },
    {
      "epoch": 0.02103510786586643,
      "grad_norm": 0.6974284648895264,
      "learning_rate": 1.99781721687286e-05,
      "loss": 0.4665,
      "step": 1165
    },
    {
      "epoch": 0.021125387298767145,
      "grad_norm": 0.48131126165390015,
      "learning_rate": 1.9977984472228895e-05,
      "loss": 0.3469,
      "step": 1170
    },
    {
      "epoch": 0.02121566673166786,
      "grad_norm": 0.4336879849433899,
      "learning_rate": 1.9977795973072886e-05,
      "loss": 0.3365,
      "step": 1175
    },
    {
      "epoch": 0.021305946164568573,
      "grad_norm": 0.4767698049545288,
      "learning_rate": 1.9977606671275735e-05,
      "loss": 0.4261,
      "step": 1180
    },
    {
      "epoch": 0.021396225597469287,
      "grad_norm": 0.5864590406417847,
      "learning_rate": 1.9977416566852673e-05,
      "loss": 0.3919,
      "step": 1185
    },
    {
      "epoch": 0.02148650503037,
      "grad_norm": 0.4024285674095154,
      "learning_rate": 1.9977225659818992e-05,
      "loss": 0.312,
      "step": 1190
    },
    {
      "epoch": 0.021576784463270715,
      "grad_norm": 0.7056189775466919,
      "learning_rate": 1.9977033950190046e-05,
      "loss": 0.3784,
      "step": 1195
    },
    {
      "epoch": 0.02166706389617143,
      "grad_norm": 0.4819659888744354,
      "learning_rate": 1.9976841437981262e-05,
      "loss": 0.2945,
      "step": 1200
    },
    {
      "epoch": 0.021757343329072143,
      "grad_norm": 0.5364168882369995,
      "learning_rate": 1.997664812320812e-05,
      "loss": 0.3986,
      "step": 1205
    },
    {
      "epoch": 0.021847622761972857,
      "grad_norm": 0.5155913233757019,
      "learning_rate": 1.9976454005886176e-05,
      "loss": 0.353,
      "step": 1210
    },
    {
      "epoch": 0.021937902194873574,
      "grad_norm": 0.5517520308494568,
      "learning_rate": 1.997625908603104e-05,
      "loss": 0.4064,
      "step": 1215
    },
    {
      "epoch": 0.022028181627774288,
      "grad_norm": 0.4747216999530792,
      "learning_rate": 1.99760633636584e-05,
      "loss": 0.3965,
      "step": 1220
    },
    {
      "epoch": 0.022118461060675002,
      "grad_norm": 0.5541701316833496,
      "learning_rate": 1.997586683878399e-05,
      "loss": 0.4402,
      "step": 1225
    },
    {
      "epoch": 0.022208740493575716,
      "grad_norm": 0.3670824468135834,
      "learning_rate": 1.9975669511423628e-05,
      "loss": 0.4575,
      "step": 1230
    },
    {
      "epoch": 0.02229901992647643,
      "grad_norm": 0.36147984862327576,
      "learning_rate": 1.9975471381593184e-05,
      "loss": 0.2835,
      "step": 1235
    },
    {
      "epoch": 0.022389299359377144,
      "grad_norm": 0.7083765864372253,
      "learning_rate": 1.9975272449308595e-05,
      "loss": 0.4307,
      "step": 1240
    },
    {
      "epoch": 0.022479578792277858,
      "grad_norm": 0.5402193069458008,
      "learning_rate": 1.9975072714585863e-05,
      "loss": 0.3787,
      "step": 1245
    },
    {
      "epoch": 0.022569858225178572,
      "grad_norm": 0.6051380634307861,
      "learning_rate": 1.997487217744106e-05,
      "loss": 0.3636,
      "step": 1250
    },
    {
      "epoch": 0.022660137658079286,
      "grad_norm": 0.7156798243522644,
      "learning_rate": 1.9974670837890312e-05,
      "loss": 0.4672,
      "step": 1255
    },
    {
      "epoch": 0.02275041709098,
      "grad_norm": 1.068514347076416,
      "learning_rate": 1.997446869594982e-05,
      "loss": 0.4114,
      "step": 1260
    },
    {
      "epoch": 0.022840696523880717,
      "grad_norm": 0.926199197769165,
      "learning_rate": 1.9974265751635838e-05,
      "loss": 0.3667,
      "step": 1265
    },
    {
      "epoch": 0.02293097595678143,
      "grad_norm": 1.1801759004592896,
      "learning_rate": 1.9974062004964702e-05,
      "loss": 0.405,
      "step": 1270
    },
    {
      "epoch": 0.023021255389682145,
      "grad_norm": 0.38531816005706787,
      "learning_rate": 1.9973857455952795e-05,
      "loss": 0.4074,
      "step": 1275
    },
    {
      "epoch": 0.02311153482258286,
      "grad_norm": 0.48144111037254333,
      "learning_rate": 1.997365210461657e-05,
      "loss": 0.35,
      "step": 1280
    },
    {
      "epoch": 0.023201814255483573,
      "grad_norm": 0.4885618984699249,
      "learning_rate": 1.997344595097255e-05,
      "loss": 0.3415,
      "step": 1285
    },
    {
      "epoch": 0.023292093688384287,
      "grad_norm": 0.40975892543792725,
      "learning_rate": 1.997323899503732e-05,
      "loss": 0.3711,
      "step": 1290
    },
    {
      "epoch": 0.023382373121285,
      "grad_norm": 0.6145893335342407,
      "learning_rate": 1.9973031236827523e-05,
      "loss": 0.288,
      "step": 1295
    },
    {
      "epoch": 0.023472652554185715,
      "grad_norm": 0.7143402695655823,
      "learning_rate": 1.997282267635987e-05,
      "loss": 0.3985,
      "step": 1300
    },
    {
      "epoch": 0.02356293198708643,
      "grad_norm": 0.5758368372917175,
      "learning_rate": 1.997261331365115e-05,
      "loss": 0.3646,
      "step": 1305
    },
    {
      "epoch": 0.023653211419987143,
      "grad_norm": 0.7814615368843079,
      "learning_rate": 1.9972403148718193e-05,
      "loss": 0.4129,
      "step": 1310
    },
    {
      "epoch": 0.02374349085288786,
      "grad_norm": 0.404526948928833,
      "learning_rate": 1.9972192181577907e-05,
      "loss": 0.2616,
      "step": 1315
    },
    {
      "epoch": 0.023833770285788574,
      "grad_norm": 0.577202320098877,
      "learning_rate": 1.9971980412247266e-05,
      "loss": 0.4203,
      "step": 1320
    },
    {
      "epoch": 0.023924049718689288,
      "grad_norm": 0.5210227370262146,
      "learning_rate": 1.9971767840743305e-05,
      "loss": 0.4053,
      "step": 1325
    },
    {
      "epoch": 0.024014329151590002,
      "grad_norm": 0.5634742379188538,
      "learning_rate": 1.9971554467083123e-05,
      "loss": 0.3577,
      "step": 1330
    },
    {
      "epoch": 0.024104608584490716,
      "grad_norm": 0.6526023745536804,
      "learning_rate": 1.9971340291283883e-05,
      "loss": 0.3555,
      "step": 1335
    },
    {
      "epoch": 0.02419488801739143,
      "grad_norm": 0.5462536811828613,
      "learning_rate": 1.9971125313362818e-05,
      "loss": 0.3353,
      "step": 1340
    },
    {
      "epoch": 0.024285167450292144,
      "grad_norm": 0.7428454756736755,
      "learning_rate": 1.9970909533337216e-05,
      "loss": 0.391,
      "step": 1345
    },
    {
      "epoch": 0.024375446883192858,
      "grad_norm": 0.8522272706031799,
      "learning_rate": 1.9970692951224437e-05,
      "loss": 0.416,
      "step": 1350
    },
    {
      "epoch": 0.02446572631609357,
      "grad_norm": 0.5467855334281921,
      "learning_rate": 1.9970475567041907e-05,
      "loss": 0.3844,
      "step": 1355
    },
    {
      "epoch": 0.024556005748994286,
      "grad_norm": 0.5009450912475586,
      "learning_rate": 1.9970257380807107e-05,
      "loss": 0.3596,
      "step": 1360
    },
    {
      "epoch": 0.024646285181895003,
      "grad_norm": 0.9635720252990723,
      "learning_rate": 1.9970038392537593e-05,
      "loss": 0.4373,
      "step": 1365
    },
    {
      "epoch": 0.024736564614795717,
      "grad_norm": 1.1551811695098877,
      "learning_rate": 1.9969818602250982e-05,
      "loss": 0.447,
      "step": 1370
    },
    {
      "epoch": 0.02482684404769643,
      "grad_norm": 0.4225436747074127,
      "learning_rate": 1.996959800996495e-05,
      "loss": 0.3365,
      "step": 1375
    },
    {
      "epoch": 0.024917123480597145,
      "grad_norm": 0.512582540512085,
      "learning_rate": 1.996937661569724e-05,
      "loss": 0.374,
      "step": 1380
    },
    {
      "epoch": 0.02500740291349786,
      "grad_norm": 0.49112316966056824,
      "learning_rate": 1.996915441946567e-05,
      "loss": 0.2438,
      "step": 1385
    },
    {
      "epoch": 0.025097682346398573,
      "grad_norm": 0.48738208413124084,
      "learning_rate": 1.9968931421288112e-05,
      "loss": 0.29,
      "step": 1390
    },
    {
      "epoch": 0.025187961779299287,
      "grad_norm": 0.5263392925262451,
      "learning_rate": 1.9968707621182498e-05,
      "loss": 0.3913,
      "step": 1395
    },
    {
      "epoch": 0.0252782412122,
      "grad_norm": 0.4985851049423218,
      "learning_rate": 1.9968483019166838e-05,
      "loss": 0.3848,
      "step": 1400
    },
    {
      "epoch": 0.025368520645100714,
      "grad_norm": 0.5502220392227173,
      "learning_rate": 1.9968257615259192e-05,
      "loss": 0.2881,
      "step": 1405
    },
    {
      "epoch": 0.02545880007800143,
      "grad_norm": 0.7383748888969421,
      "learning_rate": 1.9968031409477703e-05,
      "loss": 0.3418,
      "step": 1410
    },
    {
      "epoch": 0.025549079510902146,
      "grad_norm": 0.4471767544746399,
      "learning_rate": 1.9967804401840558e-05,
      "loss": 0.3407,
      "step": 1415
    },
    {
      "epoch": 0.02563935894380286,
      "grad_norm": 2.0812489986419678,
      "learning_rate": 1.9967576592366023e-05,
      "loss": 0.3381,
      "step": 1420
    },
    {
      "epoch": 0.025729638376703574,
      "grad_norm": 0.6298109889030457,
      "learning_rate": 1.996734798107242e-05,
      "loss": 0.4014,
      "step": 1425
    },
    {
      "epoch": 0.025819917809604288,
      "grad_norm": 0.41603001952171326,
      "learning_rate": 1.9967118567978146e-05,
      "loss": 0.396,
      "step": 1430
    },
    {
      "epoch": 0.025910197242505,
      "grad_norm": 1.275228500366211,
      "learning_rate": 1.996688835310165e-05,
      "loss": 0.4296,
      "step": 1435
    },
    {
      "epoch": 0.026000476675405716,
      "grad_norm": 0.886063277721405,
      "learning_rate": 1.9966657336461452e-05,
      "loss": 0.4299,
      "step": 1440
    },
    {
      "epoch": 0.02609075610830643,
      "grad_norm": 0.6977846622467041,
      "learning_rate": 1.9966425518076136e-05,
      "loss": 0.4333,
      "step": 1445
    },
    {
      "epoch": 0.026181035541207143,
      "grad_norm": 0.26713240146636963,
      "learning_rate": 1.996619289796435e-05,
      "loss": 0.3639,
      "step": 1450
    },
    {
      "epoch": 0.026271314974107857,
      "grad_norm": 0.4383004605770111,
      "learning_rate": 1.9965959476144805e-05,
      "loss": 0.3157,
      "step": 1455
    },
    {
      "epoch": 0.02636159440700857,
      "grad_norm": 0.6261841058731079,
      "learning_rate": 1.9965725252636283e-05,
      "loss": 0.4359,
      "step": 1460
    },
    {
      "epoch": 0.02645187383990929,
      "grad_norm": 0.5010408759117126,
      "learning_rate": 1.9965490227457616e-05,
      "loss": 0.3469,
      "step": 1465
    },
    {
      "epoch": 0.026542153272810003,
      "grad_norm": 0.5412659645080566,
      "learning_rate": 1.9965254400627723e-05,
      "loss": 0.4394,
      "step": 1470
    },
    {
      "epoch": 0.026632432705710717,
      "grad_norm": 0.5586135387420654,
      "learning_rate": 1.9965017772165564e-05,
      "loss": 0.2828,
      "step": 1475
    },
    {
      "epoch": 0.02672271213861143,
      "grad_norm": 0.5518330931663513,
      "learning_rate": 1.996478034209018e-05,
      "loss": 0.2841,
      "step": 1480
    },
    {
      "epoch": 0.026812991571512144,
      "grad_norm": 0.39491382241249084,
      "learning_rate": 1.996454211042067e-05,
      "loss": 0.3561,
      "step": 1485
    },
    {
      "epoch": 0.02690327100441286,
      "grad_norm": 0.43156126141548157,
      "learning_rate": 1.9964303077176193e-05,
      "loss": 0.3998,
      "step": 1490
    },
    {
      "epoch": 0.026993550437313572,
      "grad_norm": 0.33977770805358887,
      "learning_rate": 1.996406324237598e-05,
      "loss": 0.3794,
      "step": 1495
    },
    {
      "epoch": 0.027083829870214286,
      "grad_norm": 0.9616397619247437,
      "learning_rate": 1.9963822606039328e-05,
      "loss": 0.4311,
      "step": 1500
    },
    {
      "epoch": 0.027174109303115,
      "grad_norm": 0.6763888597488403,
      "learning_rate": 1.996358116818559e-05,
      "loss": 0.3377,
      "step": 1505
    },
    {
      "epoch": 0.027264388736015714,
      "grad_norm": 0.40334710478782654,
      "learning_rate": 1.9963338928834188e-05,
      "loss": 0.3726,
      "step": 1510
    },
    {
      "epoch": 0.02735466816891643,
      "grad_norm": 0.5339846611022949,
      "learning_rate": 1.996309588800461e-05,
      "loss": 0.3261,
      "step": 1515
    },
    {
      "epoch": 0.027444947601817146,
      "grad_norm": 0.5830851197242737,
      "learning_rate": 1.996285204571641e-05,
      "loss": 0.4298,
      "step": 1520
    },
    {
      "epoch": 0.02753522703471786,
      "grad_norm": 0.5467507243156433,
      "learning_rate": 1.9962607401989196e-05,
      "loss": 0.4063,
      "step": 1525
    },
    {
      "epoch": 0.027625506467618573,
      "grad_norm": 0.7208880186080933,
      "learning_rate": 1.9962361956842655e-05,
      "loss": 0.2796,
      "step": 1530
    },
    {
      "epoch": 0.027715785900519287,
      "grad_norm": 0.5360137820243835,
      "learning_rate": 1.9962115710296525e-05,
      "loss": 0.3729,
      "step": 1535
    },
    {
      "epoch": 0.02780606533342,
      "grad_norm": 0.4716721773147583,
      "learning_rate": 1.996186866237062e-05,
      "loss": 0.3949,
      "step": 1540
    },
    {
      "epoch": 0.027896344766320715,
      "grad_norm": 0.5756022930145264,
      "learning_rate": 1.996162081308481e-05,
      "loss": 0.3719,
      "step": 1545
    },
    {
      "epoch": 0.02798662419922143,
      "grad_norm": 0.4410122036933899,
      "learning_rate": 1.996137216245903e-05,
      "loss": 0.3141,
      "step": 1550
    },
    {
      "epoch": 0.028076903632122143,
      "grad_norm": 0.4997660219669342,
      "learning_rate": 1.996112271051329e-05,
      "loss": 0.365,
      "step": 1555
    },
    {
      "epoch": 0.028167183065022857,
      "grad_norm": 0.47943833470344543,
      "learning_rate": 1.996087245726765e-05,
      "loss": 0.4094,
      "step": 1560
    },
    {
      "epoch": 0.028257462497923574,
      "grad_norm": 0.5912288427352905,
      "learning_rate": 1.9960621402742246e-05,
      "loss": 0.4728,
      "step": 1565
    },
    {
      "epoch": 0.02834774193082429,
      "grad_norm": 0.44753339886665344,
      "learning_rate": 1.9960369546957266e-05,
      "loss": 0.4006,
      "step": 1570
    },
    {
      "epoch": 0.028438021363725002,
      "grad_norm": 0.47656089067459106,
      "learning_rate": 1.9960116889932983e-05,
      "loss": 0.3909,
      "step": 1575
    },
    {
      "epoch": 0.028528300796625716,
      "grad_norm": 0.37938275933265686,
      "learning_rate": 1.995986343168971e-05,
      "loss": 0.3484,
      "step": 1580
    },
    {
      "epoch": 0.02861858022952643,
      "grad_norm": 0.3348734676837921,
      "learning_rate": 1.9959609172247835e-05,
      "loss": 0.2864,
      "step": 1585
    },
    {
      "epoch": 0.028708859662427144,
      "grad_norm": 0.4705940783023834,
      "learning_rate": 1.9959354111627817e-05,
      "loss": 0.3737,
      "step": 1590
    },
    {
      "epoch": 0.028799139095327858,
      "grad_norm": 0.819482684135437,
      "learning_rate": 1.9959098249850174e-05,
      "loss": 0.3681,
      "step": 1595
    },
    {
      "epoch": 0.028889418528228572,
      "grad_norm": 0.826997697353363,
      "learning_rate": 1.9958841586935484e-05,
      "loss": 0.3256,
      "step": 1600
    },
    {
      "epoch": 0.028979697961129286,
      "grad_norm": 0.9151831269264221,
      "learning_rate": 1.9958584122904398e-05,
      "loss": 0.3808,
      "step": 1605
    },
    {
      "epoch": 0.029069977394030003,
      "grad_norm": 0.39642009139060974,
      "learning_rate": 1.9958325857777625e-05,
      "loss": 0.3742,
      "step": 1610
    },
    {
      "epoch": 0.029160256826930717,
      "grad_norm": 0.5577180981636047,
      "learning_rate": 1.9958066791575943e-05,
      "loss": 0.2792,
      "step": 1615
    },
    {
      "epoch": 0.02925053625983143,
      "grad_norm": 0.4958969056606293,
      "learning_rate": 1.9957806924320188e-05,
      "loss": 0.4028,
      "step": 1620
    },
    {
      "epoch": 0.029340815692732145,
      "grad_norm": 0.34428122639656067,
      "learning_rate": 1.9957546256031266e-05,
      "loss": 0.4055,
      "step": 1625
    },
    {
      "epoch": 0.02943109512563286,
      "grad_norm": 0.48582723736763,
      "learning_rate": 1.9957284786730143e-05,
      "loss": 0.3664,
      "step": 1630
    },
    {
      "epoch": 0.029521374558533573,
      "grad_norm": 0.5791293978691101,
      "learning_rate": 1.995702251643786e-05,
      "loss": 0.2648,
      "step": 1635
    },
    {
      "epoch": 0.029611653991434287,
      "grad_norm": 0.4370667636394501,
      "learning_rate": 1.995675944517551e-05,
      "loss": 0.3348,
      "step": 1640
    },
    {
      "epoch": 0.029701933424335,
      "grad_norm": 0.4895092248916626,
      "learning_rate": 1.995649557296425e-05,
      "loss": 0.4108,
      "step": 1645
    },
    {
      "epoch": 0.029792212857235715,
      "grad_norm": 0.6840490102767944,
      "learning_rate": 1.995623089982532e-05,
      "loss": 0.3839,
      "step": 1650
    },
    {
      "epoch": 0.02988249229013643,
      "grad_norm": 0.683653712272644,
      "learning_rate": 1.9955965425779994e-05,
      "loss": 0.3478,
      "step": 1655
    },
    {
      "epoch": 0.029972771723037146,
      "grad_norm": 0.29711541533470154,
      "learning_rate": 1.9955699150849643e-05,
      "loss": 0.3065,
      "step": 1660
    },
    {
      "epoch": 0.03006305115593786,
      "grad_norm": 0.47336554527282715,
      "learning_rate": 1.9955432075055677e-05,
      "loss": 0.3156,
      "step": 1665
    },
    {
      "epoch": 0.030153330588838574,
      "grad_norm": 0.5812807083129883,
      "learning_rate": 1.9955164198419585e-05,
      "loss": 0.3848,
      "step": 1670
    },
    {
      "epoch": 0.030243610021739288,
      "grad_norm": 0.44053414463996887,
      "learning_rate": 1.9954895520962912e-05,
      "loss": 0.3991,
      "step": 1675
    },
    {
      "epoch": 0.030333889454640002,
      "grad_norm": 0.4269275367259979,
      "learning_rate": 1.9954626042707277e-05,
      "loss": 0.2343,
      "step": 1680
    },
    {
      "epoch": 0.030424168887540716,
      "grad_norm": 0.4034358263015747,
      "learning_rate": 1.9954355763674355e-05,
      "loss": 0.3132,
      "step": 1685
    },
    {
      "epoch": 0.03051444832044143,
      "grad_norm": 0.38301345705986023,
      "learning_rate": 1.9954084683885883e-05,
      "loss": 0.3286,
      "step": 1690
    },
    {
      "epoch": 0.030604727753342144,
      "grad_norm": 0.49178457260131836,
      "learning_rate": 1.995381280336367e-05,
      "loss": 0.3338,
      "step": 1695
    },
    {
      "epoch": 0.030695007186242858,
      "grad_norm": 0.5337426662445068,
      "learning_rate": 1.9953540122129592e-05,
      "loss": 0.3645,
      "step": 1700
    },
    {
      "epoch": 0.03078528661914357,
      "grad_norm": 0.4313508868217468,
      "learning_rate": 1.995326664020558e-05,
      "loss": 0.3102,
      "step": 1705
    },
    {
      "epoch": 0.03087556605204429,
      "grad_norm": 0.4087493419647217,
      "learning_rate": 1.9952992357613633e-05,
      "loss": 0.4064,
      "step": 1710
    },
    {
      "epoch": 0.030965845484945003,
      "grad_norm": 1.2327971458435059,
      "learning_rate": 1.9952717274375814e-05,
      "loss": 0.3662,
      "step": 1715
    },
    {
      "epoch": 0.031056124917845717,
      "grad_norm": 0.5065000057220459,
      "learning_rate": 1.9952441390514255e-05,
      "loss": 0.2931,
      "step": 1720
    },
    {
      "epoch": 0.03114640435074643,
      "grad_norm": 0.45024505257606506,
      "learning_rate": 1.995216470605115e-05,
      "loss": 0.4285,
      "step": 1725
    },
    {
      "epoch": 0.031236683783647145,
      "grad_norm": 0.3390149176120758,
      "learning_rate": 1.995188722100875e-05,
      "loss": 0.4599,
      "step": 1730
    },
    {
      "epoch": 0.03132696321654786,
      "grad_norm": 0.3781318962574005,
      "learning_rate": 1.9951608935409382e-05,
      "loss": 0.3843,
      "step": 1735
    },
    {
      "epoch": 0.031417242649448576,
      "grad_norm": 0.5293467044830322,
      "learning_rate": 1.9951329849275432e-05,
      "loss": 0.3104,
      "step": 1740
    },
    {
      "epoch": 0.03150752208234929,
      "grad_norm": 1.1434433460235596,
      "learning_rate": 1.9951049962629346e-05,
      "loss": 0.4619,
      "step": 1745
    },
    {
      "epoch": 0.031597801515250004,
      "grad_norm": 0.5122627019882202,
      "learning_rate": 1.9950769275493645e-05,
      "loss": 0.401,
      "step": 1750
    },
    {
      "epoch": 0.03168808094815072,
      "grad_norm": 0.9754158854484558,
      "learning_rate": 1.99504877878909e-05,
      "loss": 0.4172,
      "step": 1755
    },
    {
      "epoch": 0.03177836038105143,
      "grad_norm": 0.5180284976959229,
      "learning_rate": 1.9950205499843766e-05,
      "loss": 0.28,
      "step": 1760
    },
    {
      "epoch": 0.031868639813952146,
      "grad_norm": 0.5530410408973694,
      "learning_rate": 1.994992241137494e-05,
      "loss": 0.3866,
      "step": 1765
    },
    {
      "epoch": 0.03195891924685286,
      "grad_norm": 1.103765845298767,
      "learning_rate": 1.99496385225072e-05,
      "loss": 0.3099,
      "step": 1770
    },
    {
      "epoch": 0.032049198679753574,
      "grad_norm": 0.5604039430618286,
      "learning_rate": 1.9949353833263377e-05,
      "loss": 0.3052,
      "step": 1775
    },
    {
      "epoch": 0.03213947811265429,
      "grad_norm": 0.49411916732788086,
      "learning_rate": 1.9949068343666384e-05,
      "loss": 0.2955,
      "step": 1780
    },
    {
      "epoch": 0.032229757545555,
      "grad_norm": 1.085943579673767,
      "learning_rate": 1.9948782053739174e-05,
      "loss": 0.4603,
      "step": 1785
    },
    {
      "epoch": 0.032320036978455716,
      "grad_norm": 0.4952246844768524,
      "learning_rate": 1.9948494963504786e-05,
      "loss": 0.4334,
      "step": 1790
    },
    {
      "epoch": 0.03241031641135643,
      "grad_norm": 0.45003145933151245,
      "learning_rate": 1.994820707298631e-05,
      "loss": 0.4105,
      "step": 1795
    },
    {
      "epoch": 0.032500595844257144,
      "grad_norm": 0.525973916053772,
      "learning_rate": 1.9947918382206906e-05,
      "loss": 0.3267,
      "step": 1800
    },
    {
      "epoch": 0.03259087527715786,
      "grad_norm": 0.5335582494735718,
      "learning_rate": 1.9947628891189796e-05,
      "loss": 0.4224,
      "step": 1805
    },
    {
      "epoch": 0.03268115471005857,
      "grad_norm": 0.6103760004043579,
      "learning_rate": 1.9947338599958267e-05,
      "loss": 0.2956,
      "step": 1810
    },
    {
      "epoch": 0.032771434142959285,
      "grad_norm": 0.5102728009223938,
      "learning_rate": 1.9947047508535674e-05,
      "loss": 0.2478,
      "step": 1815
    },
    {
      "epoch": 0.03286171357586,
      "grad_norm": 0.42971786856651306,
      "learning_rate": 1.994675561694543e-05,
      "loss": 0.3884,
      "step": 1820
    },
    {
      "epoch": 0.03295199300876071,
      "grad_norm": 0.6633021831512451,
      "learning_rate": 1.9946462925211018e-05,
      "loss": 0.3489,
      "step": 1825
    },
    {
      "epoch": 0.03304227244166143,
      "grad_norm": 0.38056522607803345,
      "learning_rate": 1.994616943335598e-05,
      "loss": 0.4348,
      "step": 1830
    },
    {
      "epoch": 0.03313255187456215,
      "grad_norm": 0.6503346562385559,
      "learning_rate": 1.9945875141403924e-05,
      "loss": 0.3851,
      "step": 1835
    },
    {
      "epoch": 0.03322283130746286,
      "grad_norm": 0.3538060188293457,
      "learning_rate": 1.9945580049378527e-05,
      "loss": 0.3571,
      "step": 1840
    },
    {
      "epoch": 0.033313110740363576,
      "grad_norm": 0.5887143015861511,
      "learning_rate": 1.994528415730353e-05,
      "loss": 0.4758,
      "step": 1845
    },
    {
      "epoch": 0.03340339017326429,
      "grad_norm": 0.5996758937835693,
      "learning_rate": 1.994498746520273e-05,
      "loss": 0.3693,
      "step": 1850
    },
    {
      "epoch": 0.033493669606165004,
      "grad_norm": 1.738226056098938,
      "learning_rate": 1.9944689973099997e-05,
      "loss": 0.4114,
      "step": 1855
    },
    {
      "epoch": 0.03358394903906572,
      "grad_norm": 0.3167506754398346,
      "learning_rate": 1.9944391681019257e-05,
      "loss": 0.3131,
      "step": 1860
    },
    {
      "epoch": 0.03367422847196643,
      "grad_norm": 0.5447887778282166,
      "learning_rate": 1.9944092588984508e-05,
      "loss": 0.3885,
      "step": 1865
    },
    {
      "epoch": 0.033764507904867146,
      "grad_norm": 0.39655378460884094,
      "learning_rate": 1.9943792697019816e-05,
      "loss": 0.3641,
      "step": 1870
    },
    {
      "epoch": 0.03385478733776786,
      "grad_norm": 1.3213824033737183,
      "learning_rate": 1.9943492005149297e-05,
      "loss": 0.4111,
      "step": 1875
    },
    {
      "epoch": 0.033945066770668574,
      "grad_norm": 0.4718227982521057,
      "learning_rate": 1.994319051339714e-05,
      "loss": 0.3694,
      "step": 1880
    },
    {
      "epoch": 0.03403534620356929,
      "grad_norm": 0.5119079947471619,
      "learning_rate": 1.9942888221787604e-05,
      "loss": 0.3118,
      "step": 1885
    },
    {
      "epoch": 0.03412562563647,
      "grad_norm": 0.5912514328956604,
      "learning_rate": 1.9942585130345e-05,
      "loss": 0.3498,
      "step": 1890
    },
    {
      "epoch": 0.034215905069370715,
      "grad_norm": 1.2097115516662598,
      "learning_rate": 1.9942281239093716e-05,
      "loss": 0.3146,
      "step": 1895
    },
    {
      "epoch": 0.03430618450227143,
      "grad_norm": 0.5831785798072815,
      "learning_rate": 1.994197654805819e-05,
      "loss": 0.3613,
      "step": 1900
    },
    {
      "epoch": 0.03439646393517214,
      "grad_norm": 0.7582271099090576,
      "learning_rate": 1.9941671057262937e-05,
      "loss": 0.3127,
      "step": 1905
    },
    {
      "epoch": 0.03448674336807286,
      "grad_norm": 0.702990710735321,
      "learning_rate": 1.994136476673253e-05,
      "loss": 0.2743,
      "step": 1910
    },
    {
      "epoch": 0.03457702280097357,
      "grad_norm": 0.5540505647659302,
      "learning_rate": 1.9941057676491605e-05,
      "loss": 0.367,
      "step": 1915
    },
    {
      "epoch": 0.034667302233874285,
      "grad_norm": 0.39690011739730835,
      "learning_rate": 1.9940749786564876e-05,
      "loss": 0.2994,
      "step": 1920
    },
    {
      "epoch": 0.034757581666775,
      "grad_norm": 0.32354414463043213,
      "learning_rate": 1.9940441096977095e-05,
      "loss": 0.4053,
      "step": 1925
    },
    {
      "epoch": 0.03484786109967571,
      "grad_norm": 0.4380877912044525,
      "learning_rate": 1.994013160775311e-05,
      "loss": 0.5256,
      "step": 1930
    },
    {
      "epoch": 0.034938140532576434,
      "grad_norm": 0.5137713551521301,
      "learning_rate": 1.9939821318917804e-05,
      "loss": 0.3162,
      "step": 1935
    },
    {
      "epoch": 0.03502841996547715,
      "grad_norm": 0.4871731400489807,
      "learning_rate": 1.993951023049614e-05,
      "loss": 0.2895,
      "step": 1940
    },
    {
      "epoch": 0.03511869939837786,
      "grad_norm": 0.506823718547821,
      "learning_rate": 1.993919834251315e-05,
      "loss": 0.3808,
      "step": 1945
    },
    {
      "epoch": 0.035208978831278576,
      "grad_norm": 0.5929615497589111,
      "learning_rate": 1.9938885654993914e-05,
      "loss": 0.3662,
      "step": 1950
    },
    {
      "epoch": 0.03529925826417929,
      "grad_norm": 0.34095823764801025,
      "learning_rate": 1.993857216796359e-05,
      "loss": 0.3991,
      "step": 1955
    },
    {
      "epoch": 0.035389537697080004,
      "grad_norm": 0.46202680468559265,
      "learning_rate": 1.99382578814474e-05,
      "loss": 0.4335,
      "step": 1960
    },
    {
      "epoch": 0.03547981712998072,
      "grad_norm": 0.5477923154830933,
      "learning_rate": 1.993794279547062e-05,
      "loss": 0.4666,
      "step": 1965
    },
    {
      "epoch": 0.03557009656288143,
      "grad_norm": 1.0245890617370605,
      "learning_rate": 1.9937626910058597e-05,
      "loss": 0.3747,
      "step": 1970
    },
    {
      "epoch": 0.035660375995782145,
      "grad_norm": 1.0427522659301758,
      "learning_rate": 1.993731022523674e-05,
      "loss": 0.3872,
      "step": 1975
    },
    {
      "epoch": 0.03575065542868286,
      "grad_norm": 0.5643081665039062,
      "learning_rate": 1.993699274103053e-05,
      "loss": 0.2606,
      "step": 1980
    },
    {
      "epoch": 0.03584093486158357,
      "grad_norm": 0.4256623387336731,
      "learning_rate": 1.99366744574655e-05,
      "loss": 0.2413,
      "step": 1985
    },
    {
      "epoch": 0.03593121429448429,
      "grad_norm": 0.5165201425552368,
      "learning_rate": 1.9936355374567257e-05,
      "loss": 0.3004,
      "step": 1990
    },
    {
      "epoch": 0.036021493727385,
      "grad_norm": 0.3593834340572357,
      "learning_rate": 1.993603549236147e-05,
      "loss": 0.3883,
      "step": 1995
    },
    {
      "epoch": 0.036111773160285715,
      "grad_norm": 1.036332368850708,
      "learning_rate": 1.993571481087387e-05,
      "loss": 0.4099,
      "step": 2000
    },
    {
      "epoch": 0.03620205259318643,
      "grad_norm": 0.5267215371131897,
      "learning_rate": 1.9935393330130254e-05,
      "loss": 0.3242,
      "step": 2005
    },
    {
      "epoch": 0.03629233202608714,
      "grad_norm": 0.4293646216392517,
      "learning_rate": 1.993507105015648e-05,
      "loss": 0.3024,
      "step": 2010
    },
    {
      "epoch": 0.03638261145898786,
      "grad_norm": 0.9337967038154602,
      "learning_rate": 1.9934747970978475e-05,
      "loss": 0.3096,
      "step": 2015
    },
    {
      "epoch": 0.03647289089188857,
      "grad_norm": 0.48764199018478394,
      "learning_rate": 1.993442409262223e-05,
      "loss": 0.3644,
      "step": 2020
    },
    {
      "epoch": 0.036563170324789285,
      "grad_norm": 0.4670167565345764,
      "learning_rate": 1.9934099415113795e-05,
      "loss": 0.3789,
      "step": 2025
    },
    {
      "epoch": 0.03665344975769,
      "grad_norm": 0.4873073995113373,
      "learning_rate": 1.993377393847929e-05,
      "loss": 0.3646,
      "step": 2030
    },
    {
      "epoch": 0.03674372919059072,
      "grad_norm": 0.6277158260345459,
      "learning_rate": 1.9933447662744897e-05,
      "loss": 0.3741,
      "step": 2035
    },
    {
      "epoch": 0.036834008623491434,
      "grad_norm": 0.6023591756820679,
      "learning_rate": 1.9933120587936863e-05,
      "loss": 0.3908,
      "step": 2040
    },
    {
      "epoch": 0.03692428805639215,
      "grad_norm": 0.498462438583374,
      "learning_rate": 1.9932792714081502e-05,
      "loss": 0.3714,
      "step": 2045
    },
    {
      "epoch": 0.03701456748929286,
      "grad_norm": 0.3748553693294525,
      "learning_rate": 1.9932464041205182e-05,
      "loss": 0.4282,
      "step": 2050
    },
    {
      "epoch": 0.037104846922193575,
      "grad_norm": 0.5001702308654785,
      "learning_rate": 1.9932134569334346e-05,
      "loss": 0.3829,
      "step": 2055
    },
    {
      "epoch": 0.03719512635509429,
      "grad_norm": 0.4409927725791931,
      "learning_rate": 1.9931804298495497e-05,
      "loss": 0.3179,
      "step": 2060
    },
    {
      "epoch": 0.037285405787995,
      "grad_norm": 0.46347978711128235,
      "learning_rate": 1.9931473228715203e-05,
      "loss": 0.3888,
      "step": 2065
    },
    {
      "epoch": 0.03737568522089572,
      "grad_norm": 0.5000243782997131,
      "learning_rate": 1.9931141360020097e-05,
      "loss": 0.2796,
      "step": 2070
    },
    {
      "epoch": 0.03746596465379643,
      "grad_norm": 0.556724488735199,
      "learning_rate": 1.993080869243688e-05,
      "loss": 0.4378,
      "step": 2075
    },
    {
      "epoch": 0.037556244086697145,
      "grad_norm": 0.3234434723854065,
      "learning_rate": 1.99304752259923e-05,
      "loss": 0.3901,
      "step": 2080
    },
    {
      "epoch": 0.03764652351959786,
      "grad_norm": 0.42605525255203247,
      "learning_rate": 1.9930140960713195e-05,
      "loss": 0.378,
      "step": 2085
    },
    {
      "epoch": 0.03773680295249857,
      "grad_norm": 0.833888828754425,
      "learning_rate": 1.9929805896626446e-05,
      "loss": 0.386,
      "step": 2090
    },
    {
      "epoch": 0.03782708238539929,
      "grad_norm": 0.44782599806785583,
      "learning_rate": 1.992947003375901e-05,
      "loss": 0.3741,
      "step": 2095
    },
    {
      "epoch": 0.0379173618183,
      "grad_norm": 0.6248586773872375,
      "learning_rate": 1.9929133372137907e-05,
      "loss": 0.3939,
      "step": 2100
    },
    {
      "epoch": 0.038007641251200715,
      "grad_norm": 0.444442480802536,
      "learning_rate": 1.992879591179021e-05,
      "loss": 0.4162,
      "step": 2105
    },
    {
      "epoch": 0.03809792068410143,
      "grad_norm": 0.4670912027359009,
      "learning_rate": 1.9928457652743074e-05,
      "loss": 0.3047,
      "step": 2110
    },
    {
      "epoch": 0.03818820011700214,
      "grad_norm": 0.4275720715522766,
      "learning_rate": 1.9928118595023707e-05,
      "loss": 0.3561,
      "step": 2115
    },
    {
      "epoch": 0.03827847954990286,
      "grad_norm": 0.45719200372695923,
      "learning_rate": 1.9927778738659384e-05,
      "loss": 0.3426,
      "step": 2120
    },
    {
      "epoch": 0.03836875898280357,
      "grad_norm": 0.6436276435852051,
      "learning_rate": 1.9927438083677445e-05,
      "loss": 0.3548,
      "step": 2125
    },
    {
      "epoch": 0.03845903841570429,
      "grad_norm": 0.5185565948486328,
      "learning_rate": 1.9927096630105285e-05,
      "loss": 0.3818,
      "step": 2130
    },
    {
      "epoch": 0.038549317848605005,
      "grad_norm": 0.5306310057640076,
      "learning_rate": 1.9926754377970385e-05,
      "loss": 0.3823,
      "step": 2135
    },
    {
      "epoch": 0.03863959728150572,
      "grad_norm": 0.48837965726852417,
      "learning_rate": 1.9926411327300266e-05,
      "loss": 0.433,
      "step": 2140
    },
    {
      "epoch": 0.03872987671440643,
      "grad_norm": 0.46864208579063416,
      "learning_rate": 1.992606747812253e-05,
      "loss": 0.3655,
      "step": 2145
    },
    {
      "epoch": 0.03882015614730715,
      "grad_norm": 0.4254169464111328,
      "learning_rate": 1.9925722830464832e-05,
      "loss": 0.3287,
      "step": 2150
    },
    {
      "epoch": 0.03891043558020786,
      "grad_norm": 0.6717060208320618,
      "learning_rate": 1.9925377384354903e-05,
      "loss": 0.3643,
      "step": 2155
    },
    {
      "epoch": 0.039000715013108575,
      "grad_norm": 0.40732356905937195,
      "learning_rate": 1.9925031139820526e-05,
      "loss": 0.325,
      "step": 2160
    },
    {
      "epoch": 0.03909099444600929,
      "grad_norm": 0.48590540885925293,
      "learning_rate": 1.9924684096889556e-05,
      "loss": 0.3351,
      "step": 2165
    },
    {
      "epoch": 0.03918127387891,
      "grad_norm": 0.5301512479782104,
      "learning_rate": 1.992433625558991e-05,
      "loss": 0.3271,
      "step": 2170
    },
    {
      "epoch": 0.03927155331181072,
      "grad_norm": 0.48758119344711304,
      "learning_rate": 1.9923987615949568e-05,
      "loss": 0.3486,
      "step": 2175
    },
    {
      "epoch": 0.03936183274471143,
      "grad_norm": 0.4892788231372833,
      "learning_rate": 1.9923638177996575e-05,
      "loss": 0.4185,
      "step": 2180
    },
    {
      "epoch": 0.039452112177612145,
      "grad_norm": 0.4694100618362427,
      "learning_rate": 1.9923287941759048e-05,
      "loss": 0.374,
      "step": 2185
    },
    {
      "epoch": 0.03954239161051286,
      "grad_norm": 0.42079830169677734,
      "learning_rate": 1.9922936907265152e-05,
      "loss": 0.4314,
      "step": 2190
    },
    {
      "epoch": 0.03963267104341357,
      "grad_norm": 0.4797016978263855,
      "learning_rate": 1.9922585074543125e-05,
      "loss": 0.3264,
      "step": 2195
    },
    {
      "epoch": 0.03972295047631429,
      "grad_norm": 0.49687549471855164,
      "learning_rate": 1.992223244362128e-05,
      "loss": 0.3126,
      "step": 2200
    },
    {
      "epoch": 0.039813229909215,
      "grad_norm": 0.43118202686309814,
      "learning_rate": 1.9921879014527973e-05,
      "loss": 0.3377,
      "step": 2205
    },
    {
      "epoch": 0.039903509342115714,
      "grad_norm": 0.5688696503639221,
      "learning_rate": 1.9921524787291637e-05,
      "loss": 0.3186,
      "step": 2210
    },
    {
      "epoch": 0.03999378877501643,
      "grad_norm": 0.5296006798744202,
      "learning_rate": 1.992116976194077e-05,
      "loss": 0.3401,
      "step": 2215
    },
    {
      "epoch": 0.04008406820791714,
      "grad_norm": 0.5843896865844727,
      "learning_rate": 1.992081393850393e-05,
      "loss": 0.3502,
      "step": 2220
    },
    {
      "epoch": 0.040174347640817856,
      "grad_norm": 0.44793540239334106,
      "learning_rate": 1.9920457317009737e-05,
      "loss": 0.3152,
      "step": 2225
    },
    {
      "epoch": 0.04026462707371858,
      "grad_norm": 0.41331714391708374,
      "learning_rate": 1.9920099897486885e-05,
      "loss": 0.3549,
      "step": 2230
    },
    {
      "epoch": 0.04035490650661929,
      "grad_norm": 0.6284789443016052,
      "learning_rate": 1.9919741679964122e-05,
      "loss": 0.3694,
      "step": 2235
    },
    {
      "epoch": 0.040445185939520005,
      "grad_norm": 0.5507609844207764,
      "learning_rate": 1.9919382664470264e-05,
      "loss": 0.4005,
      "step": 2240
    },
    {
      "epoch": 0.04053546537242072,
      "grad_norm": 0.5600115656852722,
      "learning_rate": 1.991902285103419e-05,
      "loss": 0.3131,
      "step": 2245
    },
    {
      "epoch": 0.04062574480532143,
      "grad_norm": 0.42502927780151367,
      "learning_rate": 1.9918662239684848e-05,
      "loss": 0.2977,
      "step": 2250
    },
    {
      "epoch": 0.04071602423822215,
      "grad_norm": 0.5070359706878662,
      "learning_rate": 1.9918300830451243e-05,
      "loss": 0.4613,
      "step": 2255
    },
    {
      "epoch": 0.04080630367112286,
      "grad_norm": 1.3776845932006836,
      "learning_rate": 1.9917938623362447e-05,
      "loss": 0.4404,
      "step": 2260
    },
    {
      "epoch": 0.040896583104023575,
      "grad_norm": 0.45074161887168884,
      "learning_rate": 1.9917575618447604e-05,
      "loss": 0.2912,
      "step": 2265
    },
    {
      "epoch": 0.04098686253692429,
      "grad_norm": 0.41434893012046814,
      "learning_rate": 1.9917211815735906e-05,
      "loss": 0.3021,
      "step": 2270
    },
    {
      "epoch": 0.041077141969825,
      "grad_norm": 1.3090029954910278,
      "learning_rate": 1.9916847215256626e-05,
      "loss": 0.2964,
      "step": 2275
    },
    {
      "epoch": 0.04116742140272572,
      "grad_norm": 0.45356735587120056,
      "learning_rate": 1.991648181703909e-05,
      "loss": 0.286,
      "step": 2280
    },
    {
      "epoch": 0.04125770083562643,
      "grad_norm": 0.5527782440185547,
      "learning_rate": 1.9916115621112687e-05,
      "loss": 0.3823,
      "step": 2285
    },
    {
      "epoch": 0.041347980268527144,
      "grad_norm": 0.30888089537620544,
      "learning_rate": 1.9915748627506884e-05,
      "loss": 0.4878,
      "step": 2290
    },
    {
      "epoch": 0.04143825970142786,
      "grad_norm": 0.42406365275382996,
      "learning_rate": 1.9915380836251198e-05,
      "loss": 0.3513,
      "step": 2295
    },
    {
      "epoch": 0.04152853913432857,
      "grad_norm": 0.4733360707759857,
      "learning_rate": 1.991501224737521e-05,
      "loss": 0.3513,
      "step": 2300
    },
    {
      "epoch": 0.041618818567229286,
      "grad_norm": 0.38414448499679565,
      "learning_rate": 1.9914642860908583e-05,
      "loss": 0.4567,
      "step": 2305
    },
    {
      "epoch": 0.04170909800013,
      "grad_norm": 0.47997698187828064,
      "learning_rate": 1.991427267688102e-05,
      "loss": 0.309,
      "step": 2310
    },
    {
      "epoch": 0.041799377433030714,
      "grad_norm": 0.5346536636352539,
      "learning_rate": 1.9913901695322308e-05,
      "loss": 0.4237,
      "step": 2315
    },
    {
      "epoch": 0.04188965686593143,
      "grad_norm": 0.4030945897102356,
      "learning_rate": 1.9913529916262283e-05,
      "loss": 0.2438,
      "step": 2320
    },
    {
      "epoch": 0.04197993629883214,
      "grad_norm": 0.686960756778717,
      "learning_rate": 1.9913157339730854e-05,
      "loss": 0.4153,
      "step": 2325
    },
    {
      "epoch": 0.04207021573173286,
      "grad_norm": 0.4791950285434723,
      "learning_rate": 1.9912783965757995e-05,
      "loss": 0.3374,
      "step": 2330
    },
    {
      "epoch": 0.04216049516463358,
      "grad_norm": 0.4995405375957489,
      "learning_rate": 1.9912409794373736e-05,
      "loss": 0.3707,
      "step": 2335
    },
    {
      "epoch": 0.04225077459753429,
      "grad_norm": 0.35879209637641907,
      "learning_rate": 1.991203482560818e-05,
      "loss": 0.3106,
      "step": 2340
    },
    {
      "epoch": 0.042341054030435005,
      "grad_norm": 0.4058288335800171,
      "learning_rate": 1.9911659059491487e-05,
      "loss": 0.366,
      "step": 2345
    },
    {
      "epoch": 0.04243133346333572,
      "grad_norm": 0.40655022859573364,
      "learning_rate": 1.9911282496053893e-05,
      "loss": 0.3075,
      "step": 2350
    },
    {
      "epoch": 0.04252161289623643,
      "grad_norm": 0.5874953269958496,
      "learning_rate": 1.991090513532568e-05,
      "loss": 0.4062,
      "step": 2355
    },
    {
      "epoch": 0.042611892329137147,
      "grad_norm": 0.4730669856071472,
      "learning_rate": 1.991052697733721e-05,
      "loss": 0.3985,
      "step": 2360
    },
    {
      "epoch": 0.04270217176203786,
      "grad_norm": 0.4676113724708557,
      "learning_rate": 1.99101480221189e-05,
      "loss": 0.3341,
      "step": 2365
    },
    {
      "epoch": 0.042792451194938574,
      "grad_norm": 0.44648388028144836,
      "learning_rate": 1.9909768269701236e-05,
      "loss": 0.3104,
      "step": 2370
    },
    {
      "epoch": 0.04288273062783929,
      "grad_norm": 0.38011738657951355,
      "learning_rate": 1.9909387720114764e-05,
      "loss": 0.3112,
      "step": 2375
    },
    {
      "epoch": 0.04297301006074,
      "grad_norm": 0.31403690576553345,
      "learning_rate": 1.9909006373390102e-05,
      "loss": 0.2566,
      "step": 2380
    },
    {
      "epoch": 0.043063289493640716,
      "grad_norm": 0.4737754464149475,
      "learning_rate": 1.990862422955792e-05,
      "loss": 0.3378,
      "step": 2385
    },
    {
      "epoch": 0.04315356892654143,
      "grad_norm": 0.374519944190979,
      "learning_rate": 1.9908241288648963e-05,
      "loss": 0.4081,
      "step": 2390
    },
    {
      "epoch": 0.043243848359442144,
      "grad_norm": 0.42596983909606934,
      "learning_rate": 1.9907857550694034e-05,
      "loss": 0.4021,
      "step": 2395
    },
    {
      "epoch": 0.04333412779234286,
      "grad_norm": 0.490863561630249,
      "learning_rate": 1.9907473015724e-05,
      "loss": 0.372,
      "step": 2400
    },
    {
      "epoch": 0.04342440722524357,
      "grad_norm": 0.3635746240615845,
      "learning_rate": 1.9907087683769795e-05,
      "loss": 0.2196,
      "step": 2405
    },
    {
      "epoch": 0.043514686658144286,
      "grad_norm": 0.4454294443130493,
      "learning_rate": 1.9906701554862423e-05,
      "loss": 0.3902,
      "step": 2410
    },
    {
      "epoch": 0.043604966091045,
      "grad_norm": 0.325874388217926,
      "learning_rate": 1.9906314629032936e-05,
      "loss": 0.2661,
      "step": 2415
    },
    {
      "epoch": 0.043695245523945714,
      "grad_norm": 0.7062177062034607,
      "learning_rate": 1.9905926906312466e-05,
      "loss": 0.2619,
      "step": 2420
    },
    {
      "epoch": 0.04378552495684643,
      "grad_norm": 0.635353684425354,
      "learning_rate": 1.9905538386732196e-05,
      "loss": 0.3557,
      "step": 2425
    },
    {
      "epoch": 0.04387580438974715,
      "grad_norm": 0.8798350095748901,
      "learning_rate": 1.9905149070323383e-05,
      "loss": 0.3551,
      "step": 2430
    },
    {
      "epoch": 0.04396608382264786,
      "grad_norm": 0.7844729423522949,
      "learning_rate": 1.9904758957117345e-05,
      "loss": 0.364,
      "step": 2435
    },
    {
      "epoch": 0.044056363255548577,
      "grad_norm": 0.49181339144706726,
      "learning_rate": 1.9904368047145462e-05,
      "loss": 0.2548,
      "step": 2440
    },
    {
      "epoch": 0.04414664268844929,
      "grad_norm": 0.31904664635658264,
      "learning_rate": 1.9903976340439186e-05,
      "loss": 0.4465,
      "step": 2445
    },
    {
      "epoch": 0.044236922121350004,
      "grad_norm": 0.33480775356292725,
      "learning_rate": 1.990358383703002e-05,
      "loss": 0.3779,
      "step": 2450
    },
    {
      "epoch": 0.04432720155425072,
      "grad_norm": 0.541325032711029,
      "learning_rate": 1.990319053694954e-05,
      "loss": 0.393,
      "step": 2455
    },
    {
      "epoch": 0.04441748098715143,
      "grad_norm": 0.5086470246315002,
      "learning_rate": 1.9902796440229383e-05,
      "loss": 0.3949,
      "step": 2460
    },
    {
      "epoch": 0.044507760420052146,
      "grad_norm": 0.813579797744751,
      "learning_rate": 1.9902401546901253e-05,
      "loss": 0.3863,
      "step": 2465
    },
    {
      "epoch": 0.04459803985295286,
      "grad_norm": 0.42829373478889465,
      "learning_rate": 1.9902005856996918e-05,
      "loss": 0.3079,
      "step": 2470
    },
    {
      "epoch": 0.044688319285853574,
      "grad_norm": 0.6729962825775146,
      "learning_rate": 1.9901609370548204e-05,
      "loss": 0.4118,
      "step": 2475
    },
    {
      "epoch": 0.04477859871875429,
      "grad_norm": 0.3326593041419983,
      "learning_rate": 1.990121208758701e-05,
      "loss": 0.3975,
      "step": 2480
    },
    {
      "epoch": 0.044868878151655,
      "grad_norm": 0.4417182505130768,
      "learning_rate": 1.990081400814529e-05,
      "loss": 0.3568,
      "step": 2485
    },
    {
      "epoch": 0.044959157584555716,
      "grad_norm": 0.38405391573905945,
      "learning_rate": 1.990041513225507e-05,
      "loss": 0.3257,
      "step": 2490
    },
    {
      "epoch": 0.04504943701745643,
      "grad_norm": 0.48008227348327637,
      "learning_rate": 1.990001545994844e-05,
      "loss": 0.4107,
      "step": 2495
    },
    {
      "epoch": 0.045139716450357144,
      "grad_norm": 0.4501303732395172,
      "learning_rate": 1.9899614991257542e-05,
      "loss": 0.2903,
      "step": 2500
    },
    {
      "epoch": 0.04522999588325786,
      "grad_norm": 0.51317298412323,
      "learning_rate": 1.9899213726214593e-05,
      "loss": 0.3633,
      "step": 2505
    },
    {
      "epoch": 0.04532027531615857,
      "grad_norm": 0.4251177906990051,
      "learning_rate": 1.9898811664851878e-05,
      "loss": 0.3627,
      "step": 2510
    },
    {
      "epoch": 0.045410554749059286,
      "grad_norm": 0.45249614119529724,
      "learning_rate": 1.9898408807201736e-05,
      "loss": 0.3179,
      "step": 2515
    },
    {
      "epoch": 0.04550083418196,
      "grad_norm": 0.42034104466438293,
      "learning_rate": 1.989800515329657e-05,
      "loss": 0.3904,
      "step": 2520
    },
    {
      "epoch": 0.045591113614860714,
      "grad_norm": 0.6970406770706177,
      "learning_rate": 1.9897600703168857e-05,
      "loss": 0.4268,
      "step": 2525
    },
    {
      "epoch": 0.045681393047761434,
      "grad_norm": 0.4760364592075348,
      "learning_rate": 1.989719545685113e-05,
      "loss": 0.3631,
      "step": 2530
    },
    {
      "epoch": 0.04577167248066215,
      "grad_norm": 0.5618122220039368,
      "learning_rate": 1.9896789414375988e-05,
      "loss": 0.2708,
      "step": 2535
    },
    {
      "epoch": 0.04586195191356286,
      "grad_norm": 0.481415718793869,
      "learning_rate": 1.9896382575776093e-05,
      "loss": 0.3261,
      "step": 2540
    },
    {
      "epoch": 0.045952231346463576,
      "grad_norm": 0.5187461972236633,
      "learning_rate": 1.989597494108418e-05,
      "loss": 0.3711,
      "step": 2545
    },
    {
      "epoch": 0.04604251077936429,
      "grad_norm": 0.4853019714355469,
      "learning_rate": 1.9895566510333024e-05,
      "loss": 0.3891,
      "step": 2550
    },
    {
      "epoch": 0.046132790212265004,
      "grad_norm": 0.7737001776695251,
      "learning_rate": 1.9895157283555495e-05,
      "loss": 0.2272,
      "step": 2555
    },
    {
      "epoch": 0.04622306964516572,
      "grad_norm": 0.5728014707565308,
      "learning_rate": 1.9894747260784507e-05,
      "loss": 0.3846,
      "step": 2560
    },
    {
      "epoch": 0.04631334907806643,
      "grad_norm": 0.3807612657546997,
      "learning_rate": 1.9894336442053043e-05,
      "loss": 0.3339,
      "step": 2565
    },
    {
      "epoch": 0.046403628510967146,
      "grad_norm": 0.405606746673584,
      "learning_rate": 1.989392482739415e-05,
      "loss": 0.3682,
      "step": 2570
    },
    {
      "epoch": 0.04649390794386786,
      "grad_norm": 0.4326905906200409,
      "learning_rate": 1.9893512416840944e-05,
      "loss": 0.3738,
      "step": 2575
    },
    {
      "epoch": 0.046584187376768574,
      "grad_norm": 0.46612468361854553,
      "learning_rate": 1.989309921042659e-05,
      "loss": 0.2992,
      "step": 2580
    },
    {
      "epoch": 0.04667446680966929,
      "grad_norm": 0.4508918821811676,
      "learning_rate": 1.9892685208184343e-05,
      "loss": 0.3831,
      "step": 2585
    },
    {
      "epoch": 0.04676474624257,
      "grad_norm": 0.5471922159194946,
      "learning_rate": 1.9892270410147492e-05,
      "loss": 0.3873,
      "step": 2590
    },
    {
      "epoch": 0.046855025675470716,
      "grad_norm": 0.4468011260032654,
      "learning_rate": 1.9891854816349415e-05,
      "loss": 0.3566,
      "step": 2595
    },
    {
      "epoch": 0.04694530510837143,
      "grad_norm": 0.5099403262138367,
      "learning_rate": 1.9891438426823532e-05,
      "loss": 0.3611,
      "step": 2600
    },
    {
      "epoch": 0.047035584541272143,
      "grad_norm": 0.5567428469657898,
      "learning_rate": 1.989102124160335e-05,
      "loss": 0.3352,
      "step": 2605
    },
    {
      "epoch": 0.04712586397417286,
      "grad_norm": 0.4418875277042389,
      "learning_rate": 1.9890603260722424e-05,
      "loss": 0.3167,
      "step": 2610
    },
    {
      "epoch": 0.04721614340707357,
      "grad_norm": 0.47677722573280334,
      "learning_rate": 1.9890184484214375e-05,
      "loss": 0.3716,
      "step": 2615
    },
    {
      "epoch": 0.047306422839974285,
      "grad_norm": 0.5190219879150391,
      "learning_rate": 1.9889764912112893e-05,
      "loss": 0.3514,
      "step": 2620
    },
    {
      "epoch": 0.047396702272875,
      "grad_norm": 0.3794596195220947,
      "learning_rate": 1.988934454445173e-05,
      "loss": 0.3697,
      "step": 2625
    },
    {
      "epoch": 0.04748698170577572,
      "grad_norm": 0.375012069940567,
      "learning_rate": 1.9888923381264703e-05,
      "loss": 0.2698,
      "step": 2630
    },
    {
      "epoch": 0.047577261138676434,
      "grad_norm": 0.5114025473594666,
      "learning_rate": 1.9888501422585687e-05,
      "loss": 0.3313,
      "step": 2635
    },
    {
      "epoch": 0.04766754057157715,
      "grad_norm": 0.5759722590446472,
      "learning_rate": 1.9888078668448627e-05,
      "loss": 0.4041,
      "step": 2640
    },
    {
      "epoch": 0.04775782000447786,
      "grad_norm": 0.6198399662971497,
      "learning_rate": 1.988765511888753e-05,
      "loss": 0.2688,
      "step": 2645
    },
    {
      "epoch": 0.047848099437378576,
      "grad_norm": 0.4790531098842621,
      "learning_rate": 1.9887230773936474e-05,
      "loss": 0.3142,
      "step": 2650
    },
    {
      "epoch": 0.04793837887027929,
      "grad_norm": 0.4564208388328552,
      "learning_rate": 1.9886805633629587e-05,
      "loss": 0.3542,
      "step": 2655
    },
    {
      "epoch": 0.048028658303180004,
      "grad_norm": 0.3777180016040802,
      "learning_rate": 1.9886379698001074e-05,
      "loss": 0.3278,
      "step": 2660
    },
    {
      "epoch": 0.04811893773608072,
      "grad_norm": 0.34479618072509766,
      "learning_rate": 1.9885952967085188e-05,
      "loss": 0.3324,
      "step": 2665
    },
    {
      "epoch": 0.04820921716898143,
      "grad_norm": 0.5915319323539734,
      "learning_rate": 1.9885525440916268e-05,
      "loss": 0.2877,
      "step": 2670
    },
    {
      "epoch": 0.048299496601882146,
      "grad_norm": 0.4588766396045685,
      "learning_rate": 1.98850971195287e-05,
      "loss": 0.4066,
      "step": 2675
    },
    {
      "epoch": 0.04838977603478286,
      "grad_norm": 0.4266802966594696,
      "learning_rate": 1.9884668002956943e-05,
      "loss": 0.3398,
      "step": 2680
    },
    {
      "epoch": 0.048480055467683573,
      "grad_norm": 0.3396337032318115,
      "learning_rate": 1.988423809123551e-05,
      "loss": 0.3159,
      "step": 2685
    },
    {
      "epoch": 0.04857033490058429,
      "grad_norm": 1.1436978578567505,
      "learning_rate": 1.988380738439899e-05,
      "loss": 0.4754,
      "step": 2690
    },
    {
      "epoch": 0.048660614333485,
      "grad_norm": 0.47760069370269775,
      "learning_rate": 1.9883375882482027e-05,
      "loss": 0.2486,
      "step": 2695
    },
    {
      "epoch": 0.048750893766385715,
      "grad_norm": 0.5436611175537109,
      "learning_rate": 1.9882943585519338e-05,
      "loss": 0.3299,
      "step": 2700
    },
    {
      "epoch": 0.04884117319928643,
      "grad_norm": 0.482835054397583,
      "learning_rate": 1.9882510493545687e-05,
      "loss": 0.4476,
      "step": 2705
    },
    {
      "epoch": 0.04893145263218714,
      "grad_norm": 0.5951783657073975,
      "learning_rate": 1.988207660659592e-05,
      "loss": 0.3985,
      "step": 2710
    },
    {
      "epoch": 0.04902173206508786,
      "grad_norm": 1.275805115699768,
      "learning_rate": 1.9881641924704945e-05,
      "loss": 0.3928,
      "step": 2715
    },
    {
      "epoch": 0.04911201149798857,
      "grad_norm": 0.49138695001602173,
      "learning_rate": 1.988120644790772e-05,
      "loss": 0.318,
      "step": 2720
    },
    {
      "epoch": 0.04920229093088929,
      "grad_norm": 0.3850145936012268,
      "learning_rate": 1.9880770176239278e-05,
      "loss": 0.3826,
      "step": 2725
    },
    {
      "epoch": 0.049292570363790006,
      "grad_norm": 0.455759197473526,
      "learning_rate": 1.988033310973472e-05,
      "loss": 0.22,
      "step": 2730
    },
    {
      "epoch": 0.04938284979669072,
      "grad_norm": 0.39483642578125,
      "learning_rate": 1.98798952484292e-05,
      "loss": 0.3783,
      "step": 2735
    },
    {
      "epoch": 0.049473129229591434,
      "grad_norm": 0.3339937925338745,
      "learning_rate": 1.9879456592357934e-05,
      "loss": 0.3165,
      "step": 2740
    },
    {
      "epoch": 0.04956340866249215,
      "grad_norm": 0.40515002608299255,
      "learning_rate": 1.987901714155622e-05,
      "loss": 0.3021,
      "step": 2745
    },
    {
      "epoch": 0.04965368809539286,
      "grad_norm": 0.5087426900863647,
      "learning_rate": 1.9878576896059408e-05,
      "loss": 0.3483,
      "step": 2750
    },
    {
      "epoch": 0.049743967528293576,
      "grad_norm": 0.4974749982357025,
      "learning_rate": 1.9878135855902908e-05,
      "loss": 0.302,
      "step": 2755
    },
    {
      "epoch": 0.04983424696119429,
      "grad_norm": 0.48196542263031006,
      "learning_rate": 1.9877694021122194e-05,
      "loss": 0.4315,
      "step": 2760
    },
    {
      "epoch": 0.049924526394095003,
      "grad_norm": 0.511130154132843,
      "learning_rate": 1.987725139175282e-05,
      "loss": 0.35,
      "step": 2765
    },
    {
      "epoch": 0.05001480582699572,
      "grad_norm": 0.48580029606819153,
      "learning_rate": 1.987680796783038e-05,
      "loss": 0.3647,
      "step": 2770
    },
    {
      "epoch": 0.05010508525989643,
      "grad_norm": 0.538997232913971,
      "learning_rate": 1.9876363749390554e-05,
      "loss": 0.3305,
      "step": 2775
    },
    {
      "epoch": 0.050195364692797145,
      "grad_norm": 0.38143017888069153,
      "learning_rate": 1.9875918736469072e-05,
      "loss": 0.2437,
      "step": 2780
    },
    {
      "epoch": 0.05028564412569786,
      "grad_norm": 0.44034236669540405,
      "learning_rate": 1.9875472929101732e-05,
      "loss": 0.2925,
      "step": 2785
    },
    {
      "epoch": 0.05037592355859857,
      "grad_norm": 0.4986114799976349,
      "learning_rate": 1.9875026327324396e-05,
      "loss": 0.2893,
      "step": 2790
    },
    {
      "epoch": 0.05046620299149929,
      "grad_norm": 0.6320555806159973,
      "learning_rate": 1.9874578931172993e-05,
      "loss": 0.2763,
      "step": 2795
    },
    {
      "epoch": 0.0505564824244,
      "grad_norm": 0.5865118503570557,
      "learning_rate": 1.9874130740683507e-05,
      "loss": 0.4108,
      "step": 2800
    },
    {
      "epoch": 0.050646761857300715,
      "grad_norm": 0.541379988193512,
      "learning_rate": 1.9873681755891996e-05,
      "loss": 0.4155,
      "step": 2805
    },
    {
      "epoch": 0.05073704129020143,
      "grad_norm": 0.49120455980300903,
      "learning_rate": 1.9873231976834576e-05,
      "loss": 0.3032,
      "step": 2810
    },
    {
      "epoch": 0.05082732072310214,
      "grad_norm": 0.5522422194480896,
      "learning_rate": 1.987278140354743e-05,
      "loss": 0.4814,
      "step": 2815
    },
    {
      "epoch": 0.05091760015600286,
      "grad_norm": 0.34580957889556885,
      "learning_rate": 1.9872330036066803e-05,
      "loss": 0.3633,
      "step": 2820
    },
    {
      "epoch": 0.05100787958890358,
      "grad_norm": 0.5104445219039917,
      "learning_rate": 1.9871877874429e-05,
      "loss": 0.3664,
      "step": 2825
    },
    {
      "epoch": 0.05109815902180429,
      "grad_norm": 0.5858346223831177,
      "learning_rate": 1.98714249186704e-05,
      "loss": 0.3519,
      "step": 2830
    },
    {
      "epoch": 0.051188438454705006,
      "grad_norm": 0.43192100524902344,
      "learning_rate": 1.9870971168827436e-05,
      "loss": 0.3973,
      "step": 2835
    },
    {
      "epoch": 0.05127871788760572,
      "grad_norm": 0.380086213350296,
      "learning_rate": 1.987051662493661e-05,
      "loss": 0.2811,
      "step": 2840
    },
    {
      "epoch": 0.05136899732050643,
      "grad_norm": 0.7647920250892639,
      "learning_rate": 1.987006128703449e-05,
      "loss": 0.4462,
      "step": 2845
    },
    {
      "epoch": 0.05145927675340715,
      "grad_norm": 0.4594999849796295,
      "learning_rate": 1.98696051551577e-05,
      "loss": 0.2968,
      "step": 2850
    },
    {
      "epoch": 0.05154955618630786,
      "grad_norm": 0.43784573674201965,
      "learning_rate": 1.9869148229342933e-05,
      "loss": 0.3771,
      "step": 2855
    },
    {
      "epoch": 0.051639835619208575,
      "grad_norm": 0.454778254032135,
      "learning_rate": 1.9868690509626945e-05,
      "loss": 0.3421,
      "step": 2860
    },
    {
      "epoch": 0.05173011505210929,
      "grad_norm": 0.5888170599937439,
      "learning_rate": 1.9868231996046563e-05,
      "loss": 0.3305,
      "step": 2865
    },
    {
      "epoch": 0.05182039448501,
      "grad_norm": 0.3069855868816376,
      "learning_rate": 1.9867772688638663e-05,
      "loss": 0.2406,
      "step": 2870
    },
    {
      "epoch": 0.05191067391791072,
      "grad_norm": 0.3524327874183655,
      "learning_rate": 1.9867312587440197e-05,
      "loss": 0.2714,
      "step": 2875
    },
    {
      "epoch": 0.05200095335081143,
      "grad_norm": 0.574265718460083,
      "learning_rate": 1.9866851692488175e-05,
      "loss": 0.4171,
      "step": 2880
    },
    {
      "epoch": 0.052091232783712145,
      "grad_norm": 0.593289315700531,
      "learning_rate": 1.9866390003819674e-05,
      "loss": 0.4074,
      "step": 2885
    },
    {
      "epoch": 0.05218151221661286,
      "grad_norm": 0.4223598837852478,
      "learning_rate": 1.9865927521471832e-05,
      "loss": 0.2159,
      "step": 2890
    },
    {
      "epoch": 0.05227179164951357,
      "grad_norm": 0.6746003031730652,
      "learning_rate": 1.9865464245481854e-05,
      "loss": 0.4203,
      "step": 2895
    },
    {
      "epoch": 0.05236207108241429,
      "grad_norm": 0.3579443395137787,
      "learning_rate": 1.9865000175887005e-05,
      "loss": 0.2457,
      "step": 2900
    },
    {
      "epoch": 0.052452350515315,
      "grad_norm": 0.4429759979248047,
      "learning_rate": 1.9864535312724617e-05,
      "loss": 0.3532,
      "step": 2905
    },
    {
      "epoch": 0.052542629948215715,
      "grad_norm": 0.4187673032283783,
      "learning_rate": 1.9864069656032087e-05,
      "loss": 0.3091,
      "step": 2910
    },
    {
      "epoch": 0.05263290938111643,
      "grad_norm": 0.48153671622276306,
      "learning_rate": 1.9863603205846867e-05,
      "loss": 0.3176,
      "step": 2915
    },
    {
      "epoch": 0.05272318881401714,
      "grad_norm": 0.5366928577423096,
      "learning_rate": 1.986313596220649e-05,
      "loss": 0.3418,
      "step": 2920
    },
    {
      "epoch": 0.05281346824691786,
      "grad_norm": 0.3011767268180847,
      "learning_rate": 1.9862667925148537e-05,
      "loss": 0.3572,
      "step": 2925
    },
    {
      "epoch": 0.05290374767981858,
      "grad_norm": 0.41139376163482666,
      "learning_rate": 1.9862199094710655e-05,
      "loss": 0.3261,
      "step": 2930
    },
    {
      "epoch": 0.05299402711271929,
      "grad_norm": 1.1922186613082886,
      "learning_rate": 1.9861729470930562e-05,
      "loss": 0.2384,
      "step": 2935
    },
    {
      "epoch": 0.053084306545620005,
      "grad_norm": 0.5575213432312012,
      "learning_rate": 1.9861259053846034e-05,
      "loss": 0.2874,
      "step": 2940
    },
    {
      "epoch": 0.05317458597852072,
      "grad_norm": 0.6539455652236938,
      "learning_rate": 1.9860787843494917e-05,
      "loss": 0.3129,
      "step": 2945
    },
    {
      "epoch": 0.05326486541142143,
      "grad_norm": 0.5355772972106934,
      "learning_rate": 1.9860315839915107e-05,
      "loss": 0.3979,
      "step": 2950
    },
    {
      "epoch": 0.05335514484432215,
      "grad_norm": 0.3928343653678894,
      "learning_rate": 1.9859843043144586e-05,
      "loss": 0.3473,
      "step": 2955
    },
    {
      "epoch": 0.05344542427722286,
      "grad_norm": 0.33358311653137207,
      "learning_rate": 1.9859369453221374e-05,
      "loss": 0.4318,
      "step": 2960
    },
    {
      "epoch": 0.053535703710123575,
      "grad_norm": 0.4088621437549591,
      "learning_rate": 1.9858895070183575e-05,
      "loss": 0.4127,
      "step": 2965
    },
    {
      "epoch": 0.05362598314302429,
      "grad_norm": 0.419476717710495,
      "learning_rate": 1.985841989406935e-05,
      "loss": 0.2679,
      "step": 2970
    },
    {
      "epoch": 0.053716262575925,
      "grad_norm": 0.5436922311782837,
      "learning_rate": 1.985794392491692e-05,
      "loss": 0.351,
      "step": 2975
    },
    {
      "epoch": 0.05380654200882572,
      "grad_norm": 0.34839963912963867,
      "learning_rate": 1.985746716276458e-05,
      "loss": 0.4118,
      "step": 2980
    },
    {
      "epoch": 0.05389682144172643,
      "grad_norm": 0.5277071595191956,
      "learning_rate": 1.9856989607650675e-05,
      "loss": 0.3752,
      "step": 2985
    },
    {
      "epoch": 0.053987100874627145,
      "grad_norm": 0.5830458998680115,
      "learning_rate": 1.9856511259613622e-05,
      "loss": 0.3538,
      "step": 2990
    },
    {
      "epoch": 0.05407738030752786,
      "grad_norm": 0.3557010889053345,
      "learning_rate": 1.9856032118691906e-05,
      "loss": 0.2958,
      "step": 2995
    },
    {
      "epoch": 0.05416765974042857,
      "grad_norm": 0.5906873941421509,
      "learning_rate": 1.9855552184924062e-05,
      "loss": 0.3612,
      "step": 3000
    },
    {
      "epoch": 0.054257939173329287,
      "grad_norm": 0.569794237613678,
      "learning_rate": 1.98550714583487e-05,
      "loss": 0.364,
      "step": 3005
    },
    {
      "epoch": 0.05434821860623,
      "grad_norm": 0.3360729515552521,
      "learning_rate": 1.9854589939004497e-05,
      "loss": 0.3981,
      "step": 3010
    },
    {
      "epoch": 0.054438498039130714,
      "grad_norm": 0.4309172034263611,
      "learning_rate": 1.9854107626930185e-05,
      "loss": 0.3597,
      "step": 3015
    },
    {
      "epoch": 0.05452877747203143,
      "grad_norm": 0.31440845131874084,
      "learning_rate": 1.9853624522164558e-05,
      "loss": 0.4754,
      "step": 3020
    },
    {
      "epoch": 0.05461905690493215,
      "grad_norm": 0.3023746609687805,
      "learning_rate": 1.9853140624746478e-05,
      "loss": 0.3672,
      "step": 3025
    },
    {
      "epoch": 0.05470933633783286,
      "grad_norm": 0.36276915669441223,
      "learning_rate": 1.9852655934714878e-05,
      "loss": 0.3996,
      "step": 3030
    },
    {
      "epoch": 0.05479961577073358,
      "grad_norm": 0.493784636259079,
      "learning_rate": 1.9852170452108742e-05,
      "loss": 0.3379,
      "step": 3035
    },
    {
      "epoch": 0.05488989520363429,
      "grad_norm": 0.5143473148345947,
      "learning_rate": 1.9851684176967122e-05,
      "loss": 0.2968,
      "step": 3040
    },
    {
      "epoch": 0.054980174636535005,
      "grad_norm": 0.47927024960517883,
      "learning_rate": 1.9851197109329147e-05,
      "loss": 0.3886,
      "step": 3045
    },
    {
      "epoch": 0.05507045406943572,
      "grad_norm": 0.35105928778648376,
      "learning_rate": 1.9850709249233984e-05,
      "loss": 0.3185,
      "step": 3050
    },
    {
      "epoch": 0.05516073350233643,
      "grad_norm": 1.5786962509155273,
      "learning_rate": 1.985022059672088e-05,
      "loss": 0.4056,
      "step": 3055
    },
    {
      "epoch": 0.05525101293523715,
      "grad_norm": 0.4125722050666809,
      "learning_rate": 1.9849731151829154e-05,
      "loss": 0.3661,
      "step": 3060
    },
    {
      "epoch": 0.05534129236813786,
      "grad_norm": 0.4464266002178192,
      "learning_rate": 1.984924091459817e-05,
      "loss": 0.4217,
      "step": 3065
    },
    {
      "epoch": 0.055431571801038575,
      "grad_norm": 0.3512585759162903,
      "learning_rate": 1.9848749885067362e-05,
      "loss": 0.354,
      "step": 3070
    },
    {
      "epoch": 0.05552185123393929,
      "grad_norm": 0.7285350561141968,
      "learning_rate": 1.9848258063276237e-05,
      "loss": 0.3269,
      "step": 3075
    },
    {
      "epoch": 0.05561213066684,
      "grad_norm": 0.5315104126930237,
      "learning_rate": 1.984776544926435e-05,
      "loss": 0.2996,
      "step": 3080
    },
    {
      "epoch": 0.055702410099740716,
      "grad_norm": 0.40287673473358154,
      "learning_rate": 1.9847272043071336e-05,
      "loss": 0.3824,
      "step": 3085
    },
    {
      "epoch": 0.05579268953264143,
      "grad_norm": 0.47473129630088806,
      "learning_rate": 1.9846777844736883e-05,
      "loss": 0.3019,
      "step": 3090
    },
    {
      "epoch": 0.055882968965542144,
      "grad_norm": 0.30406618118286133,
      "learning_rate": 1.984628285430074e-05,
      "loss": 0.3529,
      "step": 3095
    },
    {
      "epoch": 0.05597324839844286,
      "grad_norm": 0.5010662078857422,
      "learning_rate": 1.984578707180274e-05,
      "loss": 0.3797,
      "step": 3100
    },
    {
      "epoch": 0.05606352783134357,
      "grad_norm": 0.4138856828212738,
      "learning_rate": 1.984529049728275e-05,
      "loss": 0.2762,
      "step": 3105
    },
    {
      "epoch": 0.056153807264244286,
      "grad_norm": 0.3830830454826355,
      "learning_rate": 1.9844793130780723e-05,
      "loss": 0.417,
      "step": 3110
    },
    {
      "epoch": 0.056244086697145,
      "grad_norm": 0.5126296877861023,
      "learning_rate": 1.984429497233667e-05,
      "loss": 0.3476,
      "step": 3115
    },
    {
      "epoch": 0.056334366130045714,
      "grad_norm": 0.49585217237472534,
      "learning_rate": 1.9843796021990656e-05,
      "loss": 0.3795,
      "step": 3120
    },
    {
      "epoch": 0.056424645562946435,
      "grad_norm": 0.4351014792919159,
      "learning_rate": 1.9843296279782825e-05,
      "loss": 0.3296,
      "step": 3125
    },
    {
      "epoch": 0.05651492499584715,
      "grad_norm": 0.37983760237693787,
      "learning_rate": 1.9842795745753376e-05,
      "loss": 0.351,
      "step": 3130
    },
    {
      "epoch": 0.05660520442874786,
      "grad_norm": 0.7745433449745178,
      "learning_rate": 1.9842294419942572e-05,
      "loss": 0.3034,
      "step": 3135
    },
    {
      "epoch": 0.05669548386164858,
      "grad_norm": 0.553109884262085,
      "learning_rate": 1.9841792302390747e-05,
      "loss": 0.3261,
      "step": 3140
    },
    {
      "epoch": 0.05678576329454929,
      "grad_norm": 0.6224424242973328,
      "learning_rate": 1.9841289393138287e-05,
      "loss": 0.2753,
      "step": 3145
    },
    {
      "epoch": 0.056876042727450005,
      "grad_norm": 0.6656163334846497,
      "learning_rate": 1.9840785692225645e-05,
      "loss": 0.3733,
      "step": 3150
    },
    {
      "epoch": 0.05696632216035072,
      "grad_norm": 0.4527309834957123,
      "learning_rate": 1.9840281199693345e-05,
      "loss": 0.4109,
      "step": 3155
    },
    {
      "epoch": 0.05705660159325143,
      "grad_norm": 0.41423696279525757,
      "learning_rate": 1.983977591558197e-05,
      "loss": 0.357,
      "step": 3160
    },
    {
      "epoch": 0.057146881026152146,
      "grad_norm": 0.5543522238731384,
      "learning_rate": 1.9839269839932162e-05,
      "loss": 0.393,
      "step": 3165
    },
    {
      "epoch": 0.05723716045905286,
      "grad_norm": 0.6002402305603027,
      "learning_rate": 1.9838762972784634e-05,
      "loss": 0.3379,
      "step": 3170
    },
    {
      "epoch": 0.057327439891953574,
      "grad_norm": 0.38091033697128296,
      "learning_rate": 1.9838255314180163e-05,
      "loss": 0.3197,
      "step": 3175
    },
    {
      "epoch": 0.05741771932485429,
      "grad_norm": 0.34828096628189087,
      "learning_rate": 1.9837746864159578e-05,
      "loss": 0.3275,
      "step": 3180
    },
    {
      "epoch": 0.057507998757755,
      "grad_norm": 0.5166292190551758,
      "learning_rate": 1.9837237622763788e-05,
      "loss": 0.345,
      "step": 3185
    },
    {
      "epoch": 0.057598278190655716,
      "grad_norm": 0.37860623002052307,
      "learning_rate": 1.9836727590033756e-05,
      "loss": 0.2667,
      "step": 3190
    },
    {
      "epoch": 0.05768855762355643,
      "grad_norm": 0.48768389225006104,
      "learning_rate": 1.9836216766010507e-05,
      "loss": 0.4015,
      "step": 3195
    },
    {
      "epoch": 0.057778837056457144,
      "grad_norm": 0.44820886850357056,
      "learning_rate": 1.9835705150735135e-05,
      "loss": 0.3085,
      "step": 3200
    },
    {
      "epoch": 0.05786911648935786,
      "grad_norm": 0.9818539023399353,
      "learning_rate": 1.98351927442488e-05,
      "loss": 0.353,
      "step": 3205
    },
    {
      "epoch": 0.05795939592225857,
      "grad_norm": 0.42301586270332336,
      "learning_rate": 1.9834679546592713e-05,
      "loss": 0.3963,
      "step": 3210
    },
    {
      "epoch": 0.058049675355159286,
      "grad_norm": 0.537584125995636,
      "learning_rate": 1.9834165557808162e-05,
      "loss": 0.3601,
      "step": 3215
    },
    {
      "epoch": 0.05813995478806001,
      "grad_norm": 0.5891759395599365,
      "learning_rate": 1.9833650777936495e-05,
      "loss": 0.4072,
      "step": 3220
    },
    {
      "epoch": 0.05823023422096072,
      "grad_norm": 0.5722532272338867,
      "learning_rate": 1.9833135207019122e-05,
      "loss": 0.376,
      "step": 3225
    },
    {
      "epoch": 0.058320513653861435,
      "grad_norm": 0.45028820633888245,
      "learning_rate": 1.983261884509751e-05,
      "loss": 0.287,
      "step": 3230
    },
    {
      "epoch": 0.05841079308676215,
      "grad_norm": 0.4146955609321594,
      "learning_rate": 1.983210169221321e-05,
      "loss": 0.428,
      "step": 3235
    },
    {
      "epoch": 0.05850107251966286,
      "grad_norm": 0.48254334926605225,
      "learning_rate": 1.9831583748407807e-05,
      "loss": 0.2921,
      "step": 3240
    },
    {
      "epoch": 0.058591351952563576,
      "grad_norm": 0.5707606077194214,
      "learning_rate": 1.983106501372298e-05,
      "loss": 0.326,
      "step": 3245
    },
    {
      "epoch": 0.05868163138546429,
      "grad_norm": 0.43480393290519714,
      "learning_rate": 1.983054548820045e-05,
      "loss": 0.3006,
      "step": 3250
    },
    {
      "epoch": 0.058771910818365004,
      "grad_norm": 0.29172283411026,
      "learning_rate": 1.983002517188201e-05,
      "loss": 0.3144,
      "step": 3255
    },
    {
      "epoch": 0.05886219025126572,
      "grad_norm": 0.6074573993682861,
      "learning_rate": 1.982950406480951e-05,
      "loss": 0.3375,
      "step": 3260
    },
    {
      "epoch": 0.05895246968416643,
      "grad_norm": 0.42056509852409363,
      "learning_rate": 1.9828982167024882e-05,
      "loss": 0.3143,
      "step": 3265
    },
    {
      "epoch": 0.059042749117067146,
      "grad_norm": 0.6018386483192444,
      "learning_rate": 1.98284594785701e-05,
      "loss": 0.4923,
      "step": 3270
    },
    {
      "epoch": 0.05913302854996786,
      "grad_norm": 0.32788190245628357,
      "learning_rate": 1.9827935999487215e-05,
      "loss": 0.3249,
      "step": 3275
    },
    {
      "epoch": 0.059223307982868574,
      "grad_norm": 0.31291717290878296,
      "learning_rate": 1.9827411729818332e-05,
      "loss": 0.2892,
      "step": 3280
    },
    {
      "epoch": 0.05931358741576929,
      "grad_norm": 0.5440911650657654,
      "learning_rate": 1.9826886669605633e-05,
      "loss": 0.2986,
      "step": 3285
    },
    {
      "epoch": 0.05940386684867,
      "grad_norm": 0.2649855315685272,
      "learning_rate": 1.9826360818891342e-05,
      "loss": 0.287,
      "step": 3290
    },
    {
      "epoch": 0.059494146281570716,
      "grad_norm": 0.6282209753990173,
      "learning_rate": 1.9825834177717774e-05,
      "loss": 0.3779,
      "step": 3295
    },
    {
      "epoch": 0.05958442571447143,
      "grad_norm": 0.37331846356391907,
      "learning_rate": 1.9825306746127286e-05,
      "loss": 0.3176,
      "step": 3300
    },
    {
      "epoch": 0.059674705147372144,
      "grad_norm": 0.43291202187538147,
      "learning_rate": 1.9824778524162304e-05,
      "loss": 0.4459,
      "step": 3305
    },
    {
      "epoch": 0.05976498458027286,
      "grad_norm": 1.268259882926941,
      "learning_rate": 1.9824249511865327e-05,
      "loss": 0.2672,
      "step": 3310
    },
    {
      "epoch": 0.05985526401317357,
      "grad_norm": 0.6148844957351685,
      "learning_rate": 1.98237197092789e-05,
      "loss": 0.3542,
      "step": 3315
    },
    {
      "epoch": 0.05994554344607429,
      "grad_norm": 0.43036696314811707,
      "learning_rate": 1.9823189116445655e-05,
      "loss": 0.3425,
      "step": 3320
    },
    {
      "epoch": 0.060035822878975006,
      "grad_norm": 0.45163917541503906,
      "learning_rate": 1.9822657733408263e-05,
      "loss": 0.3049,
      "step": 3325
    },
    {
      "epoch": 0.06012610231187572,
      "grad_norm": 0.32905444502830505,
      "learning_rate": 1.9822125560209475e-05,
      "loss": 0.3319,
      "step": 3330
    },
    {
      "epoch": 0.060216381744776434,
      "grad_norm": 0.44667723774909973,
      "learning_rate": 1.9821592596892095e-05,
      "loss": 0.4263,
      "step": 3335
    },
    {
      "epoch": 0.06030666117767715,
      "grad_norm": 0.36804917454719543,
      "learning_rate": 1.9821058843499005e-05,
      "loss": 0.3912,
      "step": 3340
    },
    {
      "epoch": 0.06039694061057786,
      "grad_norm": 0.4349153935909271,
      "learning_rate": 1.9820524300073138e-05,
      "loss": 0.3845,
      "step": 3345
    },
    {
      "epoch": 0.060487220043478576,
      "grad_norm": 0.32116422057151794,
      "learning_rate": 1.981998896665749e-05,
      "loss": 0.386,
      "step": 3350
    },
    {
      "epoch": 0.06057749947637929,
      "grad_norm": 0.4326367974281311,
      "learning_rate": 1.981945284329513e-05,
      "loss": 0.3779,
      "step": 3355
    },
    {
      "epoch": 0.060667778909280004,
      "grad_norm": 0.49139365553855896,
      "learning_rate": 1.981891593002918e-05,
      "loss": 0.3826,
      "step": 3360
    },
    {
      "epoch": 0.06075805834218072,
      "grad_norm": 0.27795523405075073,
      "learning_rate": 1.981837822690284e-05,
      "loss": 0.2998,
      "step": 3365
    },
    {
      "epoch": 0.06084833777508143,
      "grad_norm": 0.4365173578262329,
      "learning_rate": 1.9817839733959353e-05,
      "loss": 0.3052,
      "step": 3370
    },
    {
      "epoch": 0.060938617207982146,
      "grad_norm": 0.4508330523967743,
      "learning_rate": 1.981730045124204e-05,
      "loss": 0.3769,
      "step": 3375
    },
    {
      "epoch": 0.06102889664088286,
      "grad_norm": 0.2785418629646301,
      "learning_rate": 1.9816760378794293e-05,
      "loss": 0.3254,
      "step": 3380
    },
    {
      "epoch": 0.061119176073783574,
      "grad_norm": 0.4931766390800476,
      "learning_rate": 1.9816219516659544e-05,
      "loss": 0.4362,
      "step": 3385
    },
    {
      "epoch": 0.06120945550668429,
      "grad_norm": 0.3930407166481018,
      "learning_rate": 1.98156778648813e-05,
      "loss": 0.3377,
      "step": 3390
    },
    {
      "epoch": 0.061299734939585,
      "grad_norm": 0.542591392993927,
      "learning_rate": 1.9815135423503147e-05,
      "loss": 0.3106,
      "step": 3395
    },
    {
      "epoch": 0.061390014372485716,
      "grad_norm": 0.5310285091400146,
      "learning_rate": 1.9814592192568705e-05,
      "loss": 0.3922,
      "step": 3400
    },
    {
      "epoch": 0.06148029380538643,
      "grad_norm": 0.4742412865161896,
      "learning_rate": 1.9814048172121687e-05,
      "loss": 0.3593,
      "step": 3405
    },
    {
      "epoch": 0.06157057323828714,
      "grad_norm": 0.4899846315383911,
      "learning_rate": 1.9813503362205844e-05,
      "loss": 0.4096,
      "step": 3410
    },
    {
      "epoch": 0.06166085267118786,
      "grad_norm": 0.3952639400959015,
      "learning_rate": 1.981295776286501e-05,
      "loss": 0.3031,
      "step": 3415
    },
    {
      "epoch": 0.06175113210408858,
      "grad_norm": 0.2628712058067322,
      "learning_rate": 1.9812411374143066e-05,
      "loss": 0.39,
      "step": 3420
    },
    {
      "epoch": 0.06184141153698929,
      "grad_norm": 0.44826173782348633,
      "learning_rate": 1.9811864196083974e-05,
      "loss": 0.431,
      "step": 3425
    },
    {
      "epoch": 0.061931690969890006,
      "grad_norm": 0.3274790048599243,
      "learning_rate": 1.9811316228731745e-05,
      "loss": 0.3531,
      "step": 3430
    },
    {
      "epoch": 0.06202197040279072,
      "grad_norm": 0.4401794672012329,
      "learning_rate": 1.9810767472130462e-05,
      "loss": 0.4226,
      "step": 3435
    },
    {
      "epoch": 0.062112249835691434,
      "grad_norm": 0.3578872084617615,
      "learning_rate": 1.9810217926324266e-05,
      "loss": 0.3608,
      "step": 3440
    },
    {
      "epoch": 0.06220252926859215,
      "grad_norm": 0.5258836150169373,
      "learning_rate": 1.9809667591357364e-05,
      "loss": 0.3388,
      "step": 3445
    },
    {
      "epoch": 0.06229280870149286,
      "grad_norm": 0.4620509743690491,
      "learning_rate": 1.9809116467274027e-05,
      "loss": 0.302,
      "step": 3450
    },
    {
      "epoch": 0.062383088134393576,
      "grad_norm": 0.7103094458580017,
      "learning_rate": 1.9808564554118593e-05,
      "loss": 0.4503,
      "step": 3455
    },
    {
      "epoch": 0.06247336756729429,
      "grad_norm": 0.4419422447681427,
      "learning_rate": 1.9808011851935453e-05,
      "loss": 0.3287,
      "step": 3460
    },
    {
      "epoch": 0.062563647000195,
      "grad_norm": 0.34888213872909546,
      "learning_rate": 1.980745836076907e-05,
      "loss": 0.3216,
      "step": 3465
    },
    {
      "epoch": 0.06265392643309572,
      "grad_norm": 0.4253569543361664,
      "learning_rate": 1.9806904080663973e-05,
      "loss": 0.2493,
      "step": 3470
    },
    {
      "epoch": 0.06274420586599644,
      "grad_norm": 0.32277703285217285,
      "learning_rate": 1.9806349011664743e-05,
      "loss": 0.2501,
      "step": 3475
    },
    {
      "epoch": 0.06283448529889715,
      "grad_norm": 0.6440743803977966,
      "learning_rate": 1.9805793153816033e-05,
      "loss": 0.4372,
      "step": 3480
    },
    {
      "epoch": 0.06292476473179787,
      "grad_norm": 0.34801673889160156,
      "learning_rate": 1.980523650716256e-05,
      "loss": 0.3797,
      "step": 3485
    },
    {
      "epoch": 0.06301504416469858,
      "grad_norm": 0.789560079574585,
      "learning_rate": 1.98046790717491e-05,
      "loss": 0.3145,
      "step": 3490
    },
    {
      "epoch": 0.0631053235975993,
      "grad_norm": 0.33960312604904175,
      "learning_rate": 1.9804120847620497e-05,
      "loss": 0.3943,
      "step": 3495
    },
    {
      "epoch": 0.06319560303050001,
      "grad_norm": 0.5149698853492737,
      "learning_rate": 1.9803561834821657e-05,
      "loss": 0.2716,
      "step": 3500
    },
    {
      "epoch": 0.06328588246340072,
      "grad_norm": 0.5456732511520386,
      "learning_rate": 1.9803002033397542e-05,
      "loss": 0.4073,
      "step": 3505
    },
    {
      "epoch": 0.06337616189630144,
      "grad_norm": 0.4857495129108429,
      "learning_rate": 1.9802441443393187e-05,
      "loss": 0.3824,
      "step": 3510
    },
    {
      "epoch": 0.06346644132920215,
      "grad_norm": 0.458262175321579,
      "learning_rate": 1.9801880064853694e-05,
      "loss": 0.356,
      "step": 3515
    },
    {
      "epoch": 0.06355672076210286,
      "grad_norm": 0.4099235534667969,
      "learning_rate": 1.9801317897824214e-05,
      "loss": 0.3158,
      "step": 3520
    },
    {
      "epoch": 0.06364700019500358,
      "grad_norm": 0.45231691002845764,
      "learning_rate": 1.9800754942349967e-05,
      "loss": 0.393,
      "step": 3525
    },
    {
      "epoch": 0.06373727962790429,
      "grad_norm": 0.3657291829586029,
      "learning_rate": 1.9800191198476248e-05,
      "loss": 0.3686,
      "step": 3530
    },
    {
      "epoch": 0.063827559060805,
      "grad_norm": 0.4763554036617279,
      "learning_rate": 1.9799626666248405e-05,
      "loss": 0.3492,
      "step": 3535
    },
    {
      "epoch": 0.06391783849370572,
      "grad_norm": 0.4591434597969055,
      "learning_rate": 1.9799061345711837e-05,
      "loss": 0.2811,
      "step": 3540
    },
    {
      "epoch": 0.06400811792660643,
      "grad_norm": 0.3183038830757141,
      "learning_rate": 1.9798495236912035e-05,
      "loss": 0.3524,
      "step": 3545
    },
    {
      "epoch": 0.06409839735950715,
      "grad_norm": 0.2819090187549591,
      "learning_rate": 1.9797928339894537e-05,
      "loss": 0.313,
      "step": 3550
    },
    {
      "epoch": 0.06418867679240786,
      "grad_norm": 0.4335143268108368,
      "learning_rate": 1.9797360654704937e-05,
      "loss": 0.3756,
      "step": 3555
    },
    {
      "epoch": 0.06427895622530858,
      "grad_norm": 0.6104137301445007,
      "learning_rate": 1.9796792181388905e-05,
      "loss": 0.3316,
      "step": 3560
    },
    {
      "epoch": 0.06436923565820929,
      "grad_norm": 1.3540409803390503,
      "learning_rate": 1.9796222919992176e-05,
      "loss": 0.309,
      "step": 3565
    },
    {
      "epoch": 0.06445951509111,
      "grad_norm": 0.599658191204071,
      "learning_rate": 1.9795652870560537e-05,
      "loss": 0.452,
      "step": 3570
    },
    {
      "epoch": 0.06454979452401072,
      "grad_norm": 0.5024389028549194,
      "learning_rate": 1.9795082033139846e-05,
      "loss": 0.4017,
      "step": 3575
    },
    {
      "epoch": 0.06464007395691143,
      "grad_norm": 0.6276569962501526,
      "learning_rate": 1.979451040777602e-05,
      "loss": 0.3668,
      "step": 3580
    },
    {
      "epoch": 0.06473035338981215,
      "grad_norm": 0.4447115659713745,
      "learning_rate": 1.9793937994515046e-05,
      "loss": 0.3121,
      "step": 3585
    },
    {
      "epoch": 0.06482063282271286,
      "grad_norm": 0.36672645807266235,
      "learning_rate": 1.979336479340297e-05,
      "loss": 0.3228,
      "step": 3590
    },
    {
      "epoch": 0.06491091225561357,
      "grad_norm": 0.5025267004966736,
      "learning_rate": 1.97927908044859e-05,
      "loss": 0.3724,
      "step": 3595
    },
    {
      "epoch": 0.06500119168851429,
      "grad_norm": 0.6396102905273438,
      "learning_rate": 1.9792216027810013e-05,
      "loss": 0.3236,
      "step": 3600
    },
    {
      "epoch": 0.065091471121415,
      "grad_norm": 0.36236289143562317,
      "learning_rate": 1.979164046342154e-05,
      "loss": 0.3841,
      "step": 3605
    },
    {
      "epoch": 0.06518175055431571,
      "grad_norm": 0.45055314898490906,
      "learning_rate": 1.9791064111366783e-05,
      "loss": 0.3967,
      "step": 3610
    },
    {
      "epoch": 0.06527202998721643,
      "grad_norm": 0.4135984778404236,
      "learning_rate": 1.979048697169211e-05,
      "loss": 0.2799,
      "step": 3615
    },
    {
      "epoch": 0.06536230942011714,
      "grad_norm": 0.617309033870697,
      "learning_rate": 1.9789909044443944e-05,
      "loss": 0.2757,
      "step": 3620
    },
    {
      "epoch": 0.06545258885301786,
      "grad_norm": 0.489812970161438,
      "learning_rate": 1.978933032966877e-05,
      "loss": 0.367,
      "step": 3625
    },
    {
      "epoch": 0.06554286828591857,
      "grad_norm": 0.4110862612724304,
      "learning_rate": 1.978875082741315e-05,
      "loss": 0.3243,
      "step": 3630
    },
    {
      "epoch": 0.06563314771881928,
      "grad_norm": 0.39427581429481506,
      "learning_rate": 1.9788170537723694e-05,
      "loss": 0.3962,
      "step": 3635
    },
    {
      "epoch": 0.06572342715172,
      "grad_norm": 0.44661369919776917,
      "learning_rate": 1.9787589460647084e-05,
      "loss": 0.3356,
      "step": 3640
    },
    {
      "epoch": 0.06581370658462071,
      "grad_norm": 0.9452258944511414,
      "learning_rate": 1.978700759623007e-05,
      "loss": 0.3642,
      "step": 3645
    },
    {
      "epoch": 0.06590398601752143,
      "grad_norm": 0.38086244463920593,
      "learning_rate": 1.978642494451945e-05,
      "loss": 0.3563,
      "step": 3650
    },
    {
      "epoch": 0.06599426545042214,
      "grad_norm": 0.6124266386032104,
      "learning_rate": 1.97858415055621e-05,
      "loss": 0.3346,
      "step": 3655
    },
    {
      "epoch": 0.06608454488332285,
      "grad_norm": 0.5609608888626099,
      "learning_rate": 1.978525727940495e-05,
      "loss": 0.401,
      "step": 3660
    },
    {
      "epoch": 0.06617482431622357,
      "grad_norm": 0.46799227595329285,
      "learning_rate": 1.9784672266094994e-05,
      "loss": 0.3334,
      "step": 3665
    },
    {
      "epoch": 0.0662651037491243,
      "grad_norm": 0.41846963763237,
      "learning_rate": 1.97840864656793e-05,
      "loss": 0.3084,
      "step": 3670
    },
    {
      "epoch": 0.06635538318202501,
      "grad_norm": 0.5600510239601135,
      "learning_rate": 1.9783499878204985e-05,
      "loss": 0.3431,
      "step": 3675
    },
    {
      "epoch": 0.06644566261492572,
      "grad_norm": 1.7698122262954712,
      "learning_rate": 1.9782912503719234e-05,
      "loss": 0.3264,
      "step": 3680
    },
    {
      "epoch": 0.06653594204782644,
      "grad_norm": 0.43976593017578125,
      "learning_rate": 1.9782324342269308e-05,
      "loss": 0.3047,
      "step": 3685
    },
    {
      "epoch": 0.06662622148072715,
      "grad_norm": 0.3489255905151367,
      "learning_rate": 1.9781735393902507e-05,
      "loss": 0.3868,
      "step": 3690
    },
    {
      "epoch": 0.06671650091362787,
      "grad_norm": 0.4278669059276581,
      "learning_rate": 1.978114565866622e-05,
      "loss": 0.4227,
      "step": 3695
    },
    {
      "epoch": 0.06680678034652858,
      "grad_norm": 0.37567952275276184,
      "learning_rate": 1.9780555136607873e-05,
      "loss": 0.3084,
      "step": 3700
    },
    {
      "epoch": 0.0668970597794293,
      "grad_norm": 0.651716411113739,
      "learning_rate": 1.977996382777498e-05,
      "loss": 0.2765,
      "step": 3705
    },
    {
      "epoch": 0.06698733921233001,
      "grad_norm": 0.4037342667579651,
      "learning_rate": 1.9779371732215107e-05,
      "loss": 0.3601,
      "step": 3710
    },
    {
      "epoch": 0.06707761864523072,
      "grad_norm": 0.31218260526657104,
      "learning_rate": 1.977877884997588e-05,
      "loss": 0.3889,
      "step": 3715
    },
    {
      "epoch": 0.06716789807813144,
      "grad_norm": 0.37937313318252563,
      "learning_rate": 1.977818518110499e-05,
      "loss": 0.4928,
      "step": 3720
    },
    {
      "epoch": 0.06725817751103215,
      "grad_norm": 0.33725881576538086,
      "learning_rate": 1.97775907256502e-05,
      "loss": 0.3793,
      "step": 3725
    },
    {
      "epoch": 0.06734845694393286,
      "grad_norm": 0.387472927570343,
      "learning_rate": 1.9776995483659323e-05,
      "loss": 0.4042,
      "step": 3730
    },
    {
      "epoch": 0.06743873637683358,
      "grad_norm": 0.37021878361701965,
      "learning_rate": 1.977639945518025e-05,
      "loss": 0.2947,
      "step": 3735
    },
    {
      "epoch": 0.06752901580973429,
      "grad_norm": 0.4049207270145416,
      "learning_rate": 1.9775802640260916e-05,
      "loss": 0.317,
      "step": 3740
    },
    {
      "epoch": 0.067619295242635,
      "grad_norm": 0.4391418397426605,
      "learning_rate": 1.977520503894934e-05,
      "loss": 0.276,
      "step": 3745
    },
    {
      "epoch": 0.06770957467553572,
      "grad_norm": 0.396352618932724,
      "learning_rate": 1.977460665129359e-05,
      "loss": 0.3707,
      "step": 3750
    },
    {
      "epoch": 0.06779985410843643,
      "grad_norm": 1.2072421312332153,
      "learning_rate": 1.9774007477341802e-05,
      "loss": 0.3005,
      "step": 3755
    },
    {
      "epoch": 0.06789013354133715,
      "grad_norm": 0.5573527216911316,
      "learning_rate": 1.977340751714218e-05,
      "loss": 0.4486,
      "step": 3760
    },
    {
      "epoch": 0.06798041297423786,
      "grad_norm": 0.41838520765304565,
      "learning_rate": 1.977280677074298e-05,
      "loss": 0.3391,
      "step": 3765
    },
    {
      "epoch": 0.06807069240713857,
      "grad_norm": 0.32235589623451233,
      "learning_rate": 1.977220523819253e-05,
      "loss": 0.3726,
      "step": 3770
    },
    {
      "epoch": 0.06816097184003929,
      "grad_norm": 0.37072479724884033,
      "learning_rate": 1.9771602919539223e-05,
      "loss": 0.323,
      "step": 3775
    },
    {
      "epoch": 0.06825125127294,
      "grad_norm": 0.4548185169696808,
      "learning_rate": 1.9770999814831504e-05,
      "loss": 0.3868,
      "step": 3780
    },
    {
      "epoch": 0.06834153070584072,
      "grad_norm": 0.4475475549697876,
      "learning_rate": 1.9770395924117895e-05,
      "loss": 0.2801,
      "step": 3785
    },
    {
      "epoch": 0.06843181013874143,
      "grad_norm": 0.501424252986908,
      "learning_rate": 1.9769791247446968e-05,
      "loss": 0.3281,
      "step": 3790
    },
    {
      "epoch": 0.06852208957164214,
      "grad_norm": 0.41786015033721924,
      "learning_rate": 1.976918578486737e-05,
      "loss": 0.4034,
      "step": 3795
    },
    {
      "epoch": 0.06861236900454286,
      "grad_norm": 0.45296943187713623,
      "learning_rate": 1.9768579536427804e-05,
      "loss": 0.4663,
      "step": 3800
    },
    {
      "epoch": 0.06870264843744357,
      "grad_norm": 0.5736542344093323,
      "learning_rate": 1.976797250217704e-05,
      "loss": 0.3618,
      "step": 3805
    },
    {
      "epoch": 0.06879292787034429,
      "grad_norm": 0.3980604112148285,
      "learning_rate": 1.9767364682163905e-05,
      "loss": 0.3069,
      "step": 3810
    },
    {
      "epoch": 0.068883207303245,
      "grad_norm": 0.5466489195823669,
      "learning_rate": 1.97667560764373e-05,
      "loss": 0.3198,
      "step": 3815
    },
    {
      "epoch": 0.06897348673614571,
      "grad_norm": 0.7333075404167175,
      "learning_rate": 1.9766146685046183e-05,
      "loss": 0.3152,
      "step": 3820
    },
    {
      "epoch": 0.06906376616904643,
      "grad_norm": 0.4144575893878937,
      "learning_rate": 1.9765536508039568e-05,
      "loss": 0.3104,
      "step": 3825
    },
    {
      "epoch": 0.06915404560194714,
      "grad_norm": 0.6920064091682434,
      "learning_rate": 1.976492554546654e-05,
      "loss": 0.3001,
      "step": 3830
    },
    {
      "epoch": 0.06924432503484786,
      "grad_norm": 0.39211317896842957,
      "learning_rate": 1.9764313797376253e-05,
      "loss": 0.3743,
      "step": 3835
    },
    {
      "epoch": 0.06933460446774857,
      "grad_norm": 0.510002613067627,
      "learning_rate": 1.9763701263817916e-05,
      "loss": 0.3874,
      "step": 3840
    },
    {
      "epoch": 0.06942488390064928,
      "grad_norm": 0.5999088883399963,
      "learning_rate": 1.97630879448408e-05,
      "loss": 0.3603,
      "step": 3845
    },
    {
      "epoch": 0.06951516333355,
      "grad_norm": 0.5141642093658447,
      "learning_rate": 1.976247384049424e-05,
      "loss": 0.3159,
      "step": 3850
    },
    {
      "epoch": 0.06960544276645071,
      "grad_norm": 0.43448373675346375,
      "learning_rate": 1.9761858950827646e-05,
      "loss": 0.3758,
      "step": 3855
    },
    {
      "epoch": 0.06969572219935143,
      "grad_norm": 0.4024636447429657,
      "learning_rate": 1.9761243275890473e-05,
      "loss": 0.2689,
      "step": 3860
    },
    {
      "epoch": 0.06978600163225215,
      "grad_norm": 0.3778102695941925,
      "learning_rate": 1.9760626815732244e-05,
      "loss": 0.2949,
      "step": 3865
    },
    {
      "epoch": 0.06987628106515287,
      "grad_norm": 0.5055009126663208,
      "learning_rate": 1.976000957040256e-05,
      "loss": 0.412,
      "step": 3870
    },
    {
      "epoch": 0.06996656049805358,
      "grad_norm": 0.5268850326538086,
      "learning_rate": 1.9759391539951067e-05,
      "loss": 0.3701,
      "step": 3875
    },
    {
      "epoch": 0.0700568399309543,
      "grad_norm": 0.4896146357059479,
      "learning_rate": 1.975877272442748e-05,
      "loss": 0.2631,
      "step": 3880
    },
    {
      "epoch": 0.07014711936385501,
      "grad_norm": 0.5230270624160767,
      "learning_rate": 1.9758153123881585e-05,
      "loss": 0.3436,
      "step": 3885
    },
    {
      "epoch": 0.07023739879675572,
      "grad_norm": 0.4784453809261322,
      "learning_rate": 1.9757532738363217e-05,
      "loss": 0.4115,
      "step": 3890
    },
    {
      "epoch": 0.07032767822965644,
      "grad_norm": 0.369460791349411,
      "learning_rate": 1.9756911567922287e-05,
      "loss": 0.2817,
      "step": 3895
    },
    {
      "epoch": 0.07041795766255715,
      "grad_norm": 0.40652188658714294,
      "learning_rate": 1.9756289612608755e-05,
      "loss": 0.4009,
      "step": 3900
    },
    {
      "epoch": 0.07050823709545787,
      "grad_norm": 0.6725921630859375,
      "learning_rate": 1.9755666872472662e-05,
      "loss": 0.3781,
      "step": 3905
    },
    {
      "epoch": 0.07059851652835858,
      "grad_norm": 0.4345565140247345,
      "learning_rate": 1.97550433475641e-05,
      "loss": 0.3592,
      "step": 3910
    },
    {
      "epoch": 0.0706887959612593,
      "grad_norm": 0.46316981315612793,
      "learning_rate": 1.9754419037933226e-05,
      "loss": 0.2884,
      "step": 3915
    },
    {
      "epoch": 0.07077907539416001,
      "grad_norm": 0.42469629645347595,
      "learning_rate": 1.9753793943630263e-05,
      "loss": 0.279,
      "step": 3920
    },
    {
      "epoch": 0.07086935482706072,
      "grad_norm": 0.4613531529903412,
      "learning_rate": 1.9753168064705496e-05,
      "loss": 0.3594,
      "step": 3925
    },
    {
      "epoch": 0.07095963425996143,
      "grad_norm": 0.5785402655601501,
      "learning_rate": 1.9752541401209268e-05,
      "loss": 0.355,
      "step": 3930
    },
    {
      "epoch": 0.07104991369286215,
      "grad_norm": 0.4152916371822357,
      "learning_rate": 1.9751913953191992e-05,
      "loss": 0.3728,
      "step": 3935
    },
    {
      "epoch": 0.07114019312576286,
      "grad_norm": 0.6485270261764526,
      "learning_rate": 1.975128572070414e-05,
      "loss": 0.3343,
      "step": 3940
    },
    {
      "epoch": 0.07123047255866358,
      "grad_norm": 0.47957125306129456,
      "learning_rate": 1.9750656703796254e-05,
      "loss": 0.3742,
      "step": 3945
    },
    {
      "epoch": 0.07132075199156429,
      "grad_norm": 0.5528354644775391,
      "learning_rate": 1.9750026902518928e-05,
      "loss": 0.409,
      "step": 3950
    },
    {
      "epoch": 0.071411031424465,
      "grad_norm": 0.48165422677993774,
      "learning_rate": 1.974939631692283e-05,
      "loss": 0.3406,
      "step": 3955
    },
    {
      "epoch": 0.07150131085736572,
      "grad_norm": 0.8650749325752258,
      "learning_rate": 1.974876494705868e-05,
      "loss": 0.4059,
      "step": 3960
    },
    {
      "epoch": 0.07159159029026643,
      "grad_norm": 0.5957344770431519,
      "learning_rate": 1.974813279297727e-05,
      "loss": 0.379,
      "step": 3965
    },
    {
      "epoch": 0.07168186972316715,
      "grad_norm": 0.4109193980693817,
      "learning_rate": 1.974749985472946e-05,
      "loss": 0.3744,
      "step": 3970
    },
    {
      "epoch": 0.07177214915606786,
      "grad_norm": 0.3410075902938843,
      "learning_rate": 1.9746866132366146e-05,
      "loss": 0.352,
      "step": 3975
    },
    {
      "epoch": 0.07186242858896857,
      "grad_norm": 0.35367947816848755,
      "learning_rate": 1.9746231625938326e-05,
      "loss": 0.3135,
      "step": 3980
    },
    {
      "epoch": 0.07195270802186929,
      "grad_norm": 0.8415671586990356,
      "learning_rate": 1.9745596335497033e-05,
      "loss": 0.3666,
      "step": 3985
    },
    {
      "epoch": 0.07204298745477,
      "grad_norm": 0.4690043032169342,
      "learning_rate": 1.974496026109337e-05,
      "loss": 0.3948,
      "step": 3990
    },
    {
      "epoch": 0.07213326688767072,
      "grad_norm": 0.5731785893440247,
      "learning_rate": 1.9744323402778508e-05,
      "loss": 0.3948,
      "step": 3995
    },
    {
      "epoch": 0.07222354632057143,
      "grad_norm": 0.4461507797241211,
      "learning_rate": 1.9743685760603678e-05,
      "loss": 0.3688,
      "step": 4000
    },
    {
      "epoch": 0.07231382575347214,
      "grad_norm": 0.5973614454269409,
      "learning_rate": 1.974304733462017e-05,
      "loss": 0.3063,
      "step": 4005
    },
    {
      "epoch": 0.07240410518637286,
      "grad_norm": 0.6062004566192627,
      "learning_rate": 1.9742408124879342e-05,
      "loss": 0.3435,
      "step": 4010
    },
    {
      "epoch": 0.07249438461927357,
      "grad_norm": 0.3516961932182312,
      "learning_rate": 1.9741768131432615e-05,
      "loss": 0.3467,
      "step": 4015
    },
    {
      "epoch": 0.07258466405217429,
      "grad_norm": 0.4388677179813385,
      "learning_rate": 1.9741127354331473e-05,
      "loss": 0.4482,
      "step": 4020
    },
    {
      "epoch": 0.072674943485075,
      "grad_norm": 0.5486716628074646,
      "learning_rate": 1.9740485793627456e-05,
      "loss": 0.3167,
      "step": 4025
    },
    {
      "epoch": 0.07276522291797571,
      "grad_norm": 0.533378005027771,
      "learning_rate": 1.9739843449372183e-05,
      "loss": 0.321,
      "step": 4030
    },
    {
      "epoch": 0.07285550235087643,
      "grad_norm": 0.4790610373020172,
      "learning_rate": 1.9739200321617318e-05,
      "loss": 0.3641,
      "step": 4035
    },
    {
      "epoch": 0.07294578178377714,
      "grad_norm": 0.42860275506973267,
      "learning_rate": 1.9738556410414596e-05,
      "loss": 0.3189,
      "step": 4040
    },
    {
      "epoch": 0.07303606121667786,
      "grad_norm": 0.7600460648536682,
      "learning_rate": 1.9737911715815817e-05,
      "loss": 0.4523,
      "step": 4045
    },
    {
      "epoch": 0.07312634064957857,
      "grad_norm": 0.43815359473228455,
      "learning_rate": 1.9737266237872844e-05,
      "loss": 0.2613,
      "step": 4050
    },
    {
      "epoch": 0.07321662008247928,
      "grad_norm": 0.43807390332221985,
      "learning_rate": 1.9736619976637596e-05,
      "loss": 0.2835,
      "step": 4055
    },
    {
      "epoch": 0.07330689951538,
      "grad_norm": 0.30521079897880554,
      "learning_rate": 1.9735972932162068e-05,
      "loss": 0.3206,
      "step": 4060
    },
    {
      "epoch": 0.07339717894828073,
      "grad_norm": 0.3289305567741394,
      "learning_rate": 1.97353251044983e-05,
      "loss": 0.3101,
      "step": 4065
    },
    {
      "epoch": 0.07348745838118144,
      "grad_norm": 0.39569854736328125,
      "learning_rate": 1.9734676493698412e-05,
      "loss": 0.4014,
      "step": 4070
    },
    {
      "epoch": 0.07357773781408215,
      "grad_norm": 0.33709490299224854,
      "learning_rate": 1.9734027099814578e-05,
      "loss": 0.3654,
      "step": 4075
    },
    {
      "epoch": 0.07366801724698287,
      "grad_norm": 0.39169785380363464,
      "learning_rate": 1.9733376922899036e-05,
      "loss": 0.358,
      "step": 4080
    },
    {
      "epoch": 0.07375829667988358,
      "grad_norm": 0.5747213959693909,
      "learning_rate": 1.973272596300409e-05,
      "loss": 0.3499,
      "step": 4085
    },
    {
      "epoch": 0.0738485761127843,
      "grad_norm": 0.3984346091747284,
      "learning_rate": 1.9732074220182104e-05,
      "loss": 0.3424,
      "step": 4090
    },
    {
      "epoch": 0.07393885554568501,
      "grad_norm": 0.48990488052368164,
      "learning_rate": 1.9731421694485505e-05,
      "loss": 0.3655,
      "step": 4095
    },
    {
      "epoch": 0.07402913497858572,
      "grad_norm": 0.4573669135570526,
      "learning_rate": 1.9730768385966788e-05,
      "loss": 0.4599,
      "step": 4100
    },
    {
      "epoch": 0.07411941441148644,
      "grad_norm": 0.3597734868526459,
      "learning_rate": 1.97301142946785e-05,
      "loss": 0.3179,
      "step": 4105
    },
    {
      "epoch": 0.07420969384438715,
      "grad_norm": 0.5145891308784485,
      "learning_rate": 1.972945942067326e-05,
      "loss": 0.2926,
      "step": 4110
    },
    {
      "epoch": 0.07429997327728786,
      "grad_norm": 0.33695390820503235,
      "learning_rate": 1.9728803764003757e-05,
      "loss": 0.3327,
      "step": 4115
    },
    {
      "epoch": 0.07439025271018858,
      "grad_norm": 0.5866163372993469,
      "learning_rate": 1.9728147324722722e-05,
      "loss": 0.3584,
      "step": 4120
    },
    {
      "epoch": 0.07448053214308929,
      "grad_norm": 0.3432018458843231,
      "learning_rate": 1.9727490102882966e-05,
      "loss": 0.3631,
      "step": 4125
    },
    {
      "epoch": 0.07457081157599,
      "grad_norm": 0.33699434995651245,
      "learning_rate": 1.9726832098537355e-05,
      "loss": 0.3481,
      "step": 4130
    },
    {
      "epoch": 0.07466109100889072,
      "grad_norm": 0.3236297070980072,
      "learning_rate": 1.9726173311738825e-05,
      "loss": 0.3764,
      "step": 4135
    },
    {
      "epoch": 0.07475137044179143,
      "grad_norm": 0.44569236040115356,
      "learning_rate": 1.9725513742540366e-05,
      "loss": 0.3007,
      "step": 4140
    },
    {
      "epoch": 0.07484164987469215,
      "grad_norm": 0.48883622884750366,
      "learning_rate": 1.972485339099504e-05,
      "loss": 0.336,
      "step": 4145
    },
    {
      "epoch": 0.07493192930759286,
      "grad_norm": 0.514825701713562,
      "learning_rate": 1.9724192257155965e-05,
      "loss": 0.3236,
      "step": 4150
    },
    {
      "epoch": 0.07502220874049358,
      "grad_norm": 0.7090126872062683,
      "learning_rate": 1.9723530341076326e-05,
      "loss": 0.3345,
      "step": 4155
    },
    {
      "epoch": 0.07511248817339429,
      "grad_norm": 0.3429259657859802,
      "learning_rate": 1.9722867642809364e-05,
      "loss": 0.2971,
      "step": 4160
    },
    {
      "epoch": 0.075202767606295,
      "grad_norm": 0.4875260293483734,
      "learning_rate": 1.9722204162408393e-05,
      "loss": 0.3013,
      "step": 4165
    },
    {
      "epoch": 0.07529304703919572,
      "grad_norm": 0.49795055389404297,
      "learning_rate": 1.9721539899926788e-05,
      "loss": 0.2305,
      "step": 4170
    },
    {
      "epoch": 0.07538332647209643,
      "grad_norm": 0.7367872595787048,
      "learning_rate": 1.972087485541798e-05,
      "loss": 0.3618,
      "step": 4175
    },
    {
      "epoch": 0.07547360590499715,
      "grad_norm": 0.3421260118484497,
      "learning_rate": 1.972020902893546e-05,
      "loss": 0.3283,
      "step": 4180
    },
    {
      "epoch": 0.07556388533789786,
      "grad_norm": 0.7015025019645691,
      "learning_rate": 1.9719542420532805e-05,
      "loss": 0.2494,
      "step": 4185
    },
    {
      "epoch": 0.07565416477079857,
      "grad_norm": 0.7375475168228149,
      "learning_rate": 1.9718875030263627e-05,
      "loss": 0.3091,
      "step": 4190
    },
    {
      "epoch": 0.07574444420369929,
      "grad_norm": 0.3345068097114563,
      "learning_rate": 1.9718206858181615e-05,
      "loss": 0.428,
      "step": 4195
    },
    {
      "epoch": 0.0758347236366,
      "grad_norm": 0.6164200305938721,
      "learning_rate": 1.971753790434052e-05,
      "loss": 0.4313,
      "step": 4200
    },
    {
      "epoch": 0.07592500306950072,
      "grad_norm": 0.518981397151947,
      "learning_rate": 1.9716868168794154e-05,
      "loss": 0.309,
      "step": 4205
    },
    {
      "epoch": 0.07601528250240143,
      "grad_norm": 0.44345569610595703,
      "learning_rate": 1.9716197651596393e-05,
      "loss": 0.3035,
      "step": 4210
    },
    {
      "epoch": 0.07610556193530214,
      "grad_norm": 0.5354166626930237,
      "learning_rate": 1.9715526352801175e-05,
      "loss": 0.2773,
      "step": 4215
    },
    {
      "epoch": 0.07619584136820286,
      "grad_norm": 0.5086137652397156,
      "learning_rate": 1.9714854272462497e-05,
      "loss": 0.4168,
      "step": 4220
    },
    {
      "epoch": 0.07628612080110357,
      "grad_norm": 0.631428599357605,
      "learning_rate": 1.971418141063443e-05,
      "loss": 0.2984,
      "step": 4225
    },
    {
      "epoch": 0.07637640023400429,
      "grad_norm": 0.4155879020690918,
      "learning_rate": 1.9713507767371097e-05,
      "loss": 0.3106,
      "step": 4230
    },
    {
      "epoch": 0.076466679666905,
      "grad_norm": 0.4156801998615265,
      "learning_rate": 1.971283334272669e-05,
      "loss": 0.3847,
      "step": 4235
    },
    {
      "epoch": 0.07655695909980571,
      "grad_norm": 0.4328644275665283,
      "learning_rate": 1.9712158136755457e-05,
      "loss": 0.3131,
      "step": 4240
    },
    {
      "epoch": 0.07664723853270643,
      "grad_norm": 0.7468674182891846,
      "learning_rate": 1.9711482149511717e-05,
      "loss": 0.3608,
      "step": 4245
    },
    {
      "epoch": 0.07673751796560714,
      "grad_norm": 0.41286343336105347,
      "learning_rate": 1.9710805381049845e-05,
      "loss": 0.3158,
      "step": 4250
    },
    {
      "epoch": 0.07682779739850785,
      "grad_norm": 0.4288932681083679,
      "learning_rate": 1.971012783142429e-05,
      "loss": 0.2644,
      "step": 4255
    },
    {
      "epoch": 0.07691807683140858,
      "grad_norm": 0.7052945494651794,
      "learning_rate": 1.9709449500689543e-05,
      "loss": 0.312,
      "step": 4260
    },
    {
      "epoch": 0.0770083562643093,
      "grad_norm": 0.36367496848106384,
      "learning_rate": 1.9708770388900183e-05,
      "loss": 0.2539,
      "step": 4265
    },
    {
      "epoch": 0.07709863569721001,
      "grad_norm": 0.3144047260284424,
      "learning_rate": 1.9708090496110834e-05,
      "loss": 0.2911,
      "step": 4270
    },
    {
      "epoch": 0.07718891513011072,
      "grad_norm": 0.48131924867630005,
      "learning_rate": 1.970740982237619e-05,
      "loss": 0.2972,
      "step": 4275
    },
    {
      "epoch": 0.07727919456301144,
      "grad_norm": 0.6672709584236145,
      "learning_rate": 1.9706728367751005e-05,
      "loss": 0.3804,
      "step": 4280
    },
    {
      "epoch": 0.07736947399591215,
      "grad_norm": 0.4266987144947052,
      "learning_rate": 1.9706046132290097e-05,
      "loss": 0.3536,
      "step": 4285
    },
    {
      "epoch": 0.07745975342881287,
      "grad_norm": 0.6607972979545593,
      "learning_rate": 1.9705363116048347e-05,
      "loss": 0.3393,
      "step": 4290
    },
    {
      "epoch": 0.07755003286171358,
      "grad_norm": 0.47216519713401794,
      "learning_rate": 1.9704679319080696e-05,
      "loss": 0.3604,
      "step": 4295
    },
    {
      "epoch": 0.0776403122946143,
      "grad_norm": 0.603981614112854,
      "learning_rate": 1.9703994741442162e-05,
      "loss": 0.3582,
      "step": 4300
    },
    {
      "epoch": 0.07773059172751501,
      "grad_norm": 0.5392507314682007,
      "learning_rate": 1.9703309383187798e-05,
      "loss": 0.3195,
      "step": 4305
    },
    {
      "epoch": 0.07782087116041572,
      "grad_norm": 0.6059638261795044,
      "learning_rate": 1.970262324437275e-05,
      "loss": 0.3596,
      "step": 4310
    },
    {
      "epoch": 0.07791115059331644,
      "grad_norm": 0.45418858528137207,
      "learning_rate": 1.97019363250522e-05,
      "loss": 0.3069,
      "step": 4315
    },
    {
      "epoch": 0.07800143002621715,
      "grad_norm": 0.677732527256012,
      "learning_rate": 1.970124862528142e-05,
      "loss": 0.3266,
      "step": 4320
    },
    {
      "epoch": 0.07809170945911786,
      "grad_norm": 0.4499494135379791,
      "learning_rate": 1.970056014511572e-05,
      "loss": 0.3828,
      "step": 4325
    },
    {
      "epoch": 0.07818198889201858,
      "grad_norm": 0.41509875655174255,
      "learning_rate": 1.9699870884610485e-05,
      "loss": 0.3849,
      "step": 4330
    },
    {
      "epoch": 0.07827226832491929,
      "grad_norm": 0.4929969310760498,
      "learning_rate": 1.9699180843821166e-05,
      "loss": 0.3862,
      "step": 4335
    },
    {
      "epoch": 0.07836254775782,
      "grad_norm": 0.6417096257209778,
      "learning_rate": 1.9698490022803262e-05,
      "loss": 0.3864,
      "step": 4340
    },
    {
      "epoch": 0.07845282719072072,
      "grad_norm": 0.5347675681114197,
      "learning_rate": 1.9697798421612355e-05,
      "loss": 0.3556,
      "step": 4345
    },
    {
      "epoch": 0.07854310662362143,
      "grad_norm": 0.3889930546283722,
      "learning_rate": 1.9697106040304072e-05,
      "loss": 0.3758,
      "step": 4350
    },
    {
      "epoch": 0.07863338605652215,
      "grad_norm": 0.5241528749465942,
      "learning_rate": 1.969641287893411e-05,
      "loss": 0.3058,
      "step": 4355
    },
    {
      "epoch": 0.07872366548942286,
      "grad_norm": 0.5995569229125977,
      "learning_rate": 1.969571893755824e-05,
      "loss": 0.3353,
      "step": 4360
    },
    {
      "epoch": 0.07881394492232358,
      "grad_norm": 0.44480106234550476,
      "learning_rate": 1.969502421623227e-05,
      "loss": 0.2931,
      "step": 4365
    },
    {
      "epoch": 0.07890422435522429,
      "grad_norm": 0.39942800998687744,
      "learning_rate": 1.9694328715012094e-05,
      "loss": 0.3321,
      "step": 4370
    },
    {
      "epoch": 0.078994503788125,
      "grad_norm": 0.42548105120658875,
      "learning_rate": 1.969363243395366e-05,
      "loss": 0.3378,
      "step": 4375
    },
    {
      "epoch": 0.07908478322102572,
      "grad_norm": 0.3198913335800171,
      "learning_rate": 1.969293537311297e-05,
      "loss": 0.3251,
      "step": 4380
    },
    {
      "epoch": 0.07917506265392643,
      "grad_norm": 0.4743817150592804,
      "learning_rate": 1.969223753254611e-05,
      "loss": 0.2683,
      "step": 4385
    },
    {
      "epoch": 0.07926534208682715,
      "grad_norm": 0.3459571599960327,
      "learning_rate": 1.9691538912309203e-05,
      "loss": 0.371,
      "step": 4390
    },
    {
      "epoch": 0.07935562151972786,
      "grad_norm": 0.27832481265068054,
      "learning_rate": 1.9690839512458458e-05,
      "loss": 0.306,
      "step": 4395
    },
    {
      "epoch": 0.07944590095262857,
      "grad_norm": 0.5058069825172424,
      "learning_rate": 1.9690139333050136e-05,
      "loss": 0.3165,
      "step": 4400
    },
    {
      "epoch": 0.07953618038552929,
      "grad_norm": 0.6491962671279907,
      "learning_rate": 1.9689438374140554e-05,
      "loss": 0.392,
      "step": 4405
    },
    {
      "epoch": 0.07962645981843,
      "grad_norm": 0.6899421215057373,
      "learning_rate": 1.9688736635786107e-05,
      "loss": 0.4649,
      "step": 4410
    },
    {
      "epoch": 0.07971673925133071,
      "grad_norm": 0.3906645178794861,
      "learning_rate": 1.968803411804324e-05,
      "loss": 0.2287,
      "step": 4415
    },
    {
      "epoch": 0.07980701868423143,
      "grad_norm": 0.6027063131332397,
      "learning_rate": 1.9687330820968472e-05,
      "loss": 0.2839,
      "step": 4420
    },
    {
      "epoch": 0.07989729811713214,
      "grad_norm": 0.5959811806678772,
      "learning_rate": 1.968662674461837e-05,
      "loss": 0.3659,
      "step": 4425
    },
    {
      "epoch": 0.07998757755003286,
      "grad_norm": 0.5573135614395142,
      "learning_rate": 1.9685921889049574e-05,
      "loss": 0.4069,
      "step": 4430
    },
    {
      "epoch": 0.08007785698293357,
      "grad_norm": 0.5714733600616455,
      "learning_rate": 1.9685216254318786e-05,
      "loss": 0.2656,
      "step": 4435
    },
    {
      "epoch": 0.08016813641583428,
      "grad_norm": 0.5382790565490723,
      "learning_rate": 1.9684509840482766e-05,
      "loss": 0.312,
      "step": 4440
    },
    {
      "epoch": 0.080258415848735,
      "grad_norm": 0.5374647974967957,
      "learning_rate": 1.9683802647598344e-05,
      "loss": 0.2594,
      "step": 4445
    },
    {
      "epoch": 0.08034869528163571,
      "grad_norm": 0.6212499141693115,
      "learning_rate": 1.968309467572241e-05,
      "loss": 0.4128,
      "step": 4450
    },
    {
      "epoch": 0.08043897471453643,
      "grad_norm": 0.4466805160045624,
      "learning_rate": 1.968238592491191e-05,
      "loss": 0.3326,
      "step": 4455
    },
    {
      "epoch": 0.08052925414743715,
      "grad_norm": 0.5234296917915344,
      "learning_rate": 1.9681676395223858e-05,
      "loss": 0.4153,
      "step": 4460
    },
    {
      "epoch": 0.08061953358033787,
      "grad_norm": 0.3909855782985687,
      "learning_rate": 1.9680966086715337e-05,
      "loss": 0.3775,
      "step": 4465
    },
    {
      "epoch": 0.08070981301323858,
      "grad_norm": 0.3779798746109009,
      "learning_rate": 1.9680254999443478e-05,
      "loss": 0.3641,
      "step": 4470
    },
    {
      "epoch": 0.0808000924461393,
      "grad_norm": 0.37800610065460205,
      "learning_rate": 1.967954313346549e-05,
      "loss": 0.2642,
      "step": 4475
    },
    {
      "epoch": 0.08089037187904001,
      "grad_norm": 0.3323556184768677,
      "learning_rate": 1.967883048883863e-05,
      "loss": 0.2069,
      "step": 4480
    },
    {
      "epoch": 0.08098065131194072,
      "grad_norm": 0.40910011529922485,
      "learning_rate": 1.967811706562023e-05,
      "loss": 0.3267,
      "step": 4485
    },
    {
      "epoch": 0.08107093074484144,
      "grad_norm": 0.5210789442062378,
      "learning_rate": 1.9677402863867684e-05,
      "loss": 0.3588,
      "step": 4490
    },
    {
      "epoch": 0.08116121017774215,
      "grad_norm": 0.5192407965660095,
      "learning_rate": 1.9676687883638433e-05,
      "loss": 0.3813,
      "step": 4495
    },
    {
      "epoch": 0.08125148961064287,
      "grad_norm": 0.42497411370277405,
      "learning_rate": 1.967597212499e-05,
      "loss": 0.3897,
      "step": 4500
    },
    {
      "epoch": 0.08134176904354358,
      "grad_norm": 0.4315917193889618,
      "learning_rate": 1.9675255587979963e-05,
      "loss": 0.3845,
      "step": 4505
    },
    {
      "epoch": 0.0814320484764443,
      "grad_norm": 1.1516555547714233,
      "learning_rate": 1.9674538272665957e-05,
      "loss": 0.3392,
      "step": 4510
    },
    {
      "epoch": 0.08152232790934501,
      "grad_norm": 0.44052472710609436,
      "learning_rate": 1.967382017910569e-05,
      "loss": 0.2897,
      "step": 4515
    },
    {
      "epoch": 0.08161260734224572,
      "grad_norm": 0.5504406690597534,
      "learning_rate": 1.9673101307356922e-05,
      "loss": 0.3192,
      "step": 4520
    },
    {
      "epoch": 0.08170288677514644,
      "grad_norm": 0.7369511127471924,
      "learning_rate": 1.9672381657477483e-05,
      "loss": 0.3242,
      "step": 4525
    },
    {
      "epoch": 0.08179316620804715,
      "grad_norm": 0.35009142756462097,
      "learning_rate": 1.9671661229525268e-05,
      "loss": 0.3182,
      "step": 4530
    },
    {
      "epoch": 0.08188344564094786,
      "grad_norm": 0.6602281928062439,
      "learning_rate": 1.9670940023558227e-05,
      "loss": 0.3433,
      "step": 4535
    },
    {
      "epoch": 0.08197372507384858,
      "grad_norm": 0.4609468877315521,
      "learning_rate": 1.9670218039634372e-05,
      "loss": 0.4052,
      "step": 4540
    },
    {
      "epoch": 0.08206400450674929,
      "grad_norm": 0.36140522360801697,
      "learning_rate": 1.966949527781179e-05,
      "loss": 0.2799,
      "step": 4545
    },
    {
      "epoch": 0.08215428393965,
      "grad_norm": 0.4902835488319397,
      "learning_rate": 1.9668771738148615e-05,
      "loss": 0.3481,
      "step": 4550
    },
    {
      "epoch": 0.08224456337255072,
      "grad_norm": 0.4186972975730896,
      "learning_rate": 1.966804742070305e-05,
      "loss": 0.2783,
      "step": 4555
    },
    {
      "epoch": 0.08233484280545143,
      "grad_norm": 0.394040048122406,
      "learning_rate": 1.9667322325533367e-05,
      "loss": 0.3048,
      "step": 4560
    },
    {
      "epoch": 0.08242512223835215,
      "grad_norm": 0.40318575501441956,
      "learning_rate": 1.966659645269789e-05,
      "loss": 0.3221,
      "step": 4565
    },
    {
      "epoch": 0.08251540167125286,
      "grad_norm": 0.3136727511882782,
      "learning_rate": 1.9665869802255007e-05,
      "loss": 0.3529,
      "step": 4570
    },
    {
      "epoch": 0.08260568110415357,
      "grad_norm": 0.5056544542312622,
      "learning_rate": 1.9665142374263186e-05,
      "loss": 0.3468,
      "step": 4575
    },
    {
      "epoch": 0.08269596053705429,
      "grad_norm": 0.42035847902297974,
      "learning_rate": 1.9664414168780923e-05,
      "loss": 0.3593,
      "step": 4580
    },
    {
      "epoch": 0.082786239969955,
      "grad_norm": 0.4517004191875458,
      "learning_rate": 1.9663685185866813e-05,
      "loss": 0.3332,
      "step": 4585
    },
    {
      "epoch": 0.08287651940285572,
      "grad_norm": 0.3949716091156006,
      "learning_rate": 1.966295542557949e-05,
      "loss": 0.3173,
      "step": 4590
    },
    {
      "epoch": 0.08296679883575643,
      "grad_norm": 0.575838565826416,
      "learning_rate": 1.966222488797766e-05,
      "loss": 0.3473,
      "step": 4595
    },
    {
      "epoch": 0.08305707826865714,
      "grad_norm": 0.49253588914871216,
      "learning_rate": 1.966149357312009e-05,
      "loss": 0.2891,
      "step": 4600
    },
    {
      "epoch": 0.08314735770155786,
      "grad_norm": 0.5246073603630066,
      "learning_rate": 1.9660761481065606e-05,
      "loss": 0.4548,
      "step": 4605
    },
    {
      "epoch": 0.08323763713445857,
      "grad_norm": 0.4757160246372223,
      "learning_rate": 1.9660028611873102e-05,
      "loss": 0.3069,
      "step": 4610
    },
    {
      "epoch": 0.08332791656735929,
      "grad_norm": 0.37857431173324585,
      "learning_rate": 1.965929496560153e-05,
      "loss": 0.3433,
      "step": 4615
    },
    {
      "epoch": 0.08341819600026,
      "grad_norm": 0.44293367862701416,
      "learning_rate": 1.965856054230991e-05,
      "loss": 0.3435,
      "step": 4620
    },
    {
      "epoch": 0.08350847543316071,
      "grad_norm": 0.48219558596611023,
      "learning_rate": 1.9657825342057317e-05,
      "loss": 0.2491,
      "step": 4625
    },
    {
      "epoch": 0.08359875486606143,
      "grad_norm": 0.3333645462989807,
      "learning_rate": 1.9657089364902897e-05,
      "loss": 0.2522,
      "step": 4630
    },
    {
      "epoch": 0.08368903429896214,
      "grad_norm": 0.3774479329586029,
      "learning_rate": 1.965635261090585e-05,
      "loss": 0.2893,
      "step": 4635
    },
    {
      "epoch": 0.08377931373186286,
      "grad_norm": 0.33994877338409424,
      "learning_rate": 1.9655615080125444e-05,
      "loss": 0.2458,
      "step": 4640
    },
    {
      "epoch": 0.08386959316476357,
      "grad_norm": 0.5428542494773865,
      "learning_rate": 1.9654876772621006e-05,
      "loss": 0.3831,
      "step": 4645
    },
    {
      "epoch": 0.08395987259766428,
      "grad_norm": 0.4807756841182709,
      "learning_rate": 1.965413768845193e-05,
      "loss": 0.2955,
      "step": 4650
    },
    {
      "epoch": 0.084050152030565,
      "grad_norm": 0.33526355028152466,
      "learning_rate": 1.965339782767767e-05,
      "loss": 0.3172,
      "step": 4655
    },
    {
      "epoch": 0.08414043146346573,
      "grad_norm": 0.30477970838546753,
      "learning_rate": 1.965265719035774e-05,
      "loss": 0.3401,
      "step": 4660
    },
    {
      "epoch": 0.08423071089636644,
      "grad_norm": 0.47036126255989075,
      "learning_rate": 1.9651915776551727e-05,
      "loss": 0.3484,
      "step": 4665
    },
    {
      "epoch": 0.08432099032926715,
      "grad_norm": 0.570608913898468,
      "learning_rate": 1.965117358631926e-05,
      "loss": 0.3407,
      "step": 4670
    },
    {
      "epoch": 0.08441126976216787,
      "grad_norm": 0.4351683259010315,
      "learning_rate": 1.965043061972005e-05,
      "loss": 0.4023,
      "step": 4675
    },
    {
      "epoch": 0.08450154919506858,
      "grad_norm": 0.4088486135005951,
      "learning_rate": 1.9649686876813864e-05,
      "loss": 0.3387,
      "step": 4680
    },
    {
      "epoch": 0.0845918286279693,
      "grad_norm": 0.3737155497074127,
      "learning_rate": 1.964894235766053e-05,
      "loss": 0.3863,
      "step": 4685
    },
    {
      "epoch": 0.08468210806087001,
      "grad_norm": 0.3476511836051941,
      "learning_rate": 1.9648197062319935e-05,
      "loss": 0.3141,
      "step": 4690
    },
    {
      "epoch": 0.08477238749377072,
      "grad_norm": 0.5036412477493286,
      "learning_rate": 1.9647450990852034e-05,
      "loss": 0.2858,
      "step": 4695
    },
    {
      "epoch": 0.08486266692667144,
      "grad_norm": 0.3802044093608856,
      "learning_rate": 1.9646704143316847e-05,
      "loss": 0.3502,
      "step": 4700
    },
    {
      "epoch": 0.08495294635957215,
      "grad_norm": 0.4903644621372223,
      "learning_rate": 1.9645956519774447e-05,
      "loss": 0.2952,
      "step": 4705
    },
    {
      "epoch": 0.08504322579247287,
      "grad_norm": 0.32201284170150757,
      "learning_rate": 1.9645208120284982e-05,
      "loss": 0.3125,
      "step": 4710
    },
    {
      "epoch": 0.08513350522537358,
      "grad_norm": 0.36066821217536926,
      "learning_rate": 1.9644458944908648e-05,
      "loss": 0.3653,
      "step": 4715
    },
    {
      "epoch": 0.08522378465827429,
      "grad_norm": 0.5003206133842468,
      "learning_rate": 1.9643708993705716e-05,
      "loss": 0.3442,
      "step": 4720
    },
    {
      "epoch": 0.08531406409117501,
      "grad_norm": 0.3410700857639313,
      "learning_rate": 1.9642958266736513e-05,
      "loss": 0.3535,
      "step": 4725
    },
    {
      "epoch": 0.08540434352407572,
      "grad_norm": 0.43336379528045654,
      "learning_rate": 1.9642206764061424e-05,
      "loss": 0.3435,
      "step": 4730
    },
    {
      "epoch": 0.08549462295697643,
      "grad_norm": 0.5813950300216675,
      "learning_rate": 1.9641454485740912e-05,
      "loss": 0.2742,
      "step": 4735
    },
    {
      "epoch": 0.08558490238987715,
      "grad_norm": 0.5004847645759583,
      "learning_rate": 1.964070143183548e-05,
      "loss": 0.3111,
      "step": 4740
    },
    {
      "epoch": 0.08567518182277786,
      "grad_norm": 0.3869590759277344,
      "learning_rate": 1.963994760240572e-05,
      "loss": 0.3283,
      "step": 4745
    },
    {
      "epoch": 0.08576546125567858,
      "grad_norm": 0.41948580741882324,
      "learning_rate": 1.9639192997512258e-05,
      "loss": 0.2875,
      "step": 4750
    },
    {
      "epoch": 0.08585574068857929,
      "grad_norm": 0.4879941940307617,
      "learning_rate": 1.9638437617215808e-05,
      "loss": 0.3663,
      "step": 4755
    },
    {
      "epoch": 0.08594602012148,
      "grad_norm": 0.4754737317562103,
      "learning_rate": 1.9637681461577128e-05,
      "loss": 0.3635,
      "step": 4760
    },
    {
      "epoch": 0.08603629955438072,
      "grad_norm": 0.37280508875846863,
      "learning_rate": 1.9636924530657045e-05,
      "loss": 0.3096,
      "step": 4765
    },
    {
      "epoch": 0.08612657898728143,
      "grad_norm": 0.4379524290561676,
      "learning_rate": 1.9636166824516457e-05,
      "loss": 0.351,
      "step": 4770
    },
    {
      "epoch": 0.08621685842018215,
      "grad_norm": 0.41434362530708313,
      "learning_rate": 1.9635408343216304e-05,
      "loss": 0.3746,
      "step": 4775
    },
    {
      "epoch": 0.08630713785308286,
      "grad_norm": 0.4297976493835449,
      "learning_rate": 1.9634649086817608e-05,
      "loss": 0.4458,
      "step": 4780
    },
    {
      "epoch": 0.08639741728598357,
      "grad_norm": 0.47540801763534546,
      "learning_rate": 1.9633889055381442e-05,
      "loss": 0.3872,
      "step": 4785
    },
    {
      "epoch": 0.08648769671888429,
      "grad_norm": 0.7144162058830261,
      "learning_rate": 1.9633128248968946e-05,
      "loss": 0.408,
      "step": 4790
    },
    {
      "epoch": 0.086577976151785,
      "grad_norm": 0.4906654357910156,
      "learning_rate": 1.9632366667641326e-05,
      "loss": 0.3667,
      "step": 4795
    },
    {
      "epoch": 0.08666825558468572,
      "grad_norm": 0.57340008020401,
      "learning_rate": 1.9631604311459837e-05,
      "loss": 0.3551,
      "step": 4800
    },
    {
      "epoch": 0.08675853501758643,
      "grad_norm": 0.4374023377895355,
      "learning_rate": 1.9630841180485815e-05,
      "loss": 0.2925,
      "step": 4805
    },
    {
      "epoch": 0.08684881445048714,
      "grad_norm": 0.4600036144256592,
      "learning_rate": 1.963007727478064e-05,
      "loss": 0.4128,
      "step": 4810
    },
    {
      "epoch": 0.08693909388338786,
      "grad_norm": 0.3051811158657074,
      "learning_rate": 1.9629312594405765e-05,
      "loss": 0.3569,
      "step": 4815
    },
    {
      "epoch": 0.08702937331628857,
      "grad_norm": 0.4193210303783417,
      "learning_rate": 1.9628547139422706e-05,
      "loss": 0.3374,
      "step": 4820
    },
    {
      "epoch": 0.08711965274918929,
      "grad_norm": 0.3596745729446411,
      "learning_rate": 1.962778090989303e-05,
      "loss": 0.3388,
      "step": 4825
    },
    {
      "epoch": 0.08720993218209,
      "grad_norm": 0.5314683318138123,
      "learning_rate": 1.962701390587839e-05,
      "loss": 0.3421,
      "step": 4830
    },
    {
      "epoch": 0.08730021161499071,
      "grad_norm": 0.37680259346961975,
      "learning_rate": 1.962624612744047e-05,
      "loss": 0.2718,
      "step": 4835
    },
    {
      "epoch": 0.08739049104789143,
      "grad_norm": 0.5087005496025085,
      "learning_rate": 1.962547757464104e-05,
      "loss": 0.3556,
      "step": 4840
    },
    {
      "epoch": 0.08748077048079214,
      "grad_norm": 0.38424932956695557,
      "learning_rate": 1.962470824754192e-05,
      "loss": 0.3122,
      "step": 4845
    },
    {
      "epoch": 0.08757104991369286,
      "grad_norm": 0.3894083797931671,
      "learning_rate": 1.9623938146205006e-05,
      "loss": 0.2732,
      "step": 4850
    },
    {
      "epoch": 0.08766132934659358,
      "grad_norm": 0.34054046869277954,
      "learning_rate": 1.9623167270692236e-05,
      "loss": 0.3351,
      "step": 4855
    },
    {
      "epoch": 0.0877516087794943,
      "grad_norm": 0.5789301991462708,
      "learning_rate": 1.9622395621065628e-05,
      "loss": 0.3519,
      "step": 4860
    },
    {
      "epoch": 0.08784188821239501,
      "grad_norm": 0.4679858386516571,
      "learning_rate": 1.9621623197387254e-05,
      "loss": 0.4429,
      "step": 4865
    },
    {
      "epoch": 0.08793216764529573,
      "grad_norm": 0.5168954730033875,
      "learning_rate": 1.962084999971925e-05,
      "loss": 0.3585,
      "step": 4870
    },
    {
      "epoch": 0.08802244707819644,
      "grad_norm": 0.5853481888771057,
      "learning_rate": 1.9620076028123814e-05,
      "loss": 0.3833,
      "step": 4875
    },
    {
      "epoch": 0.08811272651109715,
      "grad_norm": 0.49064958095550537,
      "learning_rate": 1.9619301282663205e-05,
      "loss": 0.3219,
      "step": 4880
    },
    {
      "epoch": 0.08820300594399787,
      "grad_norm": 0.8201891779899597,
      "learning_rate": 1.961852576339975e-05,
      "loss": 0.338,
      "step": 4885
    },
    {
      "epoch": 0.08829328537689858,
      "grad_norm": 0.30209094285964966,
      "learning_rate": 1.9617749470395825e-05,
      "loss": 0.3283,
      "step": 4890
    },
    {
      "epoch": 0.0883835648097993,
      "grad_norm": 0.5721350312232971,
      "learning_rate": 1.961697240371389e-05,
      "loss": 0.3774,
      "step": 4895
    },
    {
      "epoch": 0.08847384424270001,
      "grad_norm": 0.6147782802581787,
      "learning_rate": 1.961619456341644e-05,
      "loss": 0.2906,
      "step": 4900
    },
    {
      "epoch": 0.08856412367560072,
      "grad_norm": 0.5066529512405396,
      "learning_rate": 1.9615415949566058e-05,
      "loss": 0.4231,
      "step": 4905
    },
    {
      "epoch": 0.08865440310850144,
      "grad_norm": 0.33306726813316345,
      "learning_rate": 1.9614636562225375e-05,
      "loss": 0.252,
      "step": 4910
    },
    {
      "epoch": 0.08874468254140215,
      "grad_norm": 0.42946088314056396,
      "learning_rate": 1.9613856401457084e-05,
      "loss": 0.3567,
      "step": 4915
    },
    {
      "epoch": 0.08883496197430286,
      "grad_norm": 0.5795415043830872,
      "learning_rate": 1.9613075467323942e-05,
      "loss": 0.2967,
      "step": 4920
    },
    {
      "epoch": 0.08892524140720358,
      "grad_norm": 0.3875221610069275,
      "learning_rate": 1.9612293759888777e-05,
      "loss": 0.2638,
      "step": 4925
    },
    {
      "epoch": 0.08901552084010429,
      "grad_norm": 0.514550507068634,
      "learning_rate": 1.9611511279214467e-05,
      "loss": 0.2981,
      "step": 4930
    },
    {
      "epoch": 0.089105800273005,
      "grad_norm": 0.4188728332519531,
      "learning_rate": 1.9610728025363953e-05,
      "loss": 0.299,
      "step": 4935
    },
    {
      "epoch": 0.08919607970590572,
      "grad_norm": 0.4870274066925049,
      "learning_rate": 1.960994399840025e-05,
      "loss": 0.341,
      "step": 4940
    },
    {
      "epoch": 0.08928635913880643,
      "grad_norm": 0.5231569409370422,
      "learning_rate": 1.9609159198386422e-05,
      "loss": 0.3235,
      "step": 4945
    },
    {
      "epoch": 0.08937663857170715,
      "grad_norm": 0.2766699492931366,
      "learning_rate": 1.9608373625385604e-05,
      "loss": 0.3831,
      "step": 4950
    },
    {
      "epoch": 0.08946691800460786,
      "grad_norm": 0.38355782628059387,
      "learning_rate": 1.9607587279460984e-05,
      "loss": 0.306,
      "step": 4955
    },
    {
      "epoch": 0.08955719743750858,
      "grad_norm": 0.572722852230072,
      "learning_rate": 1.960680016067582e-05,
      "loss": 0.3822,
      "step": 4960
    },
    {
      "epoch": 0.08964747687040929,
      "grad_norm": 0.46345752477645874,
      "learning_rate": 1.9606012269093434e-05,
      "loss": 0.4394,
      "step": 4965
    },
    {
      "epoch": 0.08973775630331,
      "grad_norm": 0.4448091983795166,
      "learning_rate": 1.9605223604777207e-05,
      "loss": 0.3849,
      "step": 4970
    },
    {
      "epoch": 0.08982803573621072,
      "grad_norm": 0.5819882154464722,
      "learning_rate": 1.9604434167790574e-05,
      "loss": 0.3321,
      "step": 4975
    },
    {
      "epoch": 0.08991831516911143,
      "grad_norm": 0.2463798224925995,
      "learning_rate": 1.960364395819704e-05,
      "loss": 0.3517,
      "step": 4980
    },
    {
      "epoch": 0.09000859460201215,
      "grad_norm": 0.4587620198726654,
      "learning_rate": 1.960285297606018e-05,
      "loss": 0.3617,
      "step": 4985
    },
    {
      "epoch": 0.09009887403491286,
      "grad_norm": 0.5834743976593018,
      "learning_rate": 1.9602061221443615e-05,
      "loss": 0.3336,
      "step": 4990
    },
    {
      "epoch": 0.09018915346781357,
      "grad_norm": 0.42822515964508057,
      "learning_rate": 1.960126869441104e-05,
      "loss": 0.2742,
      "step": 4995
    },
    {
      "epoch": 0.09027943290071429,
      "grad_norm": 0.44845589995384216,
      "learning_rate": 1.9600475395026203e-05,
      "loss": 0.3435,
      "step": 5000
    },
    {
      "epoch": 0.090369712333615,
      "grad_norm": 0.33463209867477417,
      "learning_rate": 1.9599681323352926e-05,
      "loss": 0.3159,
      "step": 5005
    },
    {
      "epoch": 0.09045999176651572,
      "grad_norm": 0.49585360288619995,
      "learning_rate": 1.9598886479455083e-05,
      "loss": 0.308,
      "step": 5010
    },
    {
      "epoch": 0.09055027119941643,
      "grad_norm": 0.5451337695121765,
      "learning_rate": 1.959809086339661e-05,
      "loss": 0.2762,
      "step": 5015
    },
    {
      "epoch": 0.09064055063231714,
      "grad_norm": 0.5406076312065125,
      "learning_rate": 1.9597294475241516e-05,
      "loss": 0.2591,
      "step": 5020
    },
    {
      "epoch": 0.09073083006521786,
      "grad_norm": 0.44968345761299133,
      "learning_rate": 1.959649731505386e-05,
      "loss": 0.3669,
      "step": 5025
    },
    {
      "epoch": 0.09082110949811857,
      "grad_norm": 0.4464029371738434,
      "learning_rate": 1.9595699382897768e-05,
      "loss": 0.306,
      "step": 5030
    },
    {
      "epoch": 0.09091138893101929,
      "grad_norm": 0.4215846061706543,
      "learning_rate": 1.9594900678837426e-05,
      "loss": 0.3202,
      "step": 5035
    },
    {
      "epoch": 0.09100166836392,
      "grad_norm": 0.5329155325889587,
      "learning_rate": 1.959410120293709e-05,
      "loss": 0.3137,
      "step": 5040
    },
    {
      "epoch": 0.09109194779682071,
      "grad_norm": 0.519690990447998,
      "learning_rate": 1.9593300955261062e-05,
      "loss": 0.2903,
      "step": 5045
    },
    {
      "epoch": 0.09118222722972143,
      "grad_norm": 0.9205771088600159,
      "learning_rate": 1.9592499935873727e-05,
      "loss": 0.3304,
      "step": 5050
    },
    {
      "epoch": 0.09127250666262215,
      "grad_norm": 0.46820029616355896,
      "learning_rate": 1.9591698144839516e-05,
      "loss": 0.4189,
      "step": 5055
    },
    {
      "epoch": 0.09136278609552287,
      "grad_norm": 0.7149403095245361,
      "learning_rate": 1.9590895582222925e-05,
      "loss": 0.336,
      "step": 5060
    },
    {
      "epoch": 0.09145306552842358,
      "grad_norm": 0.23928998410701752,
      "learning_rate": 1.9590092248088522e-05,
      "loss": 0.2935,
      "step": 5065
    },
    {
      "epoch": 0.0915433449613243,
      "grad_norm": 0.293186753988266,
      "learning_rate": 1.958928814250092e-05,
      "loss": 0.3895,
      "step": 5070
    },
    {
      "epoch": 0.09163362439422501,
      "grad_norm": 0.7533143758773804,
      "learning_rate": 1.958848326552481e-05,
      "loss": 0.3255,
      "step": 5075
    },
    {
      "epoch": 0.09172390382712572,
      "grad_norm": 0.22842998802661896,
      "learning_rate": 1.9587677617224932e-05,
      "loss": 0.2246,
      "step": 5080
    },
    {
      "epoch": 0.09181418326002644,
      "grad_norm": 0.38793644309043884,
      "learning_rate": 1.9586871197666103e-05,
      "loss": 0.3714,
      "step": 5085
    },
    {
      "epoch": 0.09190446269292715,
      "grad_norm": 0.5378062725067139,
      "learning_rate": 1.9586064006913192e-05,
      "loss": 0.3374,
      "step": 5090
    },
    {
      "epoch": 0.09199474212582787,
      "grad_norm": 0.6650637984275818,
      "learning_rate": 1.958525604503112e-05,
      "loss": 0.3621,
      "step": 5095
    },
    {
      "epoch": 0.09208502155872858,
      "grad_norm": 0.6383867263793945,
      "learning_rate": 1.95844473120849e-05,
      "loss": 0.3501,
      "step": 5100
    },
    {
      "epoch": 0.0921753009916293,
      "grad_norm": 0.5531526207923889,
      "learning_rate": 1.9583637808139574e-05,
      "loss": 0.3532,
      "step": 5105
    },
    {
      "epoch": 0.09226558042453001,
      "grad_norm": 0.44341447949409485,
      "learning_rate": 1.9582827533260267e-05,
      "loss": 0.2861,
      "step": 5110
    },
    {
      "epoch": 0.09235585985743072,
      "grad_norm": 0.26350006461143494,
      "learning_rate": 1.958201648751216e-05,
      "loss": 0.1926,
      "step": 5115
    },
    {
      "epoch": 0.09244613929033144,
      "grad_norm": 0.41844233870506287,
      "learning_rate": 1.9581204670960496e-05,
      "loss": 0.2989,
      "step": 5120
    },
    {
      "epoch": 0.09253641872323215,
      "grad_norm": 0.5063326954841614,
      "learning_rate": 1.9580392083670573e-05,
      "loss": 0.3132,
      "step": 5125
    },
    {
      "epoch": 0.09262669815613286,
      "grad_norm": 0.6237820386886597,
      "learning_rate": 1.9579578725707768e-05,
      "loss": 0.3054,
      "step": 5130
    },
    {
      "epoch": 0.09271697758903358,
      "grad_norm": 0.4341614842414856,
      "learning_rate": 1.9578764597137504e-05,
      "loss": 0.3877,
      "step": 5135
    },
    {
      "epoch": 0.09280725702193429,
      "grad_norm": 0.4137227535247803,
      "learning_rate": 1.957794969802527e-05,
      "loss": 0.4141,
      "step": 5140
    },
    {
      "epoch": 0.092897536454835,
      "grad_norm": 0.4514862596988678,
      "learning_rate": 1.9577134028436626e-05,
      "loss": 0.3094,
      "step": 5145
    },
    {
      "epoch": 0.09298781588773572,
      "grad_norm": 0.4362461566925049,
      "learning_rate": 1.957631758843718e-05,
      "loss": 0.311,
      "step": 5150
    },
    {
      "epoch": 0.09307809532063643,
      "grad_norm": 0.422057181596756,
      "learning_rate": 1.9575500378092606e-05,
      "loss": 0.2631,
      "step": 5155
    },
    {
      "epoch": 0.09316837475353715,
      "grad_norm": 0.37970930337905884,
      "learning_rate": 1.957468239746865e-05,
      "loss": 0.2921,
      "step": 5160
    },
    {
      "epoch": 0.09325865418643786,
      "grad_norm": 0.4683941900730133,
      "learning_rate": 1.9573863646631113e-05,
      "loss": 0.3497,
      "step": 5165
    },
    {
      "epoch": 0.09334893361933858,
      "grad_norm": 0.38652756810188293,
      "learning_rate": 1.957304412564585e-05,
      "loss": 0.3171,
      "step": 5170
    },
    {
      "epoch": 0.09343921305223929,
      "grad_norm": 0.36577558517456055,
      "learning_rate": 1.9572223834578793e-05,
      "loss": 0.3247,
      "step": 5175
    },
    {
      "epoch": 0.09352949248514,
      "grad_norm": 0.35349130630493164,
      "learning_rate": 1.9571402773495924e-05,
      "loss": 0.4201,
      "step": 5180
    },
    {
      "epoch": 0.09361977191804072,
      "grad_norm": 0.4183078408241272,
      "learning_rate": 1.957058094246329e-05,
      "loss": 0.3452,
      "step": 5185
    },
    {
      "epoch": 0.09371005135094143,
      "grad_norm": 0.37030965089797974,
      "learning_rate": 1.956975834154701e-05,
      "loss": 0.397,
      "step": 5190
    },
    {
      "epoch": 0.09380033078384215,
      "grad_norm": 0.4157424569129944,
      "learning_rate": 1.9568934970813248e-05,
      "loss": 0.256,
      "step": 5195
    },
    {
      "epoch": 0.09389061021674286,
      "grad_norm": 0.720859169960022,
      "learning_rate": 1.956811083032824e-05,
      "loss": 0.2821,
      "step": 5200
    },
    {
      "epoch": 0.09398088964964357,
      "grad_norm": 0.5038657784461975,
      "learning_rate": 1.956728592015828e-05,
      "loss": 0.2821,
      "step": 5205
    },
    {
      "epoch": 0.09407116908254429,
      "grad_norm": 0.5624401569366455,
      "learning_rate": 1.9566460240369733e-05,
      "loss": 0.3153,
      "step": 5210
    },
    {
      "epoch": 0.094161448515445,
      "grad_norm": 0.5510204434394836,
      "learning_rate": 1.956563379102901e-05,
      "loss": 0.3937,
      "step": 5215
    },
    {
      "epoch": 0.09425172794834571,
      "grad_norm": 0.5782774090766907,
      "learning_rate": 1.95648065722026e-05,
      "loss": 0.2752,
      "step": 5220
    },
    {
      "epoch": 0.09434200738124643,
      "grad_norm": 0.5185222029685974,
      "learning_rate": 1.9563978583957044e-05,
      "loss": 0.3778,
      "step": 5225
    },
    {
      "epoch": 0.09443228681414714,
      "grad_norm": 0.3531385660171509,
      "learning_rate": 1.9563149826358946e-05,
      "loss": 0.3069,
      "step": 5230
    },
    {
      "epoch": 0.09452256624704786,
      "grad_norm": 0.4458499848842621,
      "learning_rate": 1.956232029947498e-05,
      "loss": 0.2716,
      "step": 5235
    },
    {
      "epoch": 0.09461284567994857,
      "grad_norm": 0.5337215065956116,
      "learning_rate": 1.9561490003371867e-05,
      "loss": 0.2823,
      "step": 5240
    },
    {
      "epoch": 0.09470312511284928,
      "grad_norm": 0.601732075214386,
      "learning_rate": 1.95606589381164e-05,
      "loss": 0.2908,
      "step": 5245
    },
    {
      "epoch": 0.09479340454575,
      "grad_norm": 0.38707998394966125,
      "learning_rate": 1.9559827103775437e-05,
      "loss": 0.2424,
      "step": 5250
    },
    {
      "epoch": 0.09488368397865073,
      "grad_norm": 0.3815915286540985,
      "learning_rate": 1.9558994500415892e-05,
      "loss": 0.2925,
      "step": 5255
    },
    {
      "epoch": 0.09497396341155144,
      "grad_norm": 0.4531812071800232,
      "learning_rate": 1.9558161128104736e-05,
      "loss": 0.3384,
      "step": 5260
    },
    {
      "epoch": 0.09506424284445215,
      "grad_norm": 0.5576108694076538,
      "learning_rate": 1.9557326986909015e-05,
      "loss": 0.3105,
      "step": 5265
    },
    {
      "epoch": 0.09515452227735287,
      "grad_norm": 0.3936028480529785,
      "learning_rate": 1.9556492076895822e-05,
      "loss": 0.3366,
      "step": 5270
    },
    {
      "epoch": 0.09524480171025358,
      "grad_norm": 0.4257822632789612,
      "learning_rate": 1.955565639813233e-05,
      "loss": 0.2439,
      "step": 5275
    },
    {
      "epoch": 0.0953350811431543,
      "grad_norm": 0.47362229228019714,
      "learning_rate": 1.955481995068575e-05,
      "loss": 0.342,
      "step": 5280
    },
    {
      "epoch": 0.09542536057605501,
      "grad_norm": 0.4994874894618988,
      "learning_rate": 1.955398273462338e-05,
      "loss": 0.4446,
      "step": 5285
    },
    {
      "epoch": 0.09551564000895572,
      "grad_norm": 0.42771607637405396,
      "learning_rate": 1.9553144750012565e-05,
      "loss": 0.4141,
      "step": 5290
    },
    {
      "epoch": 0.09560591944185644,
      "grad_norm": 0.2980763018131256,
      "learning_rate": 1.9552305996920712e-05,
      "loss": 0.4213,
      "step": 5295
    },
    {
      "epoch": 0.09569619887475715,
      "grad_norm": 0.5155059695243835,
      "learning_rate": 1.955146647541529e-05,
      "loss": 0.2334,
      "step": 5300
    },
    {
      "epoch": 0.09578647830765787,
      "grad_norm": 0.3601856529712677,
      "learning_rate": 1.9550626185563838e-05,
      "loss": 0.3939,
      "step": 5305
    },
    {
      "epoch": 0.09587675774055858,
      "grad_norm": 0.29154184460639954,
      "learning_rate": 1.9549785127433948e-05,
      "loss": 0.3869,
      "step": 5310
    },
    {
      "epoch": 0.0959670371734593,
      "grad_norm": 0.49999797344207764,
      "learning_rate": 1.954894330109328e-05,
      "loss": 0.3985,
      "step": 5315
    },
    {
      "epoch": 0.09605731660636001,
      "grad_norm": 0.6494349241256714,
      "learning_rate": 1.954810070660955e-05,
      "loss": 0.3517,
      "step": 5320
    },
    {
      "epoch": 0.09614759603926072,
      "grad_norm": 0.33707407116889954,
      "learning_rate": 1.954725734405054e-05,
      "loss": 0.3102,
      "step": 5325
    },
    {
      "epoch": 0.09623787547216144,
      "grad_norm": 0.4518429636955261,
      "learning_rate": 1.954641321348409e-05,
      "loss": 0.3393,
      "step": 5330
    },
    {
      "epoch": 0.09632815490506215,
      "grad_norm": 0.511194109916687,
      "learning_rate": 1.9545568314978106e-05,
      "loss": 0.2309,
      "step": 5335
    },
    {
      "epoch": 0.09641843433796286,
      "grad_norm": 0.3844567537307739,
      "learning_rate": 1.9544722648600558e-05,
      "loss": 0.377,
      "step": 5340
    },
    {
      "epoch": 0.09650871377086358,
      "grad_norm": 0.42884543538093567,
      "learning_rate": 1.9543876214419467e-05,
      "loss": 0.3377,
      "step": 5345
    },
    {
      "epoch": 0.09659899320376429,
      "grad_norm": 0.6516596674919128,
      "learning_rate": 1.9543029012502926e-05,
      "loss": 0.2655,
      "step": 5350
    },
    {
      "epoch": 0.096689272636665,
      "grad_norm": 0.3786361515522003,
      "learning_rate": 1.9542181042919084e-05,
      "loss": 0.2778,
      "step": 5355
    },
    {
      "epoch": 0.09677955206956572,
      "grad_norm": 0.3247559368610382,
      "learning_rate": 1.9541332305736156e-05,
      "loss": 0.3328,
      "step": 5360
    },
    {
      "epoch": 0.09686983150246643,
      "grad_norm": 0.5098878741264343,
      "learning_rate": 1.9540482801022417e-05,
      "loss": 0.213,
      "step": 5365
    },
    {
      "epoch": 0.09696011093536715,
      "grad_norm": 0.4012088179588318,
      "learning_rate": 1.9539632528846206e-05,
      "loss": 0.4207,
      "step": 5370
    },
    {
      "epoch": 0.09705039036826786,
      "grad_norm": 0.36352911591529846,
      "learning_rate": 1.9538781489275912e-05,
      "loss": 0.3404,
      "step": 5375
    },
    {
      "epoch": 0.09714066980116857,
      "grad_norm": 0.2575864791870117,
      "learning_rate": 1.9537929682380002e-05,
      "loss": 0.3838,
      "step": 5380
    },
    {
      "epoch": 0.09723094923406929,
      "grad_norm": 0.3757660388946533,
      "learning_rate": 1.9537077108227e-05,
      "loss": 0.3248,
      "step": 5385
    },
    {
      "epoch": 0.09732122866697,
      "grad_norm": 0.552474319934845,
      "learning_rate": 1.9536223766885484e-05,
      "loss": 0.3292,
      "step": 5390
    },
    {
      "epoch": 0.09741150809987072,
      "grad_norm": 0.596549391746521,
      "learning_rate": 1.95353696584241e-05,
      "loss": 0.2818,
      "step": 5395
    },
    {
      "epoch": 0.09750178753277143,
      "grad_norm": 0.3117164671421051,
      "learning_rate": 1.9534514782911556e-05,
      "loss": 0.2427,
      "step": 5400
    },
    {
      "epoch": 0.09759206696567214,
      "grad_norm": 0.3582278788089752,
      "learning_rate": 1.953365914041662e-05,
      "loss": 0.2953,
      "step": 5405
    },
    {
      "epoch": 0.09768234639857286,
      "grad_norm": 0.5204953551292419,
      "learning_rate": 1.953280273100812e-05,
      "loss": 0.304,
      "step": 5410
    },
    {
      "epoch": 0.09777262583147357,
      "grad_norm": 0.5520836710929871,
      "learning_rate": 1.9531945554754957e-05,
      "loss": 0.332,
      "step": 5415
    },
    {
      "epoch": 0.09786290526437429,
      "grad_norm": 0.426142156124115,
      "learning_rate": 1.9531087611726074e-05,
      "loss": 0.3299,
      "step": 5420
    },
    {
      "epoch": 0.097953184697275,
      "grad_norm": 0.43062669038772583,
      "learning_rate": 1.9530228901990493e-05,
      "loss": 0.3585,
      "step": 5425
    },
    {
      "epoch": 0.09804346413017571,
      "grad_norm": 0.570884644985199,
      "learning_rate": 1.9529369425617286e-05,
      "loss": 0.3148,
      "step": 5430
    },
    {
      "epoch": 0.09813374356307643,
      "grad_norm": 0.4310964345932007,
      "learning_rate": 1.9528509182675593e-05,
      "loss": 0.3101,
      "step": 5435
    },
    {
      "epoch": 0.09822402299597714,
      "grad_norm": 0.631697416305542,
      "learning_rate": 1.9527648173234616e-05,
      "loss": 0.2834,
      "step": 5440
    },
    {
      "epoch": 0.09831430242887786,
      "grad_norm": 0.481629341840744,
      "learning_rate": 1.952678639736362e-05,
      "loss": 0.3348,
      "step": 5445
    },
    {
      "epoch": 0.09840458186177858,
      "grad_norm": 0.3552111089229584,
      "learning_rate": 1.952592385513192e-05,
      "loss": 0.2887,
      "step": 5450
    },
    {
      "epoch": 0.0984948612946793,
      "grad_norm": 0.3959607183933258,
      "learning_rate": 1.952506054660891e-05,
      "loss": 0.3694,
      "step": 5455
    },
    {
      "epoch": 0.09858514072758001,
      "grad_norm": 0.6186146140098572,
      "learning_rate": 1.9524196471864028e-05,
      "loss": 0.3707,
      "step": 5460
    },
    {
      "epoch": 0.09867542016048073,
      "grad_norm": 0.27796289324760437,
      "learning_rate": 1.9523331630966787e-05,
      "loss": 0.3947,
      "step": 5465
    },
    {
      "epoch": 0.09876569959338144,
      "grad_norm": 0.4462321102619171,
      "learning_rate": 1.9522466023986762e-05,
      "loss": 0.276,
      "step": 5470
    },
    {
      "epoch": 0.09885597902628215,
      "grad_norm": 0.40568867325782776,
      "learning_rate": 1.952159965099358e-05,
      "loss": 0.323,
      "step": 5475
    },
    {
      "epoch": 0.09894625845918287,
      "grad_norm": 0.658328652381897,
      "learning_rate": 1.952073251205693e-05,
      "loss": 0.3548,
      "step": 5480
    },
    {
      "epoch": 0.09903653789208358,
      "grad_norm": 0.2920197546482086,
      "learning_rate": 1.951986460724658e-05,
      "loss": 0.2942,
      "step": 5485
    },
    {
      "epoch": 0.0991268173249843,
      "grad_norm": 0.5438705086708069,
      "learning_rate": 1.951899593663233e-05,
      "loss": 0.3396,
      "step": 5490
    },
    {
      "epoch": 0.09921709675788501,
      "grad_norm": 0.5157396793365479,
      "learning_rate": 1.9518126500284072e-05,
      "loss": 0.4084,
      "step": 5495
    },
    {
      "epoch": 0.09930737619078572,
      "grad_norm": 0.4184419810771942,
      "learning_rate": 1.951725629827174e-05,
      "loss": 0.3242,
      "step": 5500
    },
    {
      "epoch": 0.09939765562368644,
      "grad_norm": 0.329950749874115,
      "learning_rate": 1.9516385330665332e-05,
      "loss": 0.411,
      "step": 5505
    },
    {
      "epoch": 0.09948793505658715,
      "grad_norm": 0.5271588563919067,
      "learning_rate": 1.951551359753492e-05,
      "loss": 0.3677,
      "step": 5510
    },
    {
      "epoch": 0.09957821448948787,
      "grad_norm": 0.40672919154167175,
      "learning_rate": 1.9514641098950623e-05,
      "loss": 0.2817,
      "step": 5515
    },
    {
      "epoch": 0.09966849392238858,
      "grad_norm": 0.4978141784667969,
      "learning_rate": 1.9513767834982625e-05,
      "loss": 0.309,
      "step": 5520
    },
    {
      "epoch": 0.09975877335528929,
      "grad_norm": 0.42941680550575256,
      "learning_rate": 1.951289380570118e-05,
      "loss": 0.2661,
      "step": 5525
    },
    {
      "epoch": 0.09984905278819001,
      "grad_norm": 0.3147401809692383,
      "learning_rate": 1.951201901117659e-05,
      "loss": 0.3315,
      "step": 5530
    },
    {
      "epoch": 0.09993933222109072,
      "grad_norm": 0.43563732504844666,
      "learning_rate": 1.9511143451479228e-05,
      "loss": 0.3314,
      "step": 5535
    },
    {
      "epoch": 0.10002961165399143,
      "grad_norm": 0.556304395198822,
      "learning_rate": 1.9510267126679535e-05,
      "loss": 0.3468,
      "step": 5540
    },
    {
      "epoch": 0.10011989108689215,
      "grad_norm": 0.3554510474205017,
      "learning_rate": 1.9509390036847992e-05,
      "loss": 0.3373,
      "step": 5545
    },
    {
      "epoch": 0.10021017051979286,
      "grad_norm": 0.4136534333229065,
      "learning_rate": 1.950851218205516e-05,
      "loss": 0.2907,
      "step": 5550
    },
    {
      "epoch": 0.10030044995269358,
      "grad_norm": 0.41162267327308655,
      "learning_rate": 1.9507633562371662e-05,
      "loss": 0.3779,
      "step": 5555
    },
    {
      "epoch": 0.10039072938559429,
      "grad_norm": 0.6030270457267761,
      "learning_rate": 1.950675417786817e-05,
      "loss": 0.3726,
      "step": 5560
    },
    {
      "epoch": 0.100481008818495,
      "grad_norm": 0.45536723732948303,
      "learning_rate": 1.950587402861542e-05,
      "loss": 0.2769,
      "step": 5565
    },
    {
      "epoch": 0.10057128825139572,
      "grad_norm": 0.3023357689380646,
      "learning_rate": 1.9504993114684224e-05,
      "loss": 0.2517,
      "step": 5570
    },
    {
      "epoch": 0.10066156768429643,
      "grad_norm": 0.3225719928741455,
      "learning_rate": 1.950411143614544e-05,
      "loss": 0.2659,
      "step": 5575
    },
    {
      "epoch": 0.10075184711719715,
      "grad_norm": 0.787666916847229,
      "learning_rate": 1.9503228993069986e-05,
      "loss": 0.2757,
      "step": 5580
    },
    {
      "epoch": 0.10084212655009786,
      "grad_norm": 0.47917357087135315,
      "learning_rate": 1.9502345785528863e-05,
      "loss": 0.3791,
      "step": 5585
    },
    {
      "epoch": 0.10093240598299857,
      "grad_norm": 0.3646545708179474,
      "learning_rate": 1.9501461813593106e-05,
      "loss": 0.3377,
      "step": 5590
    },
    {
      "epoch": 0.10102268541589929,
      "grad_norm": 0.42772263288497925,
      "learning_rate": 1.9500577077333828e-05,
      "loss": 0.3481,
      "step": 5595
    },
    {
      "epoch": 0.1011129648488,
      "grad_norm": 0.3945727050304413,
      "learning_rate": 1.94996915768222e-05,
      "loss": 0.2901,
      "step": 5600
    },
    {
      "epoch": 0.10120324428170072,
      "grad_norm": 0.3048461377620697,
      "learning_rate": 1.9498805312129456e-05,
      "loss": 0.2429,
      "step": 5605
    },
    {
      "epoch": 0.10129352371460143,
      "grad_norm": 0.43724963068962097,
      "learning_rate": 1.9497918283326887e-05,
      "loss": 0.2989,
      "step": 5610
    },
    {
      "epoch": 0.10138380314750214,
      "grad_norm": 0.38626641035079956,
      "learning_rate": 1.9497030490485848e-05,
      "loss": 0.3071,
      "step": 5615
    },
    {
      "epoch": 0.10147408258040286,
      "grad_norm": 0.4728689193725586,
      "learning_rate": 1.9496141933677757e-05,
      "loss": 0.3376,
      "step": 5620
    },
    {
      "epoch": 0.10156436201330357,
      "grad_norm": 0.7466672658920288,
      "learning_rate": 1.949525261297409e-05,
      "loss": 0.2819,
      "step": 5625
    },
    {
      "epoch": 0.10165464144620429,
      "grad_norm": 0.6607660055160522,
      "learning_rate": 1.9494362528446387e-05,
      "loss": 0.3414,
      "step": 5630
    },
    {
      "epoch": 0.101744920879105,
      "grad_norm": 0.4114571213722229,
      "learning_rate": 1.949347168016625e-05,
      "loss": 0.4251,
      "step": 5635
    },
    {
      "epoch": 0.10183520031200571,
      "grad_norm": 0.4187646508216858,
      "learning_rate": 1.949258006820534e-05,
      "loss": 0.3238,
      "step": 5640
    },
    {
      "epoch": 0.10192547974490643,
      "grad_norm": 0.3874375820159912,
      "learning_rate": 1.9491687692635385e-05,
      "loss": 0.3899,
      "step": 5645
    },
    {
      "epoch": 0.10201575917780716,
      "grad_norm": 0.24559728801250458,
      "learning_rate": 1.9490794553528164e-05,
      "loss": 0.3407,
      "step": 5650
    },
    {
      "epoch": 0.10210603861070787,
      "grad_norm": 0.4480688273906708,
      "learning_rate": 1.9489900650955524e-05,
      "loss": 0.3278,
      "step": 5655
    },
    {
      "epoch": 0.10219631804360858,
      "grad_norm": 0.6497907042503357,
      "learning_rate": 1.9489005984989376e-05,
      "loss": 0.36,
      "step": 5660
    },
    {
      "epoch": 0.1022865974765093,
      "grad_norm": 0.37227800488471985,
      "learning_rate": 1.9488110555701686e-05,
      "loss": 0.3999,
      "step": 5665
    },
    {
      "epoch": 0.10237687690941001,
      "grad_norm": 0.35861340165138245,
      "learning_rate": 1.9487214363164492e-05,
      "loss": 0.2449,
      "step": 5670
    },
    {
      "epoch": 0.10246715634231073,
      "grad_norm": 0.29419082403182983,
      "learning_rate": 1.9486317407449878e-05,
      "loss": 0.3671,
      "step": 5675
    },
    {
      "epoch": 0.10255743577521144,
      "grad_norm": 0.4491279721260071,
      "learning_rate": 1.948541968863e-05,
      "loss": 0.3232,
      "step": 5680
    },
    {
      "epoch": 0.10264771520811215,
      "grad_norm": 0.4294513463973999,
      "learning_rate": 1.9484521206777077e-05,
      "loss": 0.3929,
      "step": 5685
    },
    {
      "epoch": 0.10273799464101287,
      "grad_norm": 0.43613892793655396,
      "learning_rate": 1.948362196196338e-05,
      "loss": 0.3833,
      "step": 5690
    },
    {
      "epoch": 0.10282827407391358,
      "grad_norm": 0.3967756927013397,
      "learning_rate": 1.9482721954261247e-05,
      "loss": 0.344,
      "step": 5695
    },
    {
      "epoch": 0.1029185535068143,
      "grad_norm": 0.29174667596817017,
      "learning_rate": 1.948182118374308e-05,
      "loss": 0.3903,
      "step": 5700
    },
    {
      "epoch": 0.10300883293971501,
      "grad_norm": 0.42230311036109924,
      "learning_rate": 1.948091965048134e-05,
      "loss": 0.2445,
      "step": 5705
    },
    {
      "epoch": 0.10309911237261572,
      "grad_norm": 0.502432107925415,
      "learning_rate": 1.9480017354548547e-05,
      "loss": 0.2777,
      "step": 5710
    },
    {
      "epoch": 0.10318939180551644,
      "grad_norm": 0.40217089653015137,
      "learning_rate": 1.9479114296017283e-05,
      "loss": 0.3711,
      "step": 5715
    },
    {
      "epoch": 0.10327967123841715,
      "grad_norm": 0.6035251021385193,
      "learning_rate": 1.9478210474960193e-05,
      "loss": 0.3647,
      "step": 5720
    },
    {
      "epoch": 0.10336995067131786,
      "grad_norm": 0.3311302065849304,
      "learning_rate": 1.9477305891449987e-05,
      "loss": 0.2223,
      "step": 5725
    },
    {
      "epoch": 0.10346023010421858,
      "grad_norm": 0.3849733769893646,
      "learning_rate": 1.9476400545559426e-05,
      "loss": 0.3043,
      "step": 5730
    },
    {
      "epoch": 0.10355050953711929,
      "grad_norm": 2.3564114570617676,
      "learning_rate": 1.947549443736134e-05,
      "loss": 0.3054,
      "step": 5735
    },
    {
      "epoch": 0.10364078897002,
      "grad_norm": 0.3749711215496063,
      "learning_rate": 1.947458756692862e-05,
      "loss": 0.2085,
      "step": 5740
    },
    {
      "epoch": 0.10373106840292072,
      "grad_norm": 0.40689751505851746,
      "learning_rate": 1.947367993433422e-05,
      "loss": 0.4027,
      "step": 5745
    },
    {
      "epoch": 0.10382134783582143,
      "grad_norm": 0.434182733297348,
      "learning_rate": 1.9472771539651147e-05,
      "loss": 0.337,
      "step": 5750
    },
    {
      "epoch": 0.10391162726872215,
      "grad_norm": 0.4543397128582001,
      "learning_rate": 1.947186238295248e-05,
      "loss": 0.273,
      "step": 5755
    },
    {
      "epoch": 0.10400190670162286,
      "grad_norm": 0.5350884795188904,
      "learning_rate": 1.9470952464311348e-05,
      "loss": 0.3684,
      "step": 5760
    },
    {
      "epoch": 0.10409218613452358,
      "grad_norm": 0.5613234639167786,
      "learning_rate": 1.9470041783800956e-05,
      "loss": 0.3381,
      "step": 5765
    },
    {
      "epoch": 0.10418246556742429,
      "grad_norm": 0.5730423331260681,
      "learning_rate": 1.946913034149455e-05,
      "loss": 0.2829,
      "step": 5770
    },
    {
      "epoch": 0.104272745000325,
      "grad_norm": 0.524937093257904,
      "learning_rate": 1.946821813746546e-05,
      "loss": 0.3181,
      "step": 5775
    },
    {
      "epoch": 0.10436302443322572,
      "grad_norm": 0.3746713697910309,
      "learning_rate": 1.946730517178706e-05,
      "loss": 0.3931,
      "step": 5780
    },
    {
      "epoch": 0.10445330386612643,
      "grad_norm": 0.6415619850158691,
      "learning_rate": 1.946639144453279e-05,
      "loss": 0.3812,
      "step": 5785
    },
    {
      "epoch": 0.10454358329902715,
      "grad_norm": 0.4309931993484497,
      "learning_rate": 1.9465476955776158e-05,
      "loss": 0.2501,
      "step": 5790
    },
    {
      "epoch": 0.10463386273192786,
      "grad_norm": 0.5179385542869568,
      "learning_rate": 1.9464561705590727e-05,
      "loss": 0.3465,
      "step": 5795
    },
    {
      "epoch": 0.10472414216482857,
      "grad_norm": 0.6764097213745117,
      "learning_rate": 1.946364569405012e-05,
      "loss": 0.4371,
      "step": 5800
    },
    {
      "epoch": 0.10481442159772929,
      "grad_norm": 0.4080183207988739,
      "learning_rate": 1.9462728921228025e-05,
      "loss": 0.3404,
      "step": 5805
    },
    {
      "epoch": 0.10490470103063,
      "grad_norm": 0.45554742217063904,
      "learning_rate": 1.9461811387198185e-05,
      "loss": 0.4058,
      "step": 5810
    },
    {
      "epoch": 0.10499498046353072,
      "grad_norm": 0.4290477931499481,
      "learning_rate": 1.946089309203442e-05,
      "loss": 0.3682,
      "step": 5815
    },
    {
      "epoch": 0.10508525989643143,
      "grad_norm": 0.42258429527282715,
      "learning_rate": 1.9459974035810587e-05,
      "loss": 0.3636,
      "step": 5820
    },
    {
      "epoch": 0.10517553932933214,
      "grad_norm": 0.30179986357688904,
      "learning_rate": 1.9459054218600622e-05,
      "loss": 0.404,
      "step": 5825
    },
    {
      "epoch": 0.10526581876223286,
      "grad_norm": 0.3837675154209137,
      "learning_rate": 1.945813364047852e-05,
      "loss": 0.3033,
      "step": 5830
    },
    {
      "epoch": 0.10535609819513357,
      "grad_norm": 0.4009370505809784,
      "learning_rate": 1.945721230151834e-05,
      "loss": 0.2867,
      "step": 5835
    },
    {
      "epoch": 0.10544637762803429,
      "grad_norm": 0.5982357263565063,
      "learning_rate": 1.9456290201794186e-05,
      "loss": 0.3247,
      "step": 5840
    },
    {
      "epoch": 0.105536657060935,
      "grad_norm": 0.44127723574638367,
      "learning_rate": 1.945536734138024e-05,
      "loss": 0.3373,
      "step": 5845
    },
    {
      "epoch": 0.10562693649383573,
      "grad_norm": 0.42917925119400024,
      "learning_rate": 1.9454443720350743e-05,
      "loss": 0.3158,
      "step": 5850
    },
    {
      "epoch": 0.10571721592673644,
      "grad_norm": 0.39992135763168335,
      "learning_rate": 1.9453519338779983e-05,
      "loss": 0.3198,
      "step": 5855
    },
    {
      "epoch": 0.10580749535963715,
      "grad_norm": 0.31888100504875183,
      "learning_rate": 1.945259419674233e-05,
      "loss": 0.4444,
      "step": 5860
    },
    {
      "epoch": 0.10589777479253787,
      "grad_norm": 0.31540992856025696,
      "learning_rate": 1.9451668294312197e-05,
      "loss": 0.3988,
      "step": 5865
    },
    {
      "epoch": 0.10598805422543858,
      "grad_norm": 0.7145553827285767,
      "learning_rate": 1.945074163156407e-05,
      "loss": 0.3445,
      "step": 5870
    },
    {
      "epoch": 0.1060783336583393,
      "grad_norm": 0.38850831985473633,
      "learning_rate": 1.9449814208572495e-05,
      "loss": 0.3442,
      "step": 5875
    },
    {
      "epoch": 0.10616861309124001,
      "grad_norm": 0.511945903301239,
      "learning_rate": 1.9448886025412074e-05,
      "loss": 0.2274,
      "step": 5880
    },
    {
      "epoch": 0.10625889252414072,
      "grad_norm": 0.4505237638950348,
      "learning_rate": 1.9447957082157473e-05,
      "loss": 0.3868,
      "step": 5885
    },
    {
      "epoch": 0.10634917195704144,
      "grad_norm": 0.36377763748168945,
      "learning_rate": 1.9447027378883413e-05,
      "loss": 0.2994,
      "step": 5890
    },
    {
      "epoch": 0.10643945138994215,
      "grad_norm": 0.4371629059314728,
      "learning_rate": 1.944609691566469e-05,
      "loss": 0.3161,
      "step": 5895
    },
    {
      "epoch": 0.10652973082284287,
      "grad_norm": 0.32444894313812256,
      "learning_rate": 1.944516569257615e-05,
      "loss": 0.2538,
      "step": 5900
    },
    {
      "epoch": 0.10662001025574358,
      "grad_norm": 0.4282093048095703,
      "learning_rate": 1.9444233709692704e-05,
      "loss": 0.4371,
      "step": 5905
    },
    {
      "epoch": 0.1067102896886443,
      "grad_norm": 0.4953030049800873,
      "learning_rate": 1.944330096708932e-05,
      "loss": 0.3598,
      "step": 5910
    },
    {
      "epoch": 0.10680056912154501,
      "grad_norm": 0.5108286142349243,
      "learning_rate": 1.9442367464841034e-05,
      "loss": 0.3352,
      "step": 5915
    },
    {
      "epoch": 0.10689084855444572,
      "grad_norm": 0.44767871499061584,
      "learning_rate": 1.9441433203022935e-05,
      "loss": 0.2807,
      "step": 5920
    },
    {
      "epoch": 0.10698112798734644,
      "grad_norm": 0.4025225341320038,
      "learning_rate": 1.944049818171019e-05,
      "loss": 0.3045,
      "step": 5925
    },
    {
      "epoch": 0.10707140742024715,
      "grad_norm": 0.5402584671974182,
      "learning_rate": 1.9439562400977996e-05,
      "loss": 0.3899,
      "step": 5930
    },
    {
      "epoch": 0.10716168685314786,
      "grad_norm": 0.37331920862197876,
      "learning_rate": 1.9438625860901644e-05,
      "loss": 0.3665,
      "step": 5935
    },
    {
      "epoch": 0.10725196628604858,
      "grad_norm": 0.592233419418335,
      "learning_rate": 1.9437688561556463e-05,
      "loss": 0.284,
      "step": 5940
    },
    {
      "epoch": 0.10734224571894929,
      "grad_norm": 0.5333718657493591,
      "learning_rate": 1.9436750503017858e-05,
      "loss": 0.2664,
      "step": 5945
    },
    {
      "epoch": 0.10743252515185,
      "grad_norm": 0.5840323567390442,
      "learning_rate": 1.9435811685361284e-05,
      "loss": 0.348,
      "step": 5950
    },
    {
      "epoch": 0.10752280458475072,
      "grad_norm": 0.43222957849502563,
      "learning_rate": 1.943487210866227e-05,
      "loss": 0.3971,
      "step": 5955
    },
    {
      "epoch": 0.10761308401765143,
      "grad_norm": 0.2581470012664795,
      "learning_rate": 1.943393177299639e-05,
      "loss": 0.3098,
      "step": 5960
    },
    {
      "epoch": 0.10770336345055215,
      "grad_norm": 0.3755950927734375,
      "learning_rate": 1.943299067843929e-05,
      "loss": 0.3571,
      "step": 5965
    },
    {
      "epoch": 0.10779364288345286,
      "grad_norm": 0.468671590089798,
      "learning_rate": 1.9432048825066677e-05,
      "loss": 0.3714,
      "step": 5970
    },
    {
      "epoch": 0.10788392231635358,
      "grad_norm": 0.7344030141830444,
      "learning_rate": 1.9431106212954312e-05,
      "loss": 0.3145,
      "step": 5975
    },
    {
      "epoch": 0.10797420174925429,
      "grad_norm": 0.36533960700035095,
      "learning_rate": 1.943016284217802e-05,
      "loss": 0.2636,
      "step": 5980
    },
    {
      "epoch": 0.108064481182155,
      "grad_norm": 0.545117199420929,
      "learning_rate": 1.9429218712813695e-05,
      "loss": 0.4378,
      "step": 5985
    },
    {
      "epoch": 0.10815476061505572,
      "grad_norm": 0.3776969611644745,
      "learning_rate": 1.9428273824937284e-05,
      "loss": 0.3666,
      "step": 5990
    },
    {
      "epoch": 0.10824504004795643,
      "grad_norm": 0.36028388142585754,
      "learning_rate": 1.9427328178624787e-05,
      "loss": 0.4163,
      "step": 5995
    },
    {
      "epoch": 0.10833531948085715,
      "grad_norm": 0.44015151262283325,
      "learning_rate": 1.9426381773952285e-05,
      "loss": 0.3104,
      "step": 6000
    },
    {
      "epoch": 0.10842559891375786,
      "grad_norm": 0.4195164442062378,
      "learning_rate": 1.9425434610995905e-05,
      "loss": 0.317,
      "step": 6005
    },
    {
      "epoch": 0.10851587834665857,
      "grad_norm": 0.42691734433174133,
      "learning_rate": 1.9424486689831844e-05,
      "loss": 0.2489,
      "step": 6010
    },
    {
      "epoch": 0.10860615777955929,
      "grad_norm": 0.6212683916091919,
      "learning_rate": 1.9423538010536343e-05,
      "loss": 0.3868,
      "step": 6015
    },
    {
      "epoch": 0.10869643721246,
      "grad_norm": 0.5525115132331848,
      "learning_rate": 1.942258857318573e-05,
      "loss": 0.2956,
      "step": 6020
    },
    {
      "epoch": 0.10878671664536071,
      "grad_norm": 0.28082871437072754,
      "learning_rate": 1.9421638377856378e-05,
      "loss": 0.2908,
      "step": 6025
    },
    {
      "epoch": 0.10887699607826143,
      "grad_norm": 0.46432995796203613,
      "learning_rate": 1.9420687424624714e-05,
      "loss": 0.3011,
      "step": 6030
    },
    {
      "epoch": 0.10896727551116214,
      "grad_norm": 0.3892320990562439,
      "learning_rate": 1.941973571356725e-05,
      "loss": 0.289,
      "step": 6035
    },
    {
      "epoch": 0.10905755494406286,
      "grad_norm": 0.4740193784236908,
      "learning_rate": 1.941878324476053e-05,
      "loss": 0.3497,
      "step": 6040
    },
    {
      "epoch": 0.10914783437696358,
      "grad_norm": 0.4852452576160431,
      "learning_rate": 1.9417830018281178e-05,
      "loss": 0.3478,
      "step": 6045
    },
    {
      "epoch": 0.1092381138098643,
      "grad_norm": 0.44724610447883606,
      "learning_rate": 1.941687603420588e-05,
      "loss": 0.3867,
      "step": 6050
    },
    {
      "epoch": 0.10932839324276501,
      "grad_norm": 0.45649048686027527,
      "learning_rate": 1.941592129261137e-05,
      "loss": 0.3203,
      "step": 6055
    },
    {
      "epoch": 0.10941867267566573,
      "grad_norm": 0.5838635563850403,
      "learning_rate": 1.9414965793574452e-05,
      "loss": 0.3297,
      "step": 6060
    },
    {
      "epoch": 0.10950895210856644,
      "grad_norm": 0.3139464855194092,
      "learning_rate": 1.941400953717199e-05,
      "loss": 0.3547,
      "step": 6065
    },
    {
      "epoch": 0.10959923154146715,
      "grad_norm": 0.3852726221084595,
      "learning_rate": 1.9413052523480904e-05,
      "loss": 0.417,
      "step": 6070
    },
    {
      "epoch": 0.10968951097436787,
      "grad_norm": 0.35865509510040283,
      "learning_rate": 1.9412094752578186e-05,
      "loss": 0.3131,
      "step": 6075
    },
    {
      "epoch": 0.10977979040726858,
      "grad_norm": 0.6715008020401001,
      "learning_rate": 1.9411136224540875e-05,
      "loss": 0.3649,
      "step": 6080
    },
    {
      "epoch": 0.1098700698401693,
      "grad_norm": 0.5033209919929504,
      "learning_rate": 1.9410176939446082e-05,
      "loss": 0.375,
      "step": 6085
    },
    {
      "epoch": 0.10996034927307001,
      "grad_norm": 0.4022418260574341,
      "learning_rate": 1.940921689737097e-05,
      "loss": 0.2064,
      "step": 6090
    },
    {
      "epoch": 0.11005062870597072,
      "grad_norm": 0.39626574516296387,
      "learning_rate": 1.940825609839278e-05,
      "loss": 0.3444,
      "step": 6095
    },
    {
      "epoch": 0.11014090813887144,
      "grad_norm": 0.33554548025131226,
      "learning_rate": 1.940729454258878e-05,
      "loss": 0.3455,
      "step": 6100
    },
    {
      "epoch": 0.11023118757177215,
      "grad_norm": 0.5526272654533386,
      "learning_rate": 1.9406332230036335e-05,
      "loss": 0.3254,
      "step": 6105
    },
    {
      "epoch": 0.11032146700467287,
      "grad_norm": 0.3943984806537628,
      "learning_rate": 1.9405369160812855e-05,
      "loss": 0.312,
      "step": 6110
    },
    {
      "epoch": 0.11041174643757358,
      "grad_norm": 0.4909297823905945,
      "learning_rate": 1.940440533499581e-05,
      "loss": 0.272,
      "step": 6115
    },
    {
      "epoch": 0.1105020258704743,
      "grad_norm": 0.33837640285491943,
      "learning_rate": 1.940344075266273e-05,
      "loss": 0.3183,
      "step": 6120
    },
    {
      "epoch": 0.11059230530337501,
      "grad_norm": 0.3912183940410614,
      "learning_rate": 1.9402475413891212e-05,
      "loss": 0.2911,
      "step": 6125
    },
    {
      "epoch": 0.11068258473627572,
      "grad_norm": 0.5623686909675598,
      "learning_rate": 1.940150931875891e-05,
      "loss": 0.4202,
      "step": 6130
    },
    {
      "epoch": 0.11077286416917644,
      "grad_norm": 0.4066951274871826,
      "learning_rate": 1.9400542467343537e-05,
      "loss": 0.3526,
      "step": 6135
    },
    {
      "epoch": 0.11086314360207715,
      "grad_norm": 0.43217983841896057,
      "learning_rate": 1.939957485972287e-05,
      "loss": 0.2482,
      "step": 6140
    },
    {
      "epoch": 0.11095342303497786,
      "grad_norm": 0.5239768028259277,
      "learning_rate": 1.9398606495974752e-05,
      "loss": 0.3912,
      "step": 6145
    },
    {
      "epoch": 0.11104370246787858,
      "grad_norm": 0.356083482503891,
      "learning_rate": 1.9397637376177073e-05,
      "loss": 0.3193,
      "step": 6150
    },
    {
      "epoch": 0.11113398190077929,
      "grad_norm": 0.4758167564868927,
      "learning_rate": 1.9396667500407793e-05,
      "loss": 0.3422,
      "step": 6155
    },
    {
      "epoch": 0.11122426133368,
      "grad_norm": 0.37192463874816895,
      "learning_rate": 1.9395696868744936e-05,
      "loss": 0.3409,
      "step": 6160
    },
    {
      "epoch": 0.11131454076658072,
      "grad_norm": 0.7285003066062927,
      "learning_rate": 1.9394725481266578e-05,
      "loss": 0.3798,
      "step": 6165
    },
    {
      "epoch": 0.11140482019948143,
      "grad_norm": 0.47070422768592834,
      "learning_rate": 1.939375333805086e-05,
      "loss": 0.4061,
      "step": 6170
    },
    {
      "epoch": 0.11149509963238215,
      "grad_norm": 0.35136669874191284,
      "learning_rate": 1.939278043917599e-05,
      "loss": 0.2757,
      "step": 6175
    },
    {
      "epoch": 0.11158537906528286,
      "grad_norm": 0.5866798758506775,
      "learning_rate": 1.9391806784720222e-05,
      "loss": 0.3323,
      "step": 6180
    },
    {
      "epoch": 0.11167565849818357,
      "grad_norm": 0.41596898436546326,
      "learning_rate": 1.9390832374761882e-05,
      "loss": 0.2752,
      "step": 6185
    },
    {
      "epoch": 0.11176593793108429,
      "grad_norm": 0.21890419721603394,
      "learning_rate": 1.9389857209379356e-05,
      "loss": 0.2227,
      "step": 6190
    },
    {
      "epoch": 0.111856217363985,
      "grad_norm": 0.491132527589798,
      "learning_rate": 1.938888128865109e-05,
      "loss": 0.383,
      "step": 6195
    },
    {
      "epoch": 0.11194649679688572,
      "grad_norm": 0.4853299856185913,
      "learning_rate": 1.938790461265559e-05,
      "loss": 0.342,
      "step": 6200
    },
    {
      "epoch": 0.11203677622978643,
      "grad_norm": 0.49977195262908936,
      "learning_rate": 1.9386927181471415e-05,
      "loss": 0.3409,
      "step": 6205
    },
    {
      "epoch": 0.11212705566268714,
      "grad_norm": 0.37122440338134766,
      "learning_rate": 1.9385948995177202e-05,
      "loss": 0.2683,
      "step": 6210
    },
    {
      "epoch": 0.11221733509558786,
      "grad_norm": 1.641432523727417,
      "learning_rate": 1.938497005385163e-05,
      "loss": 0.3658,
      "step": 6215
    },
    {
      "epoch": 0.11230761452848857,
      "grad_norm": 0.32605621218681335,
      "learning_rate": 1.9383990357573456e-05,
      "loss": 0.3598,
      "step": 6220
    },
    {
      "epoch": 0.11239789396138929,
      "grad_norm": 0.6883606314659119,
      "learning_rate": 1.938300990642149e-05,
      "loss": 0.3504,
      "step": 6225
    },
    {
      "epoch": 0.11248817339429,
      "grad_norm": 0.426334023475647,
      "learning_rate": 1.9382028700474592e-05,
      "loss": 0.3739,
      "step": 6230
    },
    {
      "epoch": 0.11257845282719071,
      "grad_norm": 0.3779202401638031,
      "learning_rate": 1.93810467398117e-05,
      "loss": 0.2548,
      "step": 6235
    },
    {
      "epoch": 0.11266873226009143,
      "grad_norm": 0.8037693500518799,
      "learning_rate": 1.9380064024511808e-05,
      "loss": 0.3067,
      "step": 6240
    },
    {
      "epoch": 0.11275901169299216,
      "grad_norm": 0.5130690932273865,
      "learning_rate": 1.9379080554653963e-05,
      "loss": 0.2785,
      "step": 6245
    },
    {
      "epoch": 0.11284929112589287,
      "grad_norm": 1.2105435132980347,
      "learning_rate": 1.9378096330317278e-05,
      "loss": 0.2803,
      "step": 6250
    },
    {
      "epoch": 0.11293957055879358,
      "grad_norm": 0.36144623160362244,
      "learning_rate": 1.937711135158093e-05,
      "loss": 0.375,
      "step": 6255
    },
    {
      "epoch": 0.1130298499916943,
      "grad_norm": 0.46951165795326233,
      "learning_rate": 1.937612561852415e-05,
      "loss": 0.3502,
      "step": 6260
    },
    {
      "epoch": 0.11312012942459501,
      "grad_norm": 0.37856414914131165,
      "learning_rate": 1.9375139131226237e-05,
      "loss": 0.3715,
      "step": 6265
    },
    {
      "epoch": 0.11321040885749573,
      "grad_norm": 0.5409042835235596,
      "learning_rate": 1.9374151889766546e-05,
      "loss": 0.3416,
      "step": 6270
    },
    {
      "epoch": 0.11330068829039644,
      "grad_norm": 1.3376145362854004,
      "learning_rate": 1.937316389422449e-05,
      "loss": 0.3036,
      "step": 6275
    },
    {
      "epoch": 0.11339096772329715,
      "grad_norm": 0.49774277210235596,
      "learning_rate": 1.9372175144679552e-05,
      "loss": 0.2924,
      "step": 6280
    },
    {
      "epoch": 0.11348124715619787,
      "grad_norm": 0.5464069843292236,
      "learning_rate": 1.937118564121126e-05,
      "loss": 0.3233,
      "step": 6285
    },
    {
      "epoch": 0.11357152658909858,
      "grad_norm": 0.48345452547073364,
      "learning_rate": 1.937019538389922e-05,
      "loss": 0.308,
      "step": 6290
    },
    {
      "epoch": 0.1136618060219993,
      "grad_norm": 0.335178941488266,
      "learning_rate": 1.9369204372823092e-05,
      "loss": 0.3727,
      "step": 6295
    },
    {
      "epoch": 0.11375208545490001,
      "grad_norm": 0.5843726992607117,
      "learning_rate": 1.936821260806259e-05,
      "loss": 0.3487,
      "step": 6300
    },
    {
      "epoch": 0.11384236488780072,
      "grad_norm": 0.48966479301452637,
      "learning_rate": 1.93672200896975e-05,
      "loss": 0.4023,
      "step": 6305
    },
    {
      "epoch": 0.11393264432070144,
      "grad_norm": 0.4158240258693695,
      "learning_rate": 1.9366226817807656e-05,
      "loss": 0.2914,
      "step": 6310
    },
    {
      "epoch": 0.11402292375360215,
      "grad_norm": 0.45833754539489746,
      "learning_rate": 1.9365232792472966e-05,
      "loss": 0.2859,
      "step": 6315
    },
    {
      "epoch": 0.11411320318650287,
      "grad_norm": 0.43124616146087646,
      "learning_rate": 1.936423801377339e-05,
      "loss": 0.3819,
      "step": 6320
    },
    {
      "epoch": 0.11420348261940358,
      "grad_norm": 0.47638633847236633,
      "learning_rate": 1.936324248178895e-05,
      "loss": 0.3102,
      "step": 6325
    },
    {
      "epoch": 0.11429376205230429,
      "grad_norm": 0.47278913855552673,
      "learning_rate": 1.936224619659973e-05,
      "loss": 0.2665,
      "step": 6330
    },
    {
      "epoch": 0.114384041485205,
      "grad_norm": 0.3465459942817688,
      "learning_rate": 1.9361249158285875e-05,
      "loss": 0.3992,
      "step": 6335
    },
    {
      "epoch": 0.11447432091810572,
      "grad_norm": 0.5552494525909424,
      "learning_rate": 1.9360251366927586e-05,
      "loss": 0.3767,
      "step": 6340
    },
    {
      "epoch": 0.11456460035100643,
      "grad_norm": 0.5205835700035095,
      "learning_rate": 1.9359252822605132e-05,
      "loss": 0.2317,
      "step": 6345
    },
    {
      "epoch": 0.11465487978390715,
      "grad_norm": 0.5803459882736206,
      "learning_rate": 1.9358253525398835e-05,
      "loss": 0.2835,
      "step": 6350
    },
    {
      "epoch": 0.11474515921680786,
      "grad_norm": 0.36763742566108704,
      "learning_rate": 1.9357253475389083e-05,
      "loss": 0.3426,
      "step": 6355
    },
    {
      "epoch": 0.11483543864970858,
      "grad_norm": 0.31475019454956055,
      "learning_rate": 1.9356252672656323e-05,
      "loss": 0.285,
      "step": 6360
    },
    {
      "epoch": 0.11492571808260929,
      "grad_norm": 0.4656864106655121,
      "learning_rate": 1.9355251117281063e-05,
      "loss": 0.3868,
      "step": 6365
    },
    {
      "epoch": 0.11501599751551,
      "grad_norm": 0.41800594329833984,
      "learning_rate": 1.9354248809343867e-05,
      "loss": 0.2822,
      "step": 6370
    },
    {
      "epoch": 0.11510627694841072,
      "grad_norm": 0.8545289635658264,
      "learning_rate": 1.935324574892537e-05,
      "loss": 0.3509,
      "step": 6375
    },
    {
      "epoch": 0.11519655638131143,
      "grad_norm": 0.526145875453949,
      "learning_rate": 1.9352241936106254e-05,
      "loss": 0.323,
      "step": 6380
    },
    {
      "epoch": 0.11528683581421215,
      "grad_norm": 0.445852130651474,
      "learning_rate": 1.9351237370967275e-05,
      "loss": 0.3624,
      "step": 6385
    },
    {
      "epoch": 0.11537711524711286,
      "grad_norm": 0.329623818397522,
      "learning_rate": 1.9350232053589234e-05,
      "loss": 0.2851,
      "step": 6390
    },
    {
      "epoch": 0.11546739468001357,
      "grad_norm": 0.5335118174552917,
      "learning_rate": 1.934922598405301e-05,
      "loss": 0.37,
      "step": 6395
    },
    {
      "epoch": 0.11555767411291429,
      "grad_norm": 0.3303772807121277,
      "learning_rate": 1.9348219162439534e-05,
      "loss": 0.3581,
      "step": 6400
    },
    {
      "epoch": 0.115647953545815,
      "grad_norm": 0.4297807514667511,
      "learning_rate": 1.934721158882979e-05,
      "loss": 0.2859,
      "step": 6405
    },
    {
      "epoch": 0.11573823297871572,
      "grad_norm": 0.48351696133613586,
      "learning_rate": 1.934620326330484e-05,
      "loss": 0.2975,
      "step": 6410
    },
    {
      "epoch": 0.11582851241161643,
      "grad_norm": 0.36849284172058105,
      "learning_rate": 1.9345194185945785e-05,
      "loss": 0.2877,
      "step": 6415
    },
    {
      "epoch": 0.11591879184451714,
      "grad_norm": 0.3167666792869568,
      "learning_rate": 1.9344184356833805e-05,
      "loss": 0.3535,
      "step": 6420
    },
    {
      "epoch": 0.11600907127741786,
      "grad_norm": 0.6065095663070679,
      "learning_rate": 1.9343173776050137e-05,
      "loss": 0.3139,
      "step": 6425
    },
    {
      "epoch": 0.11609935071031857,
      "grad_norm": 0.5970287919044495,
      "learning_rate": 1.9342162443676063e-05,
      "loss": 0.2853,
      "step": 6430
    },
    {
      "epoch": 0.11618963014321929,
      "grad_norm": 0.7117475271224976,
      "learning_rate": 1.9341150359792952e-05,
      "loss": 0.458,
      "step": 6435
    },
    {
      "epoch": 0.11627990957612001,
      "grad_norm": 0.42741113901138306,
      "learning_rate": 1.9340137524482204e-05,
      "loss": 0.2426,
      "step": 6440
    },
    {
      "epoch": 0.11637018900902073,
      "grad_norm": 0.5931331515312195,
      "learning_rate": 1.9339123937825307e-05,
      "loss": 0.3373,
      "step": 6445
    },
    {
      "epoch": 0.11646046844192144,
      "grad_norm": 0.3439541459083557,
      "learning_rate": 1.9338109599903792e-05,
      "loss": 0.4163,
      "step": 6450
    },
    {
      "epoch": 0.11655074787482216,
      "grad_norm": 0.42862388491630554,
      "learning_rate": 1.933709451079925e-05,
      "loss": 0.3298,
      "step": 6455
    },
    {
      "epoch": 0.11664102730772287,
      "grad_norm": 0.7009595036506653,
      "learning_rate": 1.9336078670593344e-05,
      "loss": 0.3634,
      "step": 6460
    },
    {
      "epoch": 0.11673130674062358,
      "grad_norm": 0.5495311617851257,
      "learning_rate": 1.9335062079367787e-05,
      "loss": 0.2933,
      "step": 6465
    },
    {
      "epoch": 0.1168215861735243,
      "grad_norm": 0.3590311110019684,
      "learning_rate": 1.933404473720436e-05,
      "loss": 0.3476,
      "step": 6470
    },
    {
      "epoch": 0.11691186560642501,
      "grad_norm": 0.6501250863075256,
      "learning_rate": 1.9333026644184898e-05,
      "loss": 0.2524,
      "step": 6475
    },
    {
      "epoch": 0.11700214503932573,
      "grad_norm": 0.37842854857444763,
      "learning_rate": 1.93320078003913e-05,
      "loss": 0.3653,
      "step": 6480
    },
    {
      "epoch": 0.11709242447222644,
      "grad_norm": 0.43740513920783997,
      "learning_rate": 1.9330988205905526e-05,
      "loss": 0.3841,
      "step": 6485
    },
    {
      "epoch": 0.11718270390512715,
      "grad_norm": 0.2887119650840759,
      "learning_rate": 1.9329967860809592e-05,
      "loss": 0.3,
      "step": 6490
    },
    {
      "epoch": 0.11727298333802787,
      "grad_norm": 0.4963028132915497,
      "learning_rate": 1.932894676518558e-05,
      "loss": 0.3703,
      "step": 6495
    },
    {
      "epoch": 0.11736326277092858,
      "grad_norm": 0.6175485849380493,
      "learning_rate": 1.9327924919115628e-05,
      "loss": 0.3587,
      "step": 6500
    },
    {
      "epoch": 0.1174535422038293,
      "grad_norm": 0.4199815094470978,
      "learning_rate": 1.932690232268194e-05,
      "loss": 0.2776,
      "step": 6505
    },
    {
      "epoch": 0.11754382163673001,
      "grad_norm": 0.3766605257987976,
      "learning_rate": 1.9325878975966768e-05,
      "loss": 0.247,
      "step": 6510
    },
    {
      "epoch": 0.11763410106963072,
      "grad_norm": 0.48902565240859985,
      "learning_rate": 1.9324854879052442e-05,
      "loss": 0.2583,
      "step": 6515
    },
    {
      "epoch": 0.11772438050253144,
      "grad_norm": 0.49462440609931946,
      "learning_rate": 1.932383003202134e-05,
      "loss": 0.3483,
      "step": 6520
    },
    {
      "epoch": 0.11781465993543215,
      "grad_norm": 0.44759970903396606,
      "learning_rate": 1.93228044349559e-05,
      "loss": 0.3257,
      "step": 6525
    },
    {
      "epoch": 0.11790493936833286,
      "grad_norm": 0.503629207611084,
      "learning_rate": 1.9321778087938627e-05,
      "loss": 0.2709,
      "step": 6530
    },
    {
      "epoch": 0.11799521880123358,
      "grad_norm": 0.4044511914253235,
      "learning_rate": 1.9320750991052085e-05,
      "loss": 0.3875,
      "step": 6535
    },
    {
      "epoch": 0.11808549823413429,
      "grad_norm": 0.4957432150840759,
      "learning_rate": 1.931972314437889e-05,
      "loss": 0.3982,
      "step": 6540
    },
    {
      "epoch": 0.118175777667035,
      "grad_norm": 1.0312913656234741,
      "learning_rate": 1.931869454800173e-05,
      "loss": 0.3709,
      "step": 6545
    },
    {
      "epoch": 0.11826605709993572,
      "grad_norm": 0.5594607591629028,
      "learning_rate": 1.9317665202003352e-05,
      "loss": 0.4227,
      "step": 6550
    },
    {
      "epoch": 0.11835633653283643,
      "grad_norm": 0.3540436923503876,
      "learning_rate": 1.931663510646655e-05,
      "loss": 0.191,
      "step": 6555
    },
    {
      "epoch": 0.11844661596573715,
      "grad_norm": 0.36071327328681946,
      "learning_rate": 1.931560426147419e-05,
      "loss": 0.4147,
      "step": 6560
    },
    {
      "epoch": 0.11853689539863786,
      "grad_norm": 0.37677809596061707,
      "learning_rate": 1.93145726671092e-05,
      "loss": 0.2762,
      "step": 6565
    },
    {
      "epoch": 0.11862717483153858,
      "grad_norm": 0.6227388978004456,
      "learning_rate": 1.9313540323454564e-05,
      "loss": 0.3244,
      "step": 6570
    },
    {
      "epoch": 0.11871745426443929,
      "grad_norm": 0.4217793643474579,
      "learning_rate": 1.9312507230593327e-05,
      "loss": 0.3036,
      "step": 6575
    },
    {
      "epoch": 0.11880773369734,
      "grad_norm": 0.6171652674674988,
      "learning_rate": 1.9311473388608588e-05,
      "loss": 0.3221,
      "step": 6580
    },
    {
      "epoch": 0.11889801313024072,
      "grad_norm": 0.6095660924911499,
      "learning_rate": 1.931043879758352e-05,
      "loss": 0.239,
      "step": 6585
    },
    {
      "epoch": 0.11898829256314143,
      "grad_norm": 0.34578561782836914,
      "learning_rate": 1.930940345760134e-05,
      "loss": 0.3711,
      "step": 6590
    },
    {
      "epoch": 0.11907857199604215,
      "grad_norm": 0.4061829149723053,
      "learning_rate": 1.9308367368745337e-05,
      "loss": 0.4484,
      "step": 6595
    },
    {
      "epoch": 0.11916885142894286,
      "grad_norm": 0.6296597719192505,
      "learning_rate": 1.930733053109886e-05,
      "loss": 0.2457,
      "step": 6600
    },
    {
      "epoch": 0.11925913086184357,
      "grad_norm": 0.7155535817146301,
      "learning_rate": 1.9306292944745313e-05,
      "loss": 0.3223,
      "step": 6605
    },
    {
      "epoch": 0.11934941029474429,
      "grad_norm": 0.4064818322658539,
      "learning_rate": 1.930525460976816e-05,
      "loss": 0.2874,
      "step": 6610
    },
    {
      "epoch": 0.119439689727645,
      "grad_norm": 0.3748913109302521,
      "learning_rate": 1.930421552625093e-05,
      "loss": 0.2791,
      "step": 6615
    },
    {
      "epoch": 0.11952996916054572,
      "grad_norm": 0.5190522074699402,
      "learning_rate": 1.9303175694277213e-05,
      "loss": 0.4314,
      "step": 6620
    },
    {
      "epoch": 0.11962024859344643,
      "grad_norm": 0.4803234040737152,
      "learning_rate": 1.9302135113930646e-05,
      "loss": 0.2915,
      "step": 6625
    },
    {
      "epoch": 0.11971052802634714,
      "grad_norm": 0.33393678069114685,
      "learning_rate": 1.9301093785294943e-05,
      "loss": 0.4128,
      "step": 6630
    },
    {
      "epoch": 0.11980080745924786,
      "grad_norm": 0.4914097189903259,
      "learning_rate": 1.9300051708453872e-05,
      "loss": 0.2867,
      "step": 6635
    },
    {
      "epoch": 0.11989108689214859,
      "grad_norm": 1.1089816093444824,
      "learning_rate": 1.9299008883491263e-05,
      "loss": 0.2758,
      "step": 6640
    },
    {
      "epoch": 0.1199813663250493,
      "grad_norm": 0.34248292446136475,
      "learning_rate": 1.9297965310490994e-05,
      "loss": 0.3728,
      "step": 6645
    },
    {
      "epoch": 0.12007164575795001,
      "grad_norm": 0.34159940481185913,
      "learning_rate": 1.929692098953702e-05,
      "loss": 0.247,
      "step": 6650
    },
    {
      "epoch": 0.12016192519085073,
      "grad_norm": 0.3600628972053528,
      "learning_rate": 1.929587592071335e-05,
      "loss": 0.3114,
      "step": 6655
    },
    {
      "epoch": 0.12025220462375144,
      "grad_norm": 0.4529002606868744,
      "learning_rate": 1.9294830104104045e-05,
      "loss": 0.3179,
      "step": 6660
    },
    {
      "epoch": 0.12034248405665215,
      "grad_norm": 0.39672356843948364,
      "learning_rate": 1.929378353979324e-05,
      "loss": 0.3081,
      "step": 6665
    },
    {
      "epoch": 0.12043276348955287,
      "grad_norm": 0.35251498222351074,
      "learning_rate": 1.929273622786512e-05,
      "loss": 0.2415,
      "step": 6670
    },
    {
      "epoch": 0.12052304292245358,
      "grad_norm": 1.1927274465560913,
      "learning_rate": 1.929168816840394e-05,
      "loss": 0.2911,
      "step": 6675
    },
    {
      "epoch": 0.1206133223553543,
      "grad_norm": 0.5608368515968323,
      "learning_rate": 1.9290639361494002e-05,
      "loss": 0.3454,
      "step": 6680
    },
    {
      "epoch": 0.12070360178825501,
      "grad_norm": 0.712186336517334,
      "learning_rate": 1.9289589807219674e-05,
      "loss": 0.3953,
      "step": 6685
    },
    {
      "epoch": 0.12079388122115572,
      "grad_norm": 0.849686861038208,
      "learning_rate": 1.928853950566539e-05,
      "loss": 0.3562,
      "step": 6690
    },
    {
      "epoch": 0.12088416065405644,
      "grad_norm": 0.7094193696975708,
      "learning_rate": 1.928748845691564e-05,
      "loss": 0.3219,
      "step": 6695
    },
    {
      "epoch": 0.12097444008695715,
      "grad_norm": 0.5030642747879028,
      "learning_rate": 1.9286436661054965e-05,
      "loss": 0.2835,
      "step": 6700
    },
    {
      "epoch": 0.12106471951985787,
      "grad_norm": 0.36728736758232117,
      "learning_rate": 1.9285384118167984e-05,
      "loss": 0.389,
      "step": 6705
    },
    {
      "epoch": 0.12115499895275858,
      "grad_norm": 0.6199520826339722,
      "learning_rate": 1.928433082833936e-05,
      "loss": 0.313,
      "step": 6710
    },
    {
      "epoch": 0.1212452783856593,
      "grad_norm": 0.6002387404441833,
      "learning_rate": 1.928327679165382e-05,
      "loss": 0.3059,
      "step": 6715
    },
    {
      "epoch": 0.12133555781856001,
      "grad_norm": 0.8976176381111145,
      "learning_rate": 1.9282222008196165e-05,
      "loss": 0.2992,
      "step": 6720
    },
    {
      "epoch": 0.12142583725146072,
      "grad_norm": 0.3749474585056305,
      "learning_rate": 1.9281166478051234e-05,
      "loss": 0.3149,
      "step": 6725
    },
    {
      "epoch": 0.12151611668436144,
      "grad_norm": 0.4321290850639343,
      "learning_rate": 1.9280110201303942e-05,
      "loss": 0.3096,
      "step": 6730
    },
    {
      "epoch": 0.12160639611726215,
      "grad_norm": 0.4031892716884613,
      "learning_rate": 1.9279053178039256e-05,
      "loss": 0.4123,
      "step": 6735
    },
    {
      "epoch": 0.12169667555016286,
      "grad_norm": 0.28211620450019836,
      "learning_rate": 1.927799540834221e-05,
      "loss": 0.3758,
      "step": 6740
    },
    {
      "epoch": 0.12178695498306358,
      "grad_norm": 0.6331525444984436,
      "learning_rate": 1.9276936892297887e-05,
      "loss": 0.2116,
      "step": 6745
    },
    {
      "epoch": 0.12187723441596429,
      "grad_norm": 0.44513705372810364,
      "learning_rate": 1.9275877629991444e-05,
      "loss": 0.2429,
      "step": 6750
    },
    {
      "epoch": 0.121967513848865,
      "grad_norm": 0.4318163990974426,
      "learning_rate": 1.9274817621508088e-05,
      "loss": 0.3567,
      "step": 6755
    },
    {
      "epoch": 0.12205779328176572,
      "grad_norm": 0.33108144998550415,
      "learning_rate": 1.9273756866933083e-05,
      "loss": 0.2401,
      "step": 6760
    },
    {
      "epoch": 0.12214807271466643,
      "grad_norm": 0.3258655369281769,
      "learning_rate": 1.927269536635177e-05,
      "loss": 0.2845,
      "step": 6765
    },
    {
      "epoch": 0.12223835214756715,
      "grad_norm": 0.48052114248275757,
      "learning_rate": 1.9271633119849535e-05,
      "loss": 0.3305,
      "step": 6770
    },
    {
      "epoch": 0.12232863158046786,
      "grad_norm": 0.4707503020763397,
      "learning_rate": 1.927057012751182e-05,
      "loss": 0.3054,
      "step": 6775
    },
    {
      "epoch": 0.12241891101336858,
      "grad_norm": 0.7184447050094604,
      "learning_rate": 1.926950638942415e-05,
      "loss": 0.2975,
      "step": 6780
    },
    {
      "epoch": 0.12250919044626929,
      "grad_norm": 0.3022684156894684,
      "learning_rate": 1.9268441905672078e-05,
      "loss": 0.3706,
      "step": 6785
    },
    {
      "epoch": 0.12259946987917,
      "grad_norm": 0.46065765619277954,
      "learning_rate": 1.9267376676341252e-05,
      "loss": 0.3397,
      "step": 6790
    },
    {
      "epoch": 0.12268974931207072,
      "grad_norm": 2.153079032897949,
      "learning_rate": 1.926631070151735e-05,
      "loss": 0.3677,
      "step": 6795
    },
    {
      "epoch": 0.12278002874497143,
      "grad_norm": 0.33272886276245117,
      "learning_rate": 1.9265243981286122e-05,
      "loss": 0.293,
      "step": 6800
    },
    {
      "epoch": 0.12287030817787215,
      "grad_norm": 0.6385965347290039,
      "learning_rate": 1.926417651573338e-05,
      "loss": 0.323,
      "step": 6805
    },
    {
      "epoch": 0.12296058761077286,
      "grad_norm": 0.503334105014801,
      "learning_rate": 1.9263108304944994e-05,
      "loss": 0.2452,
      "step": 6810
    },
    {
      "epoch": 0.12305086704367357,
      "grad_norm": 0.3952818810939789,
      "learning_rate": 1.92620393490069e-05,
      "loss": 0.2221,
      "step": 6815
    },
    {
      "epoch": 0.12314114647657429,
      "grad_norm": 0.39027824997901917,
      "learning_rate": 1.9260969648005076e-05,
      "loss": 0.31,
      "step": 6820
    },
    {
      "epoch": 0.123231425909475,
      "grad_norm": 0.6210625767707825,
      "learning_rate": 1.925989920202558e-05,
      "loss": 0.3762,
      "step": 6825
    },
    {
      "epoch": 0.12332170534237571,
      "grad_norm": 0.5680192708969116,
      "learning_rate": 1.925882801115452e-05,
      "loss": 0.3463,
      "step": 6830
    },
    {
      "epoch": 0.12341198477527643,
      "grad_norm": 0.29992368817329407,
      "learning_rate": 1.9257756075478063e-05,
      "loss": 0.2801,
      "step": 6835
    },
    {
      "epoch": 0.12350226420817716,
      "grad_norm": 0.31016963720321655,
      "learning_rate": 1.925668339508244e-05,
      "loss": 0.2648,
      "step": 6840
    },
    {
      "epoch": 0.12359254364107787,
      "grad_norm": 0.5393102765083313,
      "learning_rate": 1.9255609970053943e-05,
      "loss": 0.3645,
      "step": 6845
    },
    {
      "epoch": 0.12368282307397858,
      "grad_norm": 0.48195183277130127,
      "learning_rate": 1.9254535800478918e-05,
      "loss": 0.354,
      "step": 6850
    },
    {
      "epoch": 0.1237731025068793,
      "grad_norm": 0.44904544949531555,
      "learning_rate": 1.9253460886443773e-05,
      "loss": 0.327,
      "step": 6855
    },
    {
      "epoch": 0.12386338193978001,
      "grad_norm": 0.46062836050987244,
      "learning_rate": 1.925238522803498e-05,
      "loss": 0.358,
      "step": 6860
    },
    {
      "epoch": 0.12395366137268073,
      "grad_norm": 0.38590455055236816,
      "learning_rate": 1.9251308825339063e-05,
      "loss": 0.3627,
      "step": 6865
    },
    {
      "epoch": 0.12404394080558144,
      "grad_norm": 0.6193512082099915,
      "learning_rate": 1.925023167844262e-05,
      "loss": 0.3356,
      "step": 6870
    },
    {
      "epoch": 0.12413422023848215,
      "grad_norm": 0.4015476703643799,
      "learning_rate": 1.9249153787432294e-05,
      "loss": 0.3249,
      "step": 6875
    },
    {
      "epoch": 0.12422449967138287,
      "grad_norm": 0.5982859134674072,
      "learning_rate": 1.924807515239479e-05,
      "loss": 0.2964,
      "step": 6880
    },
    {
      "epoch": 0.12431477910428358,
      "grad_norm": 0.34201470017433167,
      "learning_rate": 1.924699577341688e-05,
      "loss": 0.3193,
      "step": 6885
    },
    {
      "epoch": 0.1244050585371843,
      "grad_norm": 0.5029416084289551,
      "learning_rate": 1.9245915650585396e-05,
      "loss": 0.3764,
      "step": 6890
    },
    {
      "epoch": 0.12449533797008501,
      "grad_norm": 0.5241310000419617,
      "learning_rate": 1.924483478398722e-05,
      "loss": 0.4296,
      "step": 6895
    },
    {
      "epoch": 0.12458561740298572,
      "grad_norm": 0.7433327436447144,
      "learning_rate": 1.9243753173709302e-05,
      "loss": 0.4092,
      "step": 6900
    },
    {
      "epoch": 0.12467589683588644,
      "grad_norm": 0.3439296782016754,
      "learning_rate": 1.924267081983865e-05,
      "loss": 0.2702,
      "step": 6905
    },
    {
      "epoch": 0.12476617626878715,
      "grad_norm": 0.3778502345085144,
      "learning_rate": 1.924158772246233e-05,
      "loss": 0.3099,
      "step": 6910
    },
    {
      "epoch": 0.12485645570168787,
      "grad_norm": 0.44551828503608704,
      "learning_rate": 1.9240503881667472e-05,
      "loss": 0.3896,
      "step": 6915
    },
    {
      "epoch": 0.12494673513458858,
      "grad_norm": 0.5756784677505493,
      "learning_rate": 1.923941929754126e-05,
      "loss": 0.389,
      "step": 6920
    },
    {
      "epoch": 0.12503701456748928,
      "grad_norm": 0.624144971370697,
      "learning_rate": 1.9238333970170947e-05,
      "loss": 0.2913,
      "step": 6925
    },
    {
      "epoch": 0.12512729400039,
      "grad_norm": 0.6104403734207153,
      "learning_rate": 1.923724789964383e-05,
      "loss": 0.3229,
      "step": 6930
    },
    {
      "epoch": 0.12521757343329074,
      "grad_norm": 0.32302334904670715,
      "learning_rate": 1.9236161086047286e-05,
      "loss": 0.2511,
      "step": 6935
    },
    {
      "epoch": 0.12530785286619145,
      "grad_norm": 0.28354236483573914,
      "learning_rate": 1.923507352946873e-05,
      "loss": 0.328,
      "step": 6940
    },
    {
      "epoch": 0.12539813229909216,
      "grad_norm": 0.5260065793991089,
      "learning_rate": 1.9233985229995657e-05,
      "loss": 0.3043,
      "step": 6945
    },
    {
      "epoch": 0.12548841173199288,
      "grad_norm": 0.5684165358543396,
      "learning_rate": 1.923289618771561e-05,
      "loss": 0.3425,
      "step": 6950
    },
    {
      "epoch": 0.1255786911648936,
      "grad_norm": 0.4093371033668518,
      "learning_rate": 1.9231806402716195e-05,
      "loss": 0.3634,
      "step": 6955
    },
    {
      "epoch": 0.1256689705977943,
      "grad_norm": 0.4712226986885071,
      "learning_rate": 1.9230715875085076e-05,
      "loss": 0.2982,
      "step": 6960
    },
    {
      "epoch": 0.12575925003069502,
      "grad_norm": 0.3231732249259949,
      "learning_rate": 1.9229624604909977e-05,
      "loss": 0.2779,
      "step": 6965
    },
    {
      "epoch": 0.12584952946359573,
      "grad_norm": 0.4267129600048065,
      "learning_rate": 1.9228532592278686e-05,
      "loss": 0.2812,
      "step": 6970
    },
    {
      "epoch": 0.12593980889649645,
      "grad_norm": 0.4321097731590271,
      "learning_rate": 1.9227439837279046e-05,
      "loss": 0.3731,
      "step": 6975
    },
    {
      "epoch": 0.12603008832939716,
      "grad_norm": 0.4453035295009613,
      "learning_rate": 1.9226346339998962e-05,
      "loss": 0.3764,
      "step": 6980
    },
    {
      "epoch": 0.12612036776229787,
      "grad_norm": 0.5288938283920288,
      "learning_rate": 1.9225252100526394e-05,
      "loss": 0.3277,
      "step": 6985
    },
    {
      "epoch": 0.1262106471951986,
      "grad_norm": 0.3053990304470062,
      "learning_rate": 1.9224157118949373e-05,
      "loss": 0.2816,
      "step": 6990
    },
    {
      "epoch": 0.1263009266280993,
      "grad_norm": 0.4477319121360779,
      "learning_rate": 1.9223061395355975e-05,
      "loss": 0.3374,
      "step": 6995
    },
    {
      "epoch": 0.12639120606100002,
      "grad_norm": 0.32137957215309143,
      "learning_rate": 1.9221964929834343e-05,
      "loss": 0.3038,
      "step": 7000
    },
    {
      "epoch": 0.12648148549390073,
      "grad_norm": 0.37034428119659424,
      "learning_rate": 1.9220867722472687e-05,
      "loss": 0.3287,
      "step": 7005
    },
    {
      "epoch": 0.12657176492680144,
      "grad_norm": 0.4952510893344879,
      "learning_rate": 1.9219769773359265e-05,
      "loss": 0.2908,
      "step": 7010
    },
    {
      "epoch": 0.12666204435970216,
      "grad_norm": 0.30036431550979614,
      "learning_rate": 1.9218671082582396e-05,
      "loss": 0.3113,
      "step": 7015
    },
    {
      "epoch": 0.12675232379260287,
      "grad_norm": 0.41781020164489746,
      "learning_rate": 1.9217571650230464e-05,
      "loss": 0.3724,
      "step": 7020
    },
    {
      "epoch": 0.1268426032255036,
      "grad_norm": 0.4511464238166809,
      "learning_rate": 1.9216471476391915e-05,
      "loss": 0.2724,
      "step": 7025
    },
    {
      "epoch": 0.1269328826584043,
      "grad_norm": 0.32684236764907837,
      "learning_rate": 1.9215370561155245e-05,
      "loss": 0.3578,
      "step": 7030
    },
    {
      "epoch": 0.12702316209130501,
      "grad_norm": 0.38256338238716125,
      "learning_rate": 1.9214268904609015e-05,
      "loss": 0.2174,
      "step": 7035
    },
    {
      "epoch": 0.12711344152420573,
      "grad_norm": 0.5053243637084961,
      "learning_rate": 1.9213166506841844e-05,
      "loss": 0.2677,
      "step": 7040
    },
    {
      "epoch": 0.12720372095710644,
      "grad_norm": 0.3353807330131531,
      "learning_rate": 1.9212063367942416e-05,
      "loss": 0.2603,
      "step": 7045
    },
    {
      "epoch": 0.12729400039000716,
      "grad_norm": 0.4773527979850769,
      "learning_rate": 1.9210959487999466e-05,
      "loss": 0.3303,
      "step": 7050
    },
    {
      "epoch": 0.12738427982290787,
      "grad_norm": 0.962485134601593,
      "learning_rate": 1.92098548671018e-05,
      "loss": 0.323,
      "step": 7055
    },
    {
      "epoch": 0.12747455925580858,
      "grad_norm": 0.31148481369018555,
      "learning_rate": 1.9208749505338266e-05,
      "loss": 0.3649,
      "step": 7060
    },
    {
      "epoch": 0.1275648386887093,
      "grad_norm": 0.41241616010665894,
      "learning_rate": 1.920764340279779e-05,
      "loss": 0.3506,
      "step": 7065
    },
    {
      "epoch": 0.12765511812161,
      "grad_norm": 0.5747043490409851,
      "learning_rate": 1.920653655956935e-05,
      "loss": 0.3686,
      "step": 7070
    },
    {
      "epoch": 0.12774539755451073,
      "grad_norm": 0.4869537055492401,
      "learning_rate": 1.9205428975741983e-05,
      "loss": 0.3266,
      "step": 7075
    },
    {
      "epoch": 0.12783567698741144,
      "grad_norm": 0.6491648554801941,
      "learning_rate": 1.9204320651404782e-05,
      "loss": 0.3375,
      "step": 7080
    },
    {
      "epoch": 0.12792595642031215,
      "grad_norm": 0.3836076259613037,
      "learning_rate": 1.9203211586646908e-05,
      "loss": 0.317,
      "step": 7085
    },
    {
      "epoch": 0.12801623585321287,
      "grad_norm": 0.4446025490760803,
      "learning_rate": 1.9202101781557575e-05,
      "loss": 0.3376,
      "step": 7090
    },
    {
      "epoch": 0.12810651528611358,
      "grad_norm": 0.43301495909690857,
      "learning_rate": 1.9200991236226057e-05,
      "loss": 0.3474,
      "step": 7095
    },
    {
      "epoch": 0.1281967947190143,
      "grad_norm": 0.5248858332633972,
      "learning_rate": 1.9199879950741698e-05,
      "loss": 0.3209,
      "step": 7100
    },
    {
      "epoch": 0.128287074151915,
      "grad_norm": 0.5935579538345337,
      "learning_rate": 1.919876792519388e-05,
      "loss": 0.3652,
      "step": 7105
    },
    {
      "epoch": 0.12837735358481572,
      "grad_norm": 0.3689795732498169,
      "learning_rate": 1.9197655159672068e-05,
      "loss": 0.3509,
      "step": 7110
    },
    {
      "epoch": 0.12846763301771644,
      "grad_norm": 0.44201385974884033,
      "learning_rate": 1.919654165426577e-05,
      "loss": 0.2219,
      "step": 7115
    },
    {
      "epoch": 0.12855791245061715,
      "grad_norm": 0.4781646430492401,
      "learning_rate": 1.9195427409064563e-05,
      "loss": 0.2743,
      "step": 7120
    },
    {
      "epoch": 0.12864819188351787,
      "grad_norm": 0.3723335862159729,
      "learning_rate": 1.9194312424158077e-05,
      "loss": 0.3604,
      "step": 7125
    },
    {
      "epoch": 0.12873847131641858,
      "grad_norm": 0.43962472677230835,
      "learning_rate": 1.9193196699636004e-05,
      "loss": 0.3219,
      "step": 7130
    },
    {
      "epoch": 0.1288287507493193,
      "grad_norm": 0.5739518404006958,
      "learning_rate": 1.91920802355881e-05,
      "loss": 0.3676,
      "step": 7135
    },
    {
      "epoch": 0.12891903018222,
      "grad_norm": 0.4589199721813202,
      "learning_rate": 1.9190963032104173e-05,
      "loss": 0.2633,
      "step": 7140
    },
    {
      "epoch": 0.12900930961512072,
      "grad_norm": 0.7155197262763977,
      "learning_rate": 1.9189845089274095e-05,
      "loss": 0.406,
      "step": 7145
    },
    {
      "epoch": 0.12909958904802143,
      "grad_norm": 0.5128468871116638,
      "learning_rate": 1.9188726407187795e-05,
      "loss": 0.3197,
      "step": 7150
    },
    {
      "epoch": 0.12918986848092215,
      "grad_norm": 0.29664355516433716,
      "learning_rate": 1.9187606985935267e-05,
      "loss": 0.3227,
      "step": 7155
    },
    {
      "epoch": 0.12928014791382286,
      "grad_norm": 0.4465535581111908,
      "learning_rate": 1.9186486825606553e-05,
      "loss": 0.3069,
      "step": 7160
    },
    {
      "epoch": 0.12937042734672358,
      "grad_norm": 0.5476791262626648,
      "learning_rate": 1.918536592629177e-05,
      "loss": 0.3073,
      "step": 7165
    },
    {
      "epoch": 0.1294607067796243,
      "grad_norm": 0.47054997086524963,
      "learning_rate": 1.918424428808108e-05,
      "loss": 0.2936,
      "step": 7170
    },
    {
      "epoch": 0.129550986212525,
      "grad_norm": 0.8305136561393738,
      "learning_rate": 1.9183121911064715e-05,
      "loss": 0.3147,
      "step": 7175
    },
    {
      "epoch": 0.12964126564542572,
      "grad_norm": 0.5708081126213074,
      "learning_rate": 1.9181998795332958e-05,
      "loss": 0.2975,
      "step": 7180
    },
    {
      "epoch": 0.12973154507832643,
      "grad_norm": 0.31832578778266907,
      "learning_rate": 1.918087494097616e-05,
      "loss": 0.3107,
      "step": 7185
    },
    {
      "epoch": 0.12982182451122715,
      "grad_norm": 0.8390007019042969,
      "learning_rate": 1.917975034808472e-05,
      "loss": 0.3815,
      "step": 7190
    },
    {
      "epoch": 0.12991210394412786,
      "grad_norm": 0.37863609194755554,
      "learning_rate": 1.917862501674911e-05,
      "loss": 0.3383,
      "step": 7195
    },
    {
      "epoch": 0.13000238337702857,
      "grad_norm": 0.44151777029037476,
      "learning_rate": 1.9177498947059856e-05,
      "loss": 0.281,
      "step": 7200
    },
    {
      "epoch": 0.1300926628099293,
      "grad_norm": 0.47824159264564514,
      "learning_rate": 1.917637213910753e-05,
      "loss": 0.3079,
      "step": 7205
    },
    {
      "epoch": 0.13018294224283,
      "grad_norm": 0.37554118037223816,
      "learning_rate": 1.9175244592982795e-05,
      "loss": 0.3497,
      "step": 7210
    },
    {
      "epoch": 0.13027322167573072,
      "grad_norm": 0.5599135160446167,
      "learning_rate": 1.9174116308776337e-05,
      "loss": 0.2656,
      "step": 7215
    },
    {
      "epoch": 0.13036350110863143,
      "grad_norm": 0.44534119963645935,
      "learning_rate": 1.9172987286578922e-05,
      "loss": 0.3264,
      "step": 7220
    },
    {
      "epoch": 0.13045378054153214,
      "grad_norm": 0.3763585388660431,
      "learning_rate": 1.917185752648138e-05,
      "loss": 0.2163,
      "step": 7225
    },
    {
      "epoch": 0.13054405997443286,
      "grad_norm": 0.3645566701889038,
      "learning_rate": 1.9170727028574582e-05,
      "loss": 0.3497,
      "step": 7230
    },
    {
      "epoch": 0.13063433940733357,
      "grad_norm": 0.36997899413108826,
      "learning_rate": 1.916959579294947e-05,
      "loss": 0.3257,
      "step": 7235
    },
    {
      "epoch": 0.13072461884023429,
      "grad_norm": 0.5548878312110901,
      "learning_rate": 1.916846381969705e-05,
      "loss": 0.3302,
      "step": 7240
    },
    {
      "epoch": 0.130814898273135,
      "grad_norm": 0.39686310291290283,
      "learning_rate": 1.9167331108908376e-05,
      "loss": 0.277,
      "step": 7245
    },
    {
      "epoch": 0.1309051777060357,
      "grad_norm": 0.3491361439228058,
      "learning_rate": 1.9166197660674567e-05,
      "loss": 0.2826,
      "step": 7250
    },
    {
      "epoch": 0.13099545713893643,
      "grad_norm": 0.42074933648109436,
      "learning_rate": 1.9165063475086797e-05,
      "loss": 0.3404,
      "step": 7255
    },
    {
      "epoch": 0.13108573657183714,
      "grad_norm": 0.46928852796554565,
      "learning_rate": 1.9163928552236314e-05,
      "loss": 0.3195,
      "step": 7260
    },
    {
      "epoch": 0.13117601600473786,
      "grad_norm": 0.37827062606811523,
      "learning_rate": 1.91627928922144e-05,
      "loss": 0.3645,
      "step": 7265
    },
    {
      "epoch": 0.13126629543763857,
      "grad_norm": 0.442892849445343,
      "learning_rate": 1.9161656495112424e-05,
      "loss": 0.3934,
      "step": 7270
    },
    {
      "epoch": 0.13135657487053928,
      "grad_norm": 0.3846202790737152,
      "learning_rate": 1.916051936102179e-05,
      "loss": 0.2378,
      "step": 7275
    },
    {
      "epoch": 0.13144685430344,
      "grad_norm": 0.3885536193847656,
      "learning_rate": 1.915938149003398e-05,
      "loss": 0.193,
      "step": 7280
    },
    {
      "epoch": 0.1315371337363407,
      "grad_norm": 0.3747337758541107,
      "learning_rate": 1.9158242882240523e-05,
      "loss": 0.3629,
      "step": 7285
    },
    {
      "epoch": 0.13162741316924143,
      "grad_norm": 0.3291547894477844,
      "learning_rate": 1.9157103537733013e-05,
      "loss": 0.3123,
      "step": 7290
    },
    {
      "epoch": 0.13171769260214214,
      "grad_norm": 0.3559887707233429,
      "learning_rate": 1.91559634566031e-05,
      "loss": 0.3227,
      "step": 7295
    },
    {
      "epoch": 0.13180797203504285,
      "grad_norm": 0.4530769884586334,
      "learning_rate": 1.91548226389425e-05,
      "loss": 0.3168,
      "step": 7300
    },
    {
      "epoch": 0.13189825146794357,
      "grad_norm": 0.38989293575286865,
      "learning_rate": 1.915368108484298e-05,
      "loss": 0.2548,
      "step": 7305
    },
    {
      "epoch": 0.13198853090084428,
      "grad_norm": 0.5381086468696594,
      "learning_rate": 1.9152538794396372e-05,
      "loss": 0.3533,
      "step": 7310
    },
    {
      "epoch": 0.132078810333745,
      "grad_norm": 0.5024468898773193,
      "learning_rate": 1.9151395767694563e-05,
      "loss": 0.3122,
      "step": 7315
    },
    {
      "epoch": 0.1321690897666457,
      "grad_norm": 0.5017898082733154,
      "learning_rate": 1.9150252004829498e-05,
      "loss": 0.2025,
      "step": 7320
    },
    {
      "epoch": 0.13225936919954642,
      "grad_norm": 0.46784597635269165,
      "learning_rate": 1.914910750589319e-05,
      "loss": 0.3269,
      "step": 7325
    },
    {
      "epoch": 0.13234964863244714,
      "grad_norm": 1.0234980583190918,
      "learning_rate": 1.9147962270977705e-05,
      "loss": 0.3999,
      "step": 7330
    },
    {
      "epoch": 0.13243992806534788,
      "grad_norm": 0.3947860598564148,
      "learning_rate": 1.9146816300175166e-05,
      "loss": 0.3523,
      "step": 7335
    },
    {
      "epoch": 0.1325302074982486,
      "grad_norm": 0.7022761702537537,
      "learning_rate": 1.914566959357776e-05,
      "loss": 0.3815,
      "step": 7340
    },
    {
      "epoch": 0.1326204869311493,
      "grad_norm": 0.4152867794036865,
      "learning_rate": 1.914452215127773e-05,
      "loss": 0.2709,
      "step": 7345
    },
    {
      "epoch": 0.13271076636405002,
      "grad_norm": 0.5031547546386719,
      "learning_rate": 1.914337397336738e-05,
      "loss": 0.2726,
      "step": 7350
    },
    {
      "epoch": 0.13280104579695073,
      "grad_norm": 0.5867427587509155,
      "learning_rate": 1.9142225059939073e-05,
      "loss": 0.2936,
      "step": 7355
    },
    {
      "epoch": 0.13289132522985145,
      "grad_norm": 0.49512407183647156,
      "learning_rate": 1.914107541108523e-05,
      "loss": 0.3274,
      "step": 7360
    },
    {
      "epoch": 0.13298160466275216,
      "grad_norm": 0.4875224828720093,
      "learning_rate": 1.9139925026898335e-05,
      "loss": 0.3373,
      "step": 7365
    },
    {
      "epoch": 0.13307188409565288,
      "grad_norm": 0.3881199359893799,
      "learning_rate": 1.913877390747092e-05,
      "loss": 0.2481,
      "step": 7370
    },
    {
      "epoch": 0.1331621635285536,
      "grad_norm": 0.5740962028503418,
      "learning_rate": 1.9137622052895593e-05,
      "loss": 0.3268,
      "step": 7375
    },
    {
      "epoch": 0.1332524429614543,
      "grad_norm": 0.39869681000709534,
      "learning_rate": 1.913646946326501e-05,
      "loss": 0.3082,
      "step": 7380
    },
    {
      "epoch": 0.13334272239435502,
      "grad_norm": 0.41652965545654297,
      "learning_rate": 1.9135316138671884e-05,
      "loss": 0.4155,
      "step": 7385
    },
    {
      "epoch": 0.13343300182725573,
      "grad_norm": 0.5283089280128479,
      "learning_rate": 1.9134162079208995e-05,
      "loss": 0.2462,
      "step": 7390
    },
    {
      "epoch": 0.13352328126015645,
      "grad_norm": 0.46559688448905945,
      "learning_rate": 1.9133007284969182e-05,
      "loss": 0.3209,
      "step": 7395
    },
    {
      "epoch": 0.13361356069305716,
      "grad_norm": 0.6211048364639282,
      "learning_rate": 1.9131851756045334e-05,
      "loss": 0.2968,
      "step": 7400
    },
    {
      "epoch": 0.13370384012595787,
      "grad_norm": 0.45277780294418335,
      "learning_rate": 1.9130695492530403e-05,
      "loss": 0.3602,
      "step": 7405
    },
    {
      "epoch": 0.1337941195588586,
      "grad_norm": 0.5387540459632874,
      "learning_rate": 1.912953849451741e-05,
      "loss": 0.3366,
      "step": 7410
    },
    {
      "epoch": 0.1338843989917593,
      "grad_norm": 0.44352853298187256,
      "learning_rate": 1.9128380762099422e-05,
      "loss": 0.3103,
      "step": 7415
    },
    {
      "epoch": 0.13397467842466002,
      "grad_norm": 0.5637753009796143,
      "learning_rate": 1.9127222295369576e-05,
      "loss": 0.4029,
      "step": 7420
    },
    {
      "epoch": 0.13406495785756073,
      "grad_norm": 0.36902207136154175,
      "learning_rate": 1.9126063094421055e-05,
      "loss": 0.3426,
      "step": 7425
    },
    {
      "epoch": 0.13415523729046144,
      "grad_norm": 0.2464890480041504,
      "learning_rate": 1.912490315934711e-05,
      "loss": 0.2659,
      "step": 7430
    },
    {
      "epoch": 0.13424551672336216,
      "grad_norm": 0.454995721578598,
      "learning_rate": 1.912374249024105e-05,
      "loss": 0.3515,
      "step": 7435
    },
    {
      "epoch": 0.13433579615626287,
      "grad_norm": 0.4358680546283722,
      "learning_rate": 1.9122581087196245e-05,
      "loss": 0.3289,
      "step": 7440
    },
    {
      "epoch": 0.13442607558916359,
      "grad_norm": 0.4596993923187256,
      "learning_rate": 1.9121418950306117e-05,
      "loss": 0.2327,
      "step": 7445
    },
    {
      "epoch": 0.1345163550220643,
      "grad_norm": 0.5084929466247559,
      "learning_rate": 1.9120256079664154e-05,
      "loss": 0.3468,
      "step": 7450
    },
    {
      "epoch": 0.134606634454965,
      "grad_norm": 0.3656415343284607,
      "learning_rate": 1.9119092475363902e-05,
      "loss": 0.27,
      "step": 7455
    },
    {
      "epoch": 0.13469691388786573,
      "grad_norm": 0.40600907802581787,
      "learning_rate": 1.911792813749896e-05,
      "loss": 0.1852,
      "step": 7460
    },
    {
      "epoch": 0.13478719332076644,
      "grad_norm": 0.42754778265953064,
      "learning_rate": 1.9116763066162992e-05,
      "loss": 0.3033,
      "step": 7465
    },
    {
      "epoch": 0.13487747275366715,
      "grad_norm": 0.43957412242889404,
      "learning_rate": 1.911559726144972e-05,
      "loss": 0.2942,
      "step": 7470
    },
    {
      "epoch": 0.13496775218656787,
      "grad_norm": 0.3917752802371979,
      "learning_rate": 1.911443072345293e-05,
      "loss": 0.3116,
      "step": 7475
    },
    {
      "epoch": 0.13505803161946858,
      "grad_norm": 0.6572051048278809,
      "learning_rate": 1.9113263452266457e-05,
      "loss": 0.3136,
      "step": 7480
    },
    {
      "epoch": 0.1351483110523693,
      "grad_norm": 0.4349372386932373,
      "learning_rate": 1.9112095447984195e-05,
      "loss": 0.3361,
      "step": 7485
    },
    {
      "epoch": 0.13523859048527,
      "grad_norm": 0.43233203887939453,
      "learning_rate": 1.9110926710700106e-05,
      "loss": 0.3134,
      "step": 7490
    },
    {
      "epoch": 0.13532886991817072,
      "grad_norm": 0.6694607138633728,
      "learning_rate": 1.910975724050821e-05,
      "loss": 0.3164,
      "step": 7495
    },
    {
      "epoch": 0.13541914935107144,
      "grad_norm": 0.5502925515174866,
      "learning_rate": 1.9108587037502573e-05,
      "loss": 0.3057,
      "step": 7500
    },
    {
      "epoch": 0.13550942878397215,
      "grad_norm": 0.508446216583252,
      "learning_rate": 1.9107416101777337e-05,
      "loss": 0.3227,
      "step": 7505
    },
    {
      "epoch": 0.13559970821687287,
      "grad_norm": 0.4535052180290222,
      "learning_rate": 1.9106244433426692e-05,
      "loss": 0.2741,
      "step": 7510
    },
    {
      "epoch": 0.13568998764977358,
      "grad_norm": 0.4868471324443817,
      "learning_rate": 1.910507203254489e-05,
      "loss": 0.2351,
      "step": 7515
    },
    {
      "epoch": 0.1357802670826743,
      "grad_norm": 0.9161044955253601,
      "learning_rate": 1.9103898899226246e-05,
      "loss": 0.3767,
      "step": 7520
    },
    {
      "epoch": 0.135870546515575,
      "grad_norm": 0.5390794277191162,
      "learning_rate": 1.9102725033565125e-05,
      "loss": 0.3531,
      "step": 7525
    },
    {
      "epoch": 0.13596082594847572,
      "grad_norm": 0.42803600430488586,
      "learning_rate": 1.9101550435655964e-05,
      "loss": 0.4034,
      "step": 7530
    },
    {
      "epoch": 0.13605110538137644,
      "grad_norm": 0.4380834102630615,
      "learning_rate": 1.9100375105593237e-05,
      "loss": 0.3925,
      "step": 7535
    },
    {
      "epoch": 0.13614138481427715,
      "grad_norm": 0.53178870677948,
      "learning_rate": 1.9099199043471505e-05,
      "loss": 0.26,
      "step": 7540
    },
    {
      "epoch": 0.13623166424717786,
      "grad_norm": 0.4513871669769287,
      "learning_rate": 1.9098022249385364e-05,
      "loss": 0.2941,
      "step": 7545
    },
    {
      "epoch": 0.13632194368007858,
      "grad_norm": 0.701305091381073,
      "learning_rate": 1.9096844723429485e-05,
      "loss": 0.49,
      "step": 7550
    },
    {
      "epoch": 0.1364122231129793,
      "grad_norm": 0.37477660179138184,
      "learning_rate": 1.9095666465698587e-05,
      "loss": 0.3307,
      "step": 7555
    },
    {
      "epoch": 0.13650250254588,
      "grad_norm": 0.39110666513442993,
      "learning_rate": 1.9094487476287452e-05,
      "loss": 0.3496,
      "step": 7560
    },
    {
      "epoch": 0.13659278197878072,
      "grad_norm": 0.3477240800857544,
      "learning_rate": 1.9093307755290925e-05,
      "loss": 0.4175,
      "step": 7565
    },
    {
      "epoch": 0.13668306141168143,
      "grad_norm": 0.4026183784008026,
      "learning_rate": 1.90921273028039e-05,
      "loss": 0.25,
      "step": 7570
    },
    {
      "epoch": 0.13677334084458215,
      "grad_norm": 0.4791066348552704,
      "learning_rate": 1.9090946118921345e-05,
      "loss": 0.2891,
      "step": 7575
    },
    {
      "epoch": 0.13686362027748286,
      "grad_norm": 0.33462172746658325,
      "learning_rate": 1.908976420373827e-05,
      "loss": 0.4292,
      "step": 7580
    },
    {
      "epoch": 0.13695389971038358,
      "grad_norm": 0.42122966051101685,
      "learning_rate": 1.9088581557349753e-05,
      "loss": 0.3056,
      "step": 7585
    },
    {
      "epoch": 0.1370441791432843,
      "grad_norm": 0.4601115882396698,
      "learning_rate": 1.9087398179850933e-05,
      "loss": 0.347,
      "step": 7590
    },
    {
      "epoch": 0.137134458576185,
      "grad_norm": 0.6133005619049072,
      "learning_rate": 1.9086214071336998e-05,
      "loss": 0.3491,
      "step": 7595
    },
    {
      "epoch": 0.13722473800908572,
      "grad_norm": 0.6506732702255249,
      "learning_rate": 1.9085029231903205e-05,
      "loss": 0.4802,
      "step": 7600
    },
    {
      "epoch": 0.13731501744198643,
      "grad_norm": 0.4467129707336426,
      "learning_rate": 1.9083843661644865e-05,
      "loss": 0.343,
      "step": 7605
    },
    {
      "epoch": 0.13740529687488715,
      "grad_norm": 0.3698214292526245,
      "learning_rate": 1.9082657360657347e-05,
      "loss": 0.252,
      "step": 7610
    },
    {
      "epoch": 0.13749557630778786,
      "grad_norm": 0.35796207189559937,
      "learning_rate": 1.9081470329036086e-05,
      "loss": 0.3932,
      "step": 7615
    },
    {
      "epoch": 0.13758585574068857,
      "grad_norm": 0.5271620154380798,
      "learning_rate": 1.908028256687656e-05,
      "loss": 0.3614,
      "step": 7620
    },
    {
      "epoch": 0.1376761351735893,
      "grad_norm": 0.371025949716568,
      "learning_rate": 1.9079094074274325e-05,
      "loss": 0.2348,
      "step": 7625
    },
    {
      "epoch": 0.13776641460649,
      "grad_norm": 0.5057605504989624,
      "learning_rate": 1.9077904851324982e-05,
      "loss": 0.3497,
      "step": 7630
    },
    {
      "epoch": 0.13785669403939071,
      "grad_norm": 0.4116515517234802,
      "learning_rate": 1.9076714898124197e-05,
      "loss": 0.2773,
      "step": 7635
    },
    {
      "epoch": 0.13794697347229143,
      "grad_norm": 0.5498316287994385,
      "learning_rate": 1.9075524214767694e-05,
      "loss": 0.2589,
      "step": 7640
    },
    {
      "epoch": 0.13803725290519214,
      "grad_norm": 0.6358838081359863,
      "learning_rate": 1.9074332801351253e-05,
      "loss": 0.4033,
      "step": 7645
    },
    {
      "epoch": 0.13812753233809286,
      "grad_norm": 0.33449217677116394,
      "learning_rate": 1.9073140657970712e-05,
      "loss": 0.3144,
      "step": 7650
    },
    {
      "epoch": 0.13821781177099357,
      "grad_norm": 0.4676278233528137,
      "learning_rate": 1.9071947784721976e-05,
      "loss": 0.3036,
      "step": 7655
    },
    {
      "epoch": 0.13830809120389428,
      "grad_norm": 0.6691996455192566,
      "learning_rate": 1.9070754181701e-05,
      "loss": 0.2843,
      "step": 7660
    },
    {
      "epoch": 0.138398370636795,
      "grad_norm": 0.3309561014175415,
      "learning_rate": 1.9069559849003795e-05,
      "loss": 0.2363,
      "step": 7665
    },
    {
      "epoch": 0.1384886500696957,
      "grad_norm": 0.5446121692657471,
      "learning_rate": 1.906836478672645e-05,
      "loss": 0.2942,
      "step": 7670
    },
    {
      "epoch": 0.13857892950259643,
      "grad_norm": 0.4546838700771332,
      "learning_rate": 1.9067168994965088e-05,
      "loss": 0.2817,
      "step": 7675
    },
    {
      "epoch": 0.13866920893549714,
      "grad_norm": 0.6302188634872437,
      "learning_rate": 1.9065972473815906e-05,
      "loss": 0.4446,
      "step": 7680
    },
    {
      "epoch": 0.13875948836839785,
      "grad_norm": 0.5918726921081543,
      "learning_rate": 1.9064775223375154e-05,
      "loss": 0.3337,
      "step": 7685
    },
    {
      "epoch": 0.13884976780129857,
      "grad_norm": 0.710841178894043,
      "learning_rate": 1.9063577243739143e-05,
      "loss": 0.3233,
      "step": 7690
    },
    {
      "epoch": 0.13894004723419928,
      "grad_norm": 0.5338574647903442,
      "learning_rate": 1.9062378535004236e-05,
      "loss": 0.2823,
      "step": 7695
    },
    {
      "epoch": 0.1390303266671,
      "grad_norm": 0.6912916898727417,
      "learning_rate": 1.9061179097266876e-05,
      "loss": 0.3708,
      "step": 7700
    },
    {
      "epoch": 0.1391206061000007,
      "grad_norm": 0.5621781945228577,
      "learning_rate": 1.905997893062353e-05,
      "loss": 0.2592,
      "step": 7705
    },
    {
      "epoch": 0.13921088553290142,
      "grad_norm": 0.47019150853157043,
      "learning_rate": 1.9058778035170754e-05,
      "loss": 0.3395,
      "step": 7710
    },
    {
      "epoch": 0.13930116496580214,
      "grad_norm": 0.35848888754844666,
      "learning_rate": 1.9057576411005153e-05,
      "loss": 0.2767,
      "step": 7715
    },
    {
      "epoch": 0.13939144439870285,
      "grad_norm": 0.4225088953971863,
      "learning_rate": 1.905637405822338e-05,
      "loss": 0.3292,
      "step": 7720
    },
    {
      "epoch": 0.13948172383160357,
      "grad_norm": 0.46198078989982605,
      "learning_rate": 1.9055170976922158e-05,
      "loss": 0.3877,
      "step": 7725
    },
    {
      "epoch": 0.1395720032645043,
      "grad_norm": 0.5311920046806335,
      "learning_rate": 1.9053967167198274e-05,
      "loss": 0.2245,
      "step": 7730
    },
    {
      "epoch": 0.13966228269740502,
      "grad_norm": 0.34501591324806213,
      "learning_rate": 1.905276262914856e-05,
      "loss": 0.3565,
      "step": 7735
    },
    {
      "epoch": 0.13975256213030574,
      "grad_norm": 0.42606121301651,
      "learning_rate": 1.905155736286991e-05,
      "loss": 0.3168,
      "step": 7740
    },
    {
      "epoch": 0.13984284156320645,
      "grad_norm": 0.599910318851471,
      "learning_rate": 1.905035136845928e-05,
      "loss": 0.2913,
      "step": 7745
    },
    {
      "epoch": 0.13993312099610716,
      "grad_norm": 0.3511393368244171,
      "learning_rate": 1.9049144646013686e-05,
      "loss": 0.3479,
      "step": 7750
    },
    {
      "epoch": 0.14002340042900788,
      "grad_norm": 0.4650457799434662,
      "learning_rate": 1.90479371956302e-05,
      "loss": 0.3055,
      "step": 7755
    },
    {
      "epoch": 0.1401136798619086,
      "grad_norm": 0.5261766910552979,
      "learning_rate": 1.904672901740595e-05,
      "loss": 0.3334,
      "step": 7760
    },
    {
      "epoch": 0.1402039592948093,
      "grad_norm": 0.4383196532726288,
      "learning_rate": 1.904552011143813e-05,
      "loss": 0.2858,
      "step": 7765
    },
    {
      "epoch": 0.14029423872771002,
      "grad_norm": 0.5381696820259094,
      "learning_rate": 1.9044310477823978e-05,
      "loss": 0.244,
      "step": 7770
    },
    {
      "epoch": 0.14038451816061073,
      "grad_norm": 0.411138117313385,
      "learning_rate": 1.9043100116660812e-05,
      "loss": 0.258,
      "step": 7775
    },
    {
      "epoch": 0.14047479759351145,
      "grad_norm": 0.5190327167510986,
      "learning_rate": 1.9041889028045987e-05,
      "loss": 0.3412,
      "step": 7780
    },
    {
      "epoch": 0.14056507702641216,
      "grad_norm": 0.7907403111457825,
      "learning_rate": 1.9040677212076937e-05,
      "loss": 0.3496,
      "step": 7785
    },
    {
      "epoch": 0.14065535645931287,
      "grad_norm": 0.36397334933280945,
      "learning_rate": 1.903946466885113e-05,
      "loss": 0.2988,
      "step": 7790
    },
    {
      "epoch": 0.1407456358922136,
      "grad_norm": 0.41145774722099304,
      "learning_rate": 1.9038251398466118e-05,
      "loss": 0.3286,
      "step": 7795
    },
    {
      "epoch": 0.1408359153251143,
      "grad_norm": 0.6124861836433411,
      "learning_rate": 1.9037037401019494e-05,
      "loss": 0.3634,
      "step": 7800
    },
    {
      "epoch": 0.14092619475801502,
      "grad_norm": 0.3801920711994171,
      "learning_rate": 1.903582267660892e-05,
      "loss": 0.2512,
      "step": 7805
    },
    {
      "epoch": 0.14101647419091573,
      "grad_norm": 0.3616124987602234,
      "learning_rate": 1.9034607225332104e-05,
      "loss": 0.3029,
      "step": 7810
    },
    {
      "epoch": 0.14110675362381644,
      "grad_norm": 0.6739056706428528,
      "learning_rate": 1.9033391047286826e-05,
      "loss": 0.3886,
      "step": 7815
    },
    {
      "epoch": 0.14119703305671716,
      "grad_norm": 0.5704717040061951,
      "learning_rate": 1.9032174142570918e-05,
      "loss": 0.2179,
      "step": 7820
    },
    {
      "epoch": 0.14128731248961787,
      "grad_norm": 0.36887940764427185,
      "learning_rate": 1.903095651128227e-05,
      "loss": 0.3024,
      "step": 7825
    },
    {
      "epoch": 0.1413775919225186,
      "grad_norm": 0.413861483335495,
      "learning_rate": 1.9029738153518832e-05,
      "loss": 0.2548,
      "step": 7830
    },
    {
      "epoch": 0.1414678713554193,
      "grad_norm": 0.4529048502445221,
      "learning_rate": 1.9028519069378613e-05,
      "loss": 0.4083,
      "step": 7835
    },
    {
      "epoch": 0.14155815078832001,
      "grad_norm": 0.58521968126297,
      "learning_rate": 1.902729925895968e-05,
      "loss": 0.3763,
      "step": 7840
    },
    {
      "epoch": 0.14164843022122073,
      "grad_norm": 0.5197453498840332,
      "learning_rate": 1.9026078722360154e-05,
      "loss": 0.3415,
      "step": 7845
    },
    {
      "epoch": 0.14173870965412144,
      "grad_norm": 0.3636833727359772,
      "learning_rate": 1.9024857459678218e-05,
      "loss": 0.2997,
      "step": 7850
    },
    {
      "epoch": 0.14182898908702216,
      "grad_norm": 0.2941308915615082,
      "learning_rate": 1.902363547101212e-05,
      "loss": 0.412,
      "step": 7855
    },
    {
      "epoch": 0.14191926851992287,
      "grad_norm": 0.6289583444595337,
      "learning_rate": 1.9022412756460157e-05,
      "loss": 0.3075,
      "step": 7860
    },
    {
      "epoch": 0.14200954795282358,
      "grad_norm": 0.45096611976623535,
      "learning_rate": 1.902118931612069e-05,
      "loss": 0.3337,
      "step": 7865
    },
    {
      "epoch": 0.1420998273857243,
      "grad_norm": 0.5643873810768127,
      "learning_rate": 1.9019965150092122e-05,
      "loss": 0.3478,
      "step": 7870
    },
    {
      "epoch": 0.142190106818625,
      "grad_norm": 0.4536270499229431,
      "learning_rate": 1.9018740258472948e-05,
      "loss": 0.3698,
      "step": 7875
    },
    {
      "epoch": 0.14228038625152573,
      "grad_norm": 0.4536241590976715,
      "learning_rate": 1.9017514641361693e-05,
      "loss": 0.2808,
      "step": 7880
    },
    {
      "epoch": 0.14237066568442644,
      "grad_norm": 0.4845343232154846,
      "learning_rate": 1.9016288298856948e-05,
      "loss": 0.2832,
      "step": 7885
    },
    {
      "epoch": 0.14246094511732715,
      "grad_norm": 0.3254559338092804,
      "learning_rate": 1.9015061231057358e-05,
      "loss": 0.4254,
      "step": 7890
    },
    {
      "epoch": 0.14255122455022787,
      "grad_norm": 0.43062731623649597,
      "learning_rate": 1.9013833438061647e-05,
      "loss": 0.3528,
      "step": 7895
    },
    {
      "epoch": 0.14264150398312858,
      "grad_norm": 0.4750933349132538,
      "learning_rate": 1.9012604919968568e-05,
      "loss": 0.3383,
      "step": 7900
    },
    {
      "epoch": 0.1427317834160293,
      "grad_norm": 0.538891077041626,
      "learning_rate": 1.901137567687695e-05,
      "loss": 0.2925,
      "step": 7905
    },
    {
      "epoch": 0.14282206284893,
      "grad_norm": 0.5271896123886108,
      "learning_rate": 1.9010145708885678e-05,
      "loss": 0.3562,
      "step": 7910
    },
    {
      "epoch": 0.14291234228183072,
      "grad_norm": 0.3939216434955597,
      "learning_rate": 1.9008915016093695e-05,
      "loss": 0.2797,
      "step": 7915
    },
    {
      "epoch": 0.14300262171473144,
      "grad_norm": 1.2909743785858154,
      "learning_rate": 1.90076835986e-05,
      "loss": 0.3289,
      "step": 7920
    },
    {
      "epoch": 0.14309290114763215,
      "grad_norm": 0.6015067100524902,
      "learning_rate": 1.900645145650365e-05,
      "loss": 0.3242,
      "step": 7925
    },
    {
      "epoch": 0.14318318058053287,
      "grad_norm": 0.38563331961631775,
      "learning_rate": 1.9005218589903768e-05,
      "loss": 0.3693,
      "step": 7930
    },
    {
      "epoch": 0.14327346001343358,
      "grad_norm": 0.2794496417045593,
      "learning_rate": 1.9003984998899523e-05,
      "loss": 0.3101,
      "step": 7935
    },
    {
      "epoch": 0.1433637394463343,
      "grad_norm": 0.32866206765174866,
      "learning_rate": 1.900275068359015e-05,
      "loss": 0.262,
      "step": 7940
    },
    {
      "epoch": 0.143454018879235,
      "grad_norm": 0.44161251187324524,
      "learning_rate": 1.900151564407494e-05,
      "loss": 0.2301,
      "step": 7945
    },
    {
      "epoch": 0.14354429831213572,
      "grad_norm": 0.4842809736728668,
      "learning_rate": 1.9000279880453246e-05,
      "loss": 0.3021,
      "step": 7950
    },
    {
      "epoch": 0.14363457774503643,
      "grad_norm": 0.35649093985557556,
      "learning_rate": 1.899904339282447e-05,
      "loss": 0.3484,
      "step": 7955
    },
    {
      "epoch": 0.14372485717793715,
      "grad_norm": 0.44895628094673157,
      "learning_rate": 1.8997806181288086e-05,
      "loss": 0.3674,
      "step": 7960
    },
    {
      "epoch": 0.14381513661083786,
      "grad_norm": 0.5911328196525574,
      "learning_rate": 1.8996568245943612e-05,
      "loss": 0.3428,
      "step": 7965
    },
    {
      "epoch": 0.14390541604373858,
      "grad_norm": 0.4658902883529663,
      "learning_rate": 1.8995329586890637e-05,
      "loss": 0.2989,
      "step": 7970
    },
    {
      "epoch": 0.1439956954766393,
      "grad_norm": 0.2901282012462616,
      "learning_rate": 1.8994090204228796e-05,
      "loss": 0.3787,
      "step": 7975
    },
    {
      "epoch": 0.14408597490954,
      "grad_norm": 0.44001367688179016,
      "learning_rate": 1.8992850098057795e-05,
      "loss": 0.2329,
      "step": 7980
    },
    {
      "epoch": 0.14417625434244072,
      "grad_norm": 0.9763175845146179,
      "learning_rate": 1.8991609268477383e-05,
      "loss": 0.4526,
      "step": 7985
    },
    {
      "epoch": 0.14426653377534143,
      "grad_norm": 0.263467401266098,
      "learning_rate": 1.8990367715587383e-05,
      "loss": 0.2223,
      "step": 7990
    },
    {
      "epoch": 0.14435681320824215,
      "grad_norm": 0.723092257976532,
      "learning_rate": 1.898912543948767e-05,
      "loss": 0.3509,
      "step": 7995
    },
    {
      "epoch": 0.14444709264114286,
      "grad_norm": 0.4045044481754303,
      "learning_rate": 1.898788244027817e-05,
      "loss": 0.3391,
      "step": 8000
    },
    {
      "epoch": 0.14453737207404357,
      "grad_norm": 0.43147021532058716,
      "learning_rate": 1.8986638718058872e-05,
      "loss": 0.2449,
      "step": 8005
    },
    {
      "epoch": 0.1446276515069443,
      "grad_norm": 0.6426684856414795,
      "learning_rate": 1.8985394272929835e-05,
      "loss": 0.3077,
      "step": 8010
    },
    {
      "epoch": 0.144717930939845,
      "grad_norm": 0.4371914863586426,
      "learning_rate": 1.8984149104991154e-05,
      "loss": 0.2966,
      "step": 8015
    },
    {
      "epoch": 0.14480821037274572,
      "grad_norm": 0.6035380363464355,
      "learning_rate": 1.8982903214343e-05,
      "loss": 0.3514,
      "step": 8020
    },
    {
      "epoch": 0.14489848980564643,
      "grad_norm": 0.6385045051574707,
      "learning_rate": 1.8981656601085596e-05,
      "loss": 0.2887,
      "step": 8025
    },
    {
      "epoch": 0.14498876923854714,
      "grad_norm": 0.3629067838191986,
      "learning_rate": 1.898040926531922e-05,
      "loss": 0.3372,
      "step": 8030
    },
    {
      "epoch": 0.14507904867144786,
      "grad_norm": 0.48646870255470276,
      "learning_rate": 1.897916120714421e-05,
      "loss": 0.4035,
      "step": 8035
    },
    {
      "epoch": 0.14516932810434857,
      "grad_norm": 0.3212614357471466,
      "learning_rate": 1.8977912426660968e-05,
      "loss": 0.226,
      "step": 8040
    },
    {
      "epoch": 0.14525960753724929,
      "grad_norm": 0.49237778782844543,
      "learning_rate": 1.897666292396995e-05,
      "loss": 0.3552,
      "step": 8045
    },
    {
      "epoch": 0.14534988697015,
      "grad_norm": 0.46724480390548706,
      "learning_rate": 1.8975412699171665e-05,
      "loss": 0.3779,
      "step": 8050
    },
    {
      "epoch": 0.1454401664030507,
      "grad_norm": 0.4620061218738556,
      "learning_rate": 1.8974161752366684e-05,
      "loss": 0.3916,
      "step": 8055
    },
    {
      "epoch": 0.14553044583595143,
      "grad_norm": 0.35850104689598083,
      "learning_rate": 1.897291008365564e-05,
      "loss": 0.2794,
      "step": 8060
    },
    {
      "epoch": 0.14562072526885214,
      "grad_norm": 0.5722913146018982,
      "learning_rate": 1.897165769313922e-05,
      "loss": 0.367,
      "step": 8065
    },
    {
      "epoch": 0.14571100470175286,
      "grad_norm": 0.4498758912086487,
      "learning_rate": 1.8970404580918165e-05,
      "loss": 0.3568,
      "step": 8070
    },
    {
      "epoch": 0.14580128413465357,
      "grad_norm": 0.4813031852245331,
      "learning_rate": 1.8969150747093286e-05,
      "loss": 0.2986,
      "step": 8075
    },
    {
      "epoch": 0.14589156356755428,
      "grad_norm": 0.3807981312274933,
      "learning_rate": 1.896789619176544e-05,
      "loss": 0.3528,
      "step": 8080
    },
    {
      "epoch": 0.145981843000455,
      "grad_norm": 0.6421025395393372,
      "learning_rate": 1.896664091503555e-05,
      "loss": 0.2872,
      "step": 8085
    },
    {
      "epoch": 0.1460721224333557,
      "grad_norm": 0.3946101665496826,
      "learning_rate": 1.8965384917004587e-05,
      "loss": 0.3196,
      "step": 8090
    },
    {
      "epoch": 0.14616240186625643,
      "grad_norm": 0.5006182789802551,
      "learning_rate": 1.8964128197773598e-05,
      "loss": 0.3334,
      "step": 8095
    },
    {
      "epoch": 0.14625268129915714,
      "grad_norm": 0.5575314164161682,
      "learning_rate": 1.8962870757443667e-05,
      "loss": 0.437,
      "step": 8100
    },
    {
      "epoch": 0.14634296073205785,
      "grad_norm": 0.33846473693847656,
      "learning_rate": 1.896161259611595e-05,
      "loss": 0.3118,
      "step": 8105
    },
    {
      "epoch": 0.14643324016495857,
      "grad_norm": 0.36873605847358704,
      "learning_rate": 1.896035371389166e-05,
      "loss": 0.2947,
      "step": 8110
    },
    {
      "epoch": 0.14652351959785928,
      "grad_norm": 0.39913228154182434,
      "learning_rate": 1.8959094110872062e-05,
      "loss": 0.3383,
      "step": 8115
    },
    {
      "epoch": 0.14661379903076,
      "grad_norm": 0.4663682281970978,
      "learning_rate": 1.895783378715848e-05,
      "loss": 0.243,
      "step": 8120
    },
    {
      "epoch": 0.14670407846366074,
      "grad_norm": 0.4984837472438812,
      "learning_rate": 1.89565727428523e-05,
      "loss": 0.3388,
      "step": 8125
    },
    {
      "epoch": 0.14679435789656145,
      "grad_norm": 0.3878990411758423,
      "learning_rate": 1.8955310978054963e-05,
      "loss": 0.2641,
      "step": 8130
    },
    {
      "epoch": 0.14688463732946216,
      "grad_norm": 0.46170365810394287,
      "learning_rate": 1.895404849286797e-05,
      "loss": 0.3186,
      "step": 8135
    },
    {
      "epoch": 0.14697491676236288,
      "grad_norm": 0.46712246537208557,
      "learning_rate": 1.8952785287392876e-05,
      "loss": 0.2991,
      "step": 8140
    },
    {
      "epoch": 0.1470651961952636,
      "grad_norm": 0.4796598553657532,
      "learning_rate": 1.89515213617313e-05,
      "loss": 0.3449,
      "step": 8145
    },
    {
      "epoch": 0.1471554756281643,
      "grad_norm": 0.28505873680114746,
      "learning_rate": 1.8950256715984912e-05,
      "loss": 0.2601,
      "step": 8150
    },
    {
      "epoch": 0.14724575506106502,
      "grad_norm": 0.5526983141899109,
      "learning_rate": 1.894899135025545e-05,
      "loss": 0.2915,
      "step": 8155
    },
    {
      "epoch": 0.14733603449396573,
      "grad_norm": 0.35364168882369995,
      "learning_rate": 1.8947725264644697e-05,
      "loss": 0.2816,
      "step": 8160
    },
    {
      "epoch": 0.14742631392686645,
      "grad_norm": 0.48852768540382385,
      "learning_rate": 1.8946458459254505e-05,
      "loss": 0.3107,
      "step": 8165
    },
    {
      "epoch": 0.14751659335976716,
      "grad_norm": 0.5498285889625549,
      "learning_rate": 1.894519093418677e-05,
      "loss": 0.3314,
      "step": 8170
    },
    {
      "epoch": 0.14760687279266788,
      "grad_norm": 0.332851380109787,
      "learning_rate": 1.894392268954347e-05,
      "loss": 0.3827,
      "step": 8175
    },
    {
      "epoch": 0.1476971522255686,
      "grad_norm": 0.38158583641052246,
      "learning_rate": 1.8942653725426618e-05,
      "loss": 0.3227,
      "step": 8180
    },
    {
      "epoch": 0.1477874316584693,
      "grad_norm": 0.4075416326522827,
      "learning_rate": 1.894138404193829e-05,
      "loss": 0.4028,
      "step": 8185
    },
    {
      "epoch": 0.14787771109137002,
      "grad_norm": 0.5352085828781128,
      "learning_rate": 1.8940113639180627e-05,
      "loss": 0.2752,
      "step": 8190
    },
    {
      "epoch": 0.14796799052427073,
      "grad_norm": 0.4605783224105835,
      "learning_rate": 1.893884251725582e-05,
      "loss": 0.4026,
      "step": 8195
    },
    {
      "epoch": 0.14805826995717145,
      "grad_norm": 0.31095898151397705,
      "learning_rate": 1.893757067626613e-05,
      "loss": 0.305,
      "step": 8200
    },
    {
      "epoch": 0.14814854939007216,
      "grad_norm": 0.44574591517448425,
      "learning_rate": 1.8936298116313854e-05,
      "loss": 0.3021,
      "step": 8205
    },
    {
      "epoch": 0.14823882882297287,
      "grad_norm": 0.40220680832862854,
      "learning_rate": 1.8935024837501373e-05,
      "loss": 0.3497,
      "step": 8210
    },
    {
      "epoch": 0.1483291082558736,
      "grad_norm": 0.3617036044597626,
      "learning_rate": 1.8933750839931106e-05,
      "loss": 0.3125,
      "step": 8215
    },
    {
      "epoch": 0.1484193876887743,
      "grad_norm": 0.37707874178886414,
      "learning_rate": 1.893247612370554e-05,
      "loss": 0.3204,
      "step": 8220
    },
    {
      "epoch": 0.14850966712167502,
      "grad_norm": 0.3177054524421692,
      "learning_rate": 1.8931200688927213e-05,
      "loss": 0.2895,
      "step": 8225
    },
    {
      "epoch": 0.14859994655457573,
      "grad_norm": 0.32511448860168457,
      "learning_rate": 1.892992453569873e-05,
      "loss": 0.3439,
      "step": 8230
    },
    {
      "epoch": 0.14869022598747644,
      "grad_norm": 0.5057108402252197,
      "learning_rate": 1.892864766412274e-05,
      "loss": 0.431,
      "step": 8235
    },
    {
      "epoch": 0.14878050542037716,
      "grad_norm": 0.4230477809906006,
      "learning_rate": 1.8927370074301966e-05,
      "loss": 0.3074,
      "step": 8240
    },
    {
      "epoch": 0.14887078485327787,
      "grad_norm": 0.5735442042350769,
      "learning_rate": 1.8926091766339178e-05,
      "loss": 0.2622,
      "step": 8245
    },
    {
      "epoch": 0.14896106428617858,
      "grad_norm": 0.5898981690406799,
      "learning_rate": 1.8924812740337206e-05,
      "loss": 0.3139,
      "step": 8250
    },
    {
      "epoch": 0.1490513437190793,
      "grad_norm": 0.21331223845481873,
      "learning_rate": 1.8923532996398937e-05,
      "loss": 0.3723,
      "step": 8255
    },
    {
      "epoch": 0.14914162315198,
      "grad_norm": 0.409196674823761,
      "learning_rate": 1.892225253462732e-05,
      "loss": 0.3487,
      "step": 8260
    },
    {
      "epoch": 0.14923190258488073,
      "grad_norm": 0.3333122432231903,
      "learning_rate": 1.892097135512536e-05,
      "loss": 0.3065,
      "step": 8265
    },
    {
      "epoch": 0.14932218201778144,
      "grad_norm": 0.4181888997554779,
      "learning_rate": 1.8919689457996116e-05,
      "loss": 0.2028,
      "step": 8270
    },
    {
      "epoch": 0.14941246145068215,
      "grad_norm": 0.5751470923423767,
      "learning_rate": 1.8918406843342704e-05,
      "loss": 0.3421,
      "step": 8275
    },
    {
      "epoch": 0.14950274088358287,
      "grad_norm": 0.5706935524940491,
      "learning_rate": 1.8917123511268306e-05,
      "loss": 0.3903,
      "step": 8280
    },
    {
      "epoch": 0.14959302031648358,
      "grad_norm": 0.8409367203712463,
      "learning_rate": 1.8915839461876156e-05,
      "loss": 0.3937,
      "step": 8285
    },
    {
      "epoch": 0.1496832997493843,
      "grad_norm": 0.44183939695358276,
      "learning_rate": 1.8914554695269543e-05,
      "loss": 0.3441,
      "step": 8290
    },
    {
      "epoch": 0.149773579182285,
      "grad_norm": 0.4171811044216156,
      "learning_rate": 1.8913269211551824e-05,
      "loss": 0.284,
      "step": 8295
    },
    {
      "epoch": 0.14986385861518572,
      "grad_norm": 0.3072311580181122,
      "learning_rate": 1.89119830108264e-05,
      "loss": 0.302,
      "step": 8300
    },
    {
      "epoch": 0.14995413804808644,
      "grad_norm": 0.3270736038684845,
      "learning_rate": 1.8910696093196742e-05,
      "loss": 0.4464,
      "step": 8305
    },
    {
      "epoch": 0.15004441748098715,
      "grad_norm": 0.43450701236724854,
      "learning_rate": 1.8909408458766367e-05,
      "loss": 0.3566,
      "step": 8310
    },
    {
      "epoch": 0.15013469691388787,
      "grad_norm": 0.3389006555080414,
      "learning_rate": 1.890812010763886e-05,
      "loss": 0.2731,
      "step": 8315
    },
    {
      "epoch": 0.15022497634678858,
      "grad_norm": 0.6542930603027344,
      "learning_rate": 1.8906831039917862e-05,
      "loss": 0.3361,
      "step": 8320
    },
    {
      "epoch": 0.1503152557796893,
      "grad_norm": 0.3040352463722229,
      "learning_rate": 1.890554125570706e-05,
      "loss": 0.2567,
      "step": 8325
    },
    {
      "epoch": 0.15040553521259,
      "grad_norm": 0.5242255330085754,
      "learning_rate": 1.8904250755110216e-05,
      "loss": 0.2654,
      "step": 8330
    },
    {
      "epoch": 0.15049581464549072,
      "grad_norm": 0.4824141263961792,
      "learning_rate": 1.890295953823114e-05,
      "loss": 0.3723,
      "step": 8335
    },
    {
      "epoch": 0.15058609407839144,
      "grad_norm": 0.4082823097705841,
      "learning_rate": 1.89016676051737e-05,
      "loss": 0.2324,
      "step": 8340
    },
    {
      "epoch": 0.15067637351129215,
      "grad_norm": 0.3655155897140503,
      "learning_rate": 1.890037495604182e-05,
      "loss": 0.3749,
      "step": 8345
    },
    {
      "epoch": 0.15076665294419286,
      "grad_norm": 0.5974289774894714,
      "learning_rate": 1.889908159093949e-05,
      "loss": 0.316,
      "step": 8350
    },
    {
      "epoch": 0.15085693237709358,
      "grad_norm": 0.3974992632865906,
      "learning_rate": 1.8897787509970746e-05,
      "loss": 0.2754,
      "step": 8355
    },
    {
      "epoch": 0.1509472118099943,
      "grad_norm": 0.5423205494880676,
      "learning_rate": 1.889649271323969e-05,
      "loss": 0.3706,
      "step": 8360
    },
    {
      "epoch": 0.151037491242895,
      "grad_norm": 0.6967107057571411,
      "learning_rate": 1.8895197200850482e-05,
      "loss": 0.2848,
      "step": 8365
    },
    {
      "epoch": 0.15112777067579572,
      "grad_norm": 0.5599609017372131,
      "learning_rate": 1.889390097290733e-05,
      "loss": 0.2757,
      "step": 8370
    },
    {
      "epoch": 0.15121805010869643,
      "grad_norm": 0.4817441999912262,
      "learning_rate": 1.889260402951451e-05,
      "loss": 0.3057,
      "step": 8375
    },
    {
      "epoch": 0.15130832954159715,
      "grad_norm": 0.4420287609100342,
      "learning_rate": 1.8891306370776353e-05,
      "loss": 0.303,
      "step": 8380
    },
    {
      "epoch": 0.15139860897449786,
      "grad_norm": 0.3668675422668457,
      "learning_rate": 1.8890007996797245e-05,
      "loss": 0.258,
      "step": 8385
    },
    {
      "epoch": 0.15148888840739858,
      "grad_norm": 0.5196265578269958,
      "learning_rate": 1.888870890768163e-05,
      "loss": 0.2724,
      "step": 8390
    },
    {
      "epoch": 0.1515791678402993,
      "grad_norm": 0.3072160482406616,
      "learning_rate": 1.888740910353401e-05,
      "loss": 0.3446,
      "step": 8395
    },
    {
      "epoch": 0.1516694472732,
      "grad_norm": 0.5588591694831848,
      "learning_rate": 1.8886108584458945e-05,
      "loss": 0.3808,
      "step": 8400
    },
    {
      "epoch": 0.15175972670610072,
      "grad_norm": 0.32789307832717896,
      "learning_rate": 1.8884807350561053e-05,
      "loss": 0.2752,
      "step": 8405
    },
    {
      "epoch": 0.15185000613900143,
      "grad_norm": 1.054311990737915,
      "learning_rate": 1.8883505401945008e-05,
      "loss": 0.4017,
      "step": 8410
    },
    {
      "epoch": 0.15194028557190215,
      "grad_norm": 0.5561165809631348,
      "learning_rate": 1.8882202738715542e-05,
      "loss": 0.3239,
      "step": 8415
    },
    {
      "epoch": 0.15203056500480286,
      "grad_norm": 0.3699195086956024,
      "learning_rate": 1.8880899360977447e-05,
      "loss": 0.2967,
      "step": 8420
    },
    {
      "epoch": 0.15212084443770357,
      "grad_norm": 0.3856097161769867,
      "learning_rate": 1.887959526883557e-05,
      "loss": 0.2988,
      "step": 8425
    },
    {
      "epoch": 0.1522111238706043,
      "grad_norm": 0.33342596888542175,
      "learning_rate": 1.887829046239481e-05,
      "loss": 0.4531,
      "step": 8430
    },
    {
      "epoch": 0.152301403303505,
      "grad_norm": 0.41385820508003235,
      "learning_rate": 1.8876984941760137e-05,
      "loss": 0.2464,
      "step": 8435
    },
    {
      "epoch": 0.15239168273640571,
      "grad_norm": 0.4393940567970276,
      "learning_rate": 1.8875678707036568e-05,
      "loss": 0.3006,
      "step": 8440
    },
    {
      "epoch": 0.15248196216930643,
      "grad_norm": 0.3313063681125641,
      "learning_rate": 1.8874371758329178e-05,
      "loss": 0.3243,
      "step": 8445
    },
    {
      "epoch": 0.15257224160220714,
      "grad_norm": 0.348006010055542,
      "learning_rate": 1.8873064095743104e-05,
      "loss": 0.3317,
      "step": 8450
    },
    {
      "epoch": 0.15266252103510786,
      "grad_norm": 0.31291893124580383,
      "learning_rate": 1.8871755719383535e-05,
      "loss": 0.254,
      "step": 8455
    },
    {
      "epoch": 0.15275280046800857,
      "grad_norm": 0.44833308458328247,
      "learning_rate": 1.8870446629355723e-05,
      "loss": 0.2834,
      "step": 8460
    },
    {
      "epoch": 0.15284307990090928,
      "grad_norm": 0.40464499592781067,
      "learning_rate": 1.8869136825764977e-05,
      "loss": 0.2787,
      "step": 8465
    },
    {
      "epoch": 0.15293335933381,
      "grad_norm": 0.4436585009098053,
      "learning_rate": 1.8867826308716655e-05,
      "loss": 0.3241,
      "step": 8470
    },
    {
      "epoch": 0.1530236387667107,
      "grad_norm": 0.44052863121032715,
      "learning_rate": 1.8866515078316183e-05,
      "loss": 0.4283,
      "step": 8475
    },
    {
      "epoch": 0.15311391819961143,
      "grad_norm": 0.40572524070739746,
      "learning_rate": 1.8865203134669038e-05,
      "loss": 0.248,
      "step": 8480
    },
    {
      "epoch": 0.15320419763251214,
      "grad_norm": 0.3816758096218109,
      "learning_rate": 1.8863890477880756e-05,
      "loss": 0.3807,
      "step": 8485
    },
    {
      "epoch": 0.15329447706541285,
      "grad_norm": 0.4932968020439148,
      "learning_rate": 1.8862577108056933e-05,
      "loss": 0.3574,
      "step": 8490
    },
    {
      "epoch": 0.15338475649831357,
      "grad_norm": 0.283809632062912,
      "learning_rate": 1.886126302530322e-05,
      "loss": 0.2696,
      "step": 8495
    },
    {
      "epoch": 0.15347503593121428,
      "grad_norm": 0.6454717516899109,
      "learning_rate": 1.8859948229725324e-05,
      "loss": 0.2792,
      "step": 8500
    },
    {
      "epoch": 0.153565315364115,
      "grad_norm": 0.6588291525840759,
      "learning_rate": 1.885863272142901e-05,
      "loss": 0.2829,
      "step": 8505
    },
    {
      "epoch": 0.1536555947970157,
      "grad_norm": 0.3564765751361847,
      "learning_rate": 1.88573165005201e-05,
      "loss": 0.2167,
      "step": 8510
    },
    {
      "epoch": 0.15374587422991642,
      "grad_norm": 0.40071821212768555,
      "learning_rate": 1.885599956710448e-05,
      "loss": 0.3207,
      "step": 8515
    },
    {
      "epoch": 0.15383615366281717,
      "grad_norm": 0.47020772099494934,
      "learning_rate": 1.885468192128808e-05,
      "loss": 0.2499,
      "step": 8520
    },
    {
      "epoch": 0.15392643309571788,
      "grad_norm": 0.39165768027305603,
      "learning_rate": 1.8853363563176903e-05,
      "loss": 0.3108,
      "step": 8525
    },
    {
      "epoch": 0.1540167125286186,
      "grad_norm": 0.3537646234035492,
      "learning_rate": 1.8852044492876995e-05,
      "loss": 0.3809,
      "step": 8530
    },
    {
      "epoch": 0.1541069919615193,
      "grad_norm": 0.3931910991668701,
      "learning_rate": 1.885072471049447e-05,
      "loss": 0.3819,
      "step": 8535
    },
    {
      "epoch": 0.15419727139442002,
      "grad_norm": 0.5318612456321716,
      "learning_rate": 1.8849404216135493e-05,
      "loss": 0.2379,
      "step": 8540
    },
    {
      "epoch": 0.15428755082732074,
      "grad_norm": 0.46349745988845825,
      "learning_rate": 1.8848083009906288e-05,
      "loss": 0.4473,
      "step": 8545
    },
    {
      "epoch": 0.15437783026022145,
      "grad_norm": 0.718696117401123,
      "learning_rate": 1.8846761091913138e-05,
      "loss": 0.2779,
      "step": 8550
    },
    {
      "epoch": 0.15446810969312216,
      "grad_norm": 0.32716104388237,
      "learning_rate": 1.8845438462262378e-05,
      "loss": 0.2581,
      "step": 8555
    },
    {
      "epoch": 0.15455838912602288,
      "grad_norm": 0.33014896512031555,
      "learning_rate": 1.8844115121060408e-05,
      "loss": 0.3538,
      "step": 8560
    },
    {
      "epoch": 0.1546486685589236,
      "grad_norm": 0.7303894758224487,
      "learning_rate": 1.8842791068413676e-05,
      "loss": 0.3095,
      "step": 8565
    },
    {
      "epoch": 0.1547389479918243,
      "grad_norm": 0.661654531955719,
      "learning_rate": 1.88414663044287e-05,
      "loss": 0.2514,
      "step": 8570
    },
    {
      "epoch": 0.15482922742472502,
      "grad_norm": 0.48193421959877014,
      "learning_rate": 1.884014082921204e-05,
      "loss": 0.351,
      "step": 8575
    },
    {
      "epoch": 0.15491950685762573,
      "grad_norm": 0.6869872808456421,
      "learning_rate": 1.8838814642870328e-05,
      "loss": 0.297,
      "step": 8580
    },
    {
      "epoch": 0.15500978629052645,
      "grad_norm": 0.5479223132133484,
      "learning_rate": 1.883748774551024e-05,
      "loss": 0.2831,
      "step": 8585
    },
    {
      "epoch": 0.15510006572342716,
      "grad_norm": 0.5548229813575745,
      "learning_rate": 1.8836160137238516e-05,
      "loss": 0.3498,
      "step": 8590
    },
    {
      "epoch": 0.15519034515632787,
      "grad_norm": 0.5103094577789307,
      "learning_rate": 1.8834831818161957e-05,
      "loss": 0.3182,
      "step": 8595
    },
    {
      "epoch": 0.1552806245892286,
      "grad_norm": 0.3971729874610901,
      "learning_rate": 1.8833502788387412e-05,
      "loss": 0.3637,
      "step": 8600
    },
    {
      "epoch": 0.1553709040221293,
      "grad_norm": 0.5114766955375671,
      "learning_rate": 1.8832173048021796e-05,
      "loss": 0.3682,
      "step": 8605
    },
    {
      "epoch": 0.15546118345503002,
      "grad_norm": 0.3616827428340912,
      "learning_rate": 1.8830842597172073e-05,
      "loss": 0.3377,
      "step": 8610
    },
    {
      "epoch": 0.15555146288793073,
      "grad_norm": 0.3950643837451935,
      "learning_rate": 1.882951143594527e-05,
      "loss": 0.3024,
      "step": 8615
    },
    {
      "epoch": 0.15564174232083144,
      "grad_norm": 0.46623605489730835,
      "learning_rate": 1.8828179564448468e-05,
      "loss": 0.2826,
      "step": 8620
    },
    {
      "epoch": 0.15573202175373216,
      "grad_norm": 0.542243242263794,
      "learning_rate": 1.8826846982788805e-05,
      "loss": 0.2457,
      "step": 8625
    },
    {
      "epoch": 0.15582230118663287,
      "grad_norm": 0.4826582968235016,
      "learning_rate": 1.8825513691073484e-05,
      "loss": 0.2876,
      "step": 8630
    },
    {
      "epoch": 0.1559125806195336,
      "grad_norm": 0.3582850396633148,
      "learning_rate": 1.882417968940975e-05,
      "loss": 0.2749,
      "step": 8635
    },
    {
      "epoch": 0.1560028600524343,
      "grad_norm": 0.4192401170730591,
      "learning_rate": 1.882284497790492e-05,
      "loss": 0.337,
      "step": 8640
    },
    {
      "epoch": 0.15609313948533501,
      "grad_norm": 0.37834084033966064,
      "learning_rate": 1.882150955666636e-05,
      "loss": 0.3137,
      "step": 8645
    },
    {
      "epoch": 0.15618341891823573,
      "grad_norm": 0.5601775050163269,
      "learning_rate": 1.8820173425801493e-05,
      "loss": 0.2094,
      "step": 8650
    },
    {
      "epoch": 0.15627369835113644,
      "grad_norm": 0.4296712279319763,
      "learning_rate": 1.8818836585417804e-05,
      "loss": 0.3138,
      "step": 8655
    },
    {
      "epoch": 0.15636397778403716,
      "grad_norm": 0.2774662375450134,
      "learning_rate": 1.881749903562283e-05,
      "loss": 0.2602,
      "step": 8660
    },
    {
      "epoch": 0.15645425721693787,
      "grad_norm": 0.32397139072418213,
      "learning_rate": 1.881616077652417e-05,
      "loss": 0.2654,
      "step": 8665
    },
    {
      "epoch": 0.15654453664983858,
      "grad_norm": 0.49733802676200867,
      "learning_rate": 1.8814821808229476e-05,
      "loss": 0.2114,
      "step": 8670
    },
    {
      "epoch": 0.1566348160827393,
      "grad_norm": 0.5005702972412109,
      "learning_rate": 1.8813482130846454e-05,
      "loss": 0.3082,
      "step": 8675
    },
    {
      "epoch": 0.15672509551564,
      "grad_norm": 0.48754778504371643,
      "learning_rate": 1.8812141744482876e-05,
      "loss": 0.5073,
      "step": 8680
    },
    {
      "epoch": 0.15681537494854073,
      "grad_norm": 0.575425922870636,
      "learning_rate": 1.881080064924657e-05,
      "loss": 0.3876,
      "step": 8685
    },
    {
      "epoch": 0.15690565438144144,
      "grad_norm": 0.4367899000644684,
      "learning_rate": 1.8809458845245405e-05,
      "loss": 0.296,
      "step": 8690
    },
    {
      "epoch": 0.15699593381434215,
      "grad_norm": 0.33415520191192627,
      "learning_rate": 1.880811633258733e-05,
      "loss": 0.2484,
      "step": 8695
    },
    {
      "epoch": 0.15708621324724287,
      "grad_norm": 0.3557751774787903,
      "learning_rate": 1.8806773111380337e-05,
      "loss": 0.3323,
      "step": 8700
    },
    {
      "epoch": 0.15717649268014358,
      "grad_norm": 0.409047394990921,
      "learning_rate": 1.880542918173248e-05,
      "loss": 0.1935,
      "step": 8705
    },
    {
      "epoch": 0.1572667721130443,
      "grad_norm": 0.3447967767715454,
      "learning_rate": 1.8804084543751863e-05,
      "loss": 0.3283,
      "step": 8710
    },
    {
      "epoch": 0.157357051545945,
      "grad_norm": 0.3793751001358032,
      "learning_rate": 1.880273919754666e-05,
      "loss": 0.1691,
      "step": 8715
    },
    {
      "epoch": 0.15744733097884572,
      "grad_norm": 0.6087231040000916,
      "learning_rate": 1.880139314322509e-05,
      "loss": 0.3477,
      "step": 8720
    },
    {
      "epoch": 0.15753761041174644,
      "grad_norm": 0.6917526721954346,
      "learning_rate": 1.8800046380895435e-05,
      "loss": 0.3205,
      "step": 8725
    },
    {
      "epoch": 0.15762788984464715,
      "grad_norm": 0.45527705550193787,
      "learning_rate": 1.8798698910666028e-05,
      "loss": 0.3088,
      "step": 8730
    },
    {
      "epoch": 0.15771816927754786,
      "grad_norm": 0.5412310361862183,
      "learning_rate": 1.8797350732645268e-05,
      "loss": 0.2451,
      "step": 8735
    },
    {
      "epoch": 0.15780844871044858,
      "grad_norm": 0.3872532844543457,
      "learning_rate": 1.8796001846941604e-05,
      "loss": 0.3598,
      "step": 8740
    },
    {
      "epoch": 0.1578987281433493,
      "grad_norm": 0.40230852365493774,
      "learning_rate": 1.879465225366355e-05,
      "loss": 0.3362,
      "step": 8745
    },
    {
      "epoch": 0.15798900757625,
      "grad_norm": 0.35699185729026794,
      "learning_rate": 1.879330195291966e-05,
      "loss": 0.2791,
      "step": 8750
    },
    {
      "epoch": 0.15807928700915072,
      "grad_norm": 0.9417124390602112,
      "learning_rate": 1.879195094481856e-05,
      "loss": 0.3251,
      "step": 8755
    },
    {
      "epoch": 0.15816956644205143,
      "grad_norm": 0.6971483826637268,
      "learning_rate": 1.879059922946893e-05,
      "loss": 0.3632,
      "step": 8760
    },
    {
      "epoch": 0.15825984587495215,
      "grad_norm": 0.5676143765449524,
      "learning_rate": 1.8789246806979512e-05,
      "loss": 0.337,
      "step": 8765
    },
    {
      "epoch": 0.15835012530785286,
      "grad_norm": 0.5160952806472778,
      "learning_rate": 1.8787893677459087e-05,
      "loss": 0.2961,
      "step": 8770
    },
    {
      "epoch": 0.15844040474075358,
      "grad_norm": 0.467488557100296,
      "learning_rate": 1.878653984101651e-05,
      "loss": 0.3232,
      "step": 8775
    },
    {
      "epoch": 0.1585306841736543,
      "grad_norm": 0.4414568245410919,
      "learning_rate": 1.8785185297760687e-05,
      "loss": 0.3518,
      "step": 8780
    },
    {
      "epoch": 0.158620963606555,
      "grad_norm": 0.4487064778804779,
      "learning_rate": 1.8783830047800583e-05,
      "loss": 0.2638,
      "step": 8785
    },
    {
      "epoch": 0.15871124303945572,
      "grad_norm": 1.112725019454956,
      "learning_rate": 1.878247409124522e-05,
      "loss": 0.3202,
      "step": 8790
    },
    {
      "epoch": 0.15880152247235643,
      "grad_norm": 0.30171719193458557,
      "learning_rate": 1.8781117428203665e-05,
      "loss": 0.2899,
      "step": 8795
    },
    {
      "epoch": 0.15889180190525715,
      "grad_norm": 0.3588135838508606,
      "learning_rate": 1.877976005878506e-05,
      "loss": 0.3362,
      "step": 8800
    },
    {
      "epoch": 0.15898208133815786,
      "grad_norm": 0.49537143111228943,
      "learning_rate": 1.8778401983098594e-05,
      "loss": 0.3071,
      "step": 8805
    },
    {
      "epoch": 0.15907236077105857,
      "grad_norm": 0.5950140953063965,
      "learning_rate": 1.877704320125351e-05,
      "loss": 0.3152,
      "step": 8810
    },
    {
      "epoch": 0.1591626402039593,
      "grad_norm": 0.4290141761302948,
      "learning_rate": 1.8775683713359118e-05,
      "loss": 0.3679,
      "step": 8815
    },
    {
      "epoch": 0.15925291963686,
      "grad_norm": 0.7214063405990601,
      "learning_rate": 1.877432351952478e-05,
      "loss": 0.2864,
      "step": 8820
    },
    {
      "epoch": 0.15934319906976072,
      "grad_norm": 0.5823917388916016,
      "learning_rate": 1.8772962619859908e-05,
      "loss": 0.3488,
      "step": 8825
    },
    {
      "epoch": 0.15943347850266143,
      "grad_norm": 0.47410455346107483,
      "learning_rate": 1.8771601014473978e-05,
      "loss": 0.276,
      "step": 8830
    },
    {
      "epoch": 0.15952375793556214,
      "grad_norm": 0.3576255440711975,
      "learning_rate": 1.8770238703476522e-05,
      "loss": 0.3626,
      "step": 8835
    },
    {
      "epoch": 0.15961403736846286,
      "grad_norm": 0.5306724905967712,
      "learning_rate": 1.8768875686977125e-05,
      "loss": 0.2775,
      "step": 8840
    },
    {
      "epoch": 0.15970431680136357,
      "grad_norm": 0.4195229411125183,
      "learning_rate": 1.876751196508544e-05,
      "loss": 0.3033,
      "step": 8845
    },
    {
      "epoch": 0.15979459623426429,
      "grad_norm": 0.7627418041229248,
      "learning_rate": 1.8766147537911165e-05,
      "loss": 0.338,
      "step": 8850
    },
    {
      "epoch": 0.159884875667165,
      "grad_norm": 0.3827398121356964,
      "learning_rate": 1.8764782405564052e-05,
      "loss": 0.3547,
      "step": 8855
    },
    {
      "epoch": 0.1599751551000657,
      "grad_norm": 0.4121302664279938,
      "learning_rate": 1.876341656815392e-05,
      "loss": 0.2591,
      "step": 8860
    },
    {
      "epoch": 0.16006543453296643,
      "grad_norm": 0.3650098741054535,
      "learning_rate": 1.8762050025790643e-05,
      "loss": 0.356,
      "step": 8865
    },
    {
      "epoch": 0.16015571396586714,
      "grad_norm": 0.3953309953212738,
      "learning_rate": 1.8760682778584146e-05,
      "loss": 0.3404,
      "step": 8870
    },
    {
      "epoch": 0.16024599339876786,
      "grad_norm": 0.4996097683906555,
      "learning_rate": 1.875931482664442e-05,
      "loss": 0.377,
      "step": 8875
    },
    {
      "epoch": 0.16033627283166857,
      "grad_norm": 0.34962838888168335,
      "learning_rate": 1.87579461700815e-05,
      "loss": 0.2998,
      "step": 8880
    },
    {
      "epoch": 0.16042655226456928,
      "grad_norm": 0.3440771996974945,
      "learning_rate": 1.8756576809005488e-05,
      "loss": 0.358,
      "step": 8885
    },
    {
      "epoch": 0.16051683169747,
      "grad_norm": 0.38825514912605286,
      "learning_rate": 1.8755206743526537e-05,
      "loss": 0.2428,
      "step": 8890
    },
    {
      "epoch": 0.1606071111303707,
      "grad_norm": 0.38474076986312866,
      "learning_rate": 1.875383597375486e-05,
      "loss": 0.3534,
      "step": 8895
    },
    {
      "epoch": 0.16069739056327143,
      "grad_norm": 0.42535915970802307,
      "learning_rate": 1.8752464499800727e-05,
      "loss": 0.2535,
      "step": 8900
    },
    {
      "epoch": 0.16078766999617214,
      "grad_norm": 0.3820883333683014,
      "learning_rate": 1.875109232177446e-05,
      "loss": 0.375,
      "step": 8905
    },
    {
      "epoch": 0.16087794942907285,
      "grad_norm": 0.39074403047561646,
      "learning_rate": 1.874971943978644e-05,
      "loss": 0.3036,
      "step": 8910
    },
    {
      "epoch": 0.16096822886197357,
      "grad_norm": 0.5746060609817505,
      "learning_rate": 1.8748345853947116e-05,
      "loss": 0.4176,
      "step": 8915
    },
    {
      "epoch": 0.1610585082948743,
      "grad_norm": 0.3055366575717926,
      "learning_rate": 1.8746971564366966e-05,
      "loss": 0.3802,
      "step": 8920
    },
    {
      "epoch": 0.16114878772777502,
      "grad_norm": 0.48693153262138367,
      "learning_rate": 1.8745596571156554e-05,
      "loss": 0.3053,
      "step": 8925
    },
    {
      "epoch": 0.16123906716067574,
      "grad_norm": 0.33140549063682556,
      "learning_rate": 1.8744220874426485e-05,
      "loss": 0.3338,
      "step": 8930
    },
    {
      "epoch": 0.16132934659357645,
      "grad_norm": 0.32309263944625854,
      "learning_rate": 1.874284447428742e-05,
      "loss": 0.3479,
      "step": 8935
    },
    {
      "epoch": 0.16141962602647716,
      "grad_norm": 0.3173151910305023,
      "learning_rate": 1.874146737085009e-05,
      "loss": 0.2721,
      "step": 8940
    },
    {
      "epoch": 0.16150990545937788,
      "grad_norm": 0.5842396020889282,
      "learning_rate": 1.874008956422526e-05,
      "loss": 0.306,
      "step": 8945
    },
    {
      "epoch": 0.1616001848922786,
      "grad_norm": 0.45474880933761597,
      "learning_rate": 1.8738711054523772e-05,
      "loss": 0.2855,
      "step": 8950
    },
    {
      "epoch": 0.1616904643251793,
      "grad_norm": 0.40684881806373596,
      "learning_rate": 1.8737331841856518e-05,
      "loss": 0.3572,
      "step": 8955
    },
    {
      "epoch": 0.16178074375808002,
      "grad_norm": 0.6544089317321777,
      "learning_rate": 1.8735951926334443e-05,
      "loss": 0.3391,
      "step": 8960
    },
    {
      "epoch": 0.16187102319098073,
      "grad_norm": 0.4144783914089203,
      "learning_rate": 1.8734571308068555e-05,
      "loss": 0.3722,
      "step": 8965
    },
    {
      "epoch": 0.16196130262388145,
      "grad_norm": 0.43708932399749756,
      "learning_rate": 1.873318998716991e-05,
      "loss": 0.3814,
      "step": 8970
    },
    {
      "epoch": 0.16205158205678216,
      "grad_norm": 0.6572683453559875,
      "learning_rate": 1.8731807963749627e-05,
      "loss": 0.2317,
      "step": 8975
    },
    {
      "epoch": 0.16214186148968288,
      "grad_norm": 0.48241615295410156,
      "learning_rate": 1.8730425237918878e-05,
      "loss": 0.3372,
      "step": 8980
    },
    {
      "epoch": 0.1622321409225836,
      "grad_norm": 0.3190326392650604,
      "learning_rate": 1.8729041809788894e-05,
      "loss": 0.217,
      "step": 8985
    },
    {
      "epoch": 0.1623224203554843,
      "grad_norm": 0.24012210965156555,
      "learning_rate": 1.8727657679470964e-05,
      "loss": 0.2669,
      "step": 8990
    },
    {
      "epoch": 0.16241269978838502,
      "grad_norm": 0.5661858916282654,
      "learning_rate": 1.872627284707643e-05,
      "loss": 0.3828,
      "step": 8995
    },
    {
      "epoch": 0.16250297922128573,
      "grad_norm": 0.3285576403141022,
      "learning_rate": 1.8724887312716696e-05,
      "loss": 0.3112,
      "step": 9000
    },
    {
      "epoch": 0.16259325865418645,
      "grad_norm": 0.4387272000312805,
      "learning_rate": 1.8723501076503207e-05,
      "loss": 0.2479,
      "step": 9005
    },
    {
      "epoch": 0.16268353808708716,
      "grad_norm": 0.41906678676605225,
      "learning_rate": 1.872211413854749e-05,
      "loss": 0.267,
      "step": 9010
    },
    {
      "epoch": 0.16277381751998787,
      "grad_norm": 0.37576326727867126,
      "learning_rate": 1.8720726498961098e-05,
      "loss": 0.2704,
      "step": 9015
    },
    {
      "epoch": 0.1628640969528886,
      "grad_norm": 0.5084801316261292,
      "learning_rate": 1.8719338157855672e-05,
      "loss": 0.3471,
      "step": 9020
    },
    {
      "epoch": 0.1629543763857893,
      "grad_norm": 0.38982388377189636,
      "learning_rate": 1.8717949115342882e-05,
      "loss": 0.3095,
      "step": 9025
    },
    {
      "epoch": 0.16304465581869002,
      "grad_norm": 0.40285664796829224,
      "learning_rate": 1.8716559371534472e-05,
      "loss": 0.3278,
      "step": 9030
    },
    {
      "epoch": 0.16313493525159073,
      "grad_norm": 0.49813154339790344,
      "learning_rate": 1.871516892654224e-05,
      "loss": 0.2548,
      "step": 9035
    },
    {
      "epoch": 0.16322521468449144,
      "grad_norm": 0.2894067168235779,
      "learning_rate": 1.871377778047803e-05,
      "loss": 0.2275,
      "step": 9040
    },
    {
      "epoch": 0.16331549411739216,
      "grad_norm": 0.9235433340072632,
      "learning_rate": 1.8712385933453753e-05,
      "loss": 0.3341,
      "step": 9045
    },
    {
      "epoch": 0.16340577355029287,
      "grad_norm": 0.510308563709259,
      "learning_rate": 1.8710993385581373e-05,
      "loss": 0.2429,
      "step": 9050
    },
    {
      "epoch": 0.16349605298319358,
      "grad_norm": 0.357827365398407,
      "learning_rate": 1.8709600136972906e-05,
      "loss": 0.3234,
      "step": 9055
    },
    {
      "epoch": 0.1635863324160943,
      "grad_norm": 0.5867786407470703,
      "learning_rate": 1.870820618774044e-05,
      "loss": 0.3031,
      "step": 9060
    },
    {
      "epoch": 0.163676611848995,
      "grad_norm": 0.73820561170578,
      "learning_rate": 1.87068115379961e-05,
      "loss": 0.2481,
      "step": 9065
    },
    {
      "epoch": 0.16376689128189573,
      "grad_norm": 0.43540045619010925,
      "learning_rate": 1.8705416187852073e-05,
      "loss": 0.2251,
      "step": 9070
    },
    {
      "epoch": 0.16385717071479644,
      "grad_norm": 0.37970179319381714,
      "learning_rate": 1.870402013742061e-05,
      "loss": 0.3055,
      "step": 9075
    },
    {
      "epoch": 0.16394745014769715,
      "grad_norm": 0.3866814374923706,
      "learning_rate": 1.870262338681401e-05,
      "loss": 0.2766,
      "step": 9080
    },
    {
      "epoch": 0.16403772958059787,
      "grad_norm": 0.5278443694114685,
      "learning_rate": 1.8701225936144633e-05,
      "loss": 0.3221,
      "step": 9085
    },
    {
      "epoch": 0.16412800901349858,
      "grad_norm": 0.37457430362701416,
      "learning_rate": 1.8699827785524895e-05,
      "loss": 0.3623,
      "step": 9090
    },
    {
      "epoch": 0.1642182884463993,
      "grad_norm": 0.4798072278499603,
      "learning_rate": 1.8698428935067264e-05,
      "loss": 0.3111,
      "step": 9095
    },
    {
      "epoch": 0.1643085678793,
      "grad_norm": 0.404145747423172,
      "learning_rate": 1.869702938488427e-05,
      "loss": 0.3708,
      "step": 9100
    },
    {
      "epoch": 0.16439884731220072,
      "grad_norm": 0.38765108585357666,
      "learning_rate": 1.8695629135088494e-05,
      "loss": 0.3342,
      "step": 9105
    },
    {
      "epoch": 0.16448912674510144,
      "grad_norm": 0.48070767521858215,
      "learning_rate": 1.869422818579258e-05,
      "loss": 0.3036,
      "step": 9110
    },
    {
      "epoch": 0.16457940617800215,
      "grad_norm": 0.3652752637863159,
      "learning_rate": 1.869282653710922e-05,
      "loss": 0.3352,
      "step": 9115
    },
    {
      "epoch": 0.16466968561090287,
      "grad_norm": 0.40660160779953003,
      "learning_rate": 1.8691424189151166e-05,
      "loss": 0.2548,
      "step": 9120
    },
    {
      "epoch": 0.16475996504380358,
      "grad_norm": 0.602676272392273,
      "learning_rate": 1.8690021142031233e-05,
      "loss": 0.3861,
      "step": 9125
    },
    {
      "epoch": 0.1648502444767043,
      "grad_norm": 0.3529750108718872,
      "learning_rate": 1.868861739586228e-05,
      "loss": 0.2575,
      "step": 9130
    },
    {
      "epoch": 0.164940523909605,
      "grad_norm": 0.39472639560699463,
      "learning_rate": 1.8687212950757226e-05,
      "loss": 0.2434,
      "step": 9135
    },
    {
      "epoch": 0.16503080334250572,
      "grad_norm": 0.20535361766815186,
      "learning_rate": 1.8685807806829056e-05,
      "loss": 0.1722,
      "step": 9140
    },
    {
      "epoch": 0.16512108277540644,
      "grad_norm": 0.31685560941696167,
      "learning_rate": 1.86844019641908e-05,
      "loss": 0.2489,
      "step": 9145
    },
    {
      "epoch": 0.16521136220830715,
      "grad_norm": 0.45184728503227234,
      "learning_rate": 1.8682995422955548e-05,
      "loss": 0.294,
      "step": 9150
    },
    {
      "epoch": 0.16530164164120786,
      "grad_norm": 0.4597442150115967,
      "learning_rate": 1.8681588183236443e-05,
      "loss": 0.3423,
      "step": 9155
    },
    {
      "epoch": 0.16539192107410858,
      "grad_norm": 0.5040349364280701,
      "learning_rate": 1.8680180245146693e-05,
      "loss": 0.3271,
      "step": 9160
    },
    {
      "epoch": 0.1654822005070093,
      "grad_norm": 0.44972705841064453,
      "learning_rate": 1.8678771608799553e-05,
      "loss": 0.2843,
      "step": 9165
    },
    {
      "epoch": 0.16557247993991,
      "grad_norm": 0.6874871253967285,
      "learning_rate": 1.8677362274308338e-05,
      "loss": 0.2859,
      "step": 9170
    },
    {
      "epoch": 0.16566275937281072,
      "grad_norm": 0.6598188281059265,
      "learning_rate": 1.8675952241786416e-05,
      "loss": 0.3639,
      "step": 9175
    },
    {
      "epoch": 0.16575303880571143,
      "grad_norm": 0.4872834384441376,
      "learning_rate": 1.8674541511347218e-05,
      "loss": 0.3082,
      "step": 9180
    },
    {
      "epoch": 0.16584331823861215,
      "grad_norm": 0.40957456827163696,
      "learning_rate": 1.8673130083104227e-05,
      "loss": 0.2927,
      "step": 9185
    },
    {
      "epoch": 0.16593359767151286,
      "grad_norm": 0.42083311080932617,
      "learning_rate": 1.8671717957170977e-05,
      "loss": 0.2965,
      "step": 9190
    },
    {
      "epoch": 0.16602387710441358,
      "grad_norm": 0.7757776379585266,
      "learning_rate": 1.8670305133661067e-05,
      "loss": 0.381,
      "step": 9195
    },
    {
      "epoch": 0.1661141565373143,
      "grad_norm": 0.24977640807628632,
      "learning_rate": 1.866889161268815e-05,
      "loss": 0.2825,
      "step": 9200
    },
    {
      "epoch": 0.166204435970215,
      "grad_norm": 0.42900508642196655,
      "learning_rate": 1.866747739436593e-05,
      "loss": 0.3708,
      "step": 9205
    },
    {
      "epoch": 0.16629471540311572,
      "grad_norm": 0.4191928803920746,
      "learning_rate": 1.8666062478808176e-05,
      "loss": 0.3219,
      "step": 9210
    },
    {
      "epoch": 0.16638499483601643,
      "grad_norm": 0.4868921935558319,
      "learning_rate": 1.8664646866128702e-05,
      "loss": 0.3252,
      "step": 9215
    },
    {
      "epoch": 0.16647527426891714,
      "grad_norm": 0.3454371690750122,
      "learning_rate": 1.8663230556441384e-05,
      "loss": 0.3906,
      "step": 9220
    },
    {
      "epoch": 0.16656555370181786,
      "grad_norm": 0.24596817791461945,
      "learning_rate": 1.866181354986016e-05,
      "loss": 0.3218,
      "step": 9225
    },
    {
      "epoch": 0.16665583313471857,
      "grad_norm": 0.47529229521751404,
      "learning_rate": 1.866039584649901e-05,
      "loss": 0.314,
      "step": 9230
    },
    {
      "epoch": 0.1667461125676193,
      "grad_norm": 0.5996313691139221,
      "learning_rate": 1.865897744647198e-05,
      "loss": 0.3252,
      "step": 9235
    },
    {
      "epoch": 0.16683639200052,
      "grad_norm": 0.5580554604530334,
      "learning_rate": 1.8657558349893174e-05,
      "loss": 0.2919,
      "step": 9240
    },
    {
      "epoch": 0.16692667143342071,
      "grad_norm": 0.32417070865631104,
      "learning_rate": 1.8656138556876742e-05,
      "loss": 0.3111,
      "step": 9245
    },
    {
      "epoch": 0.16701695086632143,
      "grad_norm": 0.7458740472793579,
      "learning_rate": 1.8654718067536904e-05,
      "loss": 0.2593,
      "step": 9250
    },
    {
      "epoch": 0.16710723029922214,
      "grad_norm": 0.5567686557769775,
      "learning_rate": 1.865329688198792e-05,
      "loss": 0.3088,
      "step": 9255
    },
    {
      "epoch": 0.16719750973212286,
      "grad_norm": 0.4393288195133209,
      "learning_rate": 1.865187500034412e-05,
      "loss": 0.2979,
      "step": 9260
    },
    {
      "epoch": 0.16728778916502357,
      "grad_norm": 0.39182204008102417,
      "learning_rate": 1.865045242271988e-05,
      "loss": 0.3249,
      "step": 9265
    },
    {
      "epoch": 0.16737806859792428,
      "grad_norm": 0.34086981415748596,
      "learning_rate": 1.8649029149229638e-05,
      "loss": 0.4379,
      "step": 9270
    },
    {
      "epoch": 0.167468348030825,
      "grad_norm": 0.4125692546367645,
      "learning_rate": 1.8647605179987882e-05,
      "loss": 0.2359,
      "step": 9275
    },
    {
      "epoch": 0.1675586274637257,
      "grad_norm": 0.2924150228500366,
      "learning_rate": 1.8646180515109168e-05,
      "loss": 0.2635,
      "step": 9280
    },
    {
      "epoch": 0.16764890689662643,
      "grad_norm": 0.2948613464832306,
      "learning_rate": 1.864475515470809e-05,
      "loss": 0.2019,
      "step": 9285
    },
    {
      "epoch": 0.16773918632952714,
      "grad_norm": 0.4389468729496002,
      "learning_rate": 1.8643329098899317e-05,
      "loss": 0.3511,
      "step": 9290
    },
    {
      "epoch": 0.16782946576242785,
      "grad_norm": 0.47662729024887085,
      "learning_rate": 1.864190234779756e-05,
      "loss": 0.2472,
      "step": 9295
    },
    {
      "epoch": 0.16791974519532857,
      "grad_norm": 0.3126203417778015,
      "learning_rate": 1.8640474901517597e-05,
      "loss": 0.3962,
      "step": 9300
    },
    {
      "epoch": 0.16801002462822928,
      "grad_norm": 0.5453171730041504,
      "learning_rate": 1.8639046760174243e-05,
      "loss": 0.3613,
      "step": 9305
    },
    {
      "epoch": 0.16810030406113,
      "grad_norm": 0.4740075170993805,
      "learning_rate": 1.8637617923882392e-05,
      "loss": 0.239,
      "step": 9310
    },
    {
      "epoch": 0.16819058349403074,
      "grad_norm": 0.5277817249298096,
      "learning_rate": 1.863618839275698e-05,
      "loss": 0.2778,
      "step": 9315
    },
    {
      "epoch": 0.16828086292693145,
      "grad_norm": 0.41539040207862854,
      "learning_rate": 1.8634758166913004e-05,
      "loss": 0.3002,
      "step": 9320
    },
    {
      "epoch": 0.16837114235983217,
      "grad_norm": 0.5895034074783325,
      "learning_rate": 1.8633327246465517e-05,
      "loss": 0.3376,
      "step": 9325
    },
    {
      "epoch": 0.16846142179273288,
      "grad_norm": 0.3407139778137207,
      "learning_rate": 1.863189563152962e-05,
      "loss": 0.3832,
      "step": 9330
    },
    {
      "epoch": 0.1685517012256336,
      "grad_norm": 0.5738711357116699,
      "learning_rate": 1.8630463322220477e-05,
      "loss": 0.4042,
      "step": 9335
    },
    {
      "epoch": 0.1686419806585343,
      "grad_norm": 0.4206055998802185,
      "learning_rate": 1.8629030318653313e-05,
      "loss": 0.3159,
      "step": 9340
    },
    {
      "epoch": 0.16873226009143502,
      "grad_norm": 0.4046110510826111,
      "learning_rate": 1.8627596620943398e-05,
      "loss": 0.4248,
      "step": 9345
    },
    {
      "epoch": 0.16882253952433574,
      "grad_norm": 0.4212724268436432,
      "learning_rate": 1.8626162229206063e-05,
      "loss": 0.2942,
      "step": 9350
    },
    {
      "epoch": 0.16891281895723645,
      "grad_norm": 0.6038151383399963,
      "learning_rate": 1.8624727143556694e-05,
      "loss": 0.3216,
      "step": 9355
    },
    {
      "epoch": 0.16900309839013716,
      "grad_norm": 0.5483262538909912,
      "learning_rate": 1.8623291364110737e-05,
      "loss": 0.4853,
      "step": 9360
    },
    {
      "epoch": 0.16909337782303788,
      "grad_norm": 0.7040364146232605,
      "learning_rate": 1.8621854890983687e-05,
      "loss": 0.3221,
      "step": 9365
    },
    {
      "epoch": 0.1691836572559386,
      "grad_norm": 0.6033093929290771,
      "learning_rate": 1.8620417724291096e-05,
      "loss": 0.2782,
      "step": 9370
    },
    {
      "epoch": 0.1692739366888393,
      "grad_norm": 0.7284932732582092,
      "learning_rate": 1.8618979864148574e-05,
      "loss": 0.2881,
      "step": 9375
    },
    {
      "epoch": 0.16936421612174002,
      "grad_norm": 0.3731858432292938,
      "learning_rate": 1.861754131067179e-05,
      "loss": 0.2258,
      "step": 9380
    },
    {
      "epoch": 0.16945449555464073,
      "grad_norm": 0.6486430764198303,
      "learning_rate": 1.8616102063976465e-05,
      "loss": 0.3082,
      "step": 9385
    },
    {
      "epoch": 0.16954477498754145,
      "grad_norm": 0.6816726922988892,
      "learning_rate": 1.8614662124178373e-05,
      "loss": 0.3372,
      "step": 9390
    },
    {
      "epoch": 0.16963505442044216,
      "grad_norm": 0.39962419867515564,
      "learning_rate": 1.8613221491393348e-05,
      "loss": 0.3209,
      "step": 9395
    },
    {
      "epoch": 0.16972533385334287,
      "grad_norm": 0.4884919822216034,
      "learning_rate": 1.8611780165737275e-05,
      "loss": 0.3539,
      "step": 9400
    },
    {
      "epoch": 0.1698156132862436,
      "grad_norm": 0.6725785136222839,
      "learning_rate": 1.8610338147326105e-05,
      "loss": 0.2836,
      "step": 9405
    },
    {
      "epoch": 0.1699058927191443,
      "grad_norm": 0.4711063504219055,
      "learning_rate": 1.8608895436275833e-05,
      "loss": 0.2337,
      "step": 9410
    },
    {
      "epoch": 0.16999617215204502,
      "grad_norm": 0.635113000869751,
      "learning_rate": 1.8607452032702515e-05,
      "loss": 0.2342,
      "step": 9415
    },
    {
      "epoch": 0.17008645158494573,
      "grad_norm": 1.7193716764450073,
      "learning_rate": 1.8606007936722264e-05,
      "loss": 0.3885,
      "step": 9420
    },
    {
      "epoch": 0.17017673101784644,
      "grad_norm": 0.46928563714027405,
      "learning_rate": 1.8604563148451245e-05,
      "loss": 0.3314,
      "step": 9425
    },
    {
      "epoch": 0.17026701045074716,
      "grad_norm": 0.5596863627433777,
      "learning_rate": 1.8603117668005684e-05,
      "loss": 0.2857,
      "step": 9430
    },
    {
      "epoch": 0.17035728988364787,
      "grad_norm": 0.66194748878479,
      "learning_rate": 1.860167149550186e-05,
      "loss": 0.3214,
      "step": 9435
    },
    {
      "epoch": 0.17044756931654859,
      "grad_norm": 0.2665224075317383,
      "learning_rate": 1.8600224631056098e-05,
      "loss": 0.232,
      "step": 9440
    },
    {
      "epoch": 0.1705378487494493,
      "grad_norm": 0.4515458345413208,
      "learning_rate": 1.8598777074784796e-05,
      "loss": 0.288,
      "step": 9445
    },
    {
      "epoch": 0.17062812818235001,
      "grad_norm": 0.3401292860507965,
      "learning_rate": 1.85973288268044e-05,
      "loss": 0.2927,
      "step": 9450
    },
    {
      "epoch": 0.17071840761525073,
      "grad_norm": 0.5469954013824463,
      "learning_rate": 1.8595879887231408e-05,
      "loss": 0.3329,
      "step": 9455
    },
    {
      "epoch": 0.17080868704815144,
      "grad_norm": 0.4402863383293152,
      "learning_rate": 1.8594430256182375e-05,
      "loss": 0.2655,
      "step": 9460
    },
    {
      "epoch": 0.17089896648105216,
      "grad_norm": 0.4635493755340576,
      "learning_rate": 1.8592979933773918e-05,
      "loss": 0.3237,
      "step": 9465
    },
    {
      "epoch": 0.17098924591395287,
      "grad_norm": 0.4579942524433136,
      "learning_rate": 1.85915289201227e-05,
      "loss": 0.2661,
      "step": 9470
    },
    {
      "epoch": 0.17107952534685358,
      "grad_norm": 0.30433493852615356,
      "learning_rate": 1.859007721534545e-05,
      "loss": 0.4544,
      "step": 9475
    },
    {
      "epoch": 0.1711698047797543,
      "grad_norm": 0.5254673957824707,
      "learning_rate": 1.8588624819558938e-05,
      "loss": 0.168,
      "step": 9480
    },
    {
      "epoch": 0.171260084212655,
      "grad_norm": 0.5817158818244934,
      "learning_rate": 1.858717173288001e-05,
      "loss": 0.2764,
      "step": 9485
    },
    {
      "epoch": 0.17135036364555573,
      "grad_norm": 0.4819915294647217,
      "learning_rate": 1.8585717955425545e-05,
      "loss": 0.2993,
      "step": 9490
    },
    {
      "epoch": 0.17144064307845644,
      "grad_norm": 0.6393699645996094,
      "learning_rate": 1.85842634873125e-05,
      "loss": 0.2509,
      "step": 9495
    },
    {
      "epoch": 0.17153092251135715,
      "grad_norm": 0.4473429024219513,
      "learning_rate": 1.858280832865787e-05,
      "loss": 0.3107,
      "step": 9500
    },
    {
      "epoch": 0.17162120194425787,
      "grad_norm": 0.5762903690338135,
      "learning_rate": 1.8581352479578715e-05,
      "loss": 0.2317,
      "step": 9505
    },
    {
      "epoch": 0.17171148137715858,
      "grad_norm": 0.5538835525512695,
      "learning_rate": 1.8579895940192143e-05,
      "loss": 0.3585,
      "step": 9510
    },
    {
      "epoch": 0.1718017608100593,
      "grad_norm": 0.4441256523132324,
      "learning_rate": 1.8578438710615322e-05,
      "loss": 0.2742,
      "step": 9515
    },
    {
      "epoch": 0.17189204024296,
      "grad_norm": 0.3827473223209381,
      "learning_rate": 1.8576980790965482e-05,
      "loss": 0.3831,
      "step": 9520
    },
    {
      "epoch": 0.17198231967586072,
      "grad_norm": 0.44780096411705017,
      "learning_rate": 1.85755221813599e-05,
      "loss": 0.4024,
      "step": 9525
    },
    {
      "epoch": 0.17207259910876144,
      "grad_norm": 0.27713555097579956,
      "learning_rate": 1.8574062881915904e-05,
      "loss": 0.2225,
      "step": 9530
    },
    {
      "epoch": 0.17216287854166215,
      "grad_norm": 0.43907350301742554,
      "learning_rate": 1.857260289275089e-05,
      "loss": 0.2925,
      "step": 9535
    },
    {
      "epoch": 0.17225315797456286,
      "grad_norm": 0.5583574771881104,
      "learning_rate": 1.8571142213982302e-05,
      "loss": 0.3521,
      "step": 9540
    },
    {
      "epoch": 0.17234343740746358,
      "grad_norm": 0.3877319097518921,
      "learning_rate": 1.8569680845727644e-05,
      "loss": 0.3317,
      "step": 9545
    },
    {
      "epoch": 0.1724337168403643,
      "grad_norm": 0.622263491153717,
      "learning_rate": 1.8568218788104464e-05,
      "loss": 0.2734,
      "step": 9550
    },
    {
      "epoch": 0.172523996273265,
      "grad_norm": 0.5004008412361145,
      "learning_rate": 1.8566756041230382e-05,
      "loss": 0.3637,
      "step": 9555
    },
    {
      "epoch": 0.17261427570616572,
      "grad_norm": 0.44145500659942627,
      "learning_rate": 1.8565292605223064e-05,
      "loss": 0.2419,
      "step": 9560
    },
    {
      "epoch": 0.17270455513906643,
      "grad_norm": 0.3170686364173889,
      "learning_rate": 1.856382848020023e-05,
      "loss": 0.2014,
      "step": 9565
    },
    {
      "epoch": 0.17279483457196715,
      "grad_norm": 0.43776872754096985,
      "learning_rate": 1.8562363666279663e-05,
      "loss": 0.3425,
      "step": 9570
    },
    {
      "epoch": 0.17288511400486786,
      "grad_norm": 0.4404096007347107,
      "learning_rate": 1.8560898163579187e-05,
      "loss": 0.3461,
      "step": 9575
    },
    {
      "epoch": 0.17297539343776858,
      "grad_norm": 0.3963436782360077,
      "learning_rate": 1.8559431972216704e-05,
      "loss": 0.427,
      "step": 9580
    },
    {
      "epoch": 0.1730656728706693,
      "grad_norm": 0.37308162450790405,
      "learning_rate": 1.8557965092310147e-05,
      "loss": 0.3837,
      "step": 9585
    },
    {
      "epoch": 0.17315595230357,
      "grad_norm": 0.399116188287735,
      "learning_rate": 1.8556497523977525e-05,
      "loss": 0.3143,
      "step": 9590
    },
    {
      "epoch": 0.17324623173647072,
      "grad_norm": 0.5526614785194397,
      "learning_rate": 1.8555029267336887e-05,
      "loss": 0.2633,
      "step": 9595
    },
    {
      "epoch": 0.17333651116937143,
      "grad_norm": 0.47489047050476074,
      "learning_rate": 1.8553560322506347e-05,
      "loss": 0.3514,
      "step": 9600
    },
    {
      "epoch": 0.17342679060227215,
      "grad_norm": 0.45416170358657837,
      "learning_rate": 1.855209068960407e-05,
      "loss": 0.25,
      "step": 9605
    },
    {
      "epoch": 0.17351707003517286,
      "grad_norm": 0.5261057615280151,
      "learning_rate": 1.855062036874827e-05,
      "loss": 0.3549,
      "step": 9610
    },
    {
      "epoch": 0.17360734946807357,
      "grad_norm": 0.5666975975036621,
      "learning_rate": 1.854914936005724e-05,
      "loss": 0.3656,
      "step": 9615
    },
    {
      "epoch": 0.1736976289009743,
      "grad_norm": 0.3660118579864502,
      "learning_rate": 1.8547677663649293e-05,
      "loss": 0.3276,
      "step": 9620
    },
    {
      "epoch": 0.173787908333875,
      "grad_norm": 1.1892653703689575,
      "learning_rate": 1.8546205279642834e-05,
      "loss": 0.2777,
      "step": 9625
    },
    {
      "epoch": 0.17387818776677572,
      "grad_norm": 0.605702817440033,
      "learning_rate": 1.854473220815629e-05,
      "loss": 0.2925,
      "step": 9630
    },
    {
      "epoch": 0.17396846719967643,
      "grad_norm": 0.3500027060508728,
      "learning_rate": 1.854325844930817e-05,
      "loss": 0.268,
      "step": 9635
    },
    {
      "epoch": 0.17405874663257714,
      "grad_norm": 0.7925516963005066,
      "learning_rate": 1.8541784003217023e-05,
      "loss": 0.2649,
      "step": 9640
    },
    {
      "epoch": 0.17414902606547786,
      "grad_norm": 0.5006440877914429,
      "learning_rate": 1.8540308870001456e-05,
      "loss": 0.3122,
      "step": 9645
    },
    {
      "epoch": 0.17423930549837857,
      "grad_norm": 0.39755287766456604,
      "learning_rate": 1.8538833049780136e-05,
      "loss": 0.3285,
      "step": 9650
    },
    {
      "epoch": 0.17432958493127929,
      "grad_norm": 0.32424280047416687,
      "learning_rate": 1.853735654267178e-05,
      "loss": 0.3431,
      "step": 9655
    },
    {
      "epoch": 0.17441986436418,
      "grad_norm": 0.42777007818222046,
      "learning_rate": 1.8535879348795163e-05,
      "loss": 0.2386,
      "step": 9660
    },
    {
      "epoch": 0.1745101437970807,
      "grad_norm": 0.3951456546783447,
      "learning_rate": 1.8534401468269113e-05,
      "loss": 0.3338,
      "step": 9665
    },
    {
      "epoch": 0.17460042322998143,
      "grad_norm": 0.5320587754249573,
      "learning_rate": 1.853292290121252e-05,
      "loss": 0.289,
      "step": 9670
    },
    {
      "epoch": 0.17469070266288214,
      "grad_norm": 0.4082178771495819,
      "learning_rate": 1.8531443647744317e-05,
      "loss": 0.3192,
      "step": 9675
    },
    {
      "epoch": 0.17478098209578286,
      "grad_norm": 0.45076894760131836,
      "learning_rate": 1.85299637079835e-05,
      "loss": 0.3045,
      "step": 9680
    },
    {
      "epoch": 0.17487126152868357,
      "grad_norm": 0.39481931924819946,
      "learning_rate": 1.8528483082049124e-05,
      "loss": 0.3346,
      "step": 9685
    },
    {
      "epoch": 0.17496154096158428,
      "grad_norm": 0.3986913859844208,
      "learning_rate": 1.8527001770060292e-05,
      "loss": 0.4209,
      "step": 9690
    },
    {
      "epoch": 0.175051820394485,
      "grad_norm": 0.3154614269733429,
      "learning_rate": 1.8525519772136166e-05,
      "loss": 0.3049,
      "step": 9695
    },
    {
      "epoch": 0.1751420998273857,
      "grad_norm": 0.31176674365997314,
      "learning_rate": 1.8524037088395957e-05,
      "loss": 0.2899,
      "step": 9700
    },
    {
      "epoch": 0.17523237926028642,
      "grad_norm": 0.7537745833396912,
      "learning_rate": 1.8522553718958946e-05,
      "loss": 0.3217,
      "step": 9705
    },
    {
      "epoch": 0.17532265869318717,
      "grad_norm": 0.38892337679862976,
      "learning_rate": 1.8521069663944448e-05,
      "loss": 0.2509,
      "step": 9710
    },
    {
      "epoch": 0.17541293812608788,
      "grad_norm": 0.40880605578422546,
      "learning_rate": 1.851958492347185e-05,
      "loss": 0.3098,
      "step": 9715
    },
    {
      "epoch": 0.1755032175589886,
      "grad_norm": 0.4116984009742737,
      "learning_rate": 1.851809949766059e-05,
      "loss": 0.3527,
      "step": 9720
    },
    {
      "epoch": 0.1755934969918893,
      "grad_norm": 0.36210495233535767,
      "learning_rate": 1.851661338663016e-05,
      "loss": 0.2324,
      "step": 9725
    },
    {
      "epoch": 0.17568377642479002,
      "grad_norm": 0.43543070554733276,
      "learning_rate": 1.8515126590500104e-05,
      "loss": 0.2956,
      "step": 9730
    },
    {
      "epoch": 0.17577405585769074,
      "grad_norm": 0.5375482439994812,
      "learning_rate": 1.851363910939002e-05,
      "loss": 0.3565,
      "step": 9735
    },
    {
      "epoch": 0.17586433529059145,
      "grad_norm": 0.48285213112831116,
      "learning_rate": 1.8512150943419573e-05,
      "loss": 0.3498,
      "step": 9740
    },
    {
      "epoch": 0.17595461472349216,
      "grad_norm": 0.5422080159187317,
      "learning_rate": 1.8510662092708473e-05,
      "loss": 0.369,
      "step": 9745
    },
    {
      "epoch": 0.17604489415639288,
      "grad_norm": 0.6598948836326599,
      "learning_rate": 1.8509172557376484e-05,
      "loss": 0.336,
      "step": 9750
    },
    {
      "epoch": 0.1761351735892936,
      "grad_norm": 0.5499585270881653,
      "learning_rate": 1.8507682337543432e-05,
      "loss": 0.2605,
      "step": 9755
    },
    {
      "epoch": 0.1762254530221943,
      "grad_norm": 0.4877864420413971,
      "learning_rate": 1.8506191433329196e-05,
      "loss": 0.1437,
      "step": 9760
    },
    {
      "epoch": 0.17631573245509502,
      "grad_norm": 0.7126004695892334,
      "learning_rate": 1.8504699844853697e-05,
      "loss": 0.2657,
      "step": 9765
    },
    {
      "epoch": 0.17640601188799573,
      "grad_norm": 0.3607148230075836,
      "learning_rate": 1.8503207572236938e-05,
      "loss": 0.2516,
      "step": 9770
    },
    {
      "epoch": 0.17649629132089645,
      "grad_norm": 0.5653864741325378,
      "learning_rate": 1.850171461559895e-05,
      "loss": 0.3246,
      "step": 9775
    },
    {
      "epoch": 0.17658657075379716,
      "grad_norm": 0.35818910598754883,
      "learning_rate": 1.8500220975059833e-05,
      "loss": 0.3174,
      "step": 9780
    },
    {
      "epoch": 0.17667685018669788,
      "grad_norm": 0.4652705192565918,
      "learning_rate": 1.8498726650739747e-05,
      "loss": 0.3435,
      "step": 9785
    },
    {
      "epoch": 0.1767671296195986,
      "grad_norm": 0.27703943848609924,
      "learning_rate": 1.849723164275889e-05,
      "loss": 0.305,
      "step": 9790
    },
    {
      "epoch": 0.1768574090524993,
      "grad_norm": 0.5585147738456726,
      "learning_rate": 1.849573595123753e-05,
      "loss": 0.2885,
      "step": 9795
    },
    {
      "epoch": 0.17694768848540002,
      "grad_norm": 0.413076788187027,
      "learning_rate": 1.8494239576295982e-05,
      "loss": 0.2558,
      "step": 9800
    },
    {
      "epoch": 0.17703796791830073,
      "grad_norm": 0.545178234577179,
      "learning_rate": 1.849274251805462e-05,
      "loss": 0.265,
      "step": 9805
    },
    {
      "epoch": 0.17712824735120145,
      "grad_norm": 0.42601487040519714,
      "learning_rate": 1.8491244776633868e-05,
      "loss": 0.3653,
      "step": 9810
    },
    {
      "epoch": 0.17721852678410216,
      "grad_norm": 0.32546499371528625,
      "learning_rate": 1.848974635215421e-05,
      "loss": 0.3098,
      "step": 9815
    },
    {
      "epoch": 0.17730880621700287,
      "grad_norm": 0.8420270085334778,
      "learning_rate": 1.848824724473619e-05,
      "loss": 0.3444,
      "step": 9820
    },
    {
      "epoch": 0.1773990856499036,
      "grad_norm": 0.6668725609779358,
      "learning_rate": 1.8486747454500392e-05,
      "loss": 0.3269,
      "step": 9825
    },
    {
      "epoch": 0.1774893650828043,
      "grad_norm": 0.3488863706588745,
      "learning_rate": 1.848524698156746e-05,
      "loss": 0.3749,
      "step": 9830
    },
    {
      "epoch": 0.17757964451570502,
      "grad_norm": 0.5781786441802979,
      "learning_rate": 1.848374582605811e-05,
      "loss": 0.1893,
      "step": 9835
    },
    {
      "epoch": 0.17766992394860573,
      "grad_norm": 0.4205300211906433,
      "learning_rate": 1.848224398809309e-05,
      "loss": 0.2878,
      "step": 9840
    },
    {
      "epoch": 0.17776020338150644,
      "grad_norm": 0.5582265853881836,
      "learning_rate": 1.848074146779321e-05,
      "loss": 0.3293,
      "step": 9845
    },
    {
      "epoch": 0.17785048281440716,
      "grad_norm": 1.3087825775146484,
      "learning_rate": 1.8479238265279343e-05,
      "loss": 0.2873,
      "step": 9850
    },
    {
      "epoch": 0.17794076224730787,
      "grad_norm": 0.4465949535369873,
      "learning_rate": 1.8477734380672406e-05,
      "loss": 0.2533,
      "step": 9855
    },
    {
      "epoch": 0.17803104168020858,
      "grad_norm": 0.45469945669174194,
      "learning_rate": 1.8476229814093375e-05,
      "loss": 0.3821,
      "step": 9860
    },
    {
      "epoch": 0.1781213211131093,
      "grad_norm": 0.4225321114063263,
      "learning_rate": 1.8474724565663286e-05,
      "loss": 0.298,
      "step": 9865
    },
    {
      "epoch": 0.17821160054601,
      "grad_norm": 0.4197310507297516,
      "learning_rate": 1.8473218635503223e-05,
      "loss": 0.3406,
      "step": 9870
    },
    {
      "epoch": 0.17830187997891073,
      "grad_norm": 0.6299892663955688,
      "learning_rate": 1.8471712023734328e-05,
      "loss": 0.234,
      "step": 9875
    },
    {
      "epoch": 0.17839215941181144,
      "grad_norm": 0.8793956637382507,
      "learning_rate": 1.8470204730477794e-05,
      "loss": 0.2818,
      "step": 9880
    },
    {
      "epoch": 0.17848243884471215,
      "grad_norm": 0.2909405827522278,
      "learning_rate": 1.8468696755854876e-05,
      "loss": 0.2264,
      "step": 9885
    },
    {
      "epoch": 0.17857271827761287,
      "grad_norm": 0.43145665526390076,
      "learning_rate": 1.8467188099986875e-05,
      "loss": 0.324,
      "step": 9890
    },
    {
      "epoch": 0.17866299771051358,
      "grad_norm": 0.37526363134384155,
      "learning_rate": 1.846567876299515e-05,
      "loss": 0.381,
      "step": 9895
    },
    {
      "epoch": 0.1787532771434143,
      "grad_norm": 0.4539874792098999,
      "learning_rate": 1.8464168745001128e-05,
      "loss": 0.2594,
      "step": 9900
    },
    {
      "epoch": 0.178843556576315,
      "grad_norm": 0.45686039328575134,
      "learning_rate": 1.8462658046126267e-05,
      "loss": 0.3893,
      "step": 9905
    },
    {
      "epoch": 0.17893383600921572,
      "grad_norm": 0.45958003401756287,
      "learning_rate": 1.8461146666492097e-05,
      "loss": 0.3155,
      "step": 9910
    },
    {
      "epoch": 0.17902411544211644,
      "grad_norm": 0.5699650645256042,
      "learning_rate": 1.8459634606220194e-05,
      "loss": 0.262,
      "step": 9915
    },
    {
      "epoch": 0.17911439487501715,
      "grad_norm": 0.48587679862976074,
      "learning_rate": 1.8458121865432196e-05,
      "loss": 0.2676,
      "step": 9920
    },
    {
      "epoch": 0.17920467430791787,
      "grad_norm": 0.33711960911750793,
      "learning_rate": 1.845660844424979e-05,
      "loss": 0.2437,
      "step": 9925
    },
    {
      "epoch": 0.17929495374081858,
      "grad_norm": 0.44800063967704773,
      "learning_rate": 1.845509434279472e-05,
      "loss": 0.3275,
      "step": 9930
    },
    {
      "epoch": 0.1793852331737193,
      "grad_norm": 0.630777895450592,
      "learning_rate": 1.8453579561188786e-05,
      "loss": 0.2771,
      "step": 9935
    },
    {
      "epoch": 0.17947551260662,
      "grad_norm": 0.5990297198295593,
      "learning_rate": 1.8452064099553836e-05,
      "loss": 0.2691,
      "step": 9940
    },
    {
      "epoch": 0.17956579203952072,
      "grad_norm": 0.42177292704582214,
      "learning_rate": 1.8450547958011788e-05,
      "loss": 0.3546,
      "step": 9945
    },
    {
      "epoch": 0.17965607147242144,
      "grad_norm": 0.39384013414382935,
      "learning_rate": 1.8449031136684593e-05,
      "loss": 0.2662,
      "step": 9950
    },
    {
      "epoch": 0.17974635090532215,
      "grad_norm": 0.367742657661438,
      "learning_rate": 1.8447513635694274e-05,
      "loss": 0.2673,
      "step": 9955
    },
    {
      "epoch": 0.17983663033822286,
      "grad_norm": 0.43818777799606323,
      "learning_rate": 1.8445995455162904e-05,
      "loss": 0.2634,
      "step": 9960
    },
    {
      "epoch": 0.17992690977112358,
      "grad_norm": 0.3801169693470001,
      "learning_rate": 1.8444476595212607e-05,
      "loss": 0.2969,
      "step": 9965
    },
    {
      "epoch": 0.1800171892040243,
      "grad_norm": 0.40738508105278015,
      "learning_rate": 1.8442957055965566e-05,
      "loss": 0.2861,
      "step": 9970
    },
    {
      "epoch": 0.180107468636925,
      "grad_norm": 0.40975838899612427,
      "learning_rate": 1.8441436837544014e-05,
      "loss": 0.2714,
      "step": 9975
    },
    {
      "epoch": 0.18019774806982572,
      "grad_norm": 0.3722114562988281,
      "learning_rate": 1.8439915940070247e-05,
      "loss": 0.2468,
      "step": 9980
    },
    {
      "epoch": 0.18028802750272643,
      "grad_norm": 0.44791749119758606,
      "learning_rate": 1.8438394363666603e-05,
      "loss": 0.2855,
      "step": 9985
    },
    {
      "epoch": 0.18037830693562715,
      "grad_norm": 0.6452430486679077,
      "learning_rate": 1.8436872108455487e-05,
      "loss": 0.3599,
      "step": 9990
    },
    {
      "epoch": 0.18046858636852786,
      "grad_norm": 0.4257093369960785,
      "learning_rate": 1.8435349174559355e-05,
      "loss": 0.3286,
      "step": 9995
    },
    {
      "epoch": 0.18055886580142858,
      "grad_norm": 0.4258350431919098,
      "learning_rate": 1.8433825562100708e-05,
      "loss": 0.2843,
      "step": 10000
    },
    {
      "epoch": 0.1806491452343293,
      "grad_norm": 0.3578794002532959,
      "learning_rate": 1.8432301271202117e-05,
      "loss": 0.3294,
      "step": 10005
    },
    {
      "epoch": 0.18073942466723,
      "grad_norm": 0.47850295901298523,
      "learning_rate": 1.8430776301986196e-05,
      "loss": 0.3223,
      "step": 10010
    },
    {
      "epoch": 0.18082970410013072,
      "grad_norm": 0.44029831886291504,
      "learning_rate": 1.842925065457562e-05,
      "loss": 0.3394,
      "step": 10015
    },
    {
      "epoch": 0.18091998353303143,
      "grad_norm": 0.5738121271133423,
      "learning_rate": 1.842772432909312e-05,
      "loss": 0.3103,
      "step": 10020
    },
    {
      "epoch": 0.18101026296593214,
      "grad_norm": 0.7582994699478149,
      "learning_rate": 1.8426197325661466e-05,
      "loss": 0.2223,
      "step": 10025
    },
    {
      "epoch": 0.18110054239883286,
      "grad_norm": 0.3985866904258728,
      "learning_rate": 1.8424669644403505e-05,
      "loss": 0.3643,
      "step": 10030
    },
    {
      "epoch": 0.18119082183173357,
      "grad_norm": 0.5045907497406006,
      "learning_rate": 1.8423141285442123e-05,
      "loss": 0.2353,
      "step": 10035
    },
    {
      "epoch": 0.1812811012646343,
      "grad_norm": 0.3531908690929413,
      "learning_rate": 1.842161224890027e-05,
      "loss": 0.3548,
      "step": 10040
    },
    {
      "epoch": 0.181371380697535,
      "grad_norm": 0.34367236495018005,
      "learning_rate": 1.8420082534900934e-05,
      "loss": 0.2043,
      "step": 10045
    },
    {
      "epoch": 0.18146166013043571,
      "grad_norm": 0.6093673706054688,
      "learning_rate": 1.8418552143567184e-05,
      "loss": 0.3982,
      "step": 10050
    },
    {
      "epoch": 0.18155193956333643,
      "grad_norm": 0.5029755234718323,
      "learning_rate": 1.8417021075022122e-05,
      "loss": 0.3863,
      "step": 10055
    },
    {
      "epoch": 0.18164221899623714,
      "grad_norm": 0.46935874223709106,
      "learning_rate": 1.8415489329388914e-05,
      "loss": 0.3116,
      "step": 10060
    },
    {
      "epoch": 0.18173249842913786,
      "grad_norm": 0.9425585865974426,
      "learning_rate": 1.8413956906790776e-05,
      "loss": 0.2711,
      "step": 10065
    },
    {
      "epoch": 0.18182277786203857,
      "grad_norm": 0.43612468242645264,
      "learning_rate": 1.8412423807350978e-05,
      "loss": 0.3326,
      "step": 10070
    },
    {
      "epoch": 0.18191305729493928,
      "grad_norm": 0.39147958159446716,
      "learning_rate": 1.8410890031192852e-05,
      "loss": 0.319,
      "step": 10075
    },
    {
      "epoch": 0.18200333672784,
      "grad_norm": 0.445437490940094,
      "learning_rate": 1.8409355578439773e-05,
      "loss": 0.2219,
      "step": 10080
    },
    {
      "epoch": 0.1820936161607407,
      "grad_norm": 0.3498551547527313,
      "learning_rate": 1.840782044921518e-05,
      "loss": 0.3016,
      "step": 10085
    },
    {
      "epoch": 0.18218389559364143,
      "grad_norm": 0.5700546503067017,
      "learning_rate": 1.8406284643642562e-05,
      "loss": 0.3446,
      "step": 10090
    },
    {
      "epoch": 0.18227417502654214,
      "grad_norm": 0.3117503821849823,
      "learning_rate": 1.8404748161845467e-05,
      "loss": 0.286,
      "step": 10095
    },
    {
      "epoch": 0.18236445445944285,
      "grad_norm": 0.35964101552963257,
      "learning_rate": 1.840321100394749e-05,
      "loss": 0.2157,
      "step": 10100
    },
    {
      "epoch": 0.18245473389234357,
      "grad_norm": 0.4494312107563019,
      "learning_rate": 1.8401673170072283e-05,
      "loss": 0.294,
      "step": 10105
    },
    {
      "epoch": 0.1825450133252443,
      "grad_norm": 0.39130812883377075,
      "learning_rate": 1.8400134660343555e-05,
      "loss": 0.2839,
      "step": 10110
    },
    {
      "epoch": 0.18263529275814502,
      "grad_norm": 0.6958088874816895,
      "learning_rate": 1.839859547488507e-05,
      "loss": 0.3133,
      "step": 10115
    },
    {
      "epoch": 0.18272557219104574,
      "grad_norm": 0.5423473119735718,
      "learning_rate": 1.8397055613820644e-05,
      "loss": 0.3583,
      "step": 10120
    },
    {
      "epoch": 0.18281585162394645,
      "grad_norm": 0.521397590637207,
      "learning_rate": 1.8395515077274146e-05,
      "loss": 0.316,
      "step": 10125
    },
    {
      "epoch": 0.18290613105684717,
      "grad_norm": 0.5494081377983093,
      "learning_rate": 1.83939738653695e-05,
      "loss": 0.3184,
      "step": 10130
    },
    {
      "epoch": 0.18299641048974788,
      "grad_norm": 0.49178290367126465,
      "learning_rate": 1.839243197823069e-05,
      "loss": 0.2045,
      "step": 10135
    },
    {
      "epoch": 0.1830866899226486,
      "grad_norm": 0.3609885275363922,
      "learning_rate": 1.839088941598174e-05,
      "loss": 0.2909,
      "step": 10140
    },
    {
      "epoch": 0.1831769693555493,
      "grad_norm": 0.2735249698162079,
      "learning_rate": 1.8389346178746747e-05,
      "loss": 0.2842,
      "step": 10145
    },
    {
      "epoch": 0.18326724878845002,
      "grad_norm": 0.42014428973197937,
      "learning_rate": 1.8387802266649854e-05,
      "loss": 0.3625,
      "step": 10150
    },
    {
      "epoch": 0.18335752822135074,
      "grad_norm": 0.3797282874584198,
      "learning_rate": 1.8386257679815253e-05,
      "loss": 0.3603,
      "step": 10155
    },
    {
      "epoch": 0.18344780765425145,
      "grad_norm": 0.8530354499816895,
      "learning_rate": 1.838471241836719e-05,
      "loss": 0.3365,
      "step": 10160
    },
    {
      "epoch": 0.18353808708715216,
      "grad_norm": 0.4313885271549225,
      "learning_rate": 1.838316648242998e-05,
      "loss": 0.2984,
      "step": 10165
    },
    {
      "epoch": 0.18362836652005288,
      "grad_norm": 0.41217172145843506,
      "learning_rate": 1.8381619872127978e-05,
      "loss": 0.2754,
      "step": 10170
    },
    {
      "epoch": 0.1837186459529536,
      "grad_norm": 0.41457244753837585,
      "learning_rate": 1.8380072587585597e-05,
      "loss": 0.4513,
      "step": 10175
    },
    {
      "epoch": 0.1838089253858543,
      "grad_norm": 0.550830602645874,
      "learning_rate": 1.8378524628927302e-05,
      "loss": 0.2716,
      "step": 10180
    },
    {
      "epoch": 0.18389920481875502,
      "grad_norm": 0.7130752801895142,
      "learning_rate": 1.837697599627762e-05,
      "loss": 0.2783,
      "step": 10185
    },
    {
      "epoch": 0.18398948425165573,
      "grad_norm": 0.4837576448917389,
      "learning_rate": 1.837542668976113e-05,
      "loss": 0.1878,
      "step": 10190
    },
    {
      "epoch": 0.18407976368455645,
      "grad_norm": 0.46950626373291016,
      "learning_rate": 1.8373876709502455e-05,
      "loss": 0.3807,
      "step": 10195
    },
    {
      "epoch": 0.18417004311745716,
      "grad_norm": 0.5759735107421875,
      "learning_rate": 1.8372326055626285e-05,
      "loss": 0.3412,
      "step": 10200
    },
    {
      "epoch": 0.18426032255035787,
      "grad_norm": 0.6385390162467957,
      "learning_rate": 1.837077472825735e-05,
      "loss": 0.3626,
      "step": 10205
    },
    {
      "epoch": 0.1843506019832586,
      "grad_norm": 0.5239001512527466,
      "learning_rate": 1.8369222727520455e-05,
      "loss": 0.3327,
      "step": 10210
    },
    {
      "epoch": 0.1844408814161593,
      "grad_norm": 0.43816807866096497,
      "learning_rate": 1.836767005354044e-05,
      "loss": 0.2671,
      "step": 10215
    },
    {
      "epoch": 0.18453116084906002,
      "grad_norm": 0.8781191110610962,
      "learning_rate": 1.836611670644221e-05,
      "loss": 0.34,
      "step": 10220
    },
    {
      "epoch": 0.18462144028196073,
      "grad_norm": 0.3327060341835022,
      "learning_rate": 1.836456268635072e-05,
      "loss": 0.2622,
      "step": 10225
    },
    {
      "epoch": 0.18471171971486144,
      "grad_norm": 0.338997483253479,
      "learning_rate": 1.8363007993390973e-05,
      "loss": 0.2391,
      "step": 10230
    },
    {
      "epoch": 0.18480199914776216,
      "grad_norm": 0.3980759382247925,
      "learning_rate": 1.836145262768804e-05,
      "loss": 0.3819,
      "step": 10235
    },
    {
      "epoch": 0.18489227858066287,
      "grad_norm": 0.57769376039505,
      "learning_rate": 1.8359896589367035e-05,
      "loss": 0.2773,
      "step": 10240
    },
    {
      "epoch": 0.18498255801356359,
      "grad_norm": 0.30585551261901855,
      "learning_rate": 1.8358339878553134e-05,
      "loss": 0.3717,
      "step": 10245
    },
    {
      "epoch": 0.1850728374464643,
      "grad_norm": 0.5136041641235352,
      "learning_rate": 1.835678249537156e-05,
      "loss": 0.3142,
      "step": 10250
    },
    {
      "epoch": 0.18516311687936501,
      "grad_norm": 0.4172489047050476,
      "learning_rate": 1.8355224439947594e-05,
      "loss": 0.3688,
      "step": 10255
    },
    {
      "epoch": 0.18525339631226573,
      "grad_norm": 0.6374697685241699,
      "learning_rate": 1.8353665712406572e-05,
      "loss": 0.3623,
      "step": 10260
    },
    {
      "epoch": 0.18534367574516644,
      "grad_norm": 0.3242988586425781,
      "learning_rate": 1.835210631287388e-05,
      "loss": 0.1904,
      "step": 10265
    },
    {
      "epoch": 0.18543395517806716,
      "grad_norm": 0.46144676208496094,
      "learning_rate": 1.8350546241474957e-05,
      "loss": 0.3244,
      "step": 10270
    },
    {
      "epoch": 0.18552423461096787,
      "grad_norm": 0.46591830253601074,
      "learning_rate": 1.8348985498335306e-05,
      "loss": 0.3928,
      "step": 10275
    },
    {
      "epoch": 0.18561451404386858,
      "grad_norm": 0.4382494390010834,
      "learning_rate": 1.8347424083580473e-05,
      "loss": 0.3477,
      "step": 10280
    },
    {
      "epoch": 0.1857047934767693,
      "grad_norm": 1.1415561437606812,
      "learning_rate": 1.8345861997336067e-05,
      "loss": 0.2341,
      "step": 10285
    },
    {
      "epoch": 0.18579507290967,
      "grad_norm": 0.403601735830307,
      "learning_rate": 1.834429923972774e-05,
      "loss": 0.3129,
      "step": 10290
    },
    {
      "epoch": 0.18588535234257073,
      "grad_norm": 1.1687211990356445,
      "learning_rate": 1.8342735810881214e-05,
      "loss": 0.3114,
      "step": 10295
    },
    {
      "epoch": 0.18597563177547144,
      "grad_norm": 0.42405131459236145,
      "learning_rate": 1.8341171710922244e-05,
      "loss": 0.3974,
      "step": 10300
    },
    {
      "epoch": 0.18606591120837215,
      "grad_norm": 1.2109848260879517,
      "learning_rate": 1.833960693997666e-05,
      "loss": 0.3509,
      "step": 10305
    },
    {
      "epoch": 0.18615619064127287,
      "grad_norm": 0.6749964952468872,
      "learning_rate": 1.833804149817033e-05,
      "loss": 0.2898,
      "step": 10310
    },
    {
      "epoch": 0.18624647007417358,
      "grad_norm": 0.5378634333610535,
      "learning_rate": 1.8336475385629186e-05,
      "loss": 0.2738,
      "step": 10315
    },
    {
      "epoch": 0.1863367495070743,
      "grad_norm": 0.5143244862556458,
      "learning_rate": 1.8334908602479206e-05,
      "loss": 0.2809,
      "step": 10320
    },
    {
      "epoch": 0.186427028939975,
      "grad_norm": 0.3657784163951874,
      "learning_rate": 1.833334114884643e-05,
      "loss": 0.2773,
      "step": 10325
    },
    {
      "epoch": 0.18651730837287572,
      "grad_norm": 0.5001368522644043,
      "learning_rate": 1.8331773024856953e-05,
      "loss": 0.2747,
      "step": 10330
    },
    {
      "epoch": 0.18660758780577644,
      "grad_norm": 0.4016692340373993,
      "learning_rate": 1.833020423063691e-05,
      "loss": 0.3401,
      "step": 10335
    },
    {
      "epoch": 0.18669786723867715,
      "grad_norm": 0.47351789474487305,
      "learning_rate": 1.8328634766312505e-05,
      "loss": 0.2704,
      "step": 10340
    },
    {
      "epoch": 0.18678814667157786,
      "grad_norm": 0.39467868208885193,
      "learning_rate": 1.8327064632009986e-05,
      "loss": 0.3595,
      "step": 10345
    },
    {
      "epoch": 0.18687842610447858,
      "grad_norm": 0.5907410979270935,
      "learning_rate": 1.8325493827855664e-05,
      "loss": 0.3104,
      "step": 10350
    },
    {
      "epoch": 0.1869687055373793,
      "grad_norm": 0.4638453722000122,
      "learning_rate": 1.832392235397589e-05,
      "loss": 0.18,
      "step": 10355
    },
    {
      "epoch": 0.18705898497028,
      "grad_norm": 0.6245995163917542,
      "learning_rate": 1.8322350210497092e-05,
      "loss": 0.2667,
      "step": 10360
    },
    {
      "epoch": 0.18714926440318072,
      "grad_norm": 0.39613252878189087,
      "learning_rate": 1.832077739754572e-05,
      "loss": 0.4438,
      "step": 10365
    },
    {
      "epoch": 0.18723954383608143,
      "grad_norm": 0.5103363990783691,
      "learning_rate": 1.831920391524831e-05,
      "loss": 0.3216,
      "step": 10370
    },
    {
      "epoch": 0.18732982326898215,
      "grad_norm": 0.3647821545600891,
      "learning_rate": 1.831762976373143e-05,
      "loss": 0.2869,
      "step": 10375
    },
    {
      "epoch": 0.18742010270188286,
      "grad_norm": 0.4628177583217621,
      "learning_rate": 1.831605494312171e-05,
      "loss": 0.2764,
      "step": 10380
    },
    {
      "epoch": 0.18751038213478358,
      "grad_norm": 0.23167622089385986,
      "learning_rate": 1.8314479453545836e-05,
      "loss": 0.2092,
      "step": 10385
    },
    {
      "epoch": 0.1876006615676843,
      "grad_norm": 0.39833158254623413,
      "learning_rate": 1.8312903295130538e-05,
      "loss": 0.3064,
      "step": 10390
    },
    {
      "epoch": 0.187690941000585,
      "grad_norm": 0.5542736053466797,
      "learning_rate": 1.8311326468002613e-05,
      "loss": 0.2753,
      "step": 10395
    },
    {
      "epoch": 0.18778122043348572,
      "grad_norm": 0.46590352058410645,
      "learning_rate": 1.83097489722889e-05,
      "loss": 0.3642,
      "step": 10400
    },
    {
      "epoch": 0.18787149986638643,
      "grad_norm": 0.41070303320884705,
      "learning_rate": 1.83081708081163e-05,
      "loss": 0.3594,
      "step": 10405
    },
    {
      "epoch": 0.18796177929928715,
      "grad_norm": 0.42542794346809387,
      "learning_rate": 1.830659197561177e-05,
      "loss": 0.2966,
      "step": 10410
    },
    {
      "epoch": 0.18805205873218786,
      "grad_norm": 0.3459967076778412,
      "learning_rate": 1.8305012474902305e-05,
      "loss": 0.3753,
      "step": 10415
    },
    {
      "epoch": 0.18814233816508857,
      "grad_norm": 0.35253915190696716,
      "learning_rate": 1.8303432306114972e-05,
      "loss": 0.2909,
      "step": 10420
    },
    {
      "epoch": 0.1882326175979893,
      "grad_norm": 0.34542712569236755,
      "learning_rate": 1.8301851469376885e-05,
      "loss": 0.2579,
      "step": 10425
    },
    {
      "epoch": 0.18832289703089,
      "grad_norm": 0.4709479510784149,
      "learning_rate": 1.8300269964815203e-05,
      "loss": 0.3706,
      "step": 10430
    },
    {
      "epoch": 0.18841317646379072,
      "grad_norm": 0.4012754261493683,
      "learning_rate": 1.829868779255715e-05,
      "loss": 0.2178,
      "step": 10435
    },
    {
      "epoch": 0.18850345589669143,
      "grad_norm": 0.3635656535625458,
      "learning_rate": 1.8297104952730003e-05,
      "loss": 0.3075,
      "step": 10440
    },
    {
      "epoch": 0.18859373532959214,
      "grad_norm": 0.3134946823120117,
      "learning_rate": 1.829552144546109e-05,
      "loss": 0.2807,
      "step": 10445
    },
    {
      "epoch": 0.18868401476249286,
      "grad_norm": 0.3195219933986664,
      "learning_rate": 1.829393727087779e-05,
      "loss": 0.294,
      "step": 10450
    },
    {
      "epoch": 0.18877429419539357,
      "grad_norm": 0.5945743322372437,
      "learning_rate": 1.829235242910754e-05,
      "loss": 0.3601,
      "step": 10455
    },
    {
      "epoch": 0.18886457362829429,
      "grad_norm": 0.41216209530830383,
      "learning_rate": 1.8290766920277826e-05,
      "loss": 0.3322,
      "step": 10460
    },
    {
      "epoch": 0.188954853061195,
      "grad_norm": 0.36902883648872375,
      "learning_rate": 1.8289180744516194e-05,
      "loss": 0.2222,
      "step": 10465
    },
    {
      "epoch": 0.1890451324940957,
      "grad_norm": 0.6106470227241516,
      "learning_rate": 1.8287593901950242e-05,
      "loss": 0.4442,
      "step": 10470
    },
    {
      "epoch": 0.18913541192699643,
      "grad_norm": 0.5889748930931091,
      "learning_rate": 1.8286006392707616e-05,
      "loss": 0.3389,
      "step": 10475
    },
    {
      "epoch": 0.18922569135989714,
      "grad_norm": 0.38097482919692993,
      "learning_rate": 1.8284418216916017e-05,
      "loss": 0.3128,
      "step": 10480
    },
    {
      "epoch": 0.18931597079279786,
      "grad_norm": 0.3452233374118805,
      "learning_rate": 1.828282937470321e-05,
      "loss": 0.2697,
      "step": 10485
    },
    {
      "epoch": 0.18940625022569857,
      "grad_norm": 0.6536673307418823,
      "learning_rate": 1.8281239866197e-05,
      "loss": 0.3097,
      "step": 10490
    },
    {
      "epoch": 0.18949652965859928,
      "grad_norm": 1.0002179145812988,
      "learning_rate": 1.8279649691525255e-05,
      "loss": 0.449,
      "step": 10495
    },
    {
      "epoch": 0.1895868090915,
      "grad_norm": 0.3609928488731384,
      "learning_rate": 1.827805885081589e-05,
      "loss": 0.236,
      "step": 10500
    },
    {
      "epoch": 0.18967708852440074,
      "grad_norm": 0.4937584102153778,
      "learning_rate": 1.8276467344196876e-05,
      "loss": 0.363,
      "step": 10505
    },
    {
      "epoch": 0.18976736795730145,
      "grad_norm": 0.4397891163825989,
      "learning_rate": 1.8274875171796244e-05,
      "loss": 0.3586,
      "step": 10510
    },
    {
      "epoch": 0.18985764739020217,
      "grad_norm": 0.40915870666503906,
      "learning_rate": 1.8273282333742065e-05,
      "loss": 0.2808,
      "step": 10515
    },
    {
      "epoch": 0.18994792682310288,
      "grad_norm": 0.36115843057632446,
      "learning_rate": 1.8271688830162477e-05,
      "loss": 0.3517,
      "step": 10520
    },
    {
      "epoch": 0.1900382062560036,
      "grad_norm": 0.6014983057975769,
      "learning_rate": 1.8270094661185662e-05,
      "loss": 0.3273,
      "step": 10525
    },
    {
      "epoch": 0.1901284856889043,
      "grad_norm": 0.5097769498825073,
      "learning_rate": 1.826849982693986e-05,
      "loss": 0.435,
      "step": 10530
    },
    {
      "epoch": 0.19021876512180502,
      "grad_norm": 0.5310569405555725,
      "learning_rate": 1.8266904327553367e-05,
      "loss": 0.3127,
      "step": 10535
    },
    {
      "epoch": 0.19030904455470574,
      "grad_norm": 0.5188735127449036,
      "learning_rate": 1.8265308163154525e-05,
      "loss": 0.3436,
      "step": 10540
    },
    {
      "epoch": 0.19039932398760645,
      "grad_norm": 0.5465673804283142,
      "learning_rate": 1.8263711333871737e-05,
      "loss": 0.2674,
      "step": 10545
    },
    {
      "epoch": 0.19048960342050716,
      "grad_norm": 0.5335100889205933,
      "learning_rate": 1.8262113839833457e-05,
      "loss": 0.3317,
      "step": 10550
    },
    {
      "epoch": 0.19057988285340788,
      "grad_norm": 0.5629705190658569,
      "learning_rate": 1.826051568116819e-05,
      "loss": 0.4154,
      "step": 10555
    },
    {
      "epoch": 0.1906701622863086,
      "grad_norm": 0.4672110378742218,
      "learning_rate": 1.8258916858004493e-05,
      "loss": 0.3633,
      "step": 10560
    },
    {
      "epoch": 0.1907604417192093,
      "grad_norm": 0.6804426908493042,
      "learning_rate": 1.8257317370470983e-05,
      "loss": 0.4061,
      "step": 10565
    },
    {
      "epoch": 0.19085072115211002,
      "grad_norm": 0.44888773560523987,
      "learning_rate": 1.8255717218696333e-05,
      "loss": 0.2871,
      "step": 10570
    },
    {
      "epoch": 0.19094100058501073,
      "grad_norm": 0.4249172806739807,
      "learning_rate": 1.825411640280925e-05,
      "loss": 0.2793,
      "step": 10575
    },
    {
      "epoch": 0.19103128001791145,
      "grad_norm": 0.2748965620994568,
      "learning_rate": 1.825251492293852e-05,
      "loss": 0.382,
      "step": 10580
    },
    {
      "epoch": 0.19112155945081216,
      "grad_norm": 0.5372962355613708,
      "learning_rate": 1.8250912779212965e-05,
      "loss": 0.3791,
      "step": 10585
    },
    {
      "epoch": 0.19121183888371288,
      "grad_norm": 0.4297412633895874,
      "learning_rate": 1.8249309971761465e-05,
      "loss": 0.3343,
      "step": 10590
    },
    {
      "epoch": 0.1913021183166136,
      "grad_norm": 0.49882203340530396,
      "learning_rate": 1.8247706500712957e-05,
      "loss": 0.3254,
      "step": 10595
    },
    {
      "epoch": 0.1913923977495143,
      "grad_norm": 0.37007367610931396,
      "learning_rate": 1.824610236619643e-05,
      "loss": 0.316,
      "step": 10600
    },
    {
      "epoch": 0.19148267718241502,
      "grad_norm": 0.35884764790534973,
      "learning_rate": 1.824449756834092e-05,
      "loss": 0.2357,
      "step": 10605
    },
    {
      "epoch": 0.19157295661531573,
      "grad_norm": 0.35997143387794495,
      "learning_rate": 1.8242892107275526e-05,
      "loss": 0.3166,
      "step": 10610
    },
    {
      "epoch": 0.19166323604821645,
      "grad_norm": 0.41313913464546204,
      "learning_rate": 1.8241285983129393e-05,
      "loss": 0.3539,
      "step": 10615
    },
    {
      "epoch": 0.19175351548111716,
      "grad_norm": 0.3478839695453644,
      "learning_rate": 1.8239679196031722e-05,
      "loss": 0.3434,
      "step": 10620
    },
    {
      "epoch": 0.19184379491401787,
      "grad_norm": 0.4098747670650482,
      "learning_rate": 1.8238071746111768e-05,
      "loss": 0.3339,
      "step": 10625
    },
    {
      "epoch": 0.1919340743469186,
      "grad_norm": 0.587702751159668,
      "learning_rate": 1.8236463633498838e-05,
      "loss": 0.3052,
      "step": 10630
    },
    {
      "epoch": 0.1920243537798193,
      "grad_norm": 0.7023888230323792,
      "learning_rate": 1.8234854858322292e-05,
      "loss": 0.2935,
      "step": 10635
    },
    {
      "epoch": 0.19211463321272002,
      "grad_norm": 0.40667617321014404,
      "learning_rate": 1.823324542071155e-05,
      "loss": 0.3309,
      "step": 10640
    },
    {
      "epoch": 0.19220491264562073,
      "grad_norm": 0.3113376498222351,
      "learning_rate": 1.8231635320796072e-05,
      "loss": 0.3944,
      "step": 10645
    },
    {
      "epoch": 0.19229519207852144,
      "grad_norm": 0.38273394107818604,
      "learning_rate": 1.8230024558705384e-05,
      "loss": 0.2757,
      "step": 10650
    },
    {
      "epoch": 0.19238547151142216,
      "grad_norm": 0.38133129477500916,
      "learning_rate": 1.8228413134569058e-05,
      "loss": 0.3228,
      "step": 10655
    },
    {
      "epoch": 0.19247575094432287,
      "grad_norm": 0.5952944755554199,
      "learning_rate": 1.822680104851672e-05,
      "loss": 0.3121,
      "step": 10660
    },
    {
      "epoch": 0.19256603037722358,
      "grad_norm": 0.4904284179210663,
      "learning_rate": 1.8225188300678055e-05,
      "loss": 0.3406,
      "step": 10665
    },
    {
      "epoch": 0.1926563098101243,
      "grad_norm": 0.5586575865745544,
      "learning_rate": 1.8223574891182793e-05,
      "loss": 0.2835,
      "step": 10670
    },
    {
      "epoch": 0.192746589243025,
      "grad_norm": 0.4699321687221527,
      "learning_rate": 1.8221960820160722e-05,
      "loss": 0.3106,
      "step": 10675
    },
    {
      "epoch": 0.19283686867592573,
      "grad_norm": 0.756039559841156,
      "learning_rate": 1.822034608774168e-05,
      "loss": 0.212,
      "step": 10680
    },
    {
      "epoch": 0.19292714810882644,
      "grad_norm": 0.43197789788246155,
      "learning_rate": 1.8218730694055566e-05,
      "loss": 0.2714,
      "step": 10685
    },
    {
      "epoch": 0.19301742754172715,
      "grad_norm": 0.5693886876106262,
      "learning_rate": 1.8217114639232327e-05,
      "loss": 0.4156,
      "step": 10690
    },
    {
      "epoch": 0.19310770697462787,
      "grad_norm": 0.45394495129585266,
      "learning_rate": 1.8215497923401953e-05,
      "loss": 0.3603,
      "step": 10695
    },
    {
      "epoch": 0.19319798640752858,
      "grad_norm": 0.6002922058105469,
      "learning_rate": 1.8213880546694506e-05,
      "loss": 0.3311,
      "step": 10700
    },
    {
      "epoch": 0.1932882658404293,
      "grad_norm": 0.3669627606868744,
      "learning_rate": 1.8212262509240092e-05,
      "loss": 0.3286,
      "step": 10705
    },
    {
      "epoch": 0.19337854527333,
      "grad_norm": 0.3457637429237366,
      "learning_rate": 1.8210643811168863e-05,
      "loss": 0.3186,
      "step": 10710
    },
    {
      "epoch": 0.19346882470623072,
      "grad_norm": 0.32688283920288086,
      "learning_rate": 1.820902445261104e-05,
      "loss": 0.2353,
      "step": 10715
    },
    {
      "epoch": 0.19355910413913144,
      "grad_norm": 0.5048171281814575,
      "learning_rate": 1.8207404433696883e-05,
      "loss": 0.2595,
      "step": 10720
    },
    {
      "epoch": 0.19364938357203215,
      "grad_norm": 0.4015275835990906,
      "learning_rate": 1.8205783754556717e-05,
      "loss": 0.2992,
      "step": 10725
    },
    {
      "epoch": 0.19373966300493287,
      "grad_norm": 0.5477710366249084,
      "learning_rate": 1.820416241532091e-05,
      "loss": 0.3586,
      "step": 10730
    },
    {
      "epoch": 0.19382994243783358,
      "grad_norm": 0.4308444857597351,
      "learning_rate": 1.820254041611988e-05,
      "loss": 0.3833,
      "step": 10735
    },
    {
      "epoch": 0.1939202218707343,
      "grad_norm": 0.6709811091423035,
      "learning_rate": 1.8200917757084117e-05,
      "loss": 0.2602,
      "step": 10740
    },
    {
      "epoch": 0.194010501303635,
      "grad_norm": 0.34628912806510925,
      "learning_rate": 1.8199294438344145e-05,
      "loss": 0.3714,
      "step": 10745
    },
    {
      "epoch": 0.19410078073653572,
      "grad_norm": 0.4252835214138031,
      "learning_rate": 1.819767046003055e-05,
      "loss": 0.3544,
      "step": 10750
    },
    {
      "epoch": 0.19419106016943644,
      "grad_norm": 0.6269519329071045,
      "learning_rate": 1.819604582227397e-05,
      "loss": 0.2497,
      "step": 10755
    },
    {
      "epoch": 0.19428133960233715,
      "grad_norm": 0.3291480541229248,
      "learning_rate": 1.8194420525205094e-05,
      "loss": 0.3225,
      "step": 10760
    },
    {
      "epoch": 0.19437161903523786,
      "grad_norm": 0.33760547637939453,
      "learning_rate": 1.8192794568954667e-05,
      "loss": 0.2476,
      "step": 10765
    },
    {
      "epoch": 0.19446189846813858,
      "grad_norm": 0.5854130983352661,
      "learning_rate": 1.8191167953653483e-05,
      "loss": 0.419,
      "step": 10770
    },
    {
      "epoch": 0.1945521779010393,
      "grad_norm": 0.5488379001617432,
      "learning_rate": 1.8189540679432393e-05,
      "loss": 0.3921,
      "step": 10775
    },
    {
      "epoch": 0.19464245733394,
      "grad_norm": 0.4348668158054352,
      "learning_rate": 1.81879127464223e-05,
      "loss": 0.2644,
      "step": 10780
    },
    {
      "epoch": 0.19473273676684072,
      "grad_norm": 0.4301225244998932,
      "learning_rate": 1.8186284154754158e-05,
      "loss": 0.366,
      "step": 10785
    },
    {
      "epoch": 0.19482301619974143,
      "grad_norm": 0.43786948919296265,
      "learning_rate": 1.8184654904558976e-05,
      "loss": 0.3075,
      "step": 10790
    },
    {
      "epoch": 0.19491329563264215,
      "grad_norm": 0.440784752368927,
      "learning_rate": 1.8183024995967815e-05,
      "loss": 0.3134,
      "step": 10795
    },
    {
      "epoch": 0.19500357506554286,
      "grad_norm": 0.4838646352291107,
      "learning_rate": 1.818139442911179e-05,
      "loss": 0.2791,
      "step": 10800
    },
    {
      "epoch": 0.19509385449844358,
      "grad_norm": 0.3454958200454712,
      "learning_rate": 1.8179763204122066e-05,
      "loss": 0.383,
      "step": 10805
    },
    {
      "epoch": 0.1951841339313443,
      "grad_norm": 0.49316585063934326,
      "learning_rate": 1.8178131321129865e-05,
      "loss": 0.379,
      "step": 10810
    },
    {
      "epoch": 0.195274413364245,
      "grad_norm": 0.7945411205291748,
      "learning_rate": 1.8176498780266462e-05,
      "loss": 0.3332,
      "step": 10815
    },
    {
      "epoch": 0.19536469279714572,
      "grad_norm": 0.6220671534538269,
      "learning_rate": 1.817486558166318e-05,
      "loss": 0.3037,
      "step": 10820
    },
    {
      "epoch": 0.19545497223004643,
      "grad_norm": 0.4711354672908783,
      "learning_rate": 1.8173231725451395e-05,
      "loss": 0.2762,
      "step": 10825
    },
    {
      "epoch": 0.19554525166294714,
      "grad_norm": 0.5520678758621216,
      "learning_rate": 1.817159721176255e-05,
      "loss": 0.4785,
      "step": 10830
    },
    {
      "epoch": 0.19563553109584786,
      "grad_norm": 0.43351519107818604,
      "learning_rate": 1.816996204072812e-05,
      "loss": 0.2783,
      "step": 10835
    },
    {
      "epoch": 0.19572581052874857,
      "grad_norm": 0.4619917571544647,
      "learning_rate": 1.8168326212479643e-05,
      "loss": 0.4033,
      "step": 10840
    },
    {
      "epoch": 0.1958160899616493,
      "grad_norm": 0.3751065135002136,
      "learning_rate": 1.8166689727148715e-05,
      "loss": 0.2889,
      "step": 10845
    },
    {
      "epoch": 0.19590636939455,
      "grad_norm": 0.5061205625534058,
      "learning_rate": 1.8165052584866972e-05,
      "loss": 0.2545,
      "step": 10850
    },
    {
      "epoch": 0.19599664882745071,
      "grad_norm": 0.43692877888679504,
      "learning_rate": 1.8163414785766117e-05,
      "loss": 0.2992,
      "step": 10855
    },
    {
      "epoch": 0.19608692826035143,
      "grad_norm": 0.41942158341407776,
      "learning_rate": 1.8161776329977897e-05,
      "loss": 0.1642,
      "step": 10860
    },
    {
      "epoch": 0.19617720769325214,
      "grad_norm": 0.4778582453727722,
      "learning_rate": 1.8160137217634108e-05,
      "loss": 0.2666,
      "step": 10865
    },
    {
      "epoch": 0.19626748712615286,
      "grad_norm": 0.559712827205658,
      "learning_rate": 1.8158497448866615e-05,
      "loss": 0.3298,
      "step": 10870
    },
    {
      "epoch": 0.19635776655905357,
      "grad_norm": 0.493183970451355,
      "learning_rate": 1.8156857023807316e-05,
      "loss": 0.3691,
      "step": 10875
    },
    {
      "epoch": 0.19644804599195428,
      "grad_norm": 0.5540546178817749,
      "learning_rate": 1.815521594258818e-05,
      "loss": 0.3137,
      "step": 10880
    },
    {
      "epoch": 0.196538325424855,
      "grad_norm": 0.3892921805381775,
      "learning_rate": 1.8153574205341214e-05,
      "loss": 0.26,
      "step": 10885
    },
    {
      "epoch": 0.1966286048577557,
      "grad_norm": 0.4154384732246399,
      "learning_rate": 1.8151931812198483e-05,
      "loss": 0.2584,
      "step": 10890
    },
    {
      "epoch": 0.19671888429065643,
      "grad_norm": 0.5406801104545593,
      "learning_rate": 1.815028876329211e-05,
      "loss": 0.3055,
      "step": 10895
    },
    {
      "epoch": 0.19680916372355717,
      "grad_norm": 0.6369251012802124,
      "learning_rate": 1.8148645058754265e-05,
      "loss": 0.4461,
      "step": 10900
    },
    {
      "epoch": 0.19689944315645788,
      "grad_norm": 0.4940812885761261,
      "learning_rate": 1.814700069871717e-05,
      "loss": 0.2958,
      "step": 10905
    },
    {
      "epoch": 0.1969897225893586,
      "grad_norm": 0.4416468143463135,
      "learning_rate": 1.8145355683313108e-05,
      "loss": 0.2517,
      "step": 10910
    },
    {
      "epoch": 0.1970800020222593,
      "grad_norm": 0.3803589344024658,
      "learning_rate": 1.8143710012674396e-05,
      "loss": 0.3081,
      "step": 10915
    },
    {
      "epoch": 0.19717028145516002,
      "grad_norm": 0.6976109743118286,
      "learning_rate": 1.814206368693343e-05,
      "loss": 0.3953,
      "step": 10920
    },
    {
      "epoch": 0.19726056088806074,
      "grad_norm": 0.3147580921649933,
      "learning_rate": 1.8140416706222638e-05,
      "loss": 0.2785,
      "step": 10925
    },
    {
      "epoch": 0.19735084032096145,
      "grad_norm": 0.3766973912715912,
      "learning_rate": 1.8138769070674503e-05,
      "loss": 0.2744,
      "step": 10930
    },
    {
      "epoch": 0.19744111975386217,
      "grad_norm": 0.3725707530975342,
      "learning_rate": 1.8137120780421576e-05,
      "loss": 0.2794,
      "step": 10935
    },
    {
      "epoch": 0.19753139918676288,
      "grad_norm": 0.3790420591831207,
      "learning_rate": 1.8135471835596444e-05,
      "loss": 0.265,
      "step": 10940
    },
    {
      "epoch": 0.1976216786196636,
      "grad_norm": 0.4833362400531769,
      "learning_rate": 1.8133822236331756e-05,
      "loss": 0.2857,
      "step": 10945
    },
    {
      "epoch": 0.1977119580525643,
      "grad_norm": 0.5792388916015625,
      "learning_rate": 1.8132171982760202e-05,
      "loss": 0.3057,
      "step": 10950
    },
    {
      "epoch": 0.19780223748546502,
      "grad_norm": 0.4359922409057617,
      "learning_rate": 1.813052107501454e-05,
      "loss": 0.2782,
      "step": 10955
    },
    {
      "epoch": 0.19789251691836574,
      "grad_norm": 0.3914576768875122,
      "learning_rate": 1.8128869513227577e-05,
      "loss": 0.2832,
      "step": 10960
    },
    {
      "epoch": 0.19798279635126645,
      "grad_norm": 0.49458959698677063,
      "learning_rate": 1.812721729753216e-05,
      "loss": 0.2529,
      "step": 10965
    },
    {
      "epoch": 0.19807307578416716,
      "grad_norm": 0.40677347779273987,
      "learning_rate": 1.81255644280612e-05,
      "loss": 0.1789,
      "step": 10970
    },
    {
      "epoch": 0.19816335521706788,
      "grad_norm": 0.3918363153934479,
      "learning_rate": 1.8123910904947665e-05,
      "loss": 0.3244,
      "step": 10975
    },
    {
      "epoch": 0.1982536346499686,
      "grad_norm": 0.44329938292503357,
      "learning_rate": 1.8122256728324563e-05,
      "loss": 0.2231,
      "step": 10980
    },
    {
      "epoch": 0.1983439140828693,
      "grad_norm": 0.390706866979599,
      "learning_rate": 1.812060189832496e-05,
      "loss": 0.3747,
      "step": 10985
    },
    {
      "epoch": 0.19843419351577002,
      "grad_norm": 0.4873644709587097,
      "learning_rate": 1.811894641508198e-05,
      "loss": 0.3232,
      "step": 10990
    },
    {
      "epoch": 0.19852447294867073,
      "grad_norm": 0.32289350032806396,
      "learning_rate": 1.811729027872879e-05,
      "loss": 0.3092,
      "step": 10995
    },
    {
      "epoch": 0.19861475238157145,
      "grad_norm": 0.8177574872970581,
      "learning_rate": 1.811563348939861e-05,
      "loss": 0.3436,
      "step": 11000
    },
    {
      "epoch": 0.19870503181447216,
      "grad_norm": 0.6034993529319763,
      "learning_rate": 1.811397604722473e-05,
      "loss": 0.3432,
      "step": 11005
    },
    {
      "epoch": 0.19879531124737287,
      "grad_norm": 0.5464856028556824,
      "learning_rate": 1.8112317952340467e-05,
      "loss": 0.2737,
      "step": 11010
    },
    {
      "epoch": 0.1988855906802736,
      "grad_norm": 0.41048601269721985,
      "learning_rate": 1.8110659204879208e-05,
      "loss": 0.3888,
      "step": 11015
    },
    {
      "epoch": 0.1989758701131743,
      "grad_norm": 0.5404345989227295,
      "learning_rate": 1.8108999804974386e-05,
      "loss": 0.3413,
      "step": 11020
    },
    {
      "epoch": 0.19906614954607502,
      "grad_norm": 0.3721117079257965,
      "learning_rate": 1.810733975275949e-05,
      "loss": 0.2864,
      "step": 11025
    },
    {
      "epoch": 0.19915642897897573,
      "grad_norm": 0.4072383940219879,
      "learning_rate": 1.8105679048368056e-05,
      "loss": 0.2911,
      "step": 11030
    },
    {
      "epoch": 0.19924670841187644,
      "grad_norm": 0.4085667133331299,
      "learning_rate": 1.8104017691933674e-05,
      "loss": 0.2769,
      "step": 11035
    },
    {
      "epoch": 0.19933698784477716,
      "grad_norm": 0.529788076877594,
      "learning_rate": 1.8102355683589993e-05,
      "loss": 0.3122,
      "step": 11040
    },
    {
      "epoch": 0.19942726727767787,
      "grad_norm": 0.30550462007522583,
      "learning_rate": 1.8100693023470705e-05,
      "loss": 0.2086,
      "step": 11045
    },
    {
      "epoch": 0.19951754671057859,
      "grad_norm": 0.6182094812393188,
      "learning_rate": 1.8099029711709564e-05,
      "loss": 0.3375,
      "step": 11050
    },
    {
      "epoch": 0.1996078261434793,
      "grad_norm": 0.449313223361969,
      "learning_rate": 1.8097365748440367e-05,
      "loss": 0.2956,
      "step": 11055
    },
    {
      "epoch": 0.19969810557638001,
      "grad_norm": 3.341446876525879,
      "learning_rate": 1.8095701133796968e-05,
      "loss": 0.2941,
      "step": 11060
    },
    {
      "epoch": 0.19978838500928073,
      "grad_norm": 0.3782710134983063,
      "learning_rate": 1.8094035867913274e-05,
      "loss": 0.4157,
      "step": 11065
    },
    {
      "epoch": 0.19987866444218144,
      "grad_norm": 0.9143231511116028,
      "learning_rate": 1.8092369950923244e-05,
      "loss": 0.3257,
      "step": 11070
    },
    {
      "epoch": 0.19996894387508216,
      "grad_norm": 0.5001149773597717,
      "learning_rate": 1.809070338296089e-05,
      "loss": 0.338,
      "step": 11075
    },
    {
      "epoch": 0.20005922330798287,
      "grad_norm": 0.5527740716934204,
      "learning_rate": 1.808903616416027e-05,
      "loss": 0.4017,
      "step": 11080
    },
    {
      "epoch": 0.20014950274088358,
      "grad_norm": 0.24858571588993073,
      "learning_rate": 1.808736829465551e-05,
      "loss": 0.3122,
      "step": 11085
    },
    {
      "epoch": 0.2002397821737843,
      "grad_norm": 0.4394090473651886,
      "learning_rate": 1.8085699774580767e-05,
      "loss": 0.2676,
      "step": 11090
    },
    {
      "epoch": 0.200330061606685,
      "grad_norm": 0.4978356957435608,
      "learning_rate": 1.8084030604070268e-05,
      "loss": 0.3148,
      "step": 11095
    },
    {
      "epoch": 0.20042034103958573,
      "grad_norm": 0.5462652444839478,
      "learning_rate": 1.808236078325828e-05,
      "loss": 0.2628,
      "step": 11100
    },
    {
      "epoch": 0.20051062047248644,
      "grad_norm": 0.46849769353866577,
      "learning_rate": 1.8080690312279137e-05,
      "loss": 0.3624,
      "step": 11105
    },
    {
      "epoch": 0.20060089990538715,
      "grad_norm": 0.4754716157913208,
      "learning_rate": 1.8079019191267208e-05,
      "loss": 0.265,
      "step": 11110
    },
    {
      "epoch": 0.20069117933828787,
      "grad_norm": 0.24722929298877716,
      "learning_rate": 1.8077347420356925e-05,
      "loss": 0.335,
      "step": 11115
    },
    {
      "epoch": 0.20078145877118858,
      "grad_norm": 0.5693946480751038,
      "learning_rate": 1.8075674999682768e-05,
      "loss": 0.3621,
      "step": 11120
    },
    {
      "epoch": 0.2008717382040893,
      "grad_norm": 0.3881761133670807,
      "learning_rate": 1.8074001929379277e-05,
      "loss": 0.2863,
      "step": 11125
    },
    {
      "epoch": 0.20096201763699,
      "grad_norm": 0.46133917570114136,
      "learning_rate": 1.8072328209581033e-05,
      "loss": 0.4142,
      "step": 11130
    },
    {
      "epoch": 0.20105229706989072,
      "grad_norm": 0.4075993299484253,
      "learning_rate": 1.8070653840422676e-05,
      "loss": 0.3182,
      "step": 11135
    },
    {
      "epoch": 0.20114257650279144,
      "grad_norm": 0.5719988346099854,
      "learning_rate": 1.8068978822038895e-05,
      "loss": 0.3761,
      "step": 11140
    },
    {
      "epoch": 0.20123285593569215,
      "grad_norm": 0.45642632246017456,
      "learning_rate": 1.8067303154564438e-05,
      "loss": 0.3196,
      "step": 11145
    },
    {
      "epoch": 0.20132313536859286,
      "grad_norm": 0.6719714999198914,
      "learning_rate": 1.8065626838134097e-05,
      "loss": 0.2788,
      "step": 11150
    },
    {
      "epoch": 0.20141341480149358,
      "grad_norm": 0.4135879576206207,
      "learning_rate": 1.8063949872882715e-05,
      "loss": 0.2748,
      "step": 11155
    },
    {
      "epoch": 0.2015036942343943,
      "grad_norm": 0.542934775352478,
      "learning_rate": 1.80622722589452e-05,
      "loss": 0.3757,
      "step": 11160
    },
    {
      "epoch": 0.201593973667295,
      "grad_norm": 0.36689525842666626,
      "learning_rate": 1.80605939964565e-05,
      "loss": 0.314,
      "step": 11165
    },
    {
      "epoch": 0.20168425310019572,
      "grad_norm": 0.48210838437080383,
      "learning_rate": 1.805891508555162e-05,
      "loss": 0.3291,
      "step": 11170
    },
    {
      "epoch": 0.20177453253309643,
      "grad_norm": 0.5560554265975952,
      "learning_rate": 1.8057235526365612e-05,
      "loss": 0.3179,
      "step": 11175
    },
    {
      "epoch": 0.20186481196599715,
      "grad_norm": 0.3980230391025543,
      "learning_rate": 1.805555531903359e-05,
      "loss": 0.319,
      "step": 11180
    },
    {
      "epoch": 0.20195509139889786,
      "grad_norm": 0.6233363151550293,
      "learning_rate": 1.8053874463690713e-05,
      "loss": 0.3082,
      "step": 11185
    },
    {
      "epoch": 0.20204537083179858,
      "grad_norm": 0.4474990665912628,
      "learning_rate": 1.805219296047219e-05,
      "loss": 0.3062,
      "step": 11190
    },
    {
      "epoch": 0.2021356502646993,
      "grad_norm": 0.40642160177230835,
      "learning_rate": 1.805051080951329e-05,
      "loss": 0.317,
      "step": 11195
    },
    {
      "epoch": 0.2022259296976,
      "grad_norm": 0.4216107130050659,
      "learning_rate": 1.804882801094933e-05,
      "loss": 0.3419,
      "step": 11200
    },
    {
      "epoch": 0.20231620913050072,
      "grad_norm": 0.31105688214302063,
      "learning_rate": 1.8047144564915674e-05,
      "loss": 0.192,
      "step": 11205
    },
    {
      "epoch": 0.20240648856340143,
      "grad_norm": 0.44025835394859314,
      "learning_rate": 1.804546047154775e-05,
      "loss": 0.2557,
      "step": 11210
    },
    {
      "epoch": 0.20249676799630215,
      "grad_norm": 0.808018684387207,
      "learning_rate": 1.8043775730981027e-05,
      "loss": 0.3543,
      "step": 11215
    },
    {
      "epoch": 0.20258704742920286,
      "grad_norm": 0.45786386728286743,
      "learning_rate": 1.8042090343351026e-05,
      "loss": 0.3015,
      "step": 11220
    },
    {
      "epoch": 0.20267732686210357,
      "grad_norm": 0.4077109396457672,
      "learning_rate": 1.804040430879333e-05,
      "loss": 0.3178,
      "step": 11225
    },
    {
      "epoch": 0.2027676062950043,
      "grad_norm": 1.3298776149749756,
      "learning_rate": 1.8038717627443566e-05,
      "loss": 0.3159,
      "step": 11230
    },
    {
      "epoch": 0.202857885727905,
      "grad_norm": 0.47611549496650696,
      "learning_rate": 1.8037030299437418e-05,
      "loss": 0.1996,
      "step": 11235
    },
    {
      "epoch": 0.20294816516080572,
      "grad_norm": 0.429661363363266,
      "learning_rate": 1.803534232491062e-05,
      "loss": 0.2496,
      "step": 11240
    },
    {
      "epoch": 0.20303844459370643,
      "grad_norm": 0.5555011034011841,
      "learning_rate": 1.8033653703998947e-05,
      "loss": 0.3312,
      "step": 11245
    },
    {
      "epoch": 0.20312872402660714,
      "grad_norm": 0.6299427151679993,
      "learning_rate": 1.8031964436838253e-05,
      "loss": 0.2935,
      "step": 11250
    },
    {
      "epoch": 0.20321900345950786,
      "grad_norm": 0.4307705760002136,
      "learning_rate": 1.8030274523564412e-05,
      "loss": 0.1676,
      "step": 11255
    },
    {
      "epoch": 0.20330928289240857,
      "grad_norm": 0.3759838044643402,
      "learning_rate": 1.8028583964313372e-05,
      "loss": 0.3089,
      "step": 11260
    },
    {
      "epoch": 0.20339956232530929,
      "grad_norm": 0.381085604429245,
      "learning_rate": 1.8026892759221126e-05,
      "loss": 0.2365,
      "step": 11265
    },
    {
      "epoch": 0.20348984175821,
      "grad_norm": 0.41883227229118347,
      "learning_rate": 1.8025200908423718e-05,
      "loss": 0.2366,
      "step": 11270
    },
    {
      "epoch": 0.2035801211911107,
      "grad_norm": 0.5832628607749939,
      "learning_rate": 1.8023508412057246e-05,
      "loss": 0.3491,
      "step": 11275
    },
    {
      "epoch": 0.20367040062401143,
      "grad_norm": 0.48159259557724,
      "learning_rate": 1.8021815270257858e-05,
      "loss": 0.2536,
      "step": 11280
    },
    {
      "epoch": 0.20376068005691214,
      "grad_norm": 0.36705389618873596,
      "learning_rate": 1.802012148316176e-05,
      "loss": 0.2455,
      "step": 11285
    },
    {
      "epoch": 0.20385095948981286,
      "grad_norm": 0.5337108373641968,
      "learning_rate": 1.8018427050905197e-05,
      "loss": 0.2974,
      "step": 11290
    },
    {
      "epoch": 0.2039412389227136,
      "grad_norm": 0.7574781775474548,
      "learning_rate": 1.801673197362448e-05,
      "loss": 0.3361,
      "step": 11295
    },
    {
      "epoch": 0.2040315183556143,
      "grad_norm": 0.46422940492630005,
      "learning_rate": 1.801503625145596e-05,
      "loss": 0.3119,
      "step": 11300
    },
    {
      "epoch": 0.20412179778851502,
      "grad_norm": 0.4504513144493103,
      "learning_rate": 1.801333988453605e-05,
      "loss": 0.2062,
      "step": 11305
    },
    {
      "epoch": 0.20421207722141574,
      "grad_norm": 1.045087456703186,
      "learning_rate": 1.8011642873001206e-05,
      "loss": 0.2846,
      "step": 11310
    },
    {
      "epoch": 0.20430235665431645,
      "grad_norm": 0.5900313854217529,
      "learning_rate": 1.800994521698795e-05,
      "loss": 0.2714,
      "step": 11315
    },
    {
      "epoch": 0.20439263608721717,
      "grad_norm": 0.47045981884002686,
      "learning_rate": 1.8008246916632832e-05,
      "loss": 0.311,
      "step": 11320
    },
    {
      "epoch": 0.20448291552011788,
      "grad_norm": 0.5298904776573181,
      "learning_rate": 1.8006547972072476e-05,
      "loss": 0.3081,
      "step": 11325
    },
    {
      "epoch": 0.2045731949530186,
      "grad_norm": 0.6724792718887329,
      "learning_rate": 1.800484838344355e-05,
      "loss": 0.3135,
      "step": 11330
    },
    {
      "epoch": 0.2046634743859193,
      "grad_norm": 0.5559977293014526,
      "learning_rate": 1.8003148150882777e-05,
      "loss": 0.3187,
      "step": 11335
    },
    {
      "epoch": 0.20475375381882002,
      "grad_norm": 0.5023704171180725,
      "learning_rate": 1.8001447274526917e-05,
      "loss": 0.2417,
      "step": 11340
    },
    {
      "epoch": 0.20484403325172074,
      "grad_norm": 0.45989304780960083,
      "learning_rate": 1.7999745754512802e-05,
      "loss": 0.2501,
      "step": 11345
    },
    {
      "epoch": 0.20493431268462145,
      "grad_norm": 0.5148830413818359,
      "learning_rate": 1.7998043590977307e-05,
      "loss": 0.3235,
      "step": 11350
    },
    {
      "epoch": 0.20502459211752216,
      "grad_norm": 0.7312597632408142,
      "learning_rate": 1.7996340784057354e-05,
      "loss": 0.1839,
      "step": 11355
    },
    {
      "epoch": 0.20511487155042288,
      "grad_norm": 0.5773922801017761,
      "learning_rate": 1.7994637333889925e-05,
      "loss": 0.3008,
      "step": 11360
    },
    {
      "epoch": 0.2052051509833236,
      "grad_norm": 0.5040149092674255,
      "learning_rate": 1.799293324061205e-05,
      "loss": 0.3197,
      "step": 11365
    },
    {
      "epoch": 0.2052954304162243,
      "grad_norm": 0.3866426944732666,
      "learning_rate": 1.799122850436081e-05,
      "loss": 0.2589,
      "step": 11370
    },
    {
      "epoch": 0.20538570984912502,
      "grad_norm": 1.08524489402771,
      "learning_rate": 1.7989523125273338e-05,
      "loss": 0.3366,
      "step": 11375
    },
    {
      "epoch": 0.20547598928202573,
      "grad_norm": 0.5162799954414368,
      "learning_rate": 1.7987817103486824e-05,
      "loss": 0.2687,
      "step": 11380
    },
    {
      "epoch": 0.20556626871492645,
      "grad_norm": 0.44467848539352417,
      "learning_rate": 1.7986110439138493e-05,
      "loss": 0.261,
      "step": 11385
    },
    {
      "epoch": 0.20565654814782716,
      "grad_norm": 0.5967909693717957,
      "learning_rate": 1.7984403132365647e-05,
      "loss": 0.2886,
      "step": 11390
    },
    {
      "epoch": 0.20574682758072788,
      "grad_norm": 0.46552544832229614,
      "learning_rate": 1.798269518330562e-05,
      "loss": 0.2784,
      "step": 11395
    },
    {
      "epoch": 0.2058371070136286,
      "grad_norm": 0.5744720697402954,
      "learning_rate": 1.7980986592095804e-05,
      "loss": 0.3054,
      "step": 11400
    },
    {
      "epoch": 0.2059273864465293,
      "grad_norm": 0.39584240317344666,
      "learning_rate": 1.7979277358873646e-05,
      "loss": 0.3351,
      "step": 11405
    },
    {
      "epoch": 0.20601766587943002,
      "grad_norm": 0.4696711301803589,
      "learning_rate": 1.797756748377664e-05,
      "loss": 0.2547,
      "step": 11410
    },
    {
      "epoch": 0.20610794531233073,
      "grad_norm": 0.49084359407424927,
      "learning_rate": 1.797585696694233e-05,
      "loss": 0.3508,
      "step": 11415
    },
    {
      "epoch": 0.20619822474523145,
      "grad_norm": 0.5010688900947571,
      "learning_rate": 1.7974145808508318e-05,
      "loss": 0.3102,
      "step": 11420
    },
    {
      "epoch": 0.20628850417813216,
      "grad_norm": 0.46311667561531067,
      "learning_rate": 1.7972434008612256e-05,
      "loss": 0.3046,
      "step": 11425
    },
    {
      "epoch": 0.20637878361103287,
      "grad_norm": 0.4460143446922302,
      "learning_rate": 1.797072156739184e-05,
      "loss": 0.2743,
      "step": 11430
    },
    {
      "epoch": 0.2064690630439336,
      "grad_norm": 0.370007187128067,
      "learning_rate": 1.7969008484984827e-05,
      "loss": 0.3649,
      "step": 11435
    },
    {
      "epoch": 0.2065593424768343,
      "grad_norm": 0.375128835439682,
      "learning_rate": 1.796729476152902e-05,
      "loss": 0.3949,
      "step": 11440
    },
    {
      "epoch": 0.20664962190973502,
      "grad_norm": 0.4817360043525696,
      "learning_rate": 1.7965580397162282e-05,
      "loss": 0.3477,
      "step": 11445
    },
    {
      "epoch": 0.20673990134263573,
      "grad_norm": 0.593788206577301,
      "learning_rate": 1.7963865392022515e-05,
      "loss": 0.3004,
      "step": 11450
    },
    {
      "epoch": 0.20683018077553644,
      "grad_norm": 0.4985922574996948,
      "learning_rate": 1.7962149746247678e-05,
      "loss": 0.2844,
      "step": 11455
    },
    {
      "epoch": 0.20692046020843716,
      "grad_norm": 0.31397101283073425,
      "learning_rate": 1.7960433459975786e-05,
      "loss": 0.2237,
      "step": 11460
    },
    {
      "epoch": 0.20701073964133787,
      "grad_norm": 0.5667983889579773,
      "learning_rate": 1.79587165333449e-05,
      "loss": 0.2377,
      "step": 11465
    },
    {
      "epoch": 0.20710101907423858,
      "grad_norm": 0.39671066403388977,
      "learning_rate": 1.7956998966493133e-05,
      "loss": 0.283,
      "step": 11470
    },
    {
      "epoch": 0.2071912985071393,
      "grad_norm": 0.44250059127807617,
      "learning_rate": 1.7955280759558655e-05,
      "loss": 0.438,
      "step": 11475
    },
    {
      "epoch": 0.20728157794004,
      "grad_norm": 0.465035617351532,
      "learning_rate": 1.7953561912679678e-05,
      "loss": 0.2873,
      "step": 11480
    },
    {
      "epoch": 0.20737185737294073,
      "grad_norm": 0.34597525000572205,
      "learning_rate": 1.7951842425994473e-05,
      "loss": 0.2748,
      "step": 11485
    },
    {
      "epoch": 0.20746213680584144,
      "grad_norm": 0.4495750069618225,
      "learning_rate": 1.795012229964136e-05,
      "loss": 0.3538,
      "step": 11490
    },
    {
      "epoch": 0.20755241623874215,
      "grad_norm": 0.35256561636924744,
      "learning_rate": 1.794840153375871e-05,
      "loss": 0.3767,
      "step": 11495
    },
    {
      "epoch": 0.20764269567164287,
      "grad_norm": 0.32467296719551086,
      "learning_rate": 1.794668012848495e-05,
      "loss": 0.2314,
      "step": 11500
    },
    {
      "epoch": 0.20773297510454358,
      "grad_norm": 0.4653473198413849,
      "learning_rate": 1.7944958083958553e-05,
      "loss": 0.2843,
      "step": 11505
    },
    {
      "epoch": 0.2078232545374443,
      "grad_norm": 0.7906263470649719,
      "learning_rate": 1.7943235400318037e-05,
      "loss": 0.2777,
      "step": 11510
    },
    {
      "epoch": 0.207913533970345,
      "grad_norm": 0.6436835527420044,
      "learning_rate": 1.7941512077701992e-05,
      "loss": 0.2752,
      "step": 11515
    },
    {
      "epoch": 0.20800381340324572,
      "grad_norm": 0.44068244099617004,
      "learning_rate": 1.793978811624904e-05,
      "loss": 0.2844,
      "step": 11520
    },
    {
      "epoch": 0.20809409283614644,
      "grad_norm": 0.7325283288955688,
      "learning_rate": 1.7938063516097858e-05,
      "loss": 0.3731,
      "step": 11525
    },
    {
      "epoch": 0.20818437226904715,
      "grad_norm": 0.4470127820968628,
      "learning_rate": 1.7936338277387183e-05,
      "loss": 0.2738,
      "step": 11530
    },
    {
      "epoch": 0.20827465170194787,
      "grad_norm": 0.4569440186023712,
      "learning_rate": 1.79346124002558e-05,
      "loss": 0.3692,
      "step": 11535
    },
    {
      "epoch": 0.20836493113484858,
      "grad_norm": 0.34077921509742737,
      "learning_rate": 1.7932885884842538e-05,
      "loss": 0.3582,
      "step": 11540
    },
    {
      "epoch": 0.2084552105677493,
      "grad_norm": 0.3526357114315033,
      "learning_rate": 1.7931158731286282e-05,
      "loss": 0.3567,
      "step": 11545
    },
    {
      "epoch": 0.20854549000065,
      "grad_norm": 0.429093599319458,
      "learning_rate": 1.7929430939725972e-05,
      "loss": 0.3146,
      "step": 11550
    },
    {
      "epoch": 0.20863576943355072,
      "grad_norm": 0.48118454217910767,
      "learning_rate": 1.7927702510300596e-05,
      "loss": 0.2938,
      "step": 11555
    },
    {
      "epoch": 0.20872604886645144,
      "grad_norm": 0.35543933510780334,
      "learning_rate": 1.7925973443149193e-05,
      "loss": 0.3047,
      "step": 11560
    },
    {
      "epoch": 0.20881632829935215,
      "grad_norm": 0.9324852228164673,
      "learning_rate": 1.7924243738410853e-05,
      "loss": 0.3363,
      "step": 11565
    },
    {
      "epoch": 0.20890660773225286,
      "grad_norm": 0.5078498125076294,
      "learning_rate": 1.792251339622472e-05,
      "loss": 0.3852,
      "step": 11570
    },
    {
      "epoch": 0.20899688716515358,
      "grad_norm": 0.46870550513267517,
      "learning_rate": 1.792078241672999e-05,
      "loss": 0.3656,
      "step": 11575
    },
    {
      "epoch": 0.2090871665980543,
      "grad_norm": 0.4630391299724579,
      "learning_rate": 1.7919050800065898e-05,
      "loss": 0.2391,
      "step": 11580
    },
    {
      "epoch": 0.209177446030955,
      "grad_norm": 0.4416666030883789,
      "learning_rate": 1.791731854637175e-05,
      "loss": 0.4418,
      "step": 11585
    },
    {
      "epoch": 0.20926772546385572,
      "grad_norm": 0.43558019399642944,
      "learning_rate": 1.7915585655786888e-05,
      "loss": 0.2451,
      "step": 11590
    },
    {
      "epoch": 0.20935800489675643,
      "grad_norm": 0.5845248699188232,
      "learning_rate": 1.791385212845071e-05,
      "loss": 0.3297,
      "step": 11595
    },
    {
      "epoch": 0.20944828432965715,
      "grad_norm": 0.41678881645202637,
      "learning_rate": 1.791211796450267e-05,
      "loss": 0.2795,
      "step": 11600
    },
    {
      "epoch": 0.20953856376255786,
      "grad_norm": 0.453337699174881,
      "learning_rate": 1.7910383164082266e-05,
      "loss": 0.2509,
      "step": 11605
    },
    {
      "epoch": 0.20962884319545858,
      "grad_norm": 0.46536174416542053,
      "learning_rate": 1.790864772732905e-05,
      "loss": 0.2221,
      "step": 11610
    },
    {
      "epoch": 0.2097191226283593,
      "grad_norm": 0.41316506266593933,
      "learning_rate": 1.7906911654382625e-05,
      "loss": 0.2622,
      "step": 11615
    },
    {
      "epoch": 0.20980940206126,
      "grad_norm": 0.8957769870758057,
      "learning_rate": 1.790517494538265e-05,
      "loss": 0.301,
      "step": 11620
    },
    {
      "epoch": 0.20989968149416072,
      "grad_norm": 0.5003489255905151,
      "learning_rate": 1.7903437600468822e-05,
      "loss": 0.2682,
      "step": 11625
    },
    {
      "epoch": 0.20998996092706143,
      "grad_norm": 0.3303603231906891,
      "learning_rate": 1.7901699619780905e-05,
      "loss": 0.2477,
      "step": 11630
    },
    {
      "epoch": 0.21008024035996214,
      "grad_norm": 0.46241119503974915,
      "learning_rate": 1.7899961003458705e-05,
      "loss": 0.2684,
      "step": 11635
    },
    {
      "epoch": 0.21017051979286286,
      "grad_norm": 0.4073708951473236,
      "learning_rate": 1.789822175164208e-05,
      "loss": 0.2705,
      "step": 11640
    },
    {
      "epoch": 0.21026079922576357,
      "grad_norm": 0.8670423030853271,
      "learning_rate": 1.7896481864470942e-05,
      "loss": 0.3147,
      "step": 11645
    },
    {
      "epoch": 0.2103510786586643,
      "grad_norm": 0.45808741450309753,
      "learning_rate": 1.7894741342085246e-05,
      "loss": 0.2539,
      "step": 11650
    },
    {
      "epoch": 0.210441358091565,
      "grad_norm": 0.4415580928325653,
      "learning_rate": 1.7893000184625013e-05,
      "loss": 0.3669,
      "step": 11655
    },
    {
      "epoch": 0.21053163752446571,
      "grad_norm": 0.2889484763145447,
      "learning_rate": 1.7891258392230303e-05,
      "loss": 0.2558,
      "step": 11660
    },
    {
      "epoch": 0.21062191695736643,
      "grad_norm": 0.3614408075809479,
      "learning_rate": 1.788951596504123e-05,
      "loss": 0.2535,
      "step": 11665
    },
    {
      "epoch": 0.21071219639026714,
      "grad_norm": 0.33568477630615234,
      "learning_rate": 1.788777290319796e-05,
      "loss": 0.1984,
      "step": 11670
    },
    {
      "epoch": 0.21080247582316786,
      "grad_norm": 0.46689480543136597,
      "learning_rate": 1.788602920684071e-05,
      "loss": 0.3371,
      "step": 11675
    },
    {
      "epoch": 0.21089275525606857,
      "grad_norm": 0.7152404189109802,
      "learning_rate": 1.788428487610975e-05,
      "loss": 0.303,
      "step": 11680
    },
    {
      "epoch": 0.21098303468896928,
      "grad_norm": 0.3899504244327545,
      "learning_rate": 1.788253991114539e-05,
      "loss": 0.2941,
      "step": 11685
    },
    {
      "epoch": 0.21107331412187,
      "grad_norm": 0.45941299200057983,
      "learning_rate": 1.788079431208801e-05,
      "loss": 0.309,
      "step": 11690
    },
    {
      "epoch": 0.21116359355477074,
      "grad_norm": 0.560175895690918,
      "learning_rate": 1.7879048079078024e-05,
      "loss": 0.3159,
      "step": 11695
    },
    {
      "epoch": 0.21125387298767145,
      "grad_norm": 0.5322653651237488,
      "learning_rate": 1.7877301212255908e-05,
      "loss": 0.3391,
      "step": 11700
    },
    {
      "epoch": 0.21134415242057217,
      "grad_norm": 0.5604745745658875,
      "learning_rate": 1.7875553711762182e-05,
      "loss": 0.3619,
      "step": 11705
    },
    {
      "epoch": 0.21143443185347288,
      "grad_norm": 0.46025556325912476,
      "learning_rate": 1.787380557773742e-05,
      "loss": 0.2314,
      "step": 11710
    },
    {
      "epoch": 0.2115247112863736,
      "grad_norm": 0.5464009046554565,
      "learning_rate": 1.787205681032225e-05,
      "loss": 0.2467,
      "step": 11715
    },
    {
      "epoch": 0.2116149907192743,
      "grad_norm": 0.8008533120155334,
      "learning_rate": 1.7870307409657344e-05,
      "loss": 0.2839,
      "step": 11720
    },
    {
      "epoch": 0.21170527015217502,
      "grad_norm": 0.38943585753440857,
      "learning_rate": 1.7868557375883426e-05,
      "loss": 0.2827,
      "step": 11725
    },
    {
      "epoch": 0.21179554958507574,
      "grad_norm": 0.5736641883850098,
      "learning_rate": 1.786680670914128e-05,
      "loss": 0.2569,
      "step": 11730
    },
    {
      "epoch": 0.21188582901797645,
      "grad_norm": 0.3588363230228424,
      "learning_rate": 1.7865055409571733e-05,
      "loss": 0.3281,
      "step": 11735
    },
    {
      "epoch": 0.21197610845087717,
      "grad_norm": 0.4896489083766937,
      "learning_rate": 1.786330347731566e-05,
      "loss": 0.2802,
      "step": 11740
    },
    {
      "epoch": 0.21206638788377788,
      "grad_norm": 0.4987027049064636,
      "learning_rate": 1.7861550912513997e-05,
      "loss": 0.2674,
      "step": 11745
    },
    {
      "epoch": 0.2121566673166786,
      "grad_norm": 0.4357168674468994,
      "learning_rate": 1.785979771530772e-05,
      "loss": 0.2681,
      "step": 11750
    },
    {
      "epoch": 0.2122469467495793,
      "grad_norm": 0.679339587688446,
      "learning_rate": 1.7858043885837863e-05,
      "loss": 0.2868,
      "step": 11755
    },
    {
      "epoch": 0.21233722618248002,
      "grad_norm": 0.6235727071762085,
      "learning_rate": 1.785628942424551e-05,
      "loss": 0.3005,
      "step": 11760
    },
    {
      "epoch": 0.21242750561538074,
      "grad_norm": 0.6073330640792847,
      "learning_rate": 1.7854534330671793e-05,
      "loss": 0.3244,
      "step": 11765
    },
    {
      "epoch": 0.21251778504828145,
      "grad_norm": 0.5216356515884399,
      "learning_rate": 1.7852778605257895e-05,
      "loss": 0.2622,
      "step": 11770
    },
    {
      "epoch": 0.21260806448118216,
      "grad_norm": 0.7519828677177429,
      "learning_rate": 1.7851022248145057e-05,
      "loss": 0.2536,
      "step": 11775
    },
    {
      "epoch": 0.21269834391408288,
      "grad_norm": 0.6628133058547974,
      "learning_rate": 1.784926525947456e-05,
      "loss": 0.3209,
      "step": 11780
    },
    {
      "epoch": 0.2127886233469836,
      "grad_norm": 0.45866918563842773,
      "learning_rate": 1.784750763938774e-05,
      "loss": 0.1715,
      "step": 11785
    },
    {
      "epoch": 0.2128789027798843,
      "grad_norm": 0.4917178452014923,
      "learning_rate": 1.784574938802599e-05,
      "loss": 0.2321,
      "step": 11790
    },
    {
      "epoch": 0.21296918221278502,
      "grad_norm": 0.4127363860607147,
      "learning_rate": 1.7843990505530747e-05,
      "loss": 0.3582,
      "step": 11795
    },
    {
      "epoch": 0.21305946164568573,
      "grad_norm": 0.5475683808326721,
      "learning_rate": 1.7842230992043496e-05,
      "loss": 0.3792,
      "step": 11800
    },
    {
      "epoch": 0.21314974107858645,
      "grad_norm": 1.1832107305526733,
      "learning_rate": 1.7840470847705784e-05,
      "loss": 0.3258,
      "step": 11805
    },
    {
      "epoch": 0.21324002051148716,
      "grad_norm": 0.5049902200698853,
      "learning_rate": 1.7838710072659194e-05,
      "loss": 0.3705,
      "step": 11810
    },
    {
      "epoch": 0.21333029994438787,
      "grad_norm": 0.38012224435806274,
      "learning_rate": 1.7836948667045375e-05,
      "loss": 0.2989,
      "step": 11815
    },
    {
      "epoch": 0.2134205793772886,
      "grad_norm": 0.4135642647743225,
      "learning_rate": 1.7835186631006013e-05,
      "loss": 0.2915,
      "step": 11820
    },
    {
      "epoch": 0.2135108588101893,
      "grad_norm": 0.31480443477630615,
      "learning_rate": 1.7833423964682856e-05,
      "loss": 0.2658,
      "step": 11825
    },
    {
      "epoch": 0.21360113824309002,
      "grad_norm": 0.617190957069397,
      "learning_rate": 1.7831660668217697e-05,
      "loss": 0.2827,
      "step": 11830
    },
    {
      "epoch": 0.21369141767599073,
      "grad_norm": 0.5191602110862732,
      "learning_rate": 1.7829896741752375e-05,
      "loss": 0.376,
      "step": 11835
    },
    {
      "epoch": 0.21378169710889144,
      "grad_norm": 0.23170405626296997,
      "learning_rate": 1.782813218542879e-05,
      "loss": 0.2318,
      "step": 11840
    },
    {
      "epoch": 0.21387197654179216,
      "grad_norm": 0.3107124865055084,
      "learning_rate": 1.782636699938889e-05,
      "loss": 0.2315,
      "step": 11845
    },
    {
      "epoch": 0.21396225597469287,
      "grad_norm": 0.362413614988327,
      "learning_rate": 1.7824601183774665e-05,
      "loss": 0.29,
      "step": 11850
    },
    {
      "epoch": 0.21405253540759359,
      "grad_norm": 0.349123477935791,
      "learning_rate": 1.782283473872816e-05,
      "loss": 0.2097,
      "step": 11855
    },
    {
      "epoch": 0.2141428148404943,
      "grad_norm": 0.514616072177887,
      "learning_rate": 1.782106766439149e-05,
      "loss": 0.2799,
      "step": 11860
    },
    {
      "epoch": 0.214233094273395,
      "grad_norm": 0.47504672408103943,
      "learning_rate": 1.781929996090678e-05,
      "loss": 0.2358,
      "step": 11865
    },
    {
      "epoch": 0.21432337370629573,
      "grad_norm": 0.5488296747207642,
      "learning_rate": 1.7817531628416246e-05,
      "loss": 0.2715,
      "step": 11870
    },
    {
      "epoch": 0.21441365313919644,
      "grad_norm": 0.5415710806846619,
      "learning_rate": 1.781576266706213e-05,
      "loss": 0.2744,
      "step": 11875
    },
    {
      "epoch": 0.21450393257209716,
      "grad_norm": 0.4214060604572296,
      "learning_rate": 1.7813993076986733e-05,
      "loss": 0.2486,
      "step": 11880
    },
    {
      "epoch": 0.21459421200499787,
      "grad_norm": 0.27810266613960266,
      "learning_rate": 1.781222285833241e-05,
      "loss": 0.213,
      "step": 11885
    },
    {
      "epoch": 0.21468449143789858,
      "grad_norm": 0.4299297332763672,
      "learning_rate": 1.7810452011241554e-05,
      "loss": 0.2751,
      "step": 11890
    },
    {
      "epoch": 0.2147747708707993,
      "grad_norm": 0.4374350905418396,
      "learning_rate": 1.780868053585662e-05,
      "loss": 0.2703,
      "step": 11895
    },
    {
      "epoch": 0.2148650503037,
      "grad_norm": 0.39960458874702454,
      "learning_rate": 1.7806908432320113e-05,
      "loss": 0.3154,
      "step": 11900
    },
    {
      "epoch": 0.21495532973660073,
      "grad_norm": 0.3405298590660095,
      "learning_rate": 1.7805135700774588e-05,
      "loss": 0.4555,
      "step": 11905
    },
    {
      "epoch": 0.21504560916950144,
      "grad_norm": 0.6461913585662842,
      "learning_rate": 1.780336234136264e-05,
      "loss": 0.1844,
      "step": 11910
    },
    {
      "epoch": 0.21513588860240215,
      "grad_norm": 0.539243221282959,
      "learning_rate": 1.780158835422693e-05,
      "loss": 0.3345,
      "step": 11915
    },
    {
      "epoch": 0.21522616803530287,
      "grad_norm": 0.40096384286880493,
      "learning_rate": 1.779981373951016e-05,
      "loss": 0.3022,
      "step": 11920
    },
    {
      "epoch": 0.21531644746820358,
      "grad_norm": 0.5648438334465027,
      "learning_rate": 1.7798038497355082e-05,
      "loss": 0.3982,
      "step": 11925
    },
    {
      "epoch": 0.2154067269011043,
      "grad_norm": 0.3935416638851166,
      "learning_rate": 1.7796262627904504e-05,
      "loss": 0.2931,
      "step": 11930
    },
    {
      "epoch": 0.215497006334005,
      "grad_norm": 0.4855777621269226,
      "learning_rate": 1.7794486131301286e-05,
      "loss": 0.3144,
      "step": 11935
    },
    {
      "epoch": 0.21558728576690572,
      "grad_norm": 0.535211980342865,
      "learning_rate": 1.779270900768833e-05,
      "loss": 0.2997,
      "step": 11940
    },
    {
      "epoch": 0.21567756519980644,
      "grad_norm": 0.35269421339035034,
      "learning_rate": 1.779093125720859e-05,
      "loss": 0.2473,
      "step": 11945
    },
    {
      "epoch": 0.21576784463270715,
      "grad_norm": 0.33629870414733887,
      "learning_rate": 1.7789152880005075e-05,
      "loss": 0.2585,
      "step": 11950
    },
    {
      "epoch": 0.21585812406560786,
      "grad_norm": 0.27106332778930664,
      "learning_rate": 1.7787373876220847e-05,
      "loss": 0.2862,
      "step": 11955
    },
    {
      "epoch": 0.21594840349850858,
      "grad_norm": 0.6125961542129517,
      "learning_rate": 1.7785594245999007e-05,
      "loss": 0.2712,
      "step": 11960
    },
    {
      "epoch": 0.2160386829314093,
      "grad_norm": 0.3257152736186981,
      "learning_rate": 1.778381398948272e-05,
      "loss": 0.2519,
      "step": 11965
    },
    {
      "epoch": 0.21612896236431,
      "grad_norm": 0.3735145032405853,
      "learning_rate": 1.778203310681519e-05,
      "loss": 0.2877,
      "step": 11970
    },
    {
      "epoch": 0.21621924179721072,
      "grad_norm": 0.3832739293575287,
      "learning_rate": 1.7780251598139674e-05,
      "loss": 0.4543,
      "step": 11975
    },
    {
      "epoch": 0.21630952123011143,
      "grad_norm": 0.4531655013561249,
      "learning_rate": 1.777846946359949e-05,
      "loss": 0.3394,
      "step": 11980
    },
    {
      "epoch": 0.21639980066301215,
      "grad_norm": 0.46312928199768066,
      "learning_rate": 1.777668670333799e-05,
      "loss": 0.3099,
      "step": 11985
    },
    {
      "epoch": 0.21649008009591286,
      "grad_norm": 0.9675629734992981,
      "learning_rate": 1.7774903317498587e-05,
      "loss": 0.3223,
      "step": 11990
    },
    {
      "epoch": 0.21658035952881358,
      "grad_norm": 0.44589224457740784,
      "learning_rate": 1.777311930622474e-05,
      "loss": 0.3231,
      "step": 11995
    },
    {
      "epoch": 0.2166706389617143,
      "grad_norm": 0.34616875648498535,
      "learning_rate": 1.7771334669659962e-05,
      "loss": 0.2153,
      "step": 12000
    },
    {
      "epoch": 0.216760918394615,
      "grad_norm": 0.6218799948692322,
      "learning_rate": 1.7769549407947812e-05,
      "loss": 0.3489,
      "step": 12005
    },
    {
      "epoch": 0.21685119782751572,
      "grad_norm": 0.5074170231819153,
      "learning_rate": 1.7767763521231904e-05,
      "loss": 0.3349,
      "step": 12010
    },
    {
      "epoch": 0.21694147726041643,
      "grad_norm": 0.5704885125160217,
      "learning_rate": 1.7765977009655897e-05,
      "loss": 0.2668,
      "step": 12015
    },
    {
      "epoch": 0.21703175669331715,
      "grad_norm": 0.4201370179653168,
      "learning_rate": 1.7764189873363503e-05,
      "loss": 0.2427,
      "step": 12020
    },
    {
      "epoch": 0.21712203612621786,
      "grad_norm": 0.5044018030166626,
      "learning_rate": 1.7762402112498484e-05,
      "loss": 0.2947,
      "step": 12025
    },
    {
      "epoch": 0.21721231555911857,
      "grad_norm": 0.46398696303367615,
      "learning_rate": 1.7760613727204656e-05,
      "loss": 0.3881,
      "step": 12030
    },
    {
      "epoch": 0.2173025949920193,
      "grad_norm": 0.34880530834198,
      "learning_rate": 1.775882471762588e-05,
      "loss": 0.4522,
      "step": 12035
    },
    {
      "epoch": 0.21739287442492,
      "grad_norm": 0.6617854237556458,
      "learning_rate": 1.7757035083906063e-05,
      "loss": 0.2752,
      "step": 12040
    },
    {
      "epoch": 0.21748315385782072,
      "grad_norm": 0.4192894399166107,
      "learning_rate": 1.7755244826189173e-05,
      "loss": 0.3229,
      "step": 12045
    },
    {
      "epoch": 0.21757343329072143,
      "grad_norm": 0.4937689006328583,
      "learning_rate": 1.7753453944619227e-05,
      "loss": 0.2935,
      "step": 12050
    },
    {
      "epoch": 0.21766371272362214,
      "grad_norm": 0.42962369322776794,
      "learning_rate": 1.775166243934028e-05,
      "loss": 0.3625,
      "step": 12055
    },
    {
      "epoch": 0.21775399215652286,
      "grad_norm": 0.5614218711853027,
      "learning_rate": 1.774987031049645e-05,
      "loss": 0.2634,
      "step": 12060
    },
    {
      "epoch": 0.21784427158942357,
      "grad_norm": 0.3307231664657593,
      "learning_rate": 1.7748077558231905e-05,
      "loss": 0.228,
      "step": 12065
    },
    {
      "epoch": 0.21793455102232429,
      "grad_norm": 0.5150547027587891,
      "learning_rate": 1.774628418269085e-05,
      "loss": 0.2382,
      "step": 12070
    },
    {
      "epoch": 0.218024830455225,
      "grad_norm": 0.9636505246162415,
      "learning_rate": 1.7744490184017554e-05,
      "loss": 0.3017,
      "step": 12075
    },
    {
      "epoch": 0.2181151098881257,
      "grad_norm": 0.40887361764907837,
      "learning_rate": 1.7742695562356333e-05,
      "loss": 0.3922,
      "step": 12080
    },
    {
      "epoch": 0.21820538932102643,
      "grad_norm": 0.44358113408088684,
      "learning_rate": 1.7740900317851544e-05,
      "loss": 0.4168,
      "step": 12085
    },
    {
      "epoch": 0.21829566875392717,
      "grad_norm": 0.28855910897254944,
      "learning_rate": 1.7739104450647612e-05,
      "loss": 0.2614,
      "step": 12090
    },
    {
      "epoch": 0.21838594818682788,
      "grad_norm": 0.4369180202484131,
      "learning_rate": 1.773730796088899e-05,
      "loss": 0.3756,
      "step": 12095
    },
    {
      "epoch": 0.2184762276197286,
      "grad_norm": 0.6822949051856995,
      "learning_rate": 1.7735510848720204e-05,
      "loss": 0.3275,
      "step": 12100
    },
    {
      "epoch": 0.2185665070526293,
      "grad_norm": 0.3817140460014343,
      "learning_rate": 1.773371311428581e-05,
      "loss": 0.3013,
      "step": 12105
    },
    {
      "epoch": 0.21865678648553002,
      "grad_norm": 0.4731243848800659,
      "learning_rate": 1.7731914757730423e-05,
      "loss": 0.3143,
      "step": 12110
    },
    {
      "epoch": 0.21874706591843074,
      "grad_norm": 0.4177074432373047,
      "learning_rate": 1.7730115779198713e-05,
      "loss": 0.229,
      "step": 12115
    },
    {
      "epoch": 0.21883734535133145,
      "grad_norm": 0.2697894871234894,
      "learning_rate": 1.7728316178835392e-05,
      "loss": 0.3405,
      "step": 12120
    },
    {
      "epoch": 0.21892762478423217,
      "grad_norm": 0.6174373626708984,
      "learning_rate": 1.7726515956785224e-05,
      "loss": 0.2283,
      "step": 12125
    },
    {
      "epoch": 0.21901790421713288,
      "grad_norm": 0.27584612369537354,
      "learning_rate": 1.7724715113193025e-05,
      "loss": 0.2863,
      "step": 12130
    },
    {
      "epoch": 0.2191081836500336,
      "grad_norm": 0.5504199266433716,
      "learning_rate": 1.772291364820366e-05,
      "loss": 0.3787,
      "step": 12135
    },
    {
      "epoch": 0.2191984630829343,
      "grad_norm": 0.6736400127410889,
      "learning_rate": 1.7721111561962043e-05,
      "loss": 0.3155,
      "step": 12140
    },
    {
      "epoch": 0.21928874251583502,
      "grad_norm": 0.6010838150978088,
      "learning_rate": 1.7719308854613137e-05,
      "loss": 0.3521,
      "step": 12145
    },
    {
      "epoch": 0.21937902194873574,
      "grad_norm": 0.4695472717285156,
      "learning_rate": 1.7717505526301958e-05,
      "loss": 0.2443,
      "step": 12150
    },
    {
      "epoch": 0.21946930138163645,
      "grad_norm": 0.3113847076892853,
      "learning_rate": 1.7715701577173573e-05,
      "loss": 0.2631,
      "step": 12155
    },
    {
      "epoch": 0.21955958081453716,
      "grad_norm": 0.33892154693603516,
      "learning_rate": 1.771389700737309e-05,
      "loss": 0.3811,
      "step": 12160
    },
    {
      "epoch": 0.21964986024743788,
      "grad_norm": 0.6578267812728882,
      "learning_rate": 1.7712091817045685e-05,
      "loss": 0.3701,
      "step": 12165
    },
    {
      "epoch": 0.2197401396803386,
      "grad_norm": 0.32275834679603577,
      "learning_rate": 1.7710286006336566e-05,
      "loss": 0.2341,
      "step": 12170
    },
    {
      "epoch": 0.2198304191132393,
      "grad_norm": 0.5772724151611328,
      "learning_rate": 1.770847957539099e-05,
      "loss": 0.3192,
      "step": 12175
    },
    {
      "epoch": 0.21992069854614002,
      "grad_norm": 0.5542657971382141,
      "learning_rate": 1.7706672524354283e-05,
      "loss": 0.3476,
      "step": 12180
    },
    {
      "epoch": 0.22001097797904073,
      "grad_norm": 0.4902056157588959,
      "learning_rate": 1.7704864853371805e-05,
      "loss": 0.4199,
      "step": 12185
    },
    {
      "epoch": 0.22010125741194145,
      "grad_norm": 0.4195903539657593,
      "learning_rate": 1.7703056562588966e-05,
      "loss": 0.2742,
      "step": 12190
    },
    {
      "epoch": 0.22019153684484216,
      "grad_norm": 0.5244849324226379,
      "learning_rate": 1.770124765215124e-05,
      "loss": 0.3205,
      "step": 12195
    },
    {
      "epoch": 0.22028181627774288,
      "grad_norm": 0.35584715008735657,
      "learning_rate": 1.7699438122204126e-05,
      "loss": 0.2856,
      "step": 12200
    },
    {
      "epoch": 0.2203720957106436,
      "grad_norm": 0.34900498390197754,
      "learning_rate": 1.7697627972893195e-05,
      "loss": 0.2837,
      "step": 12205
    },
    {
      "epoch": 0.2204623751435443,
      "grad_norm": 0.3512410819530487,
      "learning_rate": 1.7695817204364067e-05,
      "loss": 0.2292,
      "step": 12210
    },
    {
      "epoch": 0.22055265457644502,
      "grad_norm": 0.29271936416625977,
      "learning_rate": 1.7694005816762393e-05,
      "loss": 0.2282,
      "step": 12215
    },
    {
      "epoch": 0.22064293400934573,
      "grad_norm": 0.520492672920227,
      "learning_rate": 1.7692193810233896e-05,
      "loss": 0.341,
      "step": 12220
    },
    {
      "epoch": 0.22073321344224645,
      "grad_norm": 0.4727719724178314,
      "learning_rate": 1.7690381184924335e-05,
      "loss": 0.3201,
      "step": 12225
    },
    {
      "epoch": 0.22082349287514716,
      "grad_norm": 0.7370137572288513,
      "learning_rate": 1.768856794097952e-05,
      "loss": 0.37,
      "step": 12230
    },
    {
      "epoch": 0.22091377230804787,
      "grad_norm": 0.8863621354103088,
      "learning_rate": 1.7686754078545317e-05,
      "loss": 0.2645,
      "step": 12235
    },
    {
      "epoch": 0.2210040517409486,
      "grad_norm": 0.2831362187862396,
      "learning_rate": 1.7684939597767635e-05,
      "loss": 0.2702,
      "step": 12240
    },
    {
      "epoch": 0.2210943311738493,
      "grad_norm": 0.8353209495544434,
      "learning_rate": 1.7683124498792436e-05,
      "loss": 0.2362,
      "step": 12245
    },
    {
      "epoch": 0.22118461060675002,
      "grad_norm": 0.5277425050735474,
      "learning_rate": 1.7681308781765736e-05,
      "loss": 0.252,
      "step": 12250
    },
    {
      "epoch": 0.22127489003965073,
      "grad_norm": 0.4916650950908661,
      "learning_rate": 1.7679492446833594e-05,
      "loss": 0.2815,
      "step": 12255
    },
    {
      "epoch": 0.22136516947255144,
      "grad_norm": 0.44698864221572876,
      "learning_rate": 1.7677675494142115e-05,
      "loss": 0.3273,
      "step": 12260
    },
    {
      "epoch": 0.22145544890545216,
      "grad_norm": 0.42281705141067505,
      "learning_rate": 1.7675857923837467e-05,
      "loss": 0.359,
      "step": 12265
    },
    {
      "epoch": 0.22154572833835287,
      "grad_norm": 0.3011576533317566,
      "learning_rate": 1.767403973606586e-05,
      "loss": 0.2526,
      "step": 12270
    },
    {
      "epoch": 0.22163600777125358,
      "grad_norm": 0.4375670850276947,
      "learning_rate": 1.7672220930973548e-05,
      "loss": 0.2751,
      "step": 12275
    },
    {
      "epoch": 0.2217262872041543,
      "grad_norm": 0.5508677363395691,
      "learning_rate": 1.767040150870685e-05,
      "loss": 0.3449,
      "step": 12280
    },
    {
      "epoch": 0.221816566637055,
      "grad_norm": 0.8546921610832214,
      "learning_rate": 1.7668581469412114e-05,
      "loss": 0.2519,
      "step": 12285
    },
    {
      "epoch": 0.22190684606995573,
      "grad_norm": 0.748357892036438,
      "learning_rate": 1.7666760813235756e-05,
      "loss": 0.2205,
      "step": 12290
    },
    {
      "epoch": 0.22199712550285644,
      "grad_norm": 0.6457489132881165,
      "learning_rate": 1.7664939540324236e-05,
      "loss": 0.311,
      "step": 12295
    },
    {
      "epoch": 0.22208740493575715,
      "grad_norm": 0.5729885697364807,
      "learning_rate": 1.7663117650824057e-05,
      "loss": 0.3258,
      "step": 12300
    },
    {
      "epoch": 0.22217768436865787,
      "grad_norm": 0.5058196783065796,
      "learning_rate": 1.766129514488178e-05,
      "loss": 0.3847,
      "step": 12305
    },
    {
      "epoch": 0.22226796380155858,
      "grad_norm": 0.5538528561592102,
      "learning_rate": 1.7659472022644012e-05,
      "loss": 0.2343,
      "step": 12310
    },
    {
      "epoch": 0.2223582432344593,
      "grad_norm": 0.34853270649909973,
      "learning_rate": 1.7657648284257406e-05,
      "loss": 0.3126,
      "step": 12315
    },
    {
      "epoch": 0.22244852266736,
      "grad_norm": 0.5706051588058472,
      "learning_rate": 1.7655823929868677e-05,
      "loss": 0.346,
      "step": 12320
    },
    {
      "epoch": 0.22253880210026072,
      "grad_norm": 0.6547861099243164,
      "learning_rate": 1.7653998959624573e-05,
      "loss": 0.4004,
      "step": 12325
    },
    {
      "epoch": 0.22262908153316144,
      "grad_norm": 0.5828825831413269,
      "learning_rate": 1.7652173373671905e-05,
      "loss": 0.32,
      "step": 12330
    },
    {
      "epoch": 0.22271936096606215,
      "grad_norm": 0.7183968424797058,
      "learning_rate": 1.7650347172157523e-05,
      "loss": 0.2042,
      "step": 12335
    },
    {
      "epoch": 0.22280964039896287,
      "grad_norm": 0.4506465494632721,
      "learning_rate": 1.7648520355228335e-05,
      "loss": 0.234,
      "step": 12340
    },
    {
      "epoch": 0.22289991983186358,
      "grad_norm": 0.40599194169044495,
      "learning_rate": 1.7646692923031298e-05,
      "loss": 0.3196,
      "step": 12345
    },
    {
      "epoch": 0.2229901992647643,
      "grad_norm": 0.6732317805290222,
      "learning_rate": 1.764486487571341e-05,
      "loss": 0.3119,
      "step": 12350
    },
    {
      "epoch": 0.223080478697665,
      "grad_norm": 0.5106104612350464,
      "learning_rate": 1.7643036213421724e-05,
      "loss": 0.3718,
      "step": 12355
    },
    {
      "epoch": 0.22317075813056572,
      "grad_norm": 0.29887324571609497,
      "learning_rate": 1.7641206936303348e-05,
      "loss": 0.214,
      "step": 12360
    },
    {
      "epoch": 0.22326103756346644,
      "grad_norm": 0.4310643970966339,
      "learning_rate": 1.763937704450543e-05,
      "loss": 0.2257,
      "step": 12365
    },
    {
      "epoch": 0.22335131699636715,
      "grad_norm": 0.5841691493988037,
      "learning_rate": 1.763754653817517e-05,
      "loss": 0.3455,
      "step": 12370
    },
    {
      "epoch": 0.22344159642926786,
      "grad_norm": 0.26802316308021545,
      "learning_rate": 1.7635715417459828e-05,
      "loss": 0.2342,
      "step": 12375
    },
    {
      "epoch": 0.22353187586216858,
      "grad_norm": 0.734042227268219,
      "learning_rate": 1.763388368250669e-05,
      "loss": 0.3666,
      "step": 12380
    },
    {
      "epoch": 0.2236221552950693,
      "grad_norm": 0.5031940937042236,
      "learning_rate": 1.763205133346312e-05,
      "loss": 0.3489,
      "step": 12385
    },
    {
      "epoch": 0.22371243472797,
      "grad_norm": 0.4350252151489258,
      "learning_rate": 1.7630218370476506e-05,
      "loss": 0.3381,
      "step": 12390
    },
    {
      "epoch": 0.22380271416087072,
      "grad_norm": 0.3086232542991638,
      "learning_rate": 1.7628384793694302e-05,
      "loss": 0.2875,
      "step": 12395
    },
    {
      "epoch": 0.22389299359377143,
      "grad_norm": 0.6226441860198975,
      "learning_rate": 1.7626550603264005e-05,
      "loss": 0.2564,
      "step": 12400
    },
    {
      "epoch": 0.22398327302667215,
      "grad_norm": 0.42126598954200745,
      "learning_rate": 1.7624715799333163e-05,
      "loss": 0.3476,
      "step": 12405
    },
    {
      "epoch": 0.22407355245957286,
      "grad_norm": 0.4345131814479828,
      "learning_rate": 1.762288038204937e-05,
      "loss": 0.3486,
      "step": 12410
    },
    {
      "epoch": 0.22416383189247358,
      "grad_norm": 0.41714054346084595,
      "learning_rate": 1.7621044351560276e-05,
      "loss": 0.2472,
      "step": 12415
    },
    {
      "epoch": 0.2242541113253743,
      "grad_norm": 0.3679058253765106,
      "learning_rate": 1.7619207708013575e-05,
      "loss": 0.3669,
      "step": 12420
    },
    {
      "epoch": 0.224344390758275,
      "grad_norm": 0.421440988779068,
      "learning_rate": 1.761737045155701e-05,
      "loss": 0.3501,
      "step": 12425
    },
    {
      "epoch": 0.22443467019117572,
      "grad_norm": 0.5317081809043884,
      "learning_rate": 1.761553258233838e-05,
      "loss": 0.3304,
      "step": 12430
    },
    {
      "epoch": 0.22452494962407643,
      "grad_norm": 0.4892217814922333,
      "learning_rate": 1.7613694100505515e-05,
      "loss": 0.3608,
      "step": 12435
    },
    {
      "epoch": 0.22461522905697714,
      "grad_norm": 0.590078592300415,
      "learning_rate": 1.7611855006206322e-05,
      "loss": 0.2907,
      "step": 12440
    },
    {
      "epoch": 0.22470550848987786,
      "grad_norm": 0.37204211950302124,
      "learning_rate": 1.7610015299588737e-05,
      "loss": 0.2403,
      "step": 12445
    },
    {
      "epoch": 0.22479578792277857,
      "grad_norm": 0.3007202446460724,
      "learning_rate": 1.7608174980800748e-05,
      "loss": 0.3343,
      "step": 12450
    },
    {
      "epoch": 0.2248860673556793,
      "grad_norm": 0.38273823261260986,
      "learning_rate": 1.7606334049990405e-05,
      "loss": 0.2856,
      "step": 12455
    },
    {
      "epoch": 0.22497634678858,
      "grad_norm": 0.8064092993736267,
      "learning_rate": 1.7604492507305788e-05,
      "loss": 0.2644,
      "step": 12460
    },
    {
      "epoch": 0.22506662622148071,
      "grad_norm": 0.44359609484672546,
      "learning_rate": 1.7602650352895037e-05,
      "loss": 0.3132,
      "step": 12465
    },
    {
      "epoch": 0.22515690565438143,
      "grad_norm": 0.3986911475658417,
      "learning_rate": 1.7600807586906343e-05,
      "loss": 0.2842,
      "step": 12470
    },
    {
      "epoch": 0.22524718508728214,
      "grad_norm": 0.4192790985107422,
      "learning_rate": 1.7598964209487945e-05,
      "loss": 0.2547,
      "step": 12475
    },
    {
      "epoch": 0.22533746452018286,
      "grad_norm": 0.4528927505016327,
      "learning_rate": 1.759712022078812e-05,
      "loss": 0.2174,
      "step": 12480
    },
    {
      "epoch": 0.2254277439530836,
      "grad_norm": 0.8167347311973572,
      "learning_rate": 1.759527562095521e-05,
      "loss": 0.42,
      "step": 12485
    },
    {
      "epoch": 0.2255180233859843,
      "grad_norm": 0.34179383516311646,
      "learning_rate": 1.7593430410137604e-05,
      "loss": 0.2672,
      "step": 12490
    },
    {
      "epoch": 0.22560830281888503,
      "grad_norm": 0.5470631122589111,
      "learning_rate": 1.759158458848373e-05,
      "loss": 0.3076,
      "step": 12495
    },
    {
      "epoch": 0.22569858225178574,
      "grad_norm": 0.38790619373321533,
      "learning_rate": 1.758973815614207e-05,
      "loss": 0.2712,
      "step": 12500
    },
    {
      "epoch": 0.22578886168468645,
      "grad_norm": 0.5486795902252197,
      "learning_rate": 1.7587891113261163e-05,
      "loss": 0.3046,
      "step": 12505
    },
    {
      "epoch": 0.22587914111758717,
      "grad_norm": 0.5758505463600159,
      "learning_rate": 1.758604345998958e-05,
      "loss": 0.2934,
      "step": 12510
    },
    {
      "epoch": 0.22596942055048788,
      "grad_norm": 0.5941413044929504,
      "learning_rate": 1.7584195196475956e-05,
      "loss": 0.2367,
      "step": 12515
    },
    {
      "epoch": 0.2260596999833886,
      "grad_norm": 0.4801849126815796,
      "learning_rate": 1.758234632286897e-05,
      "loss": 0.3754,
      "step": 12520
    },
    {
      "epoch": 0.2261499794162893,
      "grad_norm": 0.5735433101654053,
      "learning_rate": 1.758049683931736e-05,
      "loss": 0.3029,
      "step": 12525
    },
    {
      "epoch": 0.22624025884919002,
      "grad_norm": 0.35783183574676514,
      "learning_rate": 1.7578646745969886e-05,
      "loss": 0.2535,
      "step": 12530
    },
    {
      "epoch": 0.22633053828209074,
      "grad_norm": 0.38716718554496765,
      "learning_rate": 1.757679604297539e-05,
      "loss": 0.2672,
      "step": 12535
    },
    {
      "epoch": 0.22642081771499145,
      "grad_norm": 0.3761030435562134,
      "learning_rate": 1.7574944730482735e-05,
      "loss": 0.2515,
      "step": 12540
    },
    {
      "epoch": 0.22651109714789217,
      "grad_norm": 0.3976900279521942,
      "learning_rate": 1.757309280864085e-05,
      "loss": 0.4006,
      "step": 12545
    },
    {
      "epoch": 0.22660137658079288,
      "grad_norm": 0.46611571311950684,
      "learning_rate": 1.7571240277598716e-05,
      "loss": 0.2155,
      "step": 12550
    },
    {
      "epoch": 0.2266916560136936,
      "grad_norm": 0.44464078545570374,
      "learning_rate": 1.756938713750535e-05,
      "loss": 0.3367,
      "step": 12555
    },
    {
      "epoch": 0.2267819354465943,
      "grad_norm": 0.3445928394794464,
      "learning_rate": 1.756753338850982e-05,
      "loss": 0.3938,
      "step": 12560
    },
    {
      "epoch": 0.22687221487949502,
      "grad_norm": 0.5699447393417358,
      "learning_rate": 1.756567903076125e-05,
      "loss": 0.3994,
      "step": 12565
    },
    {
      "epoch": 0.22696249431239574,
      "grad_norm": 0.4590452015399933,
      "learning_rate": 1.7563824064408814e-05,
      "loss": 0.2082,
      "step": 12570
    },
    {
      "epoch": 0.22705277374529645,
      "grad_norm": 0.3316526412963867,
      "learning_rate": 1.7561968489601725e-05,
      "loss": 0.3287,
      "step": 12575
    },
    {
      "epoch": 0.22714305317819716,
      "grad_norm": 0.5490213632583618,
      "learning_rate": 1.756011230648925e-05,
      "loss": 0.3007,
      "step": 12580
    },
    {
      "epoch": 0.22723333261109788,
      "grad_norm": 0.6003979444503784,
      "learning_rate": 1.7558255515220707e-05,
      "loss": 0.2635,
      "step": 12585
    },
    {
      "epoch": 0.2273236120439986,
      "grad_norm": 0.5353501439094543,
      "learning_rate": 1.7556398115945463e-05,
      "loss": 0.3996,
      "step": 12590
    },
    {
      "epoch": 0.2274138914768993,
      "grad_norm": 0.3782620131969452,
      "learning_rate": 1.755454010881293e-05,
      "loss": 0.2767,
      "step": 12595
    },
    {
      "epoch": 0.22750417090980002,
      "grad_norm": 0.3071882128715515,
      "learning_rate": 1.7552681493972573e-05,
      "loss": 0.255,
      "step": 12600
    },
    {
      "epoch": 0.22759445034270073,
      "grad_norm": 0.3926979899406433,
      "learning_rate": 1.7550822271573903e-05,
      "loss": 0.1907,
      "step": 12605
    },
    {
      "epoch": 0.22768472977560145,
      "grad_norm": 0.44747939705848694,
      "learning_rate": 1.754896244176648e-05,
      "loss": 0.2919,
      "step": 12610
    },
    {
      "epoch": 0.22777500920850216,
      "grad_norm": 0.49764126539230347,
      "learning_rate": 1.7547102004699917e-05,
      "loss": 0.2617,
      "step": 12615
    },
    {
      "epoch": 0.22786528864140287,
      "grad_norm": 0.7542524337768555,
      "learning_rate": 1.7545240960523865e-05,
      "loss": 0.3394,
      "step": 12620
    },
    {
      "epoch": 0.2279555680743036,
      "grad_norm": 0.45615649223327637,
      "learning_rate": 1.7543379309388045e-05,
      "loss": 0.2543,
      "step": 12625
    },
    {
      "epoch": 0.2280458475072043,
      "grad_norm": 0.642106831073761,
      "learning_rate": 1.75415170514422e-05,
      "loss": 0.2593,
      "step": 12630
    },
    {
      "epoch": 0.22813612694010502,
      "grad_norm": 0.4519185721874237,
      "learning_rate": 1.753965418683614e-05,
      "loss": 0.2457,
      "step": 12635
    },
    {
      "epoch": 0.22822640637300573,
      "grad_norm": 0.9257187843322754,
      "learning_rate": 1.7537790715719723e-05,
      "loss": 0.3096,
      "step": 12640
    },
    {
      "epoch": 0.22831668580590644,
      "grad_norm": 0.34161514043807983,
      "learning_rate": 1.7535926638242843e-05,
      "loss": 0.2755,
      "step": 12645
    },
    {
      "epoch": 0.22840696523880716,
      "grad_norm": 0.5337136387825012,
      "learning_rate": 1.753406195455546e-05,
      "loss": 0.2925,
      "step": 12650
    },
    {
      "epoch": 0.22849724467170787,
      "grad_norm": 0.730758786201477,
      "learning_rate": 1.753219666480757e-05,
      "loss": 0.3478,
      "step": 12655
    },
    {
      "epoch": 0.22858752410460859,
      "grad_norm": 0.40018904209136963,
      "learning_rate": 1.7530330769149222e-05,
      "loss": 0.2867,
      "step": 12660
    },
    {
      "epoch": 0.2286778035375093,
      "grad_norm": 0.4671535789966583,
      "learning_rate": 1.7528464267730518e-05,
      "loss": 0.2723,
      "step": 12665
    },
    {
      "epoch": 0.22876808297041,
      "grad_norm": 0.5026867985725403,
      "learning_rate": 1.7526597160701596e-05,
      "loss": 0.3095,
      "step": 12670
    },
    {
      "epoch": 0.22885836240331073,
      "grad_norm": 0.53568035364151,
      "learning_rate": 1.752472944821266e-05,
      "loss": 0.282,
      "step": 12675
    },
    {
      "epoch": 0.22894864183621144,
      "grad_norm": 0.3272496163845062,
      "learning_rate": 1.7522861130413944e-05,
      "loss": 0.2907,
      "step": 12680
    },
    {
      "epoch": 0.22903892126911216,
      "grad_norm": 0.393658310174942,
      "learning_rate": 1.7520992207455752e-05,
      "loss": 0.3212,
      "step": 12685
    },
    {
      "epoch": 0.22912920070201287,
      "grad_norm": 0.35009339451789856,
      "learning_rate": 1.7519122679488417e-05,
      "loss": 0.252,
      "step": 12690
    },
    {
      "epoch": 0.22921948013491358,
      "grad_norm": 0.6266551613807678,
      "learning_rate": 1.7517252546662333e-05,
      "loss": 0.2781,
      "step": 12695
    },
    {
      "epoch": 0.2293097595678143,
      "grad_norm": 0.37133973836898804,
      "learning_rate": 1.7515381809127938e-05,
      "loss": 0.3435,
      "step": 12700
    },
    {
      "epoch": 0.229400039000715,
      "grad_norm": 0.46204811334609985,
      "learning_rate": 1.7513510467035717e-05,
      "loss": 0.2844,
      "step": 12705
    },
    {
      "epoch": 0.22949031843361573,
      "grad_norm": 0.3849553167819977,
      "learning_rate": 1.7511638520536208e-05,
      "loss": 0.2516,
      "step": 12710
    },
    {
      "epoch": 0.22958059786651644,
      "grad_norm": 0.4111543595790863,
      "learning_rate": 1.7509765969779996e-05,
      "loss": 0.2439,
      "step": 12715
    },
    {
      "epoch": 0.22967087729941715,
      "grad_norm": 0.3730176091194153,
      "learning_rate": 1.750789281491771e-05,
      "loss": 0.3372,
      "step": 12720
    },
    {
      "epoch": 0.22976115673231787,
      "grad_norm": 0.5229792594909668,
      "learning_rate": 1.750601905610004e-05,
      "loss": 0.3364,
      "step": 12725
    },
    {
      "epoch": 0.22985143616521858,
      "grad_norm": 0.43100789189338684,
      "learning_rate": 1.7504144693477706e-05,
      "loss": 0.2306,
      "step": 12730
    },
    {
      "epoch": 0.2299417155981193,
      "grad_norm": 0.5689313411712646,
      "learning_rate": 1.7502269727201494e-05,
      "loss": 0.2632,
      "step": 12735
    },
    {
      "epoch": 0.23003199503102,
      "grad_norm": 0.5734375715255737,
      "learning_rate": 1.7500394157422228e-05,
      "loss": 0.2136,
      "step": 12740
    },
    {
      "epoch": 0.23012227446392072,
      "grad_norm": 0.5006566643714905,
      "learning_rate": 1.7498517984290786e-05,
      "loss": 0.2878,
      "step": 12745
    },
    {
      "epoch": 0.23021255389682144,
      "grad_norm": 0.4980749487876892,
      "learning_rate": 1.749664120795809e-05,
      "loss": 0.3401,
      "step": 12750
    },
    {
      "epoch": 0.23030283332972215,
      "grad_norm": 0.8709957599639893,
      "learning_rate": 1.749476382857512e-05,
      "loss": 0.2772,
      "step": 12755
    },
    {
      "epoch": 0.23039311276262286,
      "grad_norm": 0.5566146373748779,
      "learning_rate": 1.7492885846292887e-05,
      "loss": 0.2313,
      "step": 12760
    },
    {
      "epoch": 0.23048339219552358,
      "grad_norm": 0.37837234139442444,
      "learning_rate": 1.7491007261262466e-05,
      "loss": 0.2771,
      "step": 12765
    },
    {
      "epoch": 0.2305736716284243,
      "grad_norm": 0.4560678005218506,
      "learning_rate": 1.7489128073634978e-05,
      "loss": 0.2826,
      "step": 12770
    },
    {
      "epoch": 0.230663951061325,
      "grad_norm": 0.4916342794895172,
      "learning_rate": 1.7487248283561588e-05,
      "loss": 0.339,
      "step": 12775
    },
    {
      "epoch": 0.23075423049422572,
      "grad_norm": 0.5683901906013489,
      "learning_rate": 1.748536789119351e-05,
      "loss": 0.3836,
      "step": 12780
    },
    {
      "epoch": 0.23084450992712643,
      "grad_norm": 0.31839752197265625,
      "learning_rate": 1.748348689668201e-05,
      "loss": 0.2103,
      "step": 12785
    },
    {
      "epoch": 0.23093478936002715,
      "grad_norm": 0.3599136769771576,
      "learning_rate": 1.74816053001784e-05,
      "loss": 0.3974,
      "step": 12790
    },
    {
      "epoch": 0.23102506879292786,
      "grad_norm": 0.594417154788971,
      "learning_rate": 1.747972310183404e-05,
      "loss": 0.2634,
      "step": 12795
    },
    {
      "epoch": 0.23111534822582858,
      "grad_norm": 0.7117186784744263,
      "learning_rate": 1.7477840301800336e-05,
      "loss": 0.3887,
      "step": 12800
    },
    {
      "epoch": 0.2312056276587293,
      "grad_norm": 0.3433570861816406,
      "learning_rate": 1.747595690022875e-05,
      "loss": 0.3705,
      "step": 12805
    },
    {
      "epoch": 0.23129590709163,
      "grad_norm": 0.4502484202384949,
      "learning_rate": 1.747407289727079e-05,
      "loss": 0.2893,
      "step": 12810
    },
    {
      "epoch": 0.23138618652453072,
      "grad_norm": 0.5645655989646912,
      "learning_rate": 1.7472188293078003e-05,
      "loss": 0.3901,
      "step": 12815
    },
    {
      "epoch": 0.23147646595743143,
      "grad_norm": 0.5908382534980774,
      "learning_rate": 1.7470303087801998e-05,
      "loss": 0.3332,
      "step": 12820
    },
    {
      "epoch": 0.23156674539033215,
      "grad_norm": 0.32872527837753296,
      "learning_rate": 1.7468417281594426e-05,
      "loss": 0.2765,
      "step": 12825
    },
    {
      "epoch": 0.23165702482323286,
      "grad_norm": 0.38247713446617126,
      "learning_rate": 1.746653087460698e-05,
      "loss": 0.2047,
      "step": 12830
    },
    {
      "epoch": 0.23174730425613357,
      "grad_norm": 0.435955673456192,
      "learning_rate": 1.746464386699141e-05,
      "loss": 0.2916,
      "step": 12835
    },
    {
      "epoch": 0.2318375836890343,
      "grad_norm": 0.6716747283935547,
      "learning_rate": 1.746275625889952e-05,
      "loss": 0.343,
      "step": 12840
    },
    {
      "epoch": 0.231927863121935,
      "grad_norm": 0.4624778926372528,
      "learning_rate": 1.746086805048315e-05,
      "loss": 0.1836,
      "step": 12845
    },
    {
      "epoch": 0.23201814255483572,
      "grad_norm": 0.39304831624031067,
      "learning_rate": 1.7458979241894188e-05,
      "loss": 0.3215,
      "step": 12850
    },
    {
      "epoch": 0.23210842198773643,
      "grad_norm": 0.5324734449386597,
      "learning_rate": 1.745708983328458e-05,
      "loss": 0.3876,
      "step": 12855
    },
    {
      "epoch": 0.23219870142063714,
      "grad_norm": 0.36722588539123535,
      "learning_rate": 1.7455199824806307e-05,
      "loss": 0.2532,
      "step": 12860
    },
    {
      "epoch": 0.23228898085353786,
      "grad_norm": 0.4719274342060089,
      "learning_rate": 1.745330921661142e-05,
      "loss": 0.3529,
      "step": 12865
    },
    {
      "epoch": 0.23237926028643857,
      "grad_norm": 0.6899310350418091,
      "learning_rate": 1.7451418008851992e-05,
      "loss": 0.4033,
      "step": 12870
    },
    {
      "epoch": 0.23246953971933929,
      "grad_norm": 0.36658185720443726,
      "learning_rate": 1.7449526201680165e-05,
      "loss": 0.2566,
      "step": 12875
    },
    {
      "epoch": 0.23255981915224003,
      "grad_norm": 0.6452036499977112,
      "learning_rate": 1.7447633795248117e-05,
      "loss": 0.2768,
      "step": 12880
    },
    {
      "epoch": 0.23265009858514074,
      "grad_norm": 0.7364159226417542,
      "learning_rate": 1.744574078970808e-05,
      "loss": 0.3431,
      "step": 12885
    },
    {
      "epoch": 0.23274037801804145,
      "grad_norm": 0.37923142313957214,
      "learning_rate": 1.7443847185212335e-05,
      "loss": 0.1545,
      "step": 12890
    },
    {
      "epoch": 0.23283065745094217,
      "grad_norm": 0.48003754019737244,
      "learning_rate": 1.7441952981913202e-05,
      "loss": 0.304,
      "step": 12895
    },
    {
      "epoch": 0.23292093688384288,
      "grad_norm": 0.4077073037624359,
      "learning_rate": 1.744005817996306e-05,
      "loss": 0.2717,
      "step": 12900
    },
    {
      "epoch": 0.2330112163167436,
      "grad_norm": 0.3822029232978821,
      "learning_rate": 1.7438162779514333e-05,
      "loss": 0.2555,
      "step": 12905
    },
    {
      "epoch": 0.2331014957496443,
      "grad_norm": 0.40967702865600586,
      "learning_rate": 1.743626678071949e-05,
      "loss": 0.3063,
      "step": 12910
    },
    {
      "epoch": 0.23319177518254502,
      "grad_norm": 0.8108331561088562,
      "learning_rate": 1.7434370183731057e-05,
      "loss": 0.2122,
      "step": 12915
    },
    {
      "epoch": 0.23328205461544574,
      "grad_norm": 0.599997878074646,
      "learning_rate": 1.743247298870159e-05,
      "loss": 0.2751,
      "step": 12920
    },
    {
      "epoch": 0.23337233404834645,
      "grad_norm": 0.32683104276657104,
      "learning_rate": 1.743057519578371e-05,
      "loss": 0.2915,
      "step": 12925
    },
    {
      "epoch": 0.23346261348124717,
      "grad_norm": 0.650092363357544,
      "learning_rate": 1.7428676805130084e-05,
      "loss": 0.3193,
      "step": 12930
    },
    {
      "epoch": 0.23355289291414788,
      "grad_norm": 0.7250920534133911,
      "learning_rate": 1.742677781689342e-05,
      "loss": 0.3154,
      "step": 12935
    },
    {
      "epoch": 0.2336431723470486,
      "grad_norm": 0.5595054626464844,
      "learning_rate": 1.7424878231226476e-05,
      "loss": 0.2339,
      "step": 12940
    },
    {
      "epoch": 0.2337334517799493,
      "grad_norm": 0.6523693203926086,
      "learning_rate": 1.7422978048282066e-05,
      "loss": 0.3467,
      "step": 12945
    },
    {
      "epoch": 0.23382373121285002,
      "grad_norm": 0.7626236081123352,
      "learning_rate": 1.742107726821304e-05,
      "loss": 0.3687,
      "step": 12950
    },
    {
      "epoch": 0.23391401064575074,
      "grad_norm": 0.570476770401001,
      "learning_rate": 1.7419175891172308e-05,
      "loss": 0.3433,
      "step": 12955
    },
    {
      "epoch": 0.23400429007865145,
      "grad_norm": 0.5262225866317749,
      "learning_rate": 1.7417273917312812e-05,
      "loss": 0.2786,
      "step": 12960
    },
    {
      "epoch": 0.23409456951155216,
      "grad_norm": 0.5009921789169312,
      "learning_rate": 1.7415371346787564e-05,
      "loss": 0.2433,
      "step": 12965
    },
    {
      "epoch": 0.23418484894445288,
      "grad_norm": 0.4326619803905487,
      "learning_rate": 1.74134681797496e-05,
      "loss": 0.3382,
      "step": 12970
    },
    {
      "epoch": 0.2342751283773536,
      "grad_norm": 0.7674878239631653,
      "learning_rate": 1.7411564416352028e-05,
      "loss": 0.2951,
      "step": 12975
    },
    {
      "epoch": 0.2343654078102543,
      "grad_norm": 0.45992231369018555,
      "learning_rate": 1.7409660056747983e-05,
      "loss": 0.3278,
      "step": 12980
    },
    {
      "epoch": 0.23445568724315502,
      "grad_norm": 0.5807640552520752,
      "learning_rate": 1.740775510109066e-05,
      "loss": 0.2749,
      "step": 12985
    },
    {
      "epoch": 0.23454596667605573,
      "grad_norm": 0.8449848294258118,
      "learning_rate": 1.74058495495333e-05,
      "loss": 0.3264,
      "step": 12990
    },
    {
      "epoch": 0.23463624610895645,
      "grad_norm": 0.336150199174881,
      "learning_rate": 1.7403943402229186e-05,
      "loss": 0.3483,
      "step": 12995
    },
    {
      "epoch": 0.23472652554185716,
      "grad_norm": 0.4193001985549927,
      "learning_rate": 1.740203665933166e-05,
      "loss": 0.3628,
      "step": 13000
    },
    {
      "epoch": 0.23481680497475788,
      "grad_norm": 0.2537536919116974,
      "learning_rate": 1.7400129320994105e-05,
      "loss": 0.2823,
      "step": 13005
    },
    {
      "epoch": 0.2349070844076586,
      "grad_norm": 0.5223426222801208,
      "learning_rate": 1.7398221387369945e-05,
      "loss": 0.272,
      "step": 13010
    },
    {
      "epoch": 0.2349973638405593,
      "grad_norm": 0.4388841390609741,
      "learning_rate": 1.7396312858612668e-05,
      "loss": 0.2198,
      "step": 13015
    },
    {
      "epoch": 0.23508764327346002,
      "grad_norm": 0.7844226360321045,
      "learning_rate": 1.73944037348758e-05,
      "loss": 0.3239,
      "step": 13020
    },
    {
      "epoch": 0.23517792270636073,
      "grad_norm": 0.3475538492202759,
      "learning_rate": 1.739249401631291e-05,
      "loss": 0.27,
      "step": 13025
    },
    {
      "epoch": 0.23526820213926145,
      "grad_norm": 0.5528688430786133,
      "learning_rate": 1.7390583703077627e-05,
      "loss": 0.3892,
      "step": 13030
    },
    {
      "epoch": 0.23535848157216216,
      "grad_norm": 0.3805859386920929,
      "learning_rate": 1.7388672795323617e-05,
      "loss": 0.2687,
      "step": 13035
    },
    {
      "epoch": 0.23544876100506287,
      "grad_norm": 0.4538038372993469,
      "learning_rate": 1.7386761293204606e-05,
      "loss": 0.2865,
      "step": 13040
    },
    {
      "epoch": 0.2355390404379636,
      "grad_norm": 0.4108891487121582,
      "learning_rate": 1.738484919687435e-05,
      "loss": 0.2868,
      "step": 13045
    },
    {
      "epoch": 0.2356293198708643,
      "grad_norm": 0.35251399874687195,
      "learning_rate": 1.7382936506486676e-05,
      "loss": 0.201,
      "step": 13050
    },
    {
      "epoch": 0.23571959930376502,
      "grad_norm": 0.4963095784187317,
      "learning_rate": 1.7381023222195436e-05,
      "loss": 0.2389,
      "step": 13055
    },
    {
      "epoch": 0.23580987873666573,
      "grad_norm": 1.053669810295105,
      "learning_rate": 1.737910934415454e-05,
      "loss": 0.1928,
      "step": 13060
    },
    {
      "epoch": 0.23590015816956644,
      "grad_norm": 0.34278637170791626,
      "learning_rate": 1.7377194872517953e-05,
      "loss": 0.2457,
      "step": 13065
    },
    {
      "epoch": 0.23599043760246716,
      "grad_norm": 0.35964125394821167,
      "learning_rate": 1.7375279807439675e-05,
      "loss": 0.2703,
      "step": 13070
    },
    {
      "epoch": 0.23608071703536787,
      "grad_norm": 0.773762583732605,
      "learning_rate": 1.7373364149073757e-05,
      "loss": 0.239,
      "step": 13075
    },
    {
      "epoch": 0.23617099646826858,
      "grad_norm": 0.504447877407074,
      "learning_rate": 1.7371447897574305e-05,
      "loss": 0.2171,
      "step": 13080
    },
    {
      "epoch": 0.2362612759011693,
      "grad_norm": 0.4185939431190491,
      "learning_rate": 1.7369531053095466e-05,
      "loss": 0.3082,
      "step": 13085
    },
    {
      "epoch": 0.23635155533407,
      "grad_norm": 0.4185703694820404,
      "learning_rate": 1.7367613615791437e-05,
      "loss": 0.1746,
      "step": 13090
    },
    {
      "epoch": 0.23644183476697073,
      "grad_norm": 0.5308434963226318,
      "learning_rate": 1.736569558581646e-05,
      "loss": 0.3001,
      "step": 13095
    },
    {
      "epoch": 0.23653211419987144,
      "grad_norm": 0.5180681347846985,
      "learning_rate": 1.736377696332482e-05,
      "loss": 0.2886,
      "step": 13100
    },
    {
      "epoch": 0.23662239363277215,
      "grad_norm": 0.4812846779823303,
      "learning_rate": 1.736185774847087e-05,
      "loss": 0.2846,
      "step": 13105
    },
    {
      "epoch": 0.23671267306567287,
      "grad_norm": 0.5751941204071045,
      "learning_rate": 1.7359937941408986e-05,
      "loss": 0.4113,
      "step": 13110
    },
    {
      "epoch": 0.23680295249857358,
      "grad_norm": 0.546187162399292,
      "learning_rate": 1.735801754229361e-05,
      "loss": 0.273,
      "step": 13115
    },
    {
      "epoch": 0.2368932319314743,
      "grad_norm": 0.3179701864719391,
      "learning_rate": 1.735609655127922e-05,
      "loss": 0.3454,
      "step": 13120
    },
    {
      "epoch": 0.236983511364375,
      "grad_norm": 0.39066609740257263,
      "learning_rate": 1.7354174968520345e-05,
      "loss": 0.2601,
      "step": 13125
    },
    {
      "epoch": 0.23707379079727572,
      "grad_norm": 0.494882732629776,
      "learning_rate": 1.7352252794171564e-05,
      "loss": 0.3156,
      "step": 13130
    },
    {
      "epoch": 0.23716407023017644,
      "grad_norm": 0.37384721636772156,
      "learning_rate": 1.7350330028387504e-05,
      "loss": 0.2408,
      "step": 13135
    },
    {
      "epoch": 0.23725434966307715,
      "grad_norm": 0.40844231843948364,
      "learning_rate": 1.734840667132283e-05,
      "loss": 0.1769,
      "step": 13140
    },
    {
      "epoch": 0.23734462909597787,
      "grad_norm": 0.4068164825439453,
      "learning_rate": 1.7346482723132273e-05,
      "loss": 0.2484,
      "step": 13145
    },
    {
      "epoch": 0.23743490852887858,
      "grad_norm": 1.2531694173812866,
      "learning_rate": 1.734455818397059e-05,
      "loss": 0.3211,
      "step": 13150
    },
    {
      "epoch": 0.2375251879617793,
      "grad_norm": 0.2892794907093048,
      "learning_rate": 1.7342633053992607e-05,
      "loss": 0.331,
      "step": 13155
    },
    {
      "epoch": 0.23761546739468,
      "grad_norm": 0.6131587028503418,
      "learning_rate": 1.7340707333353178e-05,
      "loss": 0.2376,
      "step": 13160
    },
    {
      "epoch": 0.23770574682758072,
      "grad_norm": 0.5176225900650024,
      "learning_rate": 1.7338781022207217e-05,
      "loss": 0.2529,
      "step": 13165
    },
    {
      "epoch": 0.23779602626048144,
      "grad_norm": 0.38245654106140137,
      "learning_rate": 1.733685412070968e-05,
      "loss": 0.2604,
      "step": 13170
    },
    {
      "epoch": 0.23788630569338215,
      "grad_norm": 1.9461857080459595,
      "learning_rate": 1.7334926629015574e-05,
      "loss": 0.3495,
      "step": 13175
    },
    {
      "epoch": 0.23797658512628286,
      "grad_norm": 0.39455553889274597,
      "learning_rate": 1.733299854727995e-05,
      "loss": 0.287,
      "step": 13180
    },
    {
      "epoch": 0.23806686455918358,
      "grad_norm": 0.5157811045646667,
      "learning_rate": 1.733106987565791e-05,
      "loss": 0.2726,
      "step": 13185
    },
    {
      "epoch": 0.2381571439920843,
      "grad_norm": 0.5824441909790039,
      "learning_rate": 1.73291406143046e-05,
      "loss": 0.3267,
      "step": 13190
    },
    {
      "epoch": 0.238247423424985,
      "grad_norm": 0.5017169117927551,
      "learning_rate": 1.7327210763375213e-05,
      "loss": 0.3701,
      "step": 13195
    },
    {
      "epoch": 0.23833770285788572,
      "grad_norm": 0.47364509105682373,
      "learning_rate": 1.7325280323024997e-05,
      "loss": 0.3399,
      "step": 13200
    },
    {
      "epoch": 0.23842798229078643,
      "grad_norm": 0.3858482241630554,
      "learning_rate": 1.7323349293409236e-05,
      "loss": 0.3277,
      "step": 13205
    },
    {
      "epoch": 0.23851826172368715,
      "grad_norm": 0.4815135598182678,
      "learning_rate": 1.732141767468327e-05,
      "loss": 0.2735,
      "step": 13210
    },
    {
      "epoch": 0.23860854115658786,
      "grad_norm": 0.5121126174926758,
      "learning_rate": 1.731948546700249e-05,
      "loss": 0.3464,
      "step": 13215
    },
    {
      "epoch": 0.23869882058948858,
      "grad_norm": 0.4351355731487274,
      "learning_rate": 1.7317552670522317e-05,
      "loss": 0.1957,
      "step": 13220
    },
    {
      "epoch": 0.2387891000223893,
      "grad_norm": 0.38218650221824646,
      "learning_rate": 1.7315619285398234e-05,
      "loss": 0.2746,
      "step": 13225
    },
    {
      "epoch": 0.23887937945529,
      "grad_norm": 0.34511899948120117,
      "learning_rate": 1.7313685311785776e-05,
      "loss": 0.3125,
      "step": 13230
    },
    {
      "epoch": 0.23896965888819072,
      "grad_norm": 0.44880589842796326,
      "learning_rate": 1.7311750749840503e-05,
      "loss": 0.393,
      "step": 13235
    },
    {
      "epoch": 0.23905993832109143,
      "grad_norm": 0.4408525824546814,
      "learning_rate": 1.730981559971805e-05,
      "loss": 0.3846,
      "step": 13240
    },
    {
      "epoch": 0.23915021775399214,
      "grad_norm": 0.4854593873023987,
      "learning_rate": 1.7307879861574074e-05,
      "loss": 0.2986,
      "step": 13245
    },
    {
      "epoch": 0.23924049718689286,
      "grad_norm": 0.51319420337677,
      "learning_rate": 1.73059435355643e-05,
      "loss": 0.3189,
      "step": 13250
    },
    {
      "epoch": 0.23933077661979357,
      "grad_norm": 0.6964123249053955,
      "learning_rate": 1.7304006621844487e-05,
      "loss": 0.2609,
      "step": 13255
    },
    {
      "epoch": 0.2394210560526943,
      "grad_norm": 0.6647042036056519,
      "learning_rate": 1.7302069120570448e-05,
      "loss": 0.3141,
      "step": 13260
    },
    {
      "epoch": 0.239511335485595,
      "grad_norm": 0.4176599085330963,
      "learning_rate": 1.7300131031898036e-05,
      "loss": 0.2714,
      "step": 13265
    },
    {
      "epoch": 0.23960161491849571,
      "grad_norm": 0.3075650930404663,
      "learning_rate": 1.729819235598316e-05,
      "loss": 0.2895,
      "step": 13270
    },
    {
      "epoch": 0.23969189435139643,
      "grad_norm": 0.48430925607681274,
      "learning_rate": 1.7296253092981775e-05,
      "loss": 0.3174,
      "step": 13275
    },
    {
      "epoch": 0.23978217378429717,
      "grad_norm": 0.45159584283828735,
      "learning_rate": 1.7294313243049874e-05,
      "loss": 0.2679,
      "step": 13280
    },
    {
      "epoch": 0.23987245321719788,
      "grad_norm": 0.332610547542572,
      "learning_rate": 1.7292372806343508e-05,
      "loss": 0.282,
      "step": 13285
    },
    {
      "epoch": 0.2399627326500986,
      "grad_norm": 0.39457467198371887,
      "learning_rate": 1.7290431783018772e-05,
      "loss": 0.2021,
      "step": 13290
    },
    {
      "epoch": 0.2400530120829993,
      "grad_norm": 0.31859877705574036,
      "learning_rate": 1.7288490173231805e-05,
      "loss": 0.336,
      "step": 13295
    },
    {
      "epoch": 0.24014329151590003,
      "grad_norm": 0.36984288692474365,
      "learning_rate": 1.7286547977138797e-05,
      "loss": 0.2781,
      "step": 13300
    },
    {
      "epoch": 0.24023357094880074,
      "grad_norm": 0.3430190682411194,
      "learning_rate": 1.7284605194895983e-05,
      "loss": 0.331,
      "step": 13305
    },
    {
      "epoch": 0.24032385038170145,
      "grad_norm": 0.41883084177970886,
      "learning_rate": 1.7282661826659643e-05,
      "loss": 0.2829,
      "step": 13310
    },
    {
      "epoch": 0.24041412981460217,
      "grad_norm": 0.4166741669178009,
      "learning_rate": 1.7280717872586107e-05,
      "loss": 0.2906,
      "step": 13315
    },
    {
      "epoch": 0.24050440924750288,
      "grad_norm": 0.6059514880180359,
      "learning_rate": 1.7278773332831758e-05,
      "loss": 0.2372,
      "step": 13320
    },
    {
      "epoch": 0.2405946886804036,
      "grad_norm": 0.4602368175983429,
      "learning_rate": 1.7276828207553017e-05,
      "loss": 0.2972,
      "step": 13325
    },
    {
      "epoch": 0.2406849681133043,
      "grad_norm": 0.4222080409526825,
      "learning_rate": 1.7274882496906353e-05,
      "loss": 0.3206,
      "step": 13330
    },
    {
      "epoch": 0.24077524754620502,
      "grad_norm": 0.4080156981945038,
      "learning_rate": 1.7272936201048285e-05,
      "loss": 0.3971,
      "step": 13335
    },
    {
      "epoch": 0.24086552697910574,
      "grad_norm": 0.6896274089813232,
      "learning_rate": 1.727098932013538e-05,
      "loss": 0.3359,
      "step": 13340
    },
    {
      "epoch": 0.24095580641200645,
      "grad_norm": 0.40516936779022217,
      "learning_rate": 1.726904185432425e-05,
      "loss": 0.2301,
      "step": 13345
    },
    {
      "epoch": 0.24104608584490717,
      "grad_norm": 0.4957811236381531,
      "learning_rate": 1.726709380377155e-05,
      "loss": 0.3243,
      "step": 13350
    },
    {
      "epoch": 0.24113636527780788,
      "grad_norm": 0.6360941529273987,
      "learning_rate": 1.7265145168633992e-05,
      "loss": 0.343,
      "step": 13355
    },
    {
      "epoch": 0.2412266447107086,
      "grad_norm": 0.5391273498535156,
      "learning_rate": 1.726319594906833e-05,
      "loss": 0.3611,
      "step": 13360
    },
    {
      "epoch": 0.2413169241436093,
      "grad_norm": 0.4269796907901764,
      "learning_rate": 1.726124614523136e-05,
      "loss": 0.2855,
      "step": 13365
    },
    {
      "epoch": 0.24140720357651002,
      "grad_norm": 0.4511514902114868,
      "learning_rate": 1.7259295757279936e-05,
      "loss": 0.3093,
      "step": 13370
    },
    {
      "epoch": 0.24149748300941073,
      "grad_norm": 0.5537800788879395,
      "learning_rate": 1.7257344785370943e-05,
      "loss": 0.294,
      "step": 13375
    },
    {
      "epoch": 0.24158776244231145,
      "grad_norm": 0.36378413438796997,
      "learning_rate": 1.7255393229661334e-05,
      "loss": 0.2353,
      "step": 13380
    },
    {
      "epoch": 0.24167804187521216,
      "grad_norm": 0.30874359607696533,
      "learning_rate": 1.7253441090308087e-05,
      "loss": 0.33,
      "step": 13385
    },
    {
      "epoch": 0.24176832130811288,
      "grad_norm": 0.5672562718391418,
      "learning_rate": 1.7251488367468242e-05,
      "loss": 0.2755,
      "step": 13390
    },
    {
      "epoch": 0.2418586007410136,
      "grad_norm": 0.6343515515327454,
      "learning_rate": 1.7249535061298884e-05,
      "loss": 0.3293,
      "step": 13395
    },
    {
      "epoch": 0.2419488801739143,
      "grad_norm": 0.3627150356769562,
      "learning_rate": 1.7247581171957136e-05,
      "loss": 0.3181,
      "step": 13400
    },
    {
      "epoch": 0.24203915960681502,
      "grad_norm": 0.4247315526008606,
      "learning_rate": 1.724562669960018e-05,
      "loss": 0.2795,
      "step": 13405
    },
    {
      "epoch": 0.24212943903971573,
      "grad_norm": 0.37419256567955017,
      "learning_rate": 1.7243671644385237e-05,
      "loss": 0.2257,
      "step": 13410
    },
    {
      "epoch": 0.24221971847261645,
      "grad_norm": 0.391397088766098,
      "learning_rate": 1.7241716006469575e-05,
      "loss": 0.2604,
      "step": 13415
    },
    {
      "epoch": 0.24230999790551716,
      "grad_norm": 0.2934398055076599,
      "learning_rate": 1.7239759786010513e-05,
      "loss": 0.2096,
      "step": 13420
    },
    {
      "epoch": 0.24240027733841787,
      "grad_norm": 0.4060172438621521,
      "learning_rate": 1.7237802983165418e-05,
      "loss": 0.3578,
      "step": 13425
    },
    {
      "epoch": 0.2424905567713186,
      "grad_norm": 0.6043990850448608,
      "learning_rate": 1.7235845598091694e-05,
      "loss": 0.3541,
      "step": 13430
    },
    {
      "epoch": 0.2425808362042193,
      "grad_norm": 0.3376462161540985,
      "learning_rate": 1.72338876309468e-05,
      "loss": 0.223,
      "step": 13435
    },
    {
      "epoch": 0.24267111563712002,
      "grad_norm": 0.4019452929496765,
      "learning_rate": 1.7231929081888245e-05,
      "loss": 0.2463,
      "step": 13440
    },
    {
      "epoch": 0.24276139507002073,
      "grad_norm": 0.4395068883895874,
      "learning_rate": 1.7229969951073572e-05,
      "loss": 0.2632,
      "step": 13445
    },
    {
      "epoch": 0.24285167450292144,
      "grad_norm": 0.5303803086280823,
      "learning_rate": 1.7228010238660385e-05,
      "loss": 0.3481,
      "step": 13450
    },
    {
      "epoch": 0.24294195393582216,
      "grad_norm": 0.43933194875717163,
      "learning_rate": 1.722604994480633e-05,
      "loss": 0.3125,
      "step": 13455
    },
    {
      "epoch": 0.24303223336872287,
      "grad_norm": 0.42659083008766174,
      "learning_rate": 1.7224089069669093e-05,
      "loss": 0.2887,
      "step": 13460
    },
    {
      "epoch": 0.24312251280162359,
      "grad_norm": 0.5992361307144165,
      "learning_rate": 1.7222127613406417e-05,
      "loss": 0.3262,
      "step": 13465
    },
    {
      "epoch": 0.2432127922345243,
      "grad_norm": 0.31876376271247864,
      "learning_rate": 1.7220165576176082e-05,
      "loss": 0.3356,
      "step": 13470
    },
    {
      "epoch": 0.243303071667425,
      "grad_norm": 0.40272951126098633,
      "learning_rate": 1.721820295813592e-05,
      "loss": 0.3032,
      "step": 13475
    },
    {
      "epoch": 0.24339335110032573,
      "grad_norm": 0.46437716484069824,
      "learning_rate": 1.7216239759443814e-05,
      "loss": 0.2207,
      "step": 13480
    },
    {
      "epoch": 0.24348363053322644,
      "grad_norm": 0.48192399740219116,
      "learning_rate": 1.7214275980257688e-05,
      "loss": 0.2173,
      "step": 13485
    },
    {
      "epoch": 0.24357390996612716,
      "grad_norm": 0.4291000962257385,
      "learning_rate": 1.721231162073551e-05,
      "loss": 0.3343,
      "step": 13490
    },
    {
      "epoch": 0.24366418939902787,
      "grad_norm": 0.2683699429035187,
      "learning_rate": 1.7210346681035302e-05,
      "loss": 0.2025,
      "step": 13495
    },
    {
      "epoch": 0.24375446883192858,
      "grad_norm": 0.3970652222633362,
      "learning_rate": 1.7208381161315125e-05,
      "loss": 0.2634,
      "step": 13500
    },
    {
      "epoch": 0.2438447482648293,
      "grad_norm": 0.5615600943565369,
      "learning_rate": 1.7206415061733096e-05,
      "loss": 0.2508,
      "step": 13505
    },
    {
      "epoch": 0.24393502769773,
      "grad_norm": 0.3180173337459564,
      "learning_rate": 1.7204448382447375e-05,
      "loss": 0.3716,
      "step": 13510
    },
    {
      "epoch": 0.24402530713063073,
      "grad_norm": 0.7189514636993408,
      "learning_rate": 1.7202481123616156e-05,
      "loss": 0.2565,
      "step": 13515
    },
    {
      "epoch": 0.24411558656353144,
      "grad_norm": 0.3452807366847992,
      "learning_rate": 1.72005132853977e-05,
      "loss": 0.2418,
      "step": 13520
    },
    {
      "epoch": 0.24420586599643215,
      "grad_norm": 0.4875531494617462,
      "learning_rate": 1.7198544867950306e-05,
      "loss": 0.2748,
      "step": 13525
    },
    {
      "epoch": 0.24429614542933287,
      "grad_norm": 0.3451898694038391,
      "learning_rate": 1.7196575871432313e-05,
      "loss": 0.2704,
      "step": 13530
    },
    {
      "epoch": 0.24438642486223358,
      "grad_norm": 0.3998042047023773,
      "learning_rate": 1.7194606296002114e-05,
      "loss": 0.3006,
      "step": 13535
    },
    {
      "epoch": 0.2444767042951343,
      "grad_norm": 0.5872548818588257,
      "learning_rate": 1.719263614181815e-05,
      "loss": 0.3101,
      "step": 13540
    },
    {
      "epoch": 0.244566983728035,
      "grad_norm": 0.5717666149139404,
      "learning_rate": 1.7190665409038904e-05,
      "loss": 0.2535,
      "step": 13545
    },
    {
      "epoch": 0.24465726316093572,
      "grad_norm": 0.38812255859375,
      "learning_rate": 1.7188694097822907e-05,
      "loss": 0.327,
      "step": 13550
    },
    {
      "epoch": 0.24474754259383644,
      "grad_norm": 0.532560408115387,
      "learning_rate": 1.7186722208328735e-05,
      "loss": 0.3805,
      "step": 13555
    },
    {
      "epoch": 0.24483782202673715,
      "grad_norm": 0.5711003541946411,
      "learning_rate": 1.7184749740715017e-05,
      "loss": 0.2442,
      "step": 13560
    },
    {
      "epoch": 0.24492810145963786,
      "grad_norm": 0.4644785225391388,
      "learning_rate": 1.7182776695140416e-05,
      "loss": 0.2651,
      "step": 13565
    },
    {
      "epoch": 0.24501838089253858,
      "grad_norm": 0.3790689706802368,
      "learning_rate": 1.718080307176366e-05,
      "loss": 0.1993,
      "step": 13570
    },
    {
      "epoch": 0.2451086603254393,
      "grad_norm": 0.36735817790031433,
      "learning_rate": 1.7178828870743502e-05,
      "loss": 0.2514,
      "step": 13575
    },
    {
      "epoch": 0.24519893975834,
      "grad_norm": 0.4536728858947754,
      "learning_rate": 1.7176854092238756e-05,
      "loss": 0.232,
      "step": 13580
    },
    {
      "epoch": 0.24528921919124072,
      "grad_norm": 0.4569501578807831,
      "learning_rate": 1.7174878736408283e-05,
      "loss": 0.29,
      "step": 13585
    },
    {
      "epoch": 0.24537949862414143,
      "grad_norm": 0.4566640853881836,
      "learning_rate": 1.717290280341098e-05,
      "loss": 0.3562,
      "step": 13590
    },
    {
      "epoch": 0.24546977805704215,
      "grad_norm": 0.295539528131485,
      "learning_rate": 1.7170926293405795e-05,
      "loss": 0.2617,
      "step": 13595
    },
    {
      "epoch": 0.24556005748994286,
      "grad_norm": 0.7204494476318359,
      "learning_rate": 1.716894920655173e-05,
      "loss": 0.3241,
      "step": 13600
    },
    {
      "epoch": 0.24565033692284358,
      "grad_norm": 0.5700797438621521,
      "learning_rate": 1.7166971543007826e-05,
      "loss": 0.3685,
      "step": 13605
    },
    {
      "epoch": 0.2457406163557443,
      "grad_norm": 0.3525966703891754,
      "learning_rate": 1.716499330293317e-05,
      "loss": 0.2742,
      "step": 13610
    },
    {
      "epoch": 0.245830895788645,
      "grad_norm": 0.6157240867614746,
      "learning_rate": 1.71630144864869e-05,
      "loss": 0.4179,
      "step": 13615
    },
    {
      "epoch": 0.24592117522154572,
      "grad_norm": 0.37160009145736694,
      "learning_rate": 1.716103509382819e-05,
      "loss": 0.2981,
      "step": 13620
    },
    {
      "epoch": 0.24601145465444643,
      "grad_norm": 0.4578617215156555,
      "learning_rate": 1.7159055125116273e-05,
      "loss": 0.3848,
      "step": 13625
    },
    {
      "epoch": 0.24610173408734715,
      "grad_norm": 0.36614182591438293,
      "learning_rate": 1.7157074580510423e-05,
      "loss": 0.3155,
      "step": 13630
    },
    {
      "epoch": 0.24619201352024786,
      "grad_norm": 0.6067430973052979,
      "learning_rate": 1.715509346016996e-05,
      "loss": 0.224,
      "step": 13635
    },
    {
      "epoch": 0.24628229295314857,
      "grad_norm": 0.31626978516578674,
      "learning_rate": 1.715311176425425e-05,
      "loss": 0.2568,
      "step": 13640
    },
    {
      "epoch": 0.2463725723860493,
      "grad_norm": 0.3033553659915924,
      "learning_rate": 1.7151129492922708e-05,
      "loss": 0.2641,
      "step": 13645
    },
    {
      "epoch": 0.24646285181895,
      "grad_norm": 0.38154780864715576,
      "learning_rate": 1.7149146646334793e-05,
      "loss": 0.2752,
      "step": 13650
    },
    {
      "epoch": 0.24655313125185072,
      "grad_norm": 0.3966079652309418,
      "learning_rate": 1.7147163224650007e-05,
      "loss": 0.3328,
      "step": 13655
    },
    {
      "epoch": 0.24664341068475143,
      "grad_norm": 0.36520111560821533,
      "learning_rate": 1.7145179228027906e-05,
      "loss": 0.213,
      "step": 13660
    },
    {
      "epoch": 0.24673369011765214,
      "grad_norm": 0.4331825375556946,
      "learning_rate": 1.7143194656628088e-05,
      "loss": 0.3735,
      "step": 13665
    },
    {
      "epoch": 0.24682396955055286,
      "grad_norm": 0.453413188457489,
      "learning_rate": 1.7141209510610193e-05,
      "loss": 0.3221,
      "step": 13670
    },
    {
      "epoch": 0.2469142489834536,
      "grad_norm": 0.4321773946285248,
      "learning_rate": 1.7139223790133915e-05,
      "loss": 0.3019,
      "step": 13675
    },
    {
      "epoch": 0.2470045284163543,
      "grad_norm": 1.5241320133209229,
      "learning_rate": 1.7137237495358993e-05,
      "loss": 0.2381,
      "step": 13680
    },
    {
      "epoch": 0.24709480784925503,
      "grad_norm": 0.3903169631958008,
      "learning_rate": 1.7135250626445207e-05,
      "loss": 0.299,
      "step": 13685
    },
    {
      "epoch": 0.24718508728215574,
      "grad_norm": 0.3825934827327728,
      "learning_rate": 1.713326318355238e-05,
      "loss": 0.3858,
      "step": 13690
    },
    {
      "epoch": 0.24727536671505645,
      "grad_norm": 0.2576453685760498,
      "learning_rate": 1.7131275166840403e-05,
      "loss": 0.2984,
      "step": 13695
    },
    {
      "epoch": 0.24736564614795717,
      "grad_norm": 1.0343531370162964,
      "learning_rate": 1.712928657646918e-05,
      "loss": 0.3072,
      "step": 13700
    },
    {
      "epoch": 0.24745592558085788,
      "grad_norm": 0.45805513858795166,
      "learning_rate": 1.7127297412598693e-05,
      "loss": 0.2331,
      "step": 13705
    },
    {
      "epoch": 0.2475462050137586,
      "grad_norm": 0.20073974132537842,
      "learning_rate": 1.7125307675388947e-05,
      "loss": 0.2494,
      "step": 13710
    },
    {
      "epoch": 0.2476364844466593,
      "grad_norm": 0.3512426018714905,
      "learning_rate": 1.7123317365000004e-05,
      "loss": 0.3414,
      "step": 13715
    },
    {
      "epoch": 0.24772676387956002,
      "grad_norm": 0.25139570236206055,
      "learning_rate": 1.7121326481591972e-05,
      "loss": 0.2823,
      "step": 13720
    },
    {
      "epoch": 0.24781704331246074,
      "grad_norm": 0.5818011164665222,
      "learning_rate": 1.7119335025325e-05,
      "loss": 0.3291,
      "step": 13725
    },
    {
      "epoch": 0.24790732274536145,
      "grad_norm": 0.4541693329811096,
      "learning_rate": 1.711734299635929e-05,
      "loss": 0.2056,
      "step": 13730
    },
    {
      "epoch": 0.24799760217826217,
      "grad_norm": 0.3896014094352722,
      "learning_rate": 1.7115350394855083e-05,
      "loss": 0.2645,
      "step": 13735
    },
    {
      "epoch": 0.24808788161116288,
      "grad_norm": 0.42318210005760193,
      "learning_rate": 1.7113357220972668e-05,
      "loss": 0.2799,
      "step": 13740
    },
    {
      "epoch": 0.2481781610440636,
      "grad_norm": 0.48783060908317566,
      "learning_rate": 1.7111363474872385e-05,
      "loss": 0.2642,
      "step": 13745
    },
    {
      "epoch": 0.2482684404769643,
      "grad_norm": 0.4491179287433624,
      "learning_rate": 1.710936915671462e-05,
      "loss": 0.4675,
      "step": 13750
    },
    {
      "epoch": 0.24835871990986502,
      "grad_norm": 0.40137410163879395,
      "learning_rate": 1.710737426665979e-05,
      "loss": 0.2452,
      "step": 13755
    },
    {
      "epoch": 0.24844899934276574,
      "grad_norm": 0.5091173052787781,
      "learning_rate": 1.710537880486838e-05,
      "loss": 0.2596,
      "step": 13760
    },
    {
      "epoch": 0.24853927877566645,
      "grad_norm": 0.43215659260749817,
      "learning_rate": 1.71033827715009e-05,
      "loss": 0.2368,
      "step": 13765
    },
    {
      "epoch": 0.24862955820856716,
      "grad_norm": 0.5628123879432678,
      "learning_rate": 1.7101386166717928e-05,
      "loss": 0.2523,
      "step": 13770
    },
    {
      "epoch": 0.24871983764146788,
      "grad_norm": 0.376011461019516,
      "learning_rate": 1.709938899068007e-05,
      "loss": 0.3567,
      "step": 13775
    },
    {
      "epoch": 0.2488101170743686,
      "grad_norm": 0.4350323975086212,
      "learning_rate": 1.7097391243547985e-05,
      "loss": 0.2366,
      "step": 13780
    },
    {
      "epoch": 0.2489003965072693,
      "grad_norm": 0.49537816643714905,
      "learning_rate": 1.7095392925482378e-05,
      "loss": 0.252,
      "step": 13785
    },
    {
      "epoch": 0.24899067594017002,
      "grad_norm": 0.42540234327316284,
      "learning_rate": 1.7093394036644e-05,
      "loss": 0.2173,
      "step": 13790
    },
    {
      "epoch": 0.24908095537307073,
      "grad_norm": 0.4067361652851105,
      "learning_rate": 1.709139457719364e-05,
      "loss": 0.271,
      "step": 13795
    },
    {
      "epoch": 0.24917123480597145,
      "grad_norm": 0.5016900897026062,
      "learning_rate": 1.708939454729215e-05,
      "loss": 0.362,
      "step": 13800
    },
    {
      "epoch": 0.24926151423887216,
      "grad_norm": 0.5326197743415833,
      "learning_rate": 1.7087393947100413e-05,
      "loss": 0.3433,
      "step": 13805
    },
    {
      "epoch": 0.24935179367177288,
      "grad_norm": 0.4544302225112915,
      "learning_rate": 1.7085392776779363e-05,
      "loss": 0.3027,
      "step": 13810
    },
    {
      "epoch": 0.2494420731046736,
      "grad_norm": 0.5088821053504944,
      "learning_rate": 1.708339103648998e-05,
      "loss": 0.3039,
      "step": 13815
    },
    {
      "epoch": 0.2495323525375743,
      "grad_norm": 0.2824195921421051,
      "learning_rate": 1.708138872639329e-05,
      "loss": 0.1874,
      "step": 13820
    },
    {
      "epoch": 0.24962263197047502,
      "grad_norm": 0.39786961674690247,
      "learning_rate": 1.7079385846650366e-05,
      "loss": 0.2766,
      "step": 13825
    },
    {
      "epoch": 0.24971291140337573,
      "grad_norm": 0.5772325396537781,
      "learning_rate": 1.7077382397422317e-05,
      "loss": 0.2935,
      "step": 13830
    },
    {
      "epoch": 0.24980319083627645,
      "grad_norm": 0.39464205503463745,
      "learning_rate": 1.7075378378870315e-05,
      "loss": 0.2754,
      "step": 13835
    },
    {
      "epoch": 0.24989347026917716,
      "grad_norm": 0.4776792824268341,
      "learning_rate": 1.7073373791155564e-05,
      "loss": 0.3107,
      "step": 13840
    },
    {
      "epoch": 0.24998374970207787,
      "grad_norm": 0.7751340866088867,
      "learning_rate": 1.7071368634439322e-05,
      "loss": 0.3653,
      "step": 13845
    },
    {
      "epoch": 0.25007402913497856,
      "grad_norm": 0.3707751929759979,
      "learning_rate": 1.7069362908882885e-05,
      "loss": 0.2124,
      "step": 13850
    },
    {
      "epoch": 0.2501643085678793,
      "grad_norm": 0.48276251554489136,
      "learning_rate": 1.7067356614647603e-05,
      "loss": 0.2934,
      "step": 13855
    },
    {
      "epoch": 0.25025458800078,
      "grad_norm": 0.7727112174034119,
      "learning_rate": 1.7065349751894865e-05,
      "loss": 0.3284,
      "step": 13860
    },
    {
      "epoch": 0.25034486743368073,
      "grad_norm": 0.8128131628036499,
      "learning_rate": 1.706334232078611e-05,
      "loss": 0.3005,
      "step": 13865
    },
    {
      "epoch": 0.25043514686658147,
      "grad_norm": 0.20768576860427856,
      "learning_rate": 1.7061334321482822e-05,
      "loss": 0.2003,
      "step": 13870
    },
    {
      "epoch": 0.25052542629948216,
      "grad_norm": 0.3125922977924347,
      "learning_rate": 1.7059325754146528e-05,
      "loss": 0.2798,
      "step": 13875
    },
    {
      "epoch": 0.2506157057323829,
      "grad_norm": 0.5686619281768799,
      "learning_rate": 1.7057316618938803e-05,
      "loss": 0.3689,
      "step": 13880
    },
    {
      "epoch": 0.2507059851652836,
      "grad_norm": 0.5619451999664307,
      "learning_rate": 1.7055306916021266e-05,
      "loss": 0.2803,
      "step": 13885
    },
    {
      "epoch": 0.2507962645981843,
      "grad_norm": 0.5215336680412292,
      "learning_rate": 1.7053296645555587e-05,
      "loss": 0.3849,
      "step": 13890
    },
    {
      "epoch": 0.250886544031085,
      "grad_norm": 0.38564562797546387,
      "learning_rate": 1.7051285807703472e-05,
      "loss": 0.4052,
      "step": 13895
    },
    {
      "epoch": 0.25097682346398575,
      "grad_norm": 0.5514028668403625,
      "learning_rate": 1.7049274402626685e-05,
      "loss": 0.2089,
      "step": 13900
    },
    {
      "epoch": 0.25106710289688644,
      "grad_norm": 0.538511335849762,
      "learning_rate": 1.7047262430487024e-05,
      "loss": 0.2185,
      "step": 13905
    },
    {
      "epoch": 0.2511573823297872,
      "grad_norm": 0.736876368522644,
      "learning_rate": 1.7045249891446342e-05,
      "loss": 0.4343,
      "step": 13910
    },
    {
      "epoch": 0.25124766176268787,
      "grad_norm": 0.481503427028656,
      "learning_rate": 1.7043236785666523e-05,
      "loss": 0.254,
      "step": 13915
    },
    {
      "epoch": 0.2513379411955886,
      "grad_norm": 0.5032512545585632,
      "learning_rate": 1.704122311330952e-05,
      "loss": 0.2532,
      "step": 13920
    },
    {
      "epoch": 0.2514282206284893,
      "grad_norm": 0.41925984621047974,
      "learning_rate": 1.7039208874537313e-05,
      "loss": 0.3603,
      "step": 13925
    },
    {
      "epoch": 0.25151850006139004,
      "grad_norm": 0.5571647882461548,
      "learning_rate": 1.7037194069511925e-05,
      "loss": 0.2823,
      "step": 13930
    },
    {
      "epoch": 0.2516087794942907,
      "grad_norm": 0.5781228542327881,
      "learning_rate": 1.7035178698395445e-05,
      "loss": 0.379,
      "step": 13935
    },
    {
      "epoch": 0.25169905892719147,
      "grad_norm": 0.48722633719444275,
      "learning_rate": 1.7033162761349992e-05,
      "loss": 0.2829,
      "step": 13940
    },
    {
      "epoch": 0.25178933836009215,
      "grad_norm": 0.3715059161186218,
      "learning_rate": 1.7031146258537725e-05,
      "loss": 0.3258,
      "step": 13945
    },
    {
      "epoch": 0.2518796177929929,
      "grad_norm": 0.5888335704803467,
      "learning_rate": 1.7029129190120867e-05,
      "loss": 0.2693,
      "step": 13950
    },
    {
      "epoch": 0.2519698972258936,
      "grad_norm": 0.33220377564430237,
      "learning_rate": 1.702711155626167e-05,
      "loss": 0.227,
      "step": 13955
    },
    {
      "epoch": 0.2520601766587943,
      "grad_norm": 0.6103527545928955,
      "learning_rate": 1.702509335712244e-05,
      "loss": 0.2299,
      "step": 13960
    },
    {
      "epoch": 0.252150456091695,
      "grad_norm": 0.38191163539886475,
      "learning_rate": 1.7023074592865528e-05,
      "loss": 0.2695,
      "step": 13965
    },
    {
      "epoch": 0.25224073552459575,
      "grad_norm": 0.5467475652694702,
      "learning_rate": 1.702105526365333e-05,
      "loss": 0.3741,
      "step": 13970
    },
    {
      "epoch": 0.25233101495749644,
      "grad_norm": 0.6334359049797058,
      "learning_rate": 1.7019035369648278e-05,
      "loss": 0.2445,
      "step": 13975
    },
    {
      "epoch": 0.2524212943903972,
      "grad_norm": 0.4496508836746216,
      "learning_rate": 1.7017014911012868e-05,
      "loss": 0.242,
      "step": 13980
    },
    {
      "epoch": 0.25251157382329786,
      "grad_norm": 0.5502260327339172,
      "learning_rate": 1.7014993887909625e-05,
      "loss": 0.292,
      "step": 13985
    },
    {
      "epoch": 0.2526018532561986,
      "grad_norm": 0.42615535855293274,
      "learning_rate": 1.7012972300501128e-05,
      "loss": 0.2112,
      "step": 13990
    },
    {
      "epoch": 0.2526921326890993,
      "grad_norm": 0.32613807916641235,
      "learning_rate": 1.7010950148950002e-05,
      "loss": 0.2911,
      "step": 13995
    },
    {
      "epoch": 0.25278241212200003,
      "grad_norm": 0.3294132351875305,
      "learning_rate": 1.700892743341891e-05,
      "loss": 0.3252,
      "step": 14000
    },
    {
      "epoch": 0.2528726915549007,
      "grad_norm": 0.5091342329978943,
      "learning_rate": 1.7006904154070562e-05,
      "loss": 0.1856,
      "step": 14005
    },
    {
      "epoch": 0.25296297098780146,
      "grad_norm": 0.4071289300918579,
      "learning_rate": 1.7004880311067723e-05,
      "loss": 0.2937,
      "step": 14010
    },
    {
      "epoch": 0.25305325042070215,
      "grad_norm": 0.3675556480884552,
      "learning_rate": 1.7002855904573193e-05,
      "loss": 0.3034,
      "step": 14015
    },
    {
      "epoch": 0.2531435298536029,
      "grad_norm": 0.6735295057296753,
      "learning_rate": 1.700083093474982e-05,
      "loss": 0.2503,
      "step": 14020
    },
    {
      "epoch": 0.2532338092865036,
      "grad_norm": 0.7142743468284607,
      "learning_rate": 1.69988054017605e-05,
      "loss": 0.242,
      "step": 14025
    },
    {
      "epoch": 0.2533240887194043,
      "grad_norm": 0.32563072443008423,
      "learning_rate": 1.699677930576817e-05,
      "loss": 0.168,
      "step": 14030
    },
    {
      "epoch": 0.253414368152305,
      "grad_norm": 0.38879600167274475,
      "learning_rate": 1.699475264693582e-05,
      "loss": 0.3649,
      "step": 14035
    },
    {
      "epoch": 0.25350464758520574,
      "grad_norm": 0.5603982210159302,
      "learning_rate": 1.6992725425426474e-05,
      "loss": 0.3281,
      "step": 14040
    },
    {
      "epoch": 0.25359492701810643,
      "grad_norm": 0.7267544269561768,
      "learning_rate": 1.6990697641403212e-05,
      "loss": 0.3527,
      "step": 14045
    },
    {
      "epoch": 0.2536852064510072,
      "grad_norm": 0.4394630491733551,
      "learning_rate": 1.698866929502915e-05,
      "loss": 0.4412,
      "step": 14050
    },
    {
      "epoch": 0.25377548588390786,
      "grad_norm": 0.2933726906776428,
      "learning_rate": 1.6986640386467456e-05,
      "loss": 0.3036,
      "step": 14055
    },
    {
      "epoch": 0.2538657653168086,
      "grad_norm": 0.8265781402587891,
      "learning_rate": 1.6984610915881338e-05,
      "loss": 0.2785,
      "step": 14060
    },
    {
      "epoch": 0.2539560447497093,
      "grad_norm": 0.33802053332328796,
      "learning_rate": 1.698258088343406e-05,
      "loss": 0.3516,
      "step": 14065
    },
    {
      "epoch": 0.25404632418261003,
      "grad_norm": 0.29324543476104736,
      "learning_rate": 1.6980550289288915e-05,
      "loss": 0.2806,
      "step": 14070
    },
    {
      "epoch": 0.2541366036155107,
      "grad_norm": 0.41427239775657654,
      "learning_rate": 1.697851913360925e-05,
      "loss": 0.2985,
      "step": 14075
    },
    {
      "epoch": 0.25422688304841146,
      "grad_norm": 0.4136227071285248,
      "learning_rate": 1.6976487416558464e-05,
      "loss": 0.2698,
      "step": 14080
    },
    {
      "epoch": 0.25431716248131214,
      "grad_norm": 0.2818601727485657,
      "learning_rate": 1.6974455138299987e-05,
      "loss": 0.2504,
      "step": 14085
    },
    {
      "epoch": 0.2544074419142129,
      "grad_norm": 0.3512054979801178,
      "learning_rate": 1.6972422298997305e-05,
      "loss": 0.1858,
      "step": 14090
    },
    {
      "epoch": 0.25449772134711357,
      "grad_norm": 0.3316943943500519,
      "learning_rate": 1.6970388898813945e-05,
      "loss": 0.3162,
      "step": 14095
    },
    {
      "epoch": 0.2545880007800143,
      "grad_norm": 0.3768051266670227,
      "learning_rate": 1.6968354937913476e-05,
      "loss": 0.2573,
      "step": 14100
    },
    {
      "epoch": 0.254678280212915,
      "grad_norm": 0.40512460470199585,
      "learning_rate": 1.6966320416459515e-05,
      "loss": 0.2814,
      "step": 14105
    },
    {
      "epoch": 0.25476855964581574,
      "grad_norm": 0.3147158920764923,
      "learning_rate": 1.696428533461573e-05,
      "loss": 0.3551,
      "step": 14110
    },
    {
      "epoch": 0.2548588390787164,
      "grad_norm": 0.3642359972000122,
      "learning_rate": 1.696224969254582e-05,
      "loss": 0.2817,
      "step": 14115
    },
    {
      "epoch": 0.25494911851161717,
      "grad_norm": 0.5084564089775085,
      "learning_rate": 1.6960213490413546e-05,
      "loss": 0.2969,
      "step": 14120
    },
    {
      "epoch": 0.25503939794451785,
      "grad_norm": 0.4527064859867096,
      "learning_rate": 1.6958176728382704e-05,
      "loss": 0.3291,
      "step": 14125
    },
    {
      "epoch": 0.2551296773774186,
      "grad_norm": 0.5112528204917908,
      "learning_rate": 1.695613940661713e-05,
      "loss": 0.308,
      "step": 14130
    },
    {
      "epoch": 0.2552199568103193,
      "grad_norm": 0.4065954387187958,
      "learning_rate": 1.6954101525280723e-05,
      "loss": 0.3138,
      "step": 14135
    },
    {
      "epoch": 0.25531023624322,
      "grad_norm": 0.33163678646087646,
      "learning_rate": 1.6952063084537407e-05,
      "loss": 0.2584,
      "step": 14140
    },
    {
      "epoch": 0.2554005156761207,
      "grad_norm": 0.5004115104675293,
      "learning_rate": 1.6950024084551162e-05,
      "loss": 0.33,
      "step": 14145
    },
    {
      "epoch": 0.25549079510902145,
      "grad_norm": 0.5348025560379028,
      "learning_rate": 1.694798452548601e-05,
      "loss": 0.2925,
      "step": 14150
    },
    {
      "epoch": 0.25558107454192214,
      "grad_norm": 0.4292595088481903,
      "learning_rate": 1.694594440750602e-05,
      "loss": 0.3471,
      "step": 14155
    },
    {
      "epoch": 0.2556713539748229,
      "grad_norm": 0.39084112644195557,
      "learning_rate": 1.6943903730775307e-05,
      "loss": 0.2318,
      "step": 14160
    },
    {
      "epoch": 0.25576163340772357,
      "grad_norm": 0.5658642053604126,
      "learning_rate": 1.694186249545802e-05,
      "loss": 0.4035,
      "step": 14165
    },
    {
      "epoch": 0.2558519128406243,
      "grad_norm": 0.3004071116447449,
      "learning_rate": 1.693982070171837e-05,
      "loss": 0.3158,
      "step": 14170
    },
    {
      "epoch": 0.255942192273525,
      "grad_norm": 0.40406376123428345,
      "learning_rate": 1.6937778349720604e-05,
      "loss": 0.2249,
      "step": 14175
    },
    {
      "epoch": 0.25603247170642573,
      "grad_norm": 0.47171875834465027,
      "learning_rate": 1.6935735439629012e-05,
      "loss": 0.3313,
      "step": 14180
    },
    {
      "epoch": 0.2561227511393264,
      "grad_norm": 0.6781940460205078,
      "learning_rate": 1.693369197160793e-05,
      "loss": 0.2692,
      "step": 14185
    },
    {
      "epoch": 0.25621303057222716,
      "grad_norm": 0.9075122475624084,
      "learning_rate": 1.6931647945821743e-05,
      "loss": 0.3161,
      "step": 14190
    },
    {
      "epoch": 0.25630331000512785,
      "grad_norm": 0.34464186429977417,
      "learning_rate": 1.6929603362434878e-05,
      "loss": 0.2726,
      "step": 14195
    },
    {
      "epoch": 0.2563935894380286,
      "grad_norm": 0.7514514923095703,
      "learning_rate": 1.69275582216118e-05,
      "loss": 0.2552,
      "step": 14200
    },
    {
      "epoch": 0.2564838688709293,
      "grad_norm": 0.40465232729911804,
      "learning_rate": 1.692551252351704e-05,
      "loss": 0.3369,
      "step": 14205
    },
    {
      "epoch": 0.25657414830383,
      "grad_norm": 0.43503934144973755,
      "learning_rate": 1.6923466268315144e-05,
      "loss": 0.2949,
      "step": 14210
    },
    {
      "epoch": 0.2566644277367307,
      "grad_norm": 0.5278799533843994,
      "learning_rate": 1.6921419456170728e-05,
      "loss": 0.2964,
      "step": 14215
    },
    {
      "epoch": 0.25675470716963145,
      "grad_norm": 0.3669053316116333,
      "learning_rate": 1.691937208724844e-05,
      "loss": 0.2675,
      "step": 14220
    },
    {
      "epoch": 0.25684498660253213,
      "grad_norm": 0.4335341453552246,
      "learning_rate": 1.6917324161712973e-05,
      "loss": 0.3035,
      "step": 14225
    },
    {
      "epoch": 0.2569352660354329,
      "grad_norm": 0.45222485065460205,
      "learning_rate": 1.6915275679729075e-05,
      "loss": 0.2902,
      "step": 14230
    },
    {
      "epoch": 0.25702554546833356,
      "grad_norm": 0.46283671259880066,
      "learning_rate": 1.6913226641461528e-05,
      "loss": 0.3303,
      "step": 14235
    },
    {
      "epoch": 0.2571158249012343,
      "grad_norm": 0.6368562579154968,
      "learning_rate": 1.691117704707516e-05,
      "loss": 0.2445,
      "step": 14240
    },
    {
      "epoch": 0.257206104334135,
      "grad_norm": 0.44071435928344727,
      "learning_rate": 1.6909126896734847e-05,
      "loss": 0.2725,
      "step": 14245
    },
    {
      "epoch": 0.25729638376703573,
      "grad_norm": 0.40742605924606323,
      "learning_rate": 1.6907076190605507e-05,
      "loss": 0.3206,
      "step": 14250
    },
    {
      "epoch": 0.2573866631999364,
      "grad_norm": 0.5893548130989075,
      "learning_rate": 1.6905024928852112e-05,
      "loss": 0.2501,
      "step": 14255
    },
    {
      "epoch": 0.25747694263283716,
      "grad_norm": 0.3587595224380493,
      "learning_rate": 1.690297311163966e-05,
      "loss": 0.3273,
      "step": 14260
    },
    {
      "epoch": 0.25756722206573784,
      "grad_norm": 0.3252449035644531,
      "learning_rate": 1.690092073913322e-05,
      "loss": 0.2712,
      "step": 14265
    },
    {
      "epoch": 0.2576575014986386,
      "grad_norm": 0.5918602347373962,
      "learning_rate": 1.6898867811497874e-05,
      "loss": 0.1865,
      "step": 14270
    },
    {
      "epoch": 0.2577477809315393,
      "grad_norm": 1.2052277326583862,
      "learning_rate": 1.6896814328898772e-05,
      "loss": 0.2629,
      "step": 14275
    },
    {
      "epoch": 0.25783806036444,
      "grad_norm": 0.3375871181488037,
      "learning_rate": 1.6894760291501103e-05,
      "loss": 0.2149,
      "step": 14280
    },
    {
      "epoch": 0.25792833979734076,
      "grad_norm": 0.5106332302093506,
      "learning_rate": 1.6892705699470096e-05,
      "loss": 0.2918,
      "step": 14285
    },
    {
      "epoch": 0.25801861923024144,
      "grad_norm": 0.4305230975151062,
      "learning_rate": 1.6890650552971034e-05,
      "loss": 0.255,
      "step": 14290
    },
    {
      "epoch": 0.2581088986631422,
      "grad_norm": 0.5305595993995667,
      "learning_rate": 1.6888594852169236e-05,
      "loss": 0.4565,
      "step": 14295
    },
    {
      "epoch": 0.25819917809604287,
      "grad_norm": 0.4331403076648712,
      "learning_rate": 1.6886538597230062e-05,
      "loss": 0.286,
      "step": 14300
    },
    {
      "epoch": 0.2582894575289436,
      "grad_norm": 0.3630242347717285,
      "learning_rate": 1.6884481788318933e-05,
      "loss": 0.2657,
      "step": 14305
    },
    {
      "epoch": 0.2583797369618443,
      "grad_norm": 0.3561517298221588,
      "learning_rate": 1.6882424425601295e-05,
      "loss": 0.2423,
      "step": 14310
    },
    {
      "epoch": 0.25847001639474504,
      "grad_norm": 0.7556095719337463,
      "learning_rate": 1.6880366509242658e-05,
      "loss": 0.3607,
      "step": 14315
    },
    {
      "epoch": 0.2585602958276457,
      "grad_norm": 0.4049709737300873,
      "learning_rate": 1.6878308039408553e-05,
      "loss": 0.2307,
      "step": 14320
    },
    {
      "epoch": 0.25865057526054647,
      "grad_norm": 0.3996032774448395,
      "learning_rate": 1.687624901626458e-05,
      "loss": 0.3906,
      "step": 14325
    },
    {
      "epoch": 0.25874085469344715,
      "grad_norm": 0.47113409638404846,
      "learning_rate": 1.6874189439976372e-05,
      "loss": 0.3392,
      "step": 14330
    },
    {
      "epoch": 0.2588311341263479,
      "grad_norm": 0.6860496997833252,
      "learning_rate": 1.6872129310709598e-05,
      "loss": 0.3603,
      "step": 14335
    },
    {
      "epoch": 0.2589214135592486,
      "grad_norm": 0.3963136076927185,
      "learning_rate": 1.6870068628629986e-05,
      "loss": 0.4027,
      "step": 14340
    },
    {
      "epoch": 0.2590116929921493,
      "grad_norm": 0.5941218733787537,
      "learning_rate": 1.6868007393903308e-05,
      "loss": 0.3303,
      "step": 14345
    },
    {
      "epoch": 0.25910197242505,
      "grad_norm": 0.46553128957748413,
      "learning_rate": 1.6865945606695365e-05,
      "loss": 0.3103,
      "step": 14350
    },
    {
      "epoch": 0.25919225185795075,
      "grad_norm": 0.44235265254974365,
      "learning_rate": 1.6863883267172023e-05,
      "loss": 0.2916,
      "step": 14355
    },
    {
      "epoch": 0.25928253129085144,
      "grad_norm": 0.3624558448791504,
      "learning_rate": 1.6861820375499175e-05,
      "loss": 0.1915,
      "step": 14360
    },
    {
      "epoch": 0.2593728107237522,
      "grad_norm": 0.3028622269630432,
      "learning_rate": 1.685975693184277e-05,
      "loss": 0.2714,
      "step": 14365
    },
    {
      "epoch": 0.25946309015665286,
      "grad_norm": 0.5631296038627625,
      "learning_rate": 1.6857692936368795e-05,
      "loss": 0.2846,
      "step": 14370
    },
    {
      "epoch": 0.2595533695895536,
      "grad_norm": 0.4793113172054291,
      "learning_rate": 1.685562838924328e-05,
      "loss": 0.3616,
      "step": 14375
    },
    {
      "epoch": 0.2596436490224543,
      "grad_norm": 0.5509387850761414,
      "learning_rate": 1.6853563290632308e-05,
      "loss": 0.317,
      "step": 14380
    },
    {
      "epoch": 0.25973392845535503,
      "grad_norm": 0.29282665252685547,
      "learning_rate": 1.6851497640702e-05,
      "loss": 0.3121,
      "step": 14385
    },
    {
      "epoch": 0.2598242078882557,
      "grad_norm": 0.5319273471832275,
      "learning_rate": 1.6849431439618522e-05,
      "loss": 0.3166,
      "step": 14390
    },
    {
      "epoch": 0.25991448732115646,
      "grad_norm": 0.7315474152565002,
      "learning_rate": 1.6847364687548084e-05,
      "loss": 0.3886,
      "step": 14395
    },
    {
      "epoch": 0.26000476675405715,
      "grad_norm": 0.391114741563797,
      "learning_rate": 1.6845297384656944e-05,
      "loss": 0.2796,
      "step": 14400
    },
    {
      "epoch": 0.2600950461869579,
      "grad_norm": 0.403923898935318,
      "learning_rate": 1.6843229531111397e-05,
      "loss": 0.267,
      "step": 14405
    },
    {
      "epoch": 0.2601853256198586,
      "grad_norm": 0.5094287991523743,
      "learning_rate": 1.6841161127077792e-05,
      "loss": 0.261,
      "step": 14410
    },
    {
      "epoch": 0.2602756050527593,
      "grad_norm": 0.8942059874534607,
      "learning_rate": 1.6839092172722514e-05,
      "loss": 0.2961,
      "step": 14415
    },
    {
      "epoch": 0.26036588448566,
      "grad_norm": 0.40858590602874756,
      "learning_rate": 1.6837022668211995e-05,
      "loss": 0.3393,
      "step": 14420
    },
    {
      "epoch": 0.26045616391856075,
      "grad_norm": 0.4373575747013092,
      "learning_rate": 1.683495261371271e-05,
      "loss": 0.3228,
      "step": 14425
    },
    {
      "epoch": 0.26054644335146143,
      "grad_norm": 0.33809569478034973,
      "learning_rate": 1.6832882009391187e-05,
      "loss": 0.2772,
      "step": 14430
    },
    {
      "epoch": 0.2606367227843622,
      "grad_norm": 0.3225255310535431,
      "learning_rate": 1.683081085541398e-05,
      "loss": 0.2417,
      "step": 14435
    },
    {
      "epoch": 0.26072700221726286,
      "grad_norm": 0.523601770401001,
      "learning_rate": 1.682873915194771e-05,
      "loss": 0.2947,
      "step": 14440
    },
    {
      "epoch": 0.2608172816501636,
      "grad_norm": 0.44271430373191833,
      "learning_rate": 1.6826666899159022e-05,
      "loss": 0.2128,
      "step": 14445
    },
    {
      "epoch": 0.2609075610830643,
      "grad_norm": 0.4109349846839905,
      "learning_rate": 1.682459409721462e-05,
      "loss": 0.3225,
      "step": 14450
    },
    {
      "epoch": 0.26099784051596503,
      "grad_norm": 0.8876522183418274,
      "learning_rate": 1.682252074628124e-05,
      "loss": 0.2871,
      "step": 14455
    },
    {
      "epoch": 0.2610881199488657,
      "grad_norm": 1.0981746912002563,
      "learning_rate": 1.6820446846525668e-05,
      "loss": 0.3108,
      "step": 14460
    },
    {
      "epoch": 0.26117839938176646,
      "grad_norm": 0.3024000823497772,
      "learning_rate": 1.6818372398114736e-05,
      "loss": 0.3046,
      "step": 14465
    },
    {
      "epoch": 0.26126867881466714,
      "grad_norm": 0.9253686666488647,
      "learning_rate": 1.6816297401215328e-05,
      "loss": 0.2024,
      "step": 14470
    },
    {
      "epoch": 0.2613589582475679,
      "grad_norm": 0.42614516615867615,
      "learning_rate": 1.6814221855994342e-05,
      "loss": 0.3555,
      "step": 14475
    },
    {
      "epoch": 0.26144923768046857,
      "grad_norm": 0.38271841406822205,
      "learning_rate": 1.681214576261876e-05,
      "loss": 0.2707,
      "step": 14480
    },
    {
      "epoch": 0.2615395171133693,
      "grad_norm": 0.3673293888568878,
      "learning_rate": 1.6810069121255577e-05,
      "loss": 0.2125,
      "step": 14485
    },
    {
      "epoch": 0.26162979654627,
      "grad_norm": 0.398209810256958,
      "learning_rate": 1.680799193207185e-05,
      "loss": 0.2979,
      "step": 14490
    },
    {
      "epoch": 0.26172007597917074,
      "grad_norm": 0.5076440572738647,
      "learning_rate": 1.6805914195234666e-05,
      "loss": 0.2757,
      "step": 14495
    },
    {
      "epoch": 0.2618103554120714,
      "grad_norm": 0.9557774066925049,
      "learning_rate": 1.680383591091117e-05,
      "loss": 0.2924,
      "step": 14500
    },
    {
      "epoch": 0.26190063484497217,
      "grad_norm": 0.6554763913154602,
      "learning_rate": 1.6801757079268546e-05,
      "loss": 0.2519,
      "step": 14505
    },
    {
      "epoch": 0.26199091427787286,
      "grad_norm": 0.6285964250564575,
      "learning_rate": 1.679967770047402e-05,
      "loss": 0.337,
      "step": 14510
    },
    {
      "epoch": 0.2620811937107736,
      "grad_norm": 0.40318724513053894,
      "learning_rate": 1.679759777469486e-05,
      "loss": 0.2813,
      "step": 14515
    },
    {
      "epoch": 0.2621714731436743,
      "grad_norm": 0.503894031047821,
      "learning_rate": 1.6795517302098382e-05,
      "loss": 0.3493,
      "step": 14520
    },
    {
      "epoch": 0.262261752576575,
      "grad_norm": 0.37055426836013794,
      "learning_rate": 1.6793436282851944e-05,
      "loss": 0.2142,
      "step": 14525
    },
    {
      "epoch": 0.2623520320094757,
      "grad_norm": 0.47502291202545166,
      "learning_rate": 1.679135471712295e-05,
      "loss": 0.2984,
      "step": 14530
    },
    {
      "epoch": 0.26244231144237645,
      "grad_norm": 0.4711866080760956,
      "learning_rate": 1.678927260507885e-05,
      "loss": 0.2904,
      "step": 14535
    },
    {
      "epoch": 0.26253259087527714,
      "grad_norm": 0.33698809146881104,
      "learning_rate": 1.6787189946887128e-05,
      "loss": 0.3327,
      "step": 14540
    },
    {
      "epoch": 0.2626228703081779,
      "grad_norm": 0.759178638458252,
      "learning_rate": 1.6785106742715323e-05,
      "loss": 0.3373,
      "step": 14545
    },
    {
      "epoch": 0.26271314974107857,
      "grad_norm": 0.3459857702255249,
      "learning_rate": 1.6783022992731015e-05,
      "loss": 0.3514,
      "step": 14550
    },
    {
      "epoch": 0.2628034291739793,
      "grad_norm": 0.4096636176109314,
      "learning_rate": 1.678093869710182e-05,
      "loss": 0.2506,
      "step": 14555
    },
    {
      "epoch": 0.26289370860688,
      "grad_norm": 0.49789562821388245,
      "learning_rate": 1.6778853855995417e-05,
      "loss": 0.3218,
      "step": 14560
    },
    {
      "epoch": 0.26298398803978074,
      "grad_norm": 0.6544811725616455,
      "learning_rate": 1.6776768469579502e-05,
      "loss": 0.3461,
      "step": 14565
    },
    {
      "epoch": 0.2630742674726814,
      "grad_norm": 0.61392742395401,
      "learning_rate": 1.6774682538021834e-05,
      "loss": 0.3235,
      "step": 14570
    },
    {
      "epoch": 0.26316454690558216,
      "grad_norm": 0.24093428254127502,
      "learning_rate": 1.677259606149022e-05,
      "loss": 0.2255,
      "step": 14575
    },
    {
      "epoch": 0.26325482633848285,
      "grad_norm": 0.36252260208129883,
      "learning_rate": 1.677050904015249e-05,
      "loss": 0.2101,
      "step": 14580
    },
    {
      "epoch": 0.2633451057713836,
      "grad_norm": 0.6923762559890747,
      "learning_rate": 1.6768421474176533e-05,
      "loss": 0.2716,
      "step": 14585
    },
    {
      "epoch": 0.2634353852042843,
      "grad_norm": 0.4598599970340729,
      "learning_rate": 1.6766333363730282e-05,
      "loss": 0.2299,
      "step": 14590
    },
    {
      "epoch": 0.263525664637185,
      "grad_norm": 0.24030975997447968,
      "learning_rate": 1.6764244708981706e-05,
      "loss": 0.2504,
      "step": 14595
    },
    {
      "epoch": 0.2636159440700857,
      "grad_norm": 0.6802058815956116,
      "learning_rate": 1.6762155510098827e-05,
      "loss": 0.2059,
      "step": 14600
    },
    {
      "epoch": 0.26370622350298645,
      "grad_norm": 0.4267932176589966,
      "learning_rate": 1.6760065767249698e-05,
      "loss": 0.2668,
      "step": 14605
    },
    {
      "epoch": 0.26379650293588713,
      "grad_norm": 0.45004796981811523,
      "learning_rate": 1.6757975480602435e-05,
      "loss": 0.273,
      "step": 14610
    },
    {
      "epoch": 0.2638867823687879,
      "grad_norm": 0.44872379302978516,
      "learning_rate": 1.6755884650325174e-05,
      "loss": 0.3135,
      "step": 14615
    },
    {
      "epoch": 0.26397706180168856,
      "grad_norm": 0.4489907920360565,
      "learning_rate": 1.6753793276586113e-05,
      "loss": 0.2998,
      "step": 14620
    },
    {
      "epoch": 0.2640673412345893,
      "grad_norm": 0.7032406330108643,
      "learning_rate": 1.675170135955349e-05,
      "loss": 0.2633,
      "step": 14625
    },
    {
      "epoch": 0.26415762066749,
      "grad_norm": 0.3216649889945984,
      "learning_rate": 1.674960889939558e-05,
      "loss": 0.2874,
      "step": 14630
    },
    {
      "epoch": 0.26424790010039073,
      "grad_norm": 0.45787546038627625,
      "learning_rate": 1.674751589628071e-05,
      "loss": 0.2568,
      "step": 14635
    },
    {
      "epoch": 0.2643381795332914,
      "grad_norm": 0.3306974470615387,
      "learning_rate": 1.6745422350377246e-05,
      "loss": 0.192,
      "step": 14640
    },
    {
      "epoch": 0.26442845896619216,
      "grad_norm": 0.559463620185852,
      "learning_rate": 1.67433282618536e-05,
      "loss": 0.3781,
      "step": 14645
    },
    {
      "epoch": 0.26451873839909285,
      "grad_norm": 0.35545146465301514,
      "learning_rate": 1.6741233630878218e-05,
      "loss": 0.2881,
      "step": 14650
    },
    {
      "epoch": 0.2646090178319936,
      "grad_norm": 0.47090232372283936,
      "learning_rate": 1.6739138457619607e-05,
      "loss": 0.2371,
      "step": 14655
    },
    {
      "epoch": 0.2646992972648943,
      "grad_norm": 0.5291762351989746,
      "learning_rate": 1.673704274224631e-05,
      "loss": 0.3067,
      "step": 14660
    },
    {
      "epoch": 0.264789576697795,
      "grad_norm": 0.522714376449585,
      "learning_rate": 1.67349464849269e-05,
      "loss": 0.3224,
      "step": 14665
    },
    {
      "epoch": 0.26487985613069576,
      "grad_norm": 0.36193498969078064,
      "learning_rate": 1.673284968583002e-05,
      "loss": 0.2753,
      "step": 14670
    },
    {
      "epoch": 0.26497013556359644,
      "grad_norm": 0.4829557240009308,
      "learning_rate": 1.6730752345124328e-05,
      "loss": 0.2871,
      "step": 14675
    },
    {
      "epoch": 0.2650604149964972,
      "grad_norm": 0.4847275912761688,
      "learning_rate": 1.672865446297855e-05,
      "loss": 0.3003,
      "step": 14680
    },
    {
      "epoch": 0.26515069442939787,
      "grad_norm": 0.5189873576164246,
      "learning_rate": 1.6726556039561446e-05,
      "loss": 0.2667,
      "step": 14685
    },
    {
      "epoch": 0.2652409738622986,
      "grad_norm": 0.32874205708503723,
      "learning_rate": 1.6724457075041815e-05,
      "loss": 0.2473,
      "step": 14690
    },
    {
      "epoch": 0.2653312532951993,
      "grad_norm": 0.3475160300731659,
      "learning_rate": 1.6722357569588498e-05,
      "loss": 0.3136,
      "step": 14695
    },
    {
      "epoch": 0.26542153272810004,
      "grad_norm": 0.40261968970298767,
      "learning_rate": 1.6720257523370395e-05,
      "loss": 0.3522,
      "step": 14700
    },
    {
      "epoch": 0.2655118121610007,
      "grad_norm": 0.5326187610626221,
      "learning_rate": 1.6718156936556434e-05,
      "loss": 0.2753,
      "step": 14705
    },
    {
      "epoch": 0.26560209159390147,
      "grad_norm": 0.3311368227005005,
      "learning_rate": 1.6716055809315594e-05,
      "loss": 0.3193,
      "step": 14710
    },
    {
      "epoch": 0.26569237102680215,
      "grad_norm": 0.2994149327278137,
      "learning_rate": 1.6713954141816894e-05,
      "loss": 0.308,
      "step": 14715
    },
    {
      "epoch": 0.2657826504597029,
      "grad_norm": 0.581963300704956,
      "learning_rate": 1.67118519342294e-05,
      "loss": 0.2923,
      "step": 14720
    },
    {
      "epoch": 0.2658729298926036,
      "grad_norm": 0.7782924175262451,
      "learning_rate": 1.6709749186722214e-05,
      "loss": 0.2698,
      "step": 14725
    },
    {
      "epoch": 0.2659632093255043,
      "grad_norm": 0.3711308240890503,
      "learning_rate": 1.6707645899464492e-05,
      "loss": 0.2136,
      "step": 14730
    },
    {
      "epoch": 0.266053488758405,
      "grad_norm": 0.2079039067029953,
      "learning_rate": 1.6705542072625428e-05,
      "loss": 0.2666,
      "step": 14735
    },
    {
      "epoch": 0.26614376819130575,
      "grad_norm": 0.6214173436164856,
      "learning_rate": 1.670343770637426e-05,
      "loss": 0.2074,
      "step": 14740
    },
    {
      "epoch": 0.26623404762420644,
      "grad_norm": 0.4272468388080597,
      "learning_rate": 1.6701332800880266e-05,
      "loss": 0.2835,
      "step": 14745
    },
    {
      "epoch": 0.2663243270571072,
      "grad_norm": 0.23270417749881744,
      "learning_rate": 1.669922735631277e-05,
      "loss": 0.2348,
      "step": 14750
    },
    {
      "epoch": 0.26641460649000787,
      "grad_norm": 0.35107874870300293,
      "learning_rate": 1.669712137284114e-05,
      "loss": 0.3177,
      "step": 14755
    },
    {
      "epoch": 0.2665048859229086,
      "grad_norm": 0.6691100001335144,
      "learning_rate": 1.6695014850634792e-05,
      "loss": 0.3361,
      "step": 14760
    },
    {
      "epoch": 0.2665951653558093,
      "grad_norm": 0.39851701259613037,
      "learning_rate": 1.6692907789863173e-05,
      "loss": 0.3069,
      "step": 14765
    },
    {
      "epoch": 0.26668544478871004,
      "grad_norm": 0.5575686097145081,
      "learning_rate": 1.6690800190695786e-05,
      "loss": 0.2871,
      "step": 14770
    },
    {
      "epoch": 0.2667757242216107,
      "grad_norm": 0.34226763248443604,
      "learning_rate": 1.668869205330217e-05,
      "loss": 0.3051,
      "step": 14775
    },
    {
      "epoch": 0.26686600365451146,
      "grad_norm": 0.4824385941028595,
      "learning_rate": 1.668658337785191e-05,
      "loss": 0.2723,
      "step": 14780
    },
    {
      "epoch": 0.26695628308741215,
      "grad_norm": 0.49718043208122253,
      "learning_rate": 1.6684474164514637e-05,
      "loss": 0.2672,
      "step": 14785
    },
    {
      "epoch": 0.2670465625203129,
      "grad_norm": 0.37770745158195496,
      "learning_rate": 1.6682364413460014e-05,
      "loss": 0.2864,
      "step": 14790
    },
    {
      "epoch": 0.2671368419532136,
      "grad_norm": 0.7862424254417419,
      "learning_rate": 1.6680254124857763e-05,
      "loss": 0.3358,
      "step": 14795
    },
    {
      "epoch": 0.2672271213861143,
      "grad_norm": 0.5095865726470947,
      "learning_rate": 1.6678143298877635e-05,
      "loss": 0.3336,
      "step": 14800
    },
    {
      "epoch": 0.267317400819015,
      "grad_norm": 0.4720807373523712,
      "learning_rate": 1.6676031935689435e-05,
      "loss": 0.24,
      "step": 14805
    },
    {
      "epoch": 0.26740768025191575,
      "grad_norm": 0.4367506206035614,
      "learning_rate": 1.6673920035463004e-05,
      "loss": 0.2783,
      "step": 14810
    },
    {
      "epoch": 0.26749795968481643,
      "grad_norm": 0.36520060896873474,
      "learning_rate": 1.667180759836823e-05,
      "loss": 0.2842,
      "step": 14815
    },
    {
      "epoch": 0.2675882391177172,
      "grad_norm": 0.3673902451992035,
      "learning_rate": 1.6669694624575042e-05,
      "loss": 0.2936,
      "step": 14820
    },
    {
      "epoch": 0.26767851855061786,
      "grad_norm": 0.4801824390888214,
      "learning_rate": 1.6667581114253416e-05,
      "loss": 0.349,
      "step": 14825
    },
    {
      "epoch": 0.2677687979835186,
      "grad_norm": 0.5937851071357727,
      "learning_rate": 1.666546706757337e-05,
      "loss": 0.2555,
      "step": 14830
    },
    {
      "epoch": 0.2678590774164193,
      "grad_norm": 0.22917994856834412,
      "learning_rate": 1.666335248470496e-05,
      "loss": 0.2769,
      "step": 14835
    },
    {
      "epoch": 0.26794935684932003,
      "grad_norm": 0.46926772594451904,
      "learning_rate": 1.6661237365818287e-05,
      "loss": 0.2867,
      "step": 14840
    },
    {
      "epoch": 0.2680396362822207,
      "grad_norm": 0.475140243768692,
      "learning_rate": 1.6659121711083504e-05,
      "loss": 0.3331,
      "step": 14845
    },
    {
      "epoch": 0.26812991571512146,
      "grad_norm": 0.38355955481529236,
      "learning_rate": 1.6657005520670794e-05,
      "loss": 0.2744,
      "step": 14850
    },
    {
      "epoch": 0.26822019514802214,
      "grad_norm": 0.4811035990715027,
      "learning_rate": 1.665488879475039e-05,
      "loss": 0.2919,
      "step": 14855
    },
    {
      "epoch": 0.2683104745809229,
      "grad_norm": 0.6225593090057373,
      "learning_rate": 1.665277153349257e-05,
      "loss": 0.2979,
      "step": 14860
    },
    {
      "epoch": 0.2684007540138236,
      "grad_norm": 0.5459991097450256,
      "learning_rate": 1.6650653737067648e-05,
      "loss": 0.3335,
      "step": 14865
    },
    {
      "epoch": 0.2684910334467243,
      "grad_norm": 0.7860650420188904,
      "learning_rate": 1.664853540564599e-05,
      "loss": 0.2629,
      "step": 14870
    },
    {
      "epoch": 0.268581312879625,
      "grad_norm": 0.5207154154777527,
      "learning_rate": 1.6646416539397998e-05,
      "loss": 0.1395,
      "step": 14875
    },
    {
      "epoch": 0.26867159231252574,
      "grad_norm": 0.295149028301239,
      "learning_rate": 1.664429713849412e-05,
      "loss": 0.3182,
      "step": 14880
    },
    {
      "epoch": 0.26876187174542643,
      "grad_norm": 0.6800479292869568,
      "learning_rate": 1.664217720310484e-05,
      "loss": 0.2211,
      "step": 14885
    },
    {
      "epoch": 0.26885215117832717,
      "grad_norm": 0.4096933603286743,
      "learning_rate": 1.6640056733400706e-05,
      "loss": 0.3122,
      "step": 14890
    },
    {
      "epoch": 0.26894243061122786,
      "grad_norm": 0.38134050369262695,
      "learning_rate": 1.663793572955228e-05,
      "loss": 0.359,
      "step": 14895
    },
    {
      "epoch": 0.2690327100441286,
      "grad_norm": 0.7740076184272766,
      "learning_rate": 1.663581419173019e-05,
      "loss": 0.2086,
      "step": 14900
    },
    {
      "epoch": 0.2691229894770293,
      "grad_norm": 0.27558088302612305,
      "learning_rate": 1.6633692120105092e-05,
      "loss": 0.2562,
      "step": 14905
    },
    {
      "epoch": 0.26921326890993,
      "grad_norm": 0.8536859154701233,
      "learning_rate": 1.6631569514847696e-05,
      "loss": 0.2939,
      "step": 14910
    },
    {
      "epoch": 0.2693035483428307,
      "grad_norm": 0.4222334325313568,
      "learning_rate": 1.662944637612875e-05,
      "loss": 0.4318,
      "step": 14915
    },
    {
      "epoch": 0.26939382777573145,
      "grad_norm": 0.34165582060813904,
      "learning_rate": 1.662732270411904e-05,
      "loss": 0.2882,
      "step": 14920
    },
    {
      "epoch": 0.26948410720863214,
      "grad_norm": 0.43178972601890564,
      "learning_rate": 1.662519849898941e-05,
      "loss": 0.1692,
      "step": 14925
    },
    {
      "epoch": 0.2695743866415329,
      "grad_norm": 0.5931707620620728,
      "learning_rate": 1.6623073760910726e-05,
      "loss": 0.3197,
      "step": 14930
    },
    {
      "epoch": 0.26966466607443357,
      "grad_norm": 0.4385814070701599,
      "learning_rate": 1.6620948490053914e-05,
      "loss": 0.2658,
      "step": 14935
    },
    {
      "epoch": 0.2697549455073343,
      "grad_norm": 0.2507951259613037,
      "learning_rate": 1.6618822686589936e-05,
      "loss": 0.2285,
      "step": 14940
    },
    {
      "epoch": 0.269845224940235,
      "grad_norm": 0.38211968541145325,
      "learning_rate": 1.6616696350689792e-05,
      "loss": 0.3515,
      "step": 14945
    },
    {
      "epoch": 0.26993550437313574,
      "grad_norm": 0.4873267710208893,
      "learning_rate": 1.661456948252454e-05,
      "loss": 0.3296,
      "step": 14950
    },
    {
      "epoch": 0.2700257838060364,
      "grad_norm": 0.7359420657157898,
      "learning_rate": 1.6612442082265265e-05,
      "loss": 0.3376,
      "step": 14955
    },
    {
      "epoch": 0.27011606323893717,
      "grad_norm": 0.5286779403686523,
      "learning_rate": 1.6610314150083096e-05,
      "loss": 0.2761,
      "step": 14960
    },
    {
      "epoch": 0.27020634267183785,
      "grad_norm": 0.3935490846633911,
      "learning_rate": 1.660818568614922e-05,
      "loss": 0.3092,
      "step": 14965
    },
    {
      "epoch": 0.2702966221047386,
      "grad_norm": 0.6011040210723877,
      "learning_rate": 1.6606056690634852e-05,
      "loss": 0.3448,
      "step": 14970
    },
    {
      "epoch": 0.2703869015376393,
      "grad_norm": 0.6015837788581848,
      "learning_rate": 1.6603927163711254e-05,
      "loss": 0.2432,
      "step": 14975
    },
    {
      "epoch": 0.27047718097054,
      "grad_norm": 0.5535976886749268,
      "learning_rate": 1.6601797105549727e-05,
      "loss": 0.2987,
      "step": 14980
    },
    {
      "epoch": 0.2705674604034407,
      "grad_norm": 0.5601411461830139,
      "learning_rate": 1.6599666516321625e-05,
      "loss": 0.3679,
      "step": 14985
    },
    {
      "epoch": 0.27065773983634145,
      "grad_norm": 0.5071304440498352,
      "learning_rate": 1.6597535396198337e-05,
      "loss": 0.2826,
      "step": 14990
    },
    {
      "epoch": 0.27074801926924214,
      "grad_norm": 0.3708009123802185,
      "learning_rate": 1.6595403745351296e-05,
      "loss": 0.2435,
      "step": 14995
    },
    {
      "epoch": 0.2708382987021429,
      "grad_norm": 0.47595512866973877,
      "learning_rate": 1.6593271563951975e-05,
      "loss": 0.3054,
      "step": 15000
    },
    {
      "epoch": 0.27092857813504356,
      "grad_norm": 0.318886399269104,
      "learning_rate": 1.659113885217189e-05,
      "loss": 0.343,
      "step": 15005
    },
    {
      "epoch": 0.2710188575679443,
      "grad_norm": 2.0091025829315186,
      "learning_rate": 1.6589005610182613e-05,
      "loss": 0.3664,
      "step": 15010
    },
    {
      "epoch": 0.271109137000845,
      "grad_norm": 0.5164380073547363,
      "learning_rate": 1.658687183815574e-05,
      "loss": 0.3674,
      "step": 15015
    },
    {
      "epoch": 0.27119941643374573,
      "grad_norm": 0.41829022765159607,
      "learning_rate": 1.6584737536262917e-05,
      "loss": 0.3181,
      "step": 15020
    },
    {
      "epoch": 0.2712896958666464,
      "grad_norm": 0.6270329356193542,
      "learning_rate": 1.6582602704675838e-05,
      "loss": 0.3107,
      "step": 15025
    },
    {
      "epoch": 0.27137997529954716,
      "grad_norm": 0.5446771383285522,
      "learning_rate": 1.6580467343566228e-05,
      "loss": 0.2632,
      "step": 15030
    },
    {
      "epoch": 0.27147025473244785,
      "grad_norm": 0.4855729043483734,
      "learning_rate": 1.6578331453105867e-05,
      "loss": 0.258,
      "step": 15035
    },
    {
      "epoch": 0.2715605341653486,
      "grad_norm": 0.4179638922214508,
      "learning_rate": 1.657619503346657e-05,
      "loss": 0.2696,
      "step": 15040
    },
    {
      "epoch": 0.2716508135982493,
      "grad_norm": 0.6478289365768433,
      "learning_rate": 1.6574058084820195e-05,
      "loss": 0.2576,
      "step": 15045
    },
    {
      "epoch": 0.27174109303115,
      "grad_norm": 0.5988754034042358,
      "learning_rate": 1.6571920607338647e-05,
      "loss": 0.3109,
      "step": 15050
    },
    {
      "epoch": 0.2718313724640507,
      "grad_norm": 0.3570278286933899,
      "learning_rate": 1.6569782601193865e-05,
      "loss": 0.2561,
      "step": 15055
    },
    {
      "epoch": 0.27192165189695144,
      "grad_norm": 0.36204129457473755,
      "learning_rate": 1.656764406655784e-05,
      "loss": 0.3706,
      "step": 15060
    },
    {
      "epoch": 0.2720119313298522,
      "grad_norm": 0.3621353209018707,
      "learning_rate": 1.6565505003602607e-05,
      "loss": 0.3523,
      "step": 15065
    },
    {
      "epoch": 0.27210221076275287,
      "grad_norm": 0.4771013557910919,
      "learning_rate": 1.6563365412500226e-05,
      "loss": 0.2619,
      "step": 15070
    },
    {
      "epoch": 0.2721924901956536,
      "grad_norm": 0.6217138171195984,
      "learning_rate": 1.656122529342282e-05,
      "loss": 0.257,
      "step": 15075
    },
    {
      "epoch": 0.2722827696285543,
      "grad_norm": 0.48531246185302734,
      "learning_rate": 1.6559084646542544e-05,
      "loss": 0.2432,
      "step": 15080
    },
    {
      "epoch": 0.27237304906145504,
      "grad_norm": 0.48256370425224304,
      "learning_rate": 1.65569434720316e-05,
      "loss": 0.3274,
      "step": 15085
    },
    {
      "epoch": 0.2724633284943557,
      "grad_norm": 0.3124198317527771,
      "learning_rate": 1.6554801770062222e-05,
      "loss": 0.328,
      "step": 15090
    },
    {
      "epoch": 0.27255360792725647,
      "grad_norm": 0.46326014399528503,
      "learning_rate": 1.65526595408067e-05,
      "loss": 0.3229,
      "step": 15095
    },
    {
      "epoch": 0.27264388736015716,
      "grad_norm": 0.39263132214546204,
      "learning_rate": 1.6550516784437366e-05,
      "loss": 0.3548,
      "step": 15100
    },
    {
      "epoch": 0.2727341667930579,
      "grad_norm": 0.4550118148326874,
      "learning_rate": 1.6548373501126577e-05,
      "loss": 0.2756,
      "step": 15105
    },
    {
      "epoch": 0.2728244462259586,
      "grad_norm": 0.4505438804626465,
      "learning_rate": 1.6546229691046755e-05,
      "loss": 0.3097,
      "step": 15110
    },
    {
      "epoch": 0.2729147256588593,
      "grad_norm": 0.3754425346851349,
      "learning_rate": 1.654408535437035e-05,
      "loss": 0.2004,
      "step": 15115
    },
    {
      "epoch": 0.27300500509176,
      "grad_norm": 0.6223003268241882,
      "learning_rate": 1.6541940491269857e-05,
      "loss": 0.2288,
      "step": 15120
    },
    {
      "epoch": 0.27309528452466075,
      "grad_norm": 0.44022637605667114,
      "learning_rate": 1.6539795101917816e-05,
      "loss": 0.2658,
      "step": 15125
    },
    {
      "epoch": 0.27318556395756144,
      "grad_norm": 0.4375525414943695,
      "learning_rate": 1.653764918648681e-05,
      "loss": 0.1978,
      "step": 15130
    },
    {
      "epoch": 0.2732758433904622,
      "grad_norm": 0.9861794710159302,
      "learning_rate": 1.6535502745149456e-05,
      "loss": 0.3321,
      "step": 15135
    },
    {
      "epoch": 0.27336612282336287,
      "grad_norm": 1.1904473304748535,
      "learning_rate": 1.6533355778078428e-05,
      "loss": 0.2882,
      "step": 15140
    },
    {
      "epoch": 0.2734564022562636,
      "grad_norm": 0.3916078507900238,
      "learning_rate": 1.6531208285446428e-05,
      "loss": 0.2744,
      "step": 15145
    },
    {
      "epoch": 0.2735466816891643,
      "grad_norm": 0.5370838642120361,
      "learning_rate": 1.652906026742621e-05,
      "loss": 0.2788,
      "step": 15150
    },
    {
      "epoch": 0.27363696112206504,
      "grad_norm": 0.5554995536804199,
      "learning_rate": 1.6526911724190564e-05,
      "loss": 0.3579,
      "step": 15155
    },
    {
      "epoch": 0.2737272405549657,
      "grad_norm": 0.43957996368408203,
      "learning_rate": 1.6524762655912323e-05,
      "loss": 0.3007,
      "step": 15160
    },
    {
      "epoch": 0.27381751998786646,
      "grad_norm": 0.5079452991485596,
      "learning_rate": 1.6522613062764364e-05,
      "loss": 0.2514,
      "step": 15165
    },
    {
      "epoch": 0.27390779942076715,
      "grad_norm": 0.39836010336875916,
      "learning_rate": 1.652046294491961e-05,
      "loss": 0.3092,
      "step": 15170
    },
    {
      "epoch": 0.2739980788536679,
      "grad_norm": 0.5519921183586121,
      "learning_rate": 1.6518312302551024e-05,
      "loss": 0.2886,
      "step": 15175
    },
    {
      "epoch": 0.2740883582865686,
      "grad_norm": 0.3608863949775696,
      "learning_rate": 1.6516161135831602e-05,
      "loss": 0.3595,
      "step": 15180
    },
    {
      "epoch": 0.2741786377194693,
      "grad_norm": 0.33774057030677795,
      "learning_rate": 1.6514009444934392e-05,
      "loss": 0.3462,
      "step": 15185
    },
    {
      "epoch": 0.27426891715237,
      "grad_norm": 0.6961424946784973,
      "learning_rate": 1.6511857230032487e-05,
      "loss": 0.3188,
      "step": 15190
    },
    {
      "epoch": 0.27435919658527075,
      "grad_norm": 0.5834682583808899,
      "learning_rate": 1.6509704491299014e-05,
      "loss": 0.2196,
      "step": 15195
    },
    {
      "epoch": 0.27444947601817143,
      "grad_norm": 0.26620805263519287,
      "learning_rate": 1.6507551228907144e-05,
      "loss": 0.2567,
      "step": 15200
    },
    {
      "epoch": 0.2745397554510722,
      "grad_norm": 0.6540822386741638,
      "learning_rate": 1.6505397443030095e-05,
      "loss": 0.3038,
      "step": 15205
    },
    {
      "epoch": 0.27463003488397286,
      "grad_norm": 0.3851265609264374,
      "learning_rate": 1.6503243133841115e-05,
      "loss": 0.2346,
      "step": 15210
    },
    {
      "epoch": 0.2747203143168736,
      "grad_norm": 0.5700597763061523,
      "learning_rate": 1.6501088301513508e-05,
      "loss": 0.2433,
      "step": 15215
    },
    {
      "epoch": 0.2748105937497743,
      "grad_norm": 0.5008401870727539,
      "learning_rate": 1.6498932946220617e-05,
      "loss": 0.2142,
      "step": 15220
    },
    {
      "epoch": 0.27490087318267503,
      "grad_norm": 0.5411930680274963,
      "learning_rate": 1.6496777068135826e-05,
      "loss": 0.334,
      "step": 15225
    },
    {
      "epoch": 0.2749911526155757,
      "grad_norm": 0.4610949456691742,
      "learning_rate": 1.6494620667432548e-05,
      "loss": 0.2834,
      "step": 15230
    },
    {
      "epoch": 0.27508143204847646,
      "grad_norm": 0.6483005881309509,
      "learning_rate": 1.6492463744284263e-05,
      "loss": 0.3432,
      "step": 15235
    },
    {
      "epoch": 0.27517171148137715,
      "grad_norm": 3.1367764472961426,
      "learning_rate": 1.6490306298864473e-05,
      "loss": 0.2535,
      "step": 15240
    },
    {
      "epoch": 0.2752619909142779,
      "grad_norm": 0.31624963879585266,
      "learning_rate": 1.6488148331346733e-05,
      "loss": 0.1753,
      "step": 15245
    },
    {
      "epoch": 0.2753522703471786,
      "grad_norm": 0.472771555185318,
      "learning_rate": 1.648598984190463e-05,
      "loss": 0.3184,
      "step": 15250
    },
    {
      "epoch": 0.2754425497800793,
      "grad_norm": 0.4292396903038025,
      "learning_rate": 1.6483830830711803e-05,
      "loss": 0.3006,
      "step": 15255
    },
    {
      "epoch": 0.27553282921298,
      "grad_norm": 0.32948029041290283,
      "learning_rate": 1.6481671297941927e-05,
      "loss": 0.2824,
      "step": 15260
    },
    {
      "epoch": 0.27562310864588074,
      "grad_norm": 0.4458978474140167,
      "learning_rate": 1.6479511243768724e-05,
      "loss": 0.3051,
      "step": 15265
    },
    {
      "epoch": 0.27571338807878143,
      "grad_norm": 0.3954915702342987,
      "learning_rate": 1.647735066836595e-05,
      "loss": 0.3075,
      "step": 15270
    },
    {
      "epoch": 0.27580366751168217,
      "grad_norm": 0.3933243453502655,
      "learning_rate": 1.6475189571907408e-05,
      "loss": 0.2309,
      "step": 15275
    },
    {
      "epoch": 0.27589394694458286,
      "grad_norm": 0.3683823049068451,
      "learning_rate": 1.6473027954566948e-05,
      "loss": 0.3026,
      "step": 15280
    },
    {
      "epoch": 0.2759842263774836,
      "grad_norm": 0.6095166802406311,
      "learning_rate": 1.647086581651845e-05,
      "loss": 0.289,
      "step": 15285
    },
    {
      "epoch": 0.2760745058103843,
      "grad_norm": 0.47206076979637146,
      "learning_rate": 1.6468703157935844e-05,
      "loss": 0.3066,
      "step": 15290
    },
    {
      "epoch": 0.276164785243285,
      "grad_norm": 0.44620397686958313,
      "learning_rate": 1.64665399789931e-05,
      "loss": 0.335,
      "step": 15295
    },
    {
      "epoch": 0.2762550646761857,
      "grad_norm": 0.49999502301216125,
      "learning_rate": 1.6464376279864235e-05,
      "loss": 0.2846,
      "step": 15300
    },
    {
      "epoch": 0.27634534410908645,
      "grad_norm": 0.486945778131485,
      "learning_rate": 1.6462212060723297e-05,
      "loss": 0.2658,
      "step": 15305
    },
    {
      "epoch": 0.27643562354198714,
      "grad_norm": 0.5022004246711731,
      "learning_rate": 1.6460047321744384e-05,
      "loss": 0.3887,
      "step": 15310
    },
    {
      "epoch": 0.2765259029748879,
      "grad_norm": 0.5504619479179382,
      "learning_rate": 1.6457882063101635e-05,
      "loss": 0.3519,
      "step": 15315
    },
    {
      "epoch": 0.27661618240778857,
      "grad_norm": 0.22563856840133667,
      "learning_rate": 1.6455716284969222e-05,
      "loss": 0.2256,
      "step": 15320
    },
    {
      "epoch": 0.2767064618406893,
      "grad_norm": 0.40650132298469543,
      "learning_rate": 1.6453549987521376e-05,
      "loss": 0.2261,
      "step": 15325
    },
    {
      "epoch": 0.27679674127359,
      "grad_norm": 0.5303844809532166,
      "learning_rate": 1.645138317093235e-05,
      "loss": 0.271,
      "step": 15330
    },
    {
      "epoch": 0.27688702070649074,
      "grad_norm": 0.4735105335712433,
      "learning_rate": 1.6449215835376462e-05,
      "loss": 0.2691,
      "step": 15335
    },
    {
      "epoch": 0.2769773001393914,
      "grad_norm": 0.4318624436855316,
      "learning_rate": 1.6447047981028042e-05,
      "loss": 0.2865,
      "step": 15340
    },
    {
      "epoch": 0.27706757957229217,
      "grad_norm": 0.5579233765602112,
      "learning_rate": 1.644487960806149e-05,
      "loss": 0.2117,
      "step": 15345
    },
    {
      "epoch": 0.27715785900519285,
      "grad_norm": 0.47623273730278015,
      "learning_rate": 1.6442710716651234e-05,
      "loss": 0.3592,
      "step": 15350
    },
    {
      "epoch": 0.2772481384380936,
      "grad_norm": 0.6098501086235046,
      "learning_rate": 1.6440541306971745e-05,
      "loss": 0.2889,
      "step": 15355
    },
    {
      "epoch": 0.2773384178709943,
      "grad_norm": 0.41777414083480835,
      "learning_rate": 1.6438371379197534e-05,
      "loss": 0.262,
      "step": 15360
    },
    {
      "epoch": 0.277428697303895,
      "grad_norm": 0.4153043031692505,
      "learning_rate": 1.6436200933503156e-05,
      "loss": 0.2855,
      "step": 15365
    },
    {
      "epoch": 0.2775189767367957,
      "grad_norm": 0.3103995621204376,
      "learning_rate": 1.6434029970063207e-05,
      "loss": 0.2276,
      "step": 15370
    },
    {
      "epoch": 0.27760925616969645,
      "grad_norm": 0.5243020057678223,
      "learning_rate": 1.6431858489052327e-05,
      "loss": 0.3601,
      "step": 15375
    },
    {
      "epoch": 0.27769953560259714,
      "grad_norm": 0.8446954488754272,
      "learning_rate": 1.6429686490645198e-05,
      "loss": 0.2727,
      "step": 15380
    },
    {
      "epoch": 0.2777898150354979,
      "grad_norm": 0.4470043182373047,
      "learning_rate": 1.642751397501654e-05,
      "loss": 0.2341,
      "step": 15385
    },
    {
      "epoch": 0.27788009446839856,
      "grad_norm": 0.3866370916366577,
      "learning_rate": 1.642534094234111e-05,
      "loss": 0.3108,
      "step": 15390
    },
    {
      "epoch": 0.2779703739012993,
      "grad_norm": 0.4725272059440613,
      "learning_rate": 1.6423167392793723e-05,
      "loss": 0.2605,
      "step": 15395
    },
    {
      "epoch": 0.2780606533342,
      "grad_norm": 0.8187476396560669,
      "learning_rate": 1.642099332654922e-05,
      "loss": 0.2855,
      "step": 15400
    },
    {
      "epoch": 0.27815093276710073,
      "grad_norm": 0.3882555663585663,
      "learning_rate": 1.641881874378249e-05,
      "loss": 0.3275,
      "step": 15405
    },
    {
      "epoch": 0.2782412122000014,
      "grad_norm": 0.4930070638656616,
      "learning_rate": 1.6416643644668454e-05,
      "loss": 0.318,
      "step": 15410
    },
    {
      "epoch": 0.27833149163290216,
      "grad_norm": 0.5323049426078796,
      "learning_rate": 1.6414468029382095e-05,
      "loss": 0.2475,
      "step": 15415
    },
    {
      "epoch": 0.27842177106580285,
      "grad_norm": 0.3736749291419983,
      "learning_rate": 1.6412291898098423e-05,
      "loss": 0.3491,
      "step": 15420
    },
    {
      "epoch": 0.2785120504987036,
      "grad_norm": 0.33540913462638855,
      "learning_rate": 1.641011525099249e-05,
      "loss": 0.2865,
      "step": 15425
    },
    {
      "epoch": 0.2786023299316043,
      "grad_norm": 0.42792782187461853,
      "learning_rate": 1.640793808823939e-05,
      "loss": 0.2633,
      "step": 15430
    },
    {
      "epoch": 0.278692609364505,
      "grad_norm": 0.602237343788147,
      "learning_rate": 1.640576041001426e-05,
      "loss": 0.3158,
      "step": 15435
    },
    {
      "epoch": 0.2787828887974057,
      "grad_norm": 0.43899089097976685,
      "learning_rate": 1.6403582216492274e-05,
      "loss": 0.4468,
      "step": 15440
    },
    {
      "epoch": 0.27887316823030645,
      "grad_norm": 0.5117487907409668,
      "learning_rate": 1.6401403507848663e-05,
      "loss": 0.2577,
      "step": 15445
    },
    {
      "epoch": 0.27896344766320713,
      "grad_norm": 0.524939775466919,
      "learning_rate": 1.6399224284258683e-05,
      "loss": 0.3625,
      "step": 15450
    },
    {
      "epoch": 0.2790537270961079,
      "grad_norm": 0.36036819219589233,
      "learning_rate": 1.6397044545897636e-05,
      "loss": 0.2324,
      "step": 15455
    },
    {
      "epoch": 0.2791440065290086,
      "grad_norm": 0.3167564868927002,
      "learning_rate": 1.6394864292940866e-05,
      "loss": 0.3252,
      "step": 15460
    },
    {
      "epoch": 0.2792342859619093,
      "grad_norm": 0.45076414942741394,
      "learning_rate": 1.6392683525563757e-05,
      "loss": 0.3069,
      "step": 15465
    },
    {
      "epoch": 0.27932456539481004,
      "grad_norm": 0.3625401556491852,
      "learning_rate": 1.639050224394174e-05,
      "loss": 0.2986,
      "step": 15470
    },
    {
      "epoch": 0.27941484482771073,
      "grad_norm": 0.4333385229110718,
      "learning_rate": 1.6388320448250278e-05,
      "loss": 0.294,
      "step": 15475
    },
    {
      "epoch": 0.27950512426061147,
      "grad_norm": 0.5874447822570801,
      "learning_rate": 1.6386138138664883e-05,
      "loss": 0.2447,
      "step": 15480
    },
    {
      "epoch": 0.27959540369351216,
      "grad_norm": 0.36848005652427673,
      "learning_rate": 1.638395531536111e-05,
      "loss": 0.1787,
      "step": 15485
    },
    {
      "epoch": 0.2796856831264129,
      "grad_norm": 0.5163210034370422,
      "learning_rate": 1.6381771978514548e-05,
      "loss": 0.278,
      "step": 15490
    },
    {
      "epoch": 0.2797759625593136,
      "grad_norm": 0.26837676763534546,
      "learning_rate": 1.6379588128300828e-05,
      "loss": 0.3203,
      "step": 15495
    },
    {
      "epoch": 0.2798662419922143,
      "grad_norm": 0.2180904746055603,
      "learning_rate": 1.637740376489563e-05,
      "loss": 0.261,
      "step": 15500
    },
    {
      "epoch": 0.279956521425115,
      "grad_norm": 0.40104246139526367,
      "learning_rate": 1.6375218888474666e-05,
      "loss": 0.3079,
      "step": 15505
    },
    {
      "epoch": 0.28004680085801575,
      "grad_norm": 0.4926336109638214,
      "learning_rate": 1.637303349921369e-05,
      "loss": 0.3486,
      "step": 15510
    },
    {
      "epoch": 0.28013708029091644,
      "grad_norm": 0.38141384720802307,
      "learning_rate": 1.637084759728851e-05,
      "loss": 0.2401,
      "step": 15515
    },
    {
      "epoch": 0.2802273597238172,
      "grad_norm": 0.31682920455932617,
      "learning_rate": 1.6368661182874964e-05,
      "loss": 0.2688,
      "step": 15520
    },
    {
      "epoch": 0.28031763915671787,
      "grad_norm": 0.6005726456642151,
      "learning_rate": 1.6366474256148925e-05,
      "loss": 0.2366,
      "step": 15525
    },
    {
      "epoch": 0.2804079185896186,
      "grad_norm": 0.33243298530578613,
      "learning_rate": 1.6364286817286323e-05,
      "loss": 0.2397,
      "step": 15530
    },
    {
      "epoch": 0.2804981980225193,
      "grad_norm": 0.38761410117149353,
      "learning_rate": 1.6362098866463122e-05,
      "loss": 0.2715,
      "step": 15535
    },
    {
      "epoch": 0.28058847745542004,
      "grad_norm": 0.5326482653617859,
      "learning_rate": 1.635991040385532e-05,
      "loss": 0.3462,
      "step": 15540
    },
    {
      "epoch": 0.2806787568883207,
      "grad_norm": 0.5653488636016846,
      "learning_rate": 1.635772142963897e-05,
      "loss": 0.3557,
      "step": 15545
    },
    {
      "epoch": 0.28076903632122147,
      "grad_norm": 0.4962272047996521,
      "learning_rate": 1.6355531943990153e-05,
      "loss": 0.266,
      "step": 15550
    },
    {
      "epoch": 0.28085931575412215,
      "grad_norm": 0.5806422233581543,
      "learning_rate": 1.6353341947085e-05,
      "loss": 0.2875,
      "step": 15555
    },
    {
      "epoch": 0.2809495951870229,
      "grad_norm": 0.429801344871521,
      "learning_rate": 1.6351151439099684e-05,
      "loss": 0.2699,
      "step": 15560
    },
    {
      "epoch": 0.2810398746199236,
      "grad_norm": 1.272374153137207,
      "learning_rate": 1.634896042021041e-05,
      "loss": 0.2556,
      "step": 15565
    },
    {
      "epoch": 0.2811301540528243,
      "grad_norm": 0.4203578233718872,
      "learning_rate": 1.6346768890593432e-05,
      "loss": 0.2648,
      "step": 15570
    },
    {
      "epoch": 0.281220433485725,
      "grad_norm": 0.6440940499305725,
      "learning_rate": 1.6344576850425044e-05,
      "loss": 0.2569,
      "step": 15575
    },
    {
      "epoch": 0.28131071291862575,
      "grad_norm": 0.40680187940597534,
      "learning_rate": 1.6342384299881574e-05,
      "loss": 0.352,
      "step": 15580
    },
    {
      "epoch": 0.28140099235152644,
      "grad_norm": 1.5566149950027466,
      "learning_rate": 1.6340191239139405e-05,
      "loss": 0.3613,
      "step": 15585
    },
    {
      "epoch": 0.2814912717844272,
      "grad_norm": 0.35331594944000244,
      "learning_rate": 1.633799766837495e-05,
      "loss": 0.2324,
      "step": 15590
    },
    {
      "epoch": 0.28158155121732786,
      "grad_norm": 0.38838991522789,
      "learning_rate": 1.6335803587764655e-05,
      "loss": 0.2894,
      "step": 15595
    },
    {
      "epoch": 0.2816718306502286,
      "grad_norm": 0.6177527904510498,
      "learning_rate": 1.6333608997485034e-05,
      "loss": 0.2121,
      "step": 15600
    },
    {
      "epoch": 0.2817621100831293,
      "grad_norm": 0.3072573244571686,
      "learning_rate": 1.633141389771262e-05,
      "loss": 0.1905,
      "step": 15605
    },
    {
      "epoch": 0.28185238951603003,
      "grad_norm": 0.5108826756477356,
      "learning_rate": 1.632921828862399e-05,
      "loss": 0.2624,
      "step": 15610
    },
    {
      "epoch": 0.2819426689489307,
      "grad_norm": 0.8060066103935242,
      "learning_rate": 1.6327022170395763e-05,
      "loss": 0.2973,
      "step": 15615
    },
    {
      "epoch": 0.28203294838183146,
      "grad_norm": 0.4345701336860657,
      "learning_rate": 1.632482554320461e-05,
      "loss": 0.2287,
      "step": 15620
    },
    {
      "epoch": 0.28212322781473215,
      "grad_norm": 0.47276219725608826,
      "learning_rate": 1.6322628407227226e-05,
      "loss": 0.2187,
      "step": 15625
    },
    {
      "epoch": 0.2822135072476329,
      "grad_norm": 0.6881159543991089,
      "learning_rate": 1.6320430762640357e-05,
      "loss": 0.3437,
      "step": 15630
    },
    {
      "epoch": 0.2823037866805336,
      "grad_norm": 0.47420379519462585,
      "learning_rate": 1.6318232609620783e-05,
      "loss": 0.262,
      "step": 15635
    },
    {
      "epoch": 0.2823940661134343,
      "grad_norm": 0.48898845911026,
      "learning_rate": 1.6316033948345336e-05,
      "loss": 0.4279,
      "step": 15640
    },
    {
      "epoch": 0.282484345546335,
      "grad_norm": 0.516494631767273,
      "learning_rate": 1.631383477899088e-05,
      "loss": 0.2792,
      "step": 15645
    },
    {
      "epoch": 0.28257462497923574,
      "grad_norm": 0.9257849454879761,
      "learning_rate": 1.6311635101734326e-05,
      "loss": 0.248,
      "step": 15650
    },
    {
      "epoch": 0.28266490441213643,
      "grad_norm": 0.41677549481391907,
      "learning_rate": 1.6309434916752612e-05,
      "loss": 0.291,
      "step": 15655
    },
    {
      "epoch": 0.2827551838450372,
      "grad_norm": 0.4060181975364685,
      "learning_rate": 1.6307234224222735e-05,
      "loss": 0.3158,
      "step": 15660
    },
    {
      "epoch": 0.28284546327793786,
      "grad_norm": 0.7808871269226074,
      "learning_rate": 1.6305033024321725e-05,
      "loss": 0.2654,
      "step": 15665
    },
    {
      "epoch": 0.2829357427108386,
      "grad_norm": 0.6203485131263733,
      "learning_rate": 1.630283131722665e-05,
      "loss": 0.2618,
      "step": 15670
    },
    {
      "epoch": 0.2830260221437393,
      "grad_norm": 0.42930033802986145,
      "learning_rate": 1.6300629103114617e-05,
      "loss": 0.3103,
      "step": 15675
    },
    {
      "epoch": 0.28311630157664003,
      "grad_norm": 0.4635891616344452,
      "learning_rate": 1.6298426382162788e-05,
      "loss": 0.3125,
      "step": 15680
    },
    {
      "epoch": 0.2832065810095407,
      "grad_norm": 0.6747693419456482,
      "learning_rate": 1.629622315454835e-05,
      "loss": 0.3134,
      "step": 15685
    },
    {
      "epoch": 0.28329686044244146,
      "grad_norm": 0.4666113257408142,
      "learning_rate": 1.6294019420448533e-05,
      "loss": 0.2255,
      "step": 15690
    },
    {
      "epoch": 0.28338713987534214,
      "grad_norm": 1.8684427738189697,
      "learning_rate": 1.6291815180040617e-05,
      "loss": 0.245,
      "step": 15695
    },
    {
      "epoch": 0.2834774193082429,
      "grad_norm": 0.4526993930339813,
      "learning_rate": 1.6289610433501916e-05,
      "loss": 0.2927,
      "step": 15700
    },
    {
      "epoch": 0.28356769874114357,
      "grad_norm": 0.44100838899612427,
      "learning_rate": 1.628740518100979e-05,
      "loss": 0.2808,
      "step": 15705
    },
    {
      "epoch": 0.2836579781740443,
      "grad_norm": 0.2732015550136566,
      "learning_rate": 1.6285199422741627e-05,
      "loss": 0.2947,
      "step": 15710
    },
    {
      "epoch": 0.283748257606945,
      "grad_norm": 0.48613548278808594,
      "learning_rate": 1.6282993158874867e-05,
      "loss": 0.3734,
      "step": 15715
    },
    {
      "epoch": 0.28383853703984574,
      "grad_norm": 0.7183789610862732,
      "learning_rate": 1.628078638958699e-05,
      "loss": 0.3986,
      "step": 15720
    },
    {
      "epoch": 0.2839288164727464,
      "grad_norm": 0.3593280613422394,
      "learning_rate": 1.6278579115055514e-05,
      "loss": 0.2467,
      "step": 15725
    },
    {
      "epoch": 0.28401909590564717,
      "grad_norm": 0.47348156571388245,
      "learning_rate": 1.6276371335458e-05,
      "loss": 0.2409,
      "step": 15730
    },
    {
      "epoch": 0.28410937533854785,
      "grad_norm": 0.7523562908172607,
      "learning_rate": 1.6274163050972044e-05,
      "loss": 0.3164,
      "step": 15735
    },
    {
      "epoch": 0.2841996547714486,
      "grad_norm": 0.41706204414367676,
      "learning_rate": 1.6271954261775285e-05,
      "loss": 0.2297,
      "step": 15740
    },
    {
      "epoch": 0.2842899342043493,
      "grad_norm": 0.4382179081439972,
      "learning_rate": 1.6269744968045412e-05,
      "loss": 0.2788,
      "step": 15745
    },
    {
      "epoch": 0.28438021363725,
      "grad_norm": 0.4931817054748535,
      "learning_rate": 1.626753516996014e-05,
      "loss": 0.315,
      "step": 15750
    },
    {
      "epoch": 0.2844704930701507,
      "grad_norm": 0.5786130428314209,
      "learning_rate": 1.626532486769723e-05,
      "loss": 0.2401,
      "step": 15755
    },
    {
      "epoch": 0.28456077250305145,
      "grad_norm": 0.5008528828620911,
      "learning_rate": 1.6263114061434492e-05,
      "loss": 0.3156,
      "step": 15760
    },
    {
      "epoch": 0.28465105193595214,
      "grad_norm": 0.24782541394233704,
      "learning_rate": 1.6260902751349762e-05,
      "loss": 0.2095,
      "step": 15765
    },
    {
      "epoch": 0.2847413313688529,
      "grad_norm": 0.3510185480117798,
      "learning_rate": 1.6258690937620932e-05,
      "loss": 0.2288,
      "step": 15770
    },
    {
      "epoch": 0.28483161080175357,
      "grad_norm": 0.5392242670059204,
      "learning_rate": 1.6256478620425914e-05,
      "loss": 0.3031,
      "step": 15775
    },
    {
      "epoch": 0.2849218902346543,
      "grad_norm": 0.4234905540943146,
      "learning_rate": 1.6254265799942685e-05,
      "loss": 0.2683,
      "step": 15780
    },
    {
      "epoch": 0.285012169667555,
      "grad_norm": 0.26604676246643066,
      "learning_rate": 1.6252052476349244e-05,
      "loss": 0.271,
      "step": 15785
    },
    {
      "epoch": 0.28510244910045573,
      "grad_norm": 0.6017569899559021,
      "learning_rate": 1.624983864982364e-05,
      "loss": 0.217,
      "step": 15790
    },
    {
      "epoch": 0.2851927285333564,
      "grad_norm": 0.41519540548324585,
      "learning_rate": 1.6247624320543955e-05,
      "loss": 0.245,
      "step": 15795
    },
    {
      "epoch": 0.28528300796625716,
      "grad_norm": 0.4786364734172821,
      "learning_rate": 1.6245409488688317e-05,
      "loss": 0.1956,
      "step": 15800
    },
    {
      "epoch": 0.28537328739915785,
      "grad_norm": 0.30143222212791443,
      "learning_rate": 1.6243194154434897e-05,
      "loss": 0.1802,
      "step": 15805
    },
    {
      "epoch": 0.2854635668320586,
      "grad_norm": 0.48613789677619934,
      "learning_rate": 1.6240978317961897e-05,
      "loss": 0.2628,
      "step": 15810
    },
    {
      "epoch": 0.2855538462649593,
      "grad_norm": 0.35156938433647156,
      "learning_rate": 1.623876197944757e-05,
      "loss": 0.2647,
      "step": 15815
    },
    {
      "epoch": 0.28564412569786,
      "grad_norm": 0.5738462805747986,
      "learning_rate": 1.62365451390702e-05,
      "loss": 0.3109,
      "step": 15820
    },
    {
      "epoch": 0.2857344051307607,
      "grad_norm": 0.489198237657547,
      "learning_rate": 1.623432779700812e-05,
      "loss": 0.3053,
      "step": 15825
    },
    {
      "epoch": 0.28582468456366145,
      "grad_norm": 0.3832317888736725,
      "learning_rate": 1.623210995343969e-05,
      "loss": 0.3054,
      "step": 15830
    },
    {
      "epoch": 0.28591496399656213,
      "grad_norm": 0.5632326006889343,
      "learning_rate": 1.6229891608543334e-05,
      "loss": 0.2865,
      "step": 15835
    },
    {
      "epoch": 0.2860052434294629,
      "grad_norm": 0.56822270154953,
      "learning_rate": 1.622767276249749e-05,
      "loss": 0.2577,
      "step": 15840
    },
    {
      "epoch": 0.28609552286236356,
      "grad_norm": 0.5073590874671936,
      "learning_rate": 1.6225453415480652e-05,
      "loss": 0.2407,
      "step": 15845
    },
    {
      "epoch": 0.2861858022952643,
      "grad_norm": 0.42480191588401794,
      "learning_rate": 1.6223233567671352e-05,
      "loss": 0.2864,
      "step": 15850
    },
    {
      "epoch": 0.28627608172816504,
      "grad_norm": 0.40834692120552063,
      "learning_rate": 1.6221013219248158e-05,
      "loss": 0.3367,
      "step": 15855
    },
    {
      "epoch": 0.28636636116106573,
      "grad_norm": 0.5145226120948792,
      "learning_rate": 1.621879237038968e-05,
      "loss": 0.3203,
      "step": 15860
    },
    {
      "epoch": 0.28645664059396647,
      "grad_norm": 0.3466629385948181,
      "learning_rate": 1.621657102127457e-05,
      "loss": 0.193,
      "step": 15865
    },
    {
      "epoch": 0.28654692002686716,
      "grad_norm": 0.5159338712692261,
      "learning_rate": 1.6214349172081525e-05,
      "loss": 0.2962,
      "step": 15870
    },
    {
      "epoch": 0.2866371994597679,
      "grad_norm": 0.48120617866516113,
      "learning_rate": 1.6212126822989268e-05,
      "loss": 0.3539,
      "step": 15875
    },
    {
      "epoch": 0.2867274788926686,
      "grad_norm": 0.5824990272521973,
      "learning_rate": 1.6209903974176574e-05,
      "loss": 0.1397,
      "step": 15880
    },
    {
      "epoch": 0.2868177583255693,
      "grad_norm": 0.3986798822879791,
      "learning_rate": 1.620768062582226e-05,
      "loss": 0.2509,
      "step": 15885
    },
    {
      "epoch": 0.28690803775847,
      "grad_norm": 0.35834360122680664,
      "learning_rate": 1.6205456778105172e-05,
      "loss": 0.3183,
      "step": 15890
    },
    {
      "epoch": 0.28699831719137076,
      "grad_norm": 0.4624342620372772,
      "learning_rate": 1.6203232431204204e-05,
      "loss": 0.3156,
      "step": 15895
    },
    {
      "epoch": 0.28708859662427144,
      "grad_norm": 0.6071195006370544,
      "learning_rate": 1.6201007585298288e-05,
      "loss": 0.3294,
      "step": 15900
    },
    {
      "epoch": 0.2871788760571722,
      "grad_norm": 0.4487033486366272,
      "learning_rate": 1.6198782240566395e-05,
      "loss": 0.3327,
      "step": 15905
    },
    {
      "epoch": 0.28726915549007287,
      "grad_norm": 0.30451515316963196,
      "learning_rate": 1.6196556397187545e-05,
      "loss": 0.2031,
      "step": 15910
    },
    {
      "epoch": 0.2873594349229736,
      "grad_norm": 0.4486877918243408,
      "learning_rate": 1.619433005534078e-05,
      "loss": 0.29,
      "step": 15915
    },
    {
      "epoch": 0.2874497143558743,
      "grad_norm": 0.40688467025756836,
      "learning_rate": 1.6192103215205205e-05,
      "loss": 0.2412,
      "step": 15920
    },
    {
      "epoch": 0.28753999378877504,
      "grad_norm": 0.5863103866577148,
      "learning_rate": 1.6189875876959947e-05,
      "loss": 0.2982,
      "step": 15925
    },
    {
      "epoch": 0.2876302732216757,
      "grad_norm": 0.6324847936630249,
      "learning_rate": 1.6187648040784178e-05,
      "loss": 0.3914,
      "step": 15930
    },
    {
      "epoch": 0.28772055265457647,
      "grad_norm": 0.4159594476222992,
      "learning_rate": 1.618541970685711e-05,
      "loss": 0.213,
      "step": 15935
    },
    {
      "epoch": 0.28781083208747715,
      "grad_norm": 0.4259512722492218,
      "learning_rate": 1.6183190875358e-05,
      "loss": 0.1838,
      "step": 15940
    },
    {
      "epoch": 0.2879011115203779,
      "grad_norm": 0.41506460309028625,
      "learning_rate": 1.6180961546466145e-05,
      "loss": 0.2815,
      "step": 15945
    },
    {
      "epoch": 0.2879913909532786,
      "grad_norm": 0.5842285752296448,
      "learning_rate": 1.617873172036087e-05,
      "loss": 0.2814,
      "step": 15950
    },
    {
      "epoch": 0.2880816703861793,
      "grad_norm": 0.26104700565338135,
      "learning_rate": 1.617650139722155e-05,
      "loss": 0.3148,
      "step": 15955
    },
    {
      "epoch": 0.28817194981908,
      "grad_norm": 0.4021354615688324,
      "learning_rate": 1.6174270577227603e-05,
      "loss": 0.3118,
      "step": 15960
    },
    {
      "epoch": 0.28826222925198075,
      "grad_norm": 0.44796377420425415,
      "learning_rate": 1.617203926055848e-05,
      "loss": 0.2368,
      "step": 15965
    },
    {
      "epoch": 0.28835250868488144,
      "grad_norm": 0.37265872955322266,
      "learning_rate": 1.616980744739367e-05,
      "loss": 0.2213,
      "step": 15970
    },
    {
      "epoch": 0.2884427881177822,
      "grad_norm": 0.47346892952919006,
      "learning_rate": 1.616757513791271e-05,
      "loss": 0.4529,
      "step": 15975
    },
    {
      "epoch": 0.28853306755068286,
      "grad_norm": 0.5431389212608337,
      "learning_rate": 1.6165342332295176e-05,
      "loss": 0.2396,
      "step": 15980
    },
    {
      "epoch": 0.2886233469835836,
      "grad_norm": 0.3073919415473938,
      "learning_rate": 1.6163109030720678e-05,
      "loss": 0.3849,
      "step": 15985
    },
    {
      "epoch": 0.2887136264164843,
      "grad_norm": 0.6999151110649109,
      "learning_rate": 1.6160875233368868e-05,
      "loss": 0.3738,
      "step": 15990
    },
    {
      "epoch": 0.28880390584938503,
      "grad_norm": 0.36075490713119507,
      "learning_rate": 1.6158640940419435e-05,
      "loss": 0.3154,
      "step": 15995
    },
    {
      "epoch": 0.2888941852822857,
      "grad_norm": 0.39156901836395264,
      "learning_rate": 1.615640615205212e-05,
      "loss": 0.3754,
      "step": 16000
    },
    {
      "epoch": 0.28898446471518646,
      "grad_norm": 0.5082807540893555,
      "learning_rate": 1.6154170868446692e-05,
      "loss": 0.2995,
      "step": 16005
    },
    {
      "epoch": 0.28907474414808715,
      "grad_norm": 0.5425965785980225,
      "learning_rate": 1.615193508978296e-05,
      "loss": 0.3891,
      "step": 16010
    },
    {
      "epoch": 0.2891650235809879,
      "grad_norm": 0.532785177230835,
      "learning_rate": 1.6149698816240782e-05,
      "loss": 0.245,
      "step": 16015
    },
    {
      "epoch": 0.2892553030138886,
      "grad_norm": 0.5923757553100586,
      "learning_rate": 1.6147462048000043e-05,
      "loss": 0.2548,
      "step": 16020
    },
    {
      "epoch": 0.2893455824467893,
      "grad_norm": 0.49339884519577026,
      "learning_rate": 1.6145224785240682e-05,
      "loss": 0.257,
      "step": 16025
    },
    {
      "epoch": 0.28943586187969,
      "grad_norm": 0.31761792302131653,
      "learning_rate": 1.614298702814267e-05,
      "loss": 0.3083,
      "step": 16030
    },
    {
      "epoch": 0.28952614131259075,
      "grad_norm": 1.4089819192886353,
      "learning_rate": 1.614074877688601e-05,
      "loss": 0.391,
      "step": 16035
    },
    {
      "epoch": 0.28961642074549143,
      "grad_norm": 0.5526888370513916,
      "learning_rate": 1.6138510031650757e-05,
      "loss": 0.2956,
      "step": 16040
    },
    {
      "epoch": 0.2897067001783922,
      "grad_norm": 0.39860251545906067,
      "learning_rate": 1.6136270792617008e-05,
      "loss": 0.2583,
      "step": 16045
    },
    {
      "epoch": 0.28979697961129286,
      "grad_norm": 0.3737854063510895,
      "learning_rate": 1.6134031059964884e-05,
      "loss": 0.2668,
      "step": 16050
    },
    {
      "epoch": 0.2898872590441936,
      "grad_norm": 0.47919437289237976,
      "learning_rate": 1.6131790833874564e-05,
      "loss": 0.277,
      "step": 16055
    },
    {
      "epoch": 0.2899775384770943,
      "grad_norm": 0.3381844460964203,
      "learning_rate": 1.612955011452625e-05,
      "loss": 0.3252,
      "step": 16060
    },
    {
      "epoch": 0.29006781790999503,
      "grad_norm": 0.881658673286438,
      "learning_rate": 1.61273089021002e-05,
      "loss": 0.3333,
      "step": 16065
    },
    {
      "epoch": 0.2901580973428957,
      "grad_norm": 0.4923168420791626,
      "learning_rate": 1.61250671967767e-05,
      "loss": 0.2895,
      "step": 16070
    },
    {
      "epoch": 0.29024837677579646,
      "grad_norm": 0.44712138175964355,
      "learning_rate": 1.6122824998736074e-05,
      "loss": 0.3409,
      "step": 16075
    },
    {
      "epoch": 0.29033865620869714,
      "grad_norm": 0.4037761092185974,
      "learning_rate": 1.612058230815869e-05,
      "loss": 0.2853,
      "step": 16080
    },
    {
      "epoch": 0.2904289356415979,
      "grad_norm": 0.3157300651073456,
      "learning_rate": 1.6118339125224966e-05,
      "loss": 0.2477,
      "step": 16085
    },
    {
      "epoch": 0.29051921507449857,
      "grad_norm": 0.3346655070781708,
      "learning_rate": 1.6116095450115344e-05,
      "loss": 0.2159,
      "step": 16090
    },
    {
      "epoch": 0.2906094945073993,
      "grad_norm": 0.4581669569015503,
      "learning_rate": 1.611385128301031e-05,
      "loss": 0.3279,
      "step": 16095
    },
    {
      "epoch": 0.2906997739403,
      "grad_norm": 0.5271456241607666,
      "learning_rate": 1.6111606624090392e-05,
      "loss": 0.2819,
      "step": 16100
    },
    {
      "epoch": 0.29079005337320074,
      "grad_norm": 0.4685978293418884,
      "learning_rate": 1.6109361473536156e-05,
      "loss": 0.2761,
      "step": 16105
    },
    {
      "epoch": 0.2908803328061014,
      "grad_norm": 0.536946713924408,
      "learning_rate": 1.610711583152821e-05,
      "loss": 0.3384,
      "step": 16110
    },
    {
      "epoch": 0.29097061223900217,
      "grad_norm": 1.053295373916626,
      "learning_rate": 1.61048696982472e-05,
      "loss": 0.3751,
      "step": 16115
    },
    {
      "epoch": 0.29106089167190285,
      "grad_norm": 0.5262178182601929,
      "learning_rate": 1.6102623073873806e-05,
      "loss": 0.2472,
      "step": 16120
    },
    {
      "epoch": 0.2911511711048036,
      "grad_norm": 0.5010016560554504,
      "learning_rate": 1.610037595858876e-05,
      "loss": 0.2458,
      "step": 16125
    },
    {
      "epoch": 0.2912414505377043,
      "grad_norm": 0.5782332420349121,
      "learning_rate": 1.6098128352572817e-05,
      "loss": 0.3115,
      "step": 16130
    },
    {
      "epoch": 0.291331729970605,
      "grad_norm": 0.4458324909210205,
      "learning_rate": 1.609588025600679e-05,
      "loss": 0.3693,
      "step": 16135
    },
    {
      "epoch": 0.2914220094035057,
      "grad_norm": 0.6408203840255737,
      "learning_rate": 1.6093631669071514e-05,
      "loss": 0.2183,
      "step": 16140
    },
    {
      "epoch": 0.29151228883640645,
      "grad_norm": 0.5350770354270935,
      "learning_rate": 1.6091382591947878e-05,
      "loss": 0.3975,
      "step": 16145
    },
    {
      "epoch": 0.29160256826930714,
      "grad_norm": 0.31011661887168884,
      "learning_rate": 1.60891330248168e-05,
      "loss": 0.2466,
      "step": 16150
    },
    {
      "epoch": 0.2916928477022079,
      "grad_norm": 0.44612425565719604,
      "learning_rate": 1.6086882967859244e-05,
      "loss": 0.2606,
      "step": 16155
    },
    {
      "epoch": 0.29178312713510857,
      "grad_norm": 0.8416286110877991,
      "learning_rate": 1.6084632421256206e-05,
      "loss": 0.3034,
      "step": 16160
    },
    {
      "epoch": 0.2918734065680093,
      "grad_norm": 0.840193510055542,
      "learning_rate": 1.6082381385188736e-05,
      "loss": 0.3395,
      "step": 16165
    },
    {
      "epoch": 0.29196368600091,
      "grad_norm": 0.5311060547828674,
      "learning_rate": 1.60801298598379e-05,
      "loss": 0.2897,
      "step": 16170
    },
    {
      "epoch": 0.29205396543381074,
      "grad_norm": 0.4871434271335602,
      "learning_rate": 1.6077877845384826e-05,
      "loss": 0.2197,
      "step": 16175
    },
    {
      "epoch": 0.2921442448667114,
      "grad_norm": 0.3499642014503479,
      "learning_rate": 1.607562534201067e-05,
      "loss": 0.2727,
      "step": 16180
    },
    {
      "epoch": 0.29223452429961216,
      "grad_norm": 0.47589465975761414,
      "learning_rate": 1.6073372349896626e-05,
      "loss": 0.3038,
      "step": 16185
    },
    {
      "epoch": 0.29232480373251285,
      "grad_norm": 0.4539521038532257,
      "learning_rate": 1.6071118869223942e-05,
      "loss": 0.2644,
      "step": 16190
    },
    {
      "epoch": 0.2924150831654136,
      "grad_norm": 0.39266568422317505,
      "learning_rate": 1.606886490017388e-05,
      "loss": 0.2343,
      "step": 16195
    },
    {
      "epoch": 0.2925053625983143,
      "grad_norm": 0.5503832697868347,
      "learning_rate": 1.6066610442927762e-05,
      "loss": 0.3114,
      "step": 16200
    },
    {
      "epoch": 0.292595642031215,
      "grad_norm": 0.3184185028076172,
      "learning_rate": 1.6064355497666942e-05,
      "loss": 0.17,
      "step": 16205
    },
    {
      "epoch": 0.2926859214641157,
      "grad_norm": 0.6293808221817017,
      "learning_rate": 1.6062100064572818e-05,
      "loss": 0.3187,
      "step": 16210
    },
    {
      "epoch": 0.29277620089701645,
      "grad_norm": 0.5355658531188965,
      "learning_rate": 1.6059844143826816e-05,
      "loss": 0.2458,
      "step": 16215
    },
    {
      "epoch": 0.29286648032991713,
      "grad_norm": 0.3671532869338989,
      "learning_rate": 1.6057587735610413e-05,
      "loss": 0.2434,
      "step": 16220
    },
    {
      "epoch": 0.2929567597628179,
      "grad_norm": 0.39543014764785767,
      "learning_rate": 1.605533084010512e-05,
      "loss": 0.333,
      "step": 16225
    },
    {
      "epoch": 0.29304703919571856,
      "grad_norm": 0.4987691342830658,
      "learning_rate": 1.6053073457492486e-05,
      "loss": 0.2069,
      "step": 16230
    },
    {
      "epoch": 0.2931373186286193,
      "grad_norm": 0.4521399736404419,
      "learning_rate": 1.6050815587954103e-05,
      "loss": 0.2345,
      "step": 16235
    },
    {
      "epoch": 0.29322759806152,
      "grad_norm": 0.35981765389442444,
      "learning_rate": 1.6048557231671602e-05,
      "loss": 0.2338,
      "step": 16240
    },
    {
      "epoch": 0.29331787749442073,
      "grad_norm": 0.6158979535102844,
      "learning_rate": 1.6046298388826646e-05,
      "loss": 0.2522,
      "step": 16245
    },
    {
      "epoch": 0.2934081569273215,
      "grad_norm": 0.36089736223220825,
      "learning_rate": 1.6044039059600946e-05,
      "loss": 0.2624,
      "step": 16250
    },
    {
      "epoch": 0.29349843636022216,
      "grad_norm": 0.7971797585487366,
      "learning_rate": 1.6041779244176253e-05,
      "loss": 0.2781,
      "step": 16255
    },
    {
      "epoch": 0.2935887157931229,
      "grad_norm": 0.44186967611312866,
      "learning_rate": 1.603951894273434e-05,
      "loss": 0.3122,
      "step": 16260
    },
    {
      "epoch": 0.2936789952260236,
      "grad_norm": 0.6414748430252075,
      "learning_rate": 1.6037258155457046e-05,
      "loss": 0.3285,
      "step": 16265
    },
    {
      "epoch": 0.29376927465892433,
      "grad_norm": 0.611411452293396,
      "learning_rate": 1.6034996882526224e-05,
      "loss": 0.2994,
      "step": 16270
    },
    {
      "epoch": 0.293859554091825,
      "grad_norm": 0.8820737600326538,
      "learning_rate": 1.6032735124123783e-05,
      "loss": 0.225,
      "step": 16275
    },
    {
      "epoch": 0.29394983352472576,
      "grad_norm": 0.45774945616722107,
      "learning_rate": 1.6030472880431665e-05,
      "loss": 0.2805,
      "step": 16280
    },
    {
      "epoch": 0.29404011295762644,
      "grad_norm": 0.2715279459953308,
      "learning_rate": 1.6028210151631848e-05,
      "loss": 0.2294,
      "step": 16285
    },
    {
      "epoch": 0.2941303923905272,
      "grad_norm": 0.4004227817058563,
      "learning_rate": 1.6025946937906353e-05,
      "loss": 0.2253,
      "step": 16290
    },
    {
      "epoch": 0.29422067182342787,
      "grad_norm": 0.6729216575622559,
      "learning_rate": 1.602368323943724e-05,
      "loss": 0.2271,
      "step": 16295
    },
    {
      "epoch": 0.2943109512563286,
      "grad_norm": 0.6926440000534058,
      "learning_rate": 1.6021419056406608e-05,
      "loss": 0.2381,
      "step": 16300
    },
    {
      "epoch": 0.2944012306892293,
      "grad_norm": 0.3082001805305481,
      "learning_rate": 1.601915438899659e-05,
      "loss": 0.2876,
      "step": 16305
    },
    {
      "epoch": 0.29449151012213004,
      "grad_norm": 0.3748997449874878,
      "learning_rate": 1.6016889237389368e-05,
      "loss": 0.3011,
      "step": 16310
    },
    {
      "epoch": 0.2945817895550307,
      "grad_norm": 0.31639403104782104,
      "learning_rate": 1.6014623601767155e-05,
      "loss": 0.2779,
      "step": 16315
    },
    {
      "epoch": 0.29467206898793147,
      "grad_norm": 0.3767625689506531,
      "learning_rate": 1.6012357482312197e-05,
      "loss": 0.3008,
      "step": 16320
    },
    {
      "epoch": 0.29476234842083215,
      "grad_norm": 1.117903232574463,
      "learning_rate": 1.60100908792068e-05,
      "loss": 0.3156,
      "step": 16325
    },
    {
      "epoch": 0.2948526278537329,
      "grad_norm": 0.7283892631530762,
      "learning_rate": 1.600782379263329e-05,
      "loss": 0.2838,
      "step": 16330
    },
    {
      "epoch": 0.2949429072866336,
      "grad_norm": 0.5270529985427856,
      "learning_rate": 1.6005556222774033e-05,
      "loss": 0.2027,
      "step": 16335
    },
    {
      "epoch": 0.2950331867195343,
      "grad_norm": 0.3388364613056183,
      "learning_rate": 1.6003288169811444e-05,
      "loss": 0.282,
      "step": 16340
    },
    {
      "epoch": 0.295123466152435,
      "grad_norm": 0.32637423276901245,
      "learning_rate": 1.600101963392797e-05,
      "loss": 0.2616,
      "step": 16345
    },
    {
      "epoch": 0.29521374558533575,
      "grad_norm": 0.4189916253089905,
      "learning_rate": 1.5998750615306102e-05,
      "loss": 0.2784,
      "step": 16350
    },
    {
      "epoch": 0.29530402501823644,
      "grad_norm": 0.41963228583335876,
      "learning_rate": 1.5996481114128355e-05,
      "loss": 0.2659,
      "step": 16355
    },
    {
      "epoch": 0.2953943044511372,
      "grad_norm": 0.7162477374076843,
      "learning_rate": 1.5994211130577305e-05,
      "loss": 0.2969,
      "step": 16360
    },
    {
      "epoch": 0.29548458388403787,
      "grad_norm": 0.43092843890190125,
      "learning_rate": 1.599194066483555e-05,
      "loss": 0.369,
      "step": 16365
    },
    {
      "epoch": 0.2955748633169386,
      "grad_norm": 0.41079744696617126,
      "learning_rate": 1.5989669717085736e-05,
      "loss": 0.28,
      "step": 16370
    },
    {
      "epoch": 0.2956651427498393,
      "grad_norm": 0.4858797788619995,
      "learning_rate": 1.598739828751054e-05,
      "loss": 0.2702,
      "step": 16375
    },
    {
      "epoch": 0.29575542218274004,
      "grad_norm": 0.61838698387146,
      "learning_rate": 1.5985126376292684e-05,
      "loss": 0.3351,
      "step": 16380
    },
    {
      "epoch": 0.2958457016156407,
      "grad_norm": 0.42351454496383667,
      "learning_rate": 1.598285398361493e-05,
      "loss": 0.2322,
      "step": 16385
    },
    {
      "epoch": 0.29593598104854146,
      "grad_norm": 0.42310693860054016,
      "learning_rate": 1.5980581109660072e-05,
      "loss": 0.3411,
      "step": 16390
    },
    {
      "epoch": 0.29602626048144215,
      "grad_norm": 0.4089198410511017,
      "learning_rate": 1.5978307754610948e-05,
      "loss": 0.2977,
      "step": 16395
    },
    {
      "epoch": 0.2961165399143429,
      "grad_norm": 0.7078156471252441,
      "learning_rate": 1.5976033918650428e-05,
      "loss": 0.2277,
      "step": 16400
    },
    {
      "epoch": 0.2962068193472436,
      "grad_norm": 1.6692484617233276,
      "learning_rate": 1.597375960196143e-05,
      "loss": 0.2924,
      "step": 16405
    },
    {
      "epoch": 0.2962970987801443,
      "grad_norm": 0.5945263504981995,
      "learning_rate": 1.5971484804726908e-05,
      "loss": 0.1733,
      "step": 16410
    },
    {
      "epoch": 0.296387378213045,
      "grad_norm": 0.4478415250778198,
      "learning_rate": 1.5969209527129847e-05,
      "loss": 0.3286,
      "step": 16415
    },
    {
      "epoch": 0.29647765764594575,
      "grad_norm": 0.42879509925842285,
      "learning_rate": 1.5966933769353282e-05,
      "loss": 0.2969,
      "step": 16420
    },
    {
      "epoch": 0.29656793707884643,
      "grad_norm": 0.43934568762779236,
      "learning_rate": 1.596465753158028e-05,
      "loss": 0.3345,
      "step": 16425
    },
    {
      "epoch": 0.2966582165117472,
      "grad_norm": 0.3786970376968384,
      "learning_rate": 1.5962380813993946e-05,
      "loss": 0.3804,
      "step": 16430
    },
    {
      "epoch": 0.29674849594464786,
      "grad_norm": 0.5921241044998169,
      "learning_rate": 1.5960103616777423e-05,
      "loss": 0.2281,
      "step": 16435
    },
    {
      "epoch": 0.2968387753775486,
      "grad_norm": 0.842401921749115,
      "learning_rate": 1.5957825940113903e-05,
      "loss": 0.3482,
      "step": 16440
    },
    {
      "epoch": 0.2969290548104493,
      "grad_norm": 0.5948904752731323,
      "learning_rate": 1.59555477841866e-05,
      "loss": 0.2591,
      "step": 16445
    },
    {
      "epoch": 0.29701933424335003,
      "grad_norm": 0.6085169911384583,
      "learning_rate": 1.595326914917878e-05,
      "loss": 0.3126,
      "step": 16450
    },
    {
      "epoch": 0.2971096136762507,
      "grad_norm": 0.3494928181171417,
      "learning_rate": 1.5950990035273745e-05,
      "loss": 0.3576,
      "step": 16455
    },
    {
      "epoch": 0.29719989310915146,
      "grad_norm": 0.5723707675933838,
      "learning_rate": 1.5948710442654823e-05,
      "loss": 0.302,
      "step": 16460
    },
    {
      "epoch": 0.29729017254205214,
      "grad_norm": 0.5860376954078674,
      "learning_rate": 1.5946430371505404e-05,
      "loss": 0.2754,
      "step": 16465
    },
    {
      "epoch": 0.2973804519749529,
      "grad_norm": 0.5320484042167664,
      "learning_rate": 1.5944149822008893e-05,
      "loss": 0.3047,
      "step": 16470
    },
    {
      "epoch": 0.2974707314078536,
      "grad_norm": 0.3392791152000427,
      "learning_rate": 1.5941868794348745e-05,
      "loss": 0.2641,
      "step": 16475
    },
    {
      "epoch": 0.2975610108407543,
      "grad_norm": 0.3529278039932251,
      "learning_rate": 1.593958728870846e-05,
      "loss": 0.2366,
      "step": 16480
    },
    {
      "epoch": 0.297651290273655,
      "grad_norm": 0.4152165353298187,
      "learning_rate": 1.5937305305271556e-05,
      "loss": 0.2023,
      "step": 16485
    },
    {
      "epoch": 0.29774156970655574,
      "grad_norm": 0.3920888900756836,
      "learning_rate": 1.593502284422161e-05,
      "loss": 0.2125,
      "step": 16490
    },
    {
      "epoch": 0.29783184913945643,
      "grad_norm": 0.4801064431667328,
      "learning_rate": 1.5932739905742228e-05,
      "loss": 0.3785,
      "step": 16495
    },
    {
      "epoch": 0.29792212857235717,
      "grad_norm": 0.9531915187835693,
      "learning_rate": 1.593045649001706e-05,
      "loss": 0.1912,
      "step": 16500
    },
    {
      "epoch": 0.29801240800525786,
      "grad_norm": 0.35262203216552734,
      "learning_rate": 1.592817259722978e-05,
      "loss": 0.2331,
      "step": 16505
    },
    {
      "epoch": 0.2981026874381586,
      "grad_norm": 0.3981858491897583,
      "learning_rate": 1.592588822756412e-05,
      "loss": 0.326,
      "step": 16510
    },
    {
      "epoch": 0.2981929668710593,
      "grad_norm": 0.45648452639579773,
      "learning_rate": 1.5923603381203837e-05,
      "loss": 0.3099,
      "step": 16515
    },
    {
      "epoch": 0.29828324630396,
      "grad_norm": 0.3206727206707001,
      "learning_rate": 1.592131805833273e-05,
      "loss": 0.2139,
      "step": 16520
    },
    {
      "epoch": 0.2983735257368607,
      "grad_norm": 0.22646181285381317,
      "learning_rate": 1.5919032259134638e-05,
      "loss": 0.3353,
      "step": 16525
    },
    {
      "epoch": 0.29846380516976145,
      "grad_norm": 0.5294449925422668,
      "learning_rate": 1.5916745983793436e-05,
      "loss": 0.2825,
      "step": 16530
    },
    {
      "epoch": 0.29855408460266214,
      "grad_norm": 0.576151430606842,
      "learning_rate": 1.591445923249304e-05,
      "loss": 0.3866,
      "step": 16535
    },
    {
      "epoch": 0.2986443640355629,
      "grad_norm": 0.4805523157119751,
      "learning_rate": 1.5912172005417402e-05,
      "loss": 0.202,
      "step": 16540
    },
    {
      "epoch": 0.29873464346846357,
      "grad_norm": 0.873942494392395,
      "learning_rate": 1.590988430275051e-05,
      "loss": 0.3125,
      "step": 16545
    },
    {
      "epoch": 0.2988249229013643,
      "grad_norm": 0.37433138489723206,
      "learning_rate": 1.590759612467639e-05,
      "loss": 0.2575,
      "step": 16550
    },
    {
      "epoch": 0.298915202334265,
      "grad_norm": 0.5413210391998291,
      "learning_rate": 1.5905307471379118e-05,
      "loss": 0.2877,
      "step": 16555
    },
    {
      "epoch": 0.29900548176716574,
      "grad_norm": 0.3787118196487427,
      "learning_rate": 1.5903018343042797e-05,
      "loss": 0.2922,
      "step": 16560
    },
    {
      "epoch": 0.2990957612000664,
      "grad_norm": 0.3813949525356293,
      "learning_rate": 1.5900728739851568e-05,
      "loss": 0.2428,
      "step": 16565
    },
    {
      "epoch": 0.29918604063296717,
      "grad_norm": 0.6623268127441406,
      "learning_rate": 1.5898438661989615e-05,
      "loss": 0.3732,
      "step": 16570
    },
    {
      "epoch": 0.29927632006586785,
      "grad_norm": 0.5135146975517273,
      "learning_rate": 1.5896148109641157e-05,
      "loss": 0.2872,
      "step": 16575
    },
    {
      "epoch": 0.2993665994987686,
      "grad_norm": 0.36938363313674927,
      "learning_rate": 1.5893857082990455e-05,
      "loss": 0.312,
      "step": 16580
    },
    {
      "epoch": 0.2994568789316693,
      "grad_norm": 0.338520884513855,
      "learning_rate": 1.58915655822218e-05,
      "loss": 0.257,
      "step": 16585
    },
    {
      "epoch": 0.29954715836457,
      "grad_norm": 0.272137850522995,
      "learning_rate": 1.5889273607519535e-05,
      "loss": 0.2715,
      "step": 16590
    },
    {
      "epoch": 0.2996374377974707,
      "grad_norm": 0.46022340655326843,
      "learning_rate": 1.588698115906802e-05,
      "loss": 0.2981,
      "step": 16595
    },
    {
      "epoch": 0.29972771723037145,
      "grad_norm": 0.6358815431594849,
      "learning_rate": 1.5884688237051675e-05,
      "loss": 0.3019,
      "step": 16600
    },
    {
      "epoch": 0.29981799666327213,
      "grad_norm": 1.5296499729156494,
      "learning_rate": 1.588239484165495e-05,
      "loss": 0.2328,
      "step": 16605
    },
    {
      "epoch": 0.2999082760961729,
      "grad_norm": 0.49699029326438904,
      "learning_rate": 1.5880100973062326e-05,
      "loss": 0.2118,
      "step": 16610
    },
    {
      "epoch": 0.29999855552907356,
      "grad_norm": 0.5322579741477966,
      "learning_rate": 1.5877806631458334e-05,
      "loss": 0.4579,
      "step": 16615
    },
    {
      "epoch": 0.3000888349619743,
      "grad_norm": 0.356161504983902,
      "learning_rate": 1.5875511817027534e-05,
      "loss": 0.2937,
      "step": 16620
    },
    {
      "epoch": 0.300179114394875,
      "grad_norm": 0.45161619782447815,
      "learning_rate": 1.5873216529954528e-05,
      "loss": 0.2722,
      "step": 16625
    },
    {
      "epoch": 0.30026939382777573,
      "grad_norm": 1.0573439598083496,
      "learning_rate": 1.5870920770423956e-05,
      "loss": 0.3443,
      "step": 16630
    },
    {
      "epoch": 0.3003596732606764,
      "grad_norm": 0.6455829739570618,
      "learning_rate": 1.586862453862049e-05,
      "loss": 0.221,
      "step": 16635
    },
    {
      "epoch": 0.30044995269357716,
      "grad_norm": 0.332810640335083,
      "learning_rate": 1.5866327834728846e-05,
      "loss": 0.1939,
      "step": 16640
    },
    {
      "epoch": 0.3005402321264779,
      "grad_norm": 0.3890936076641083,
      "learning_rate": 1.5864030658933785e-05,
      "loss": 0.2534,
      "step": 16645
    },
    {
      "epoch": 0.3006305115593786,
      "grad_norm": 0.9418467283248901,
      "learning_rate": 1.5861733011420092e-05,
      "loss": 0.296,
      "step": 16650
    },
    {
      "epoch": 0.30072079099227933,
      "grad_norm": 0.5516576170921326,
      "learning_rate": 1.5859434892372597e-05,
      "loss": 0.316,
      "step": 16655
    },
    {
      "epoch": 0.30081107042518,
      "grad_norm": 0.5210039615631104,
      "learning_rate": 1.585713630197617e-05,
      "loss": 0.3231,
      "step": 16660
    },
    {
      "epoch": 0.30090134985808076,
      "grad_norm": 0.45455679297447205,
      "learning_rate": 1.5854837240415707e-05,
      "loss": 0.1983,
      "step": 16665
    },
    {
      "epoch": 0.30099162929098144,
      "grad_norm": 0.35978347063064575,
      "learning_rate": 1.5852537707876163e-05,
      "loss": 0.3051,
      "step": 16670
    },
    {
      "epoch": 0.3010819087238822,
      "grad_norm": 0.39413386583328247,
      "learning_rate": 1.5850237704542508e-05,
      "loss": 0.3355,
      "step": 16675
    },
    {
      "epoch": 0.30117218815678287,
      "grad_norm": 0.4914714992046356,
      "learning_rate": 1.584793723059977e-05,
      "loss": 0.2743,
      "step": 16680
    },
    {
      "epoch": 0.3012624675896836,
      "grad_norm": 0.41727781295776367,
      "learning_rate": 1.5845636286232993e-05,
      "loss": 0.274,
      "step": 16685
    },
    {
      "epoch": 0.3013527470225843,
      "grad_norm": 0.5112030506134033,
      "learning_rate": 1.5843334871627285e-05,
      "loss": 0.2437,
      "step": 16690
    },
    {
      "epoch": 0.30144302645548504,
      "grad_norm": 0.4653795063495636,
      "learning_rate": 1.5841032986967768e-05,
      "loss": 0.4123,
      "step": 16695
    },
    {
      "epoch": 0.3015333058883857,
      "grad_norm": 0.2604282796382904,
      "learning_rate": 1.583873063243962e-05,
      "loss": 0.2101,
      "step": 16700
    },
    {
      "epoch": 0.30162358532128647,
      "grad_norm": 0.3000047206878662,
      "learning_rate": 1.5836427808228037e-05,
      "loss": 0.2573,
      "step": 16705
    },
    {
      "epoch": 0.30171386475418716,
      "grad_norm": 0.5702049136161804,
      "learning_rate": 1.5834124514518275e-05,
      "loss": 0.3305,
      "step": 16710
    },
    {
      "epoch": 0.3018041441870879,
      "grad_norm": 0.7986447215080261,
      "learning_rate": 1.5831820751495618e-05,
      "loss": 0.2979,
      "step": 16715
    },
    {
      "epoch": 0.3018944236199886,
      "grad_norm": 0.5973787307739258,
      "learning_rate": 1.5829516519345382e-05,
      "loss": 0.2826,
      "step": 16720
    },
    {
      "epoch": 0.3019847030528893,
      "grad_norm": 0.3181029260158539,
      "learning_rate": 1.5827211818252922e-05,
      "loss": 0.2041,
      "step": 16725
    },
    {
      "epoch": 0.30207498248579,
      "grad_norm": 0.3650952875614166,
      "learning_rate": 1.5824906648403645e-05,
      "loss": 0.1462,
      "step": 16730
    },
    {
      "epoch": 0.30216526191869075,
      "grad_norm": 0.4607314169406891,
      "learning_rate": 1.5822601009982977e-05,
      "loss": 0.2993,
      "step": 16735
    },
    {
      "epoch": 0.30225554135159144,
      "grad_norm": 0.8629481792449951,
      "learning_rate": 1.582029490317639e-05,
      "loss": 0.3386,
      "step": 16740
    },
    {
      "epoch": 0.3023458207844922,
      "grad_norm": 0.8705028295516968,
      "learning_rate": 1.5817988328169397e-05,
      "loss": 0.2005,
      "step": 16745
    },
    {
      "epoch": 0.30243610021739287,
      "grad_norm": 0.5850492715835571,
      "learning_rate": 1.5815681285147543e-05,
      "loss": 0.2795,
      "step": 16750
    },
    {
      "epoch": 0.3025263796502936,
      "grad_norm": 0.43750256299972534,
      "learning_rate": 1.5813373774296415e-05,
      "loss": 0.3561,
      "step": 16755
    },
    {
      "epoch": 0.3026166590831943,
      "grad_norm": 0.4892030656337738,
      "learning_rate": 1.5811065795801633e-05,
      "loss": 0.3041,
      "step": 16760
    },
    {
      "epoch": 0.30270693851609504,
      "grad_norm": 0.7717058658599854,
      "learning_rate": 1.580875734984886e-05,
      "loss": 0.2017,
      "step": 16765
    },
    {
      "epoch": 0.3027972179489957,
      "grad_norm": 0.4515610635280609,
      "learning_rate": 1.580644843662379e-05,
      "loss": 0.2682,
      "step": 16770
    },
    {
      "epoch": 0.30288749738189646,
      "grad_norm": 0.5138480067253113,
      "learning_rate": 1.580413905631216e-05,
      "loss": 0.2458,
      "step": 16775
    },
    {
      "epoch": 0.30297777681479715,
      "grad_norm": 0.40732017159461975,
      "learning_rate": 1.5801829209099747e-05,
      "loss": 0.2704,
      "step": 16780
    },
    {
      "epoch": 0.3030680562476979,
      "grad_norm": 0.4371160566806793,
      "learning_rate": 1.5799518895172353e-05,
      "loss": 0.1914,
      "step": 16785
    },
    {
      "epoch": 0.3031583356805986,
      "grad_norm": 0.4059670567512512,
      "learning_rate": 1.5797208114715834e-05,
      "loss": 0.2064,
      "step": 16790
    },
    {
      "epoch": 0.3032486151134993,
      "grad_norm": 0.46853598952293396,
      "learning_rate": 1.579489686791607e-05,
      "loss": 0.3301,
      "step": 16795
    },
    {
      "epoch": 0.3033388945464,
      "grad_norm": 0.4065183699131012,
      "learning_rate": 1.5792585154958983e-05,
      "loss": 0.3527,
      "step": 16800
    },
    {
      "epoch": 0.30342917397930075,
      "grad_norm": 0.5132343173027039,
      "learning_rate": 1.579027297603054e-05,
      "loss": 0.2933,
      "step": 16805
    },
    {
      "epoch": 0.30351945341220143,
      "grad_norm": 0.5387590527534485,
      "learning_rate": 1.5787960331316737e-05,
      "loss": 0.2777,
      "step": 16810
    },
    {
      "epoch": 0.3036097328451022,
      "grad_norm": 0.451153963804245,
      "learning_rate": 1.5785647221003607e-05,
      "loss": 0.2794,
      "step": 16815
    },
    {
      "epoch": 0.30370001227800286,
      "grad_norm": 0.4658501446247101,
      "learning_rate": 1.578333364527722e-05,
      "loss": 0.2532,
      "step": 16820
    },
    {
      "epoch": 0.3037902917109036,
      "grad_norm": 0.6236963868141174,
      "learning_rate": 1.5781019604323695e-05,
      "loss": 0.3406,
      "step": 16825
    },
    {
      "epoch": 0.3038805711438043,
      "grad_norm": 0.5373722314834595,
      "learning_rate": 1.5778705098329174e-05,
      "loss": 0.2902,
      "step": 16830
    },
    {
      "epoch": 0.30397085057670503,
      "grad_norm": 0.5444965958595276,
      "learning_rate": 1.577639012747984e-05,
      "loss": 0.2481,
      "step": 16835
    },
    {
      "epoch": 0.3040611300096057,
      "grad_norm": 0.5736184120178223,
      "learning_rate": 1.577407469196192e-05,
      "loss": 0.2832,
      "step": 16840
    },
    {
      "epoch": 0.30415140944250646,
      "grad_norm": 0.3974739611148834,
      "learning_rate": 1.5771758791961676e-05,
      "loss": 0.2287,
      "step": 16845
    },
    {
      "epoch": 0.30424168887540715,
      "grad_norm": 0.3344295024871826,
      "learning_rate": 1.5769442427665397e-05,
      "loss": 0.2858,
      "step": 16850
    },
    {
      "epoch": 0.3043319683083079,
      "grad_norm": 0.3950170874595642,
      "learning_rate": 1.5767125599259424e-05,
      "loss": 0.347,
      "step": 16855
    },
    {
      "epoch": 0.3044222477412086,
      "grad_norm": 0.527312159538269,
      "learning_rate": 1.5764808306930127e-05,
      "loss": 0.2666,
      "step": 16860
    },
    {
      "epoch": 0.3045125271741093,
      "grad_norm": 0.46841877698898315,
      "learning_rate": 1.576249055086392e-05,
      "loss": 0.2881,
      "step": 16865
    },
    {
      "epoch": 0.30460280660701,
      "grad_norm": 1.189628005027771,
      "learning_rate": 1.5760172331247242e-05,
      "loss": 0.257,
      "step": 16870
    },
    {
      "epoch": 0.30469308603991074,
      "grad_norm": 0.5873405337333679,
      "learning_rate": 1.575785364826658e-05,
      "loss": 0.2372,
      "step": 16875
    },
    {
      "epoch": 0.30478336547281143,
      "grad_norm": 0.5122183561325073,
      "learning_rate": 1.5755534502108458e-05,
      "loss": 0.3921,
      "step": 16880
    },
    {
      "epoch": 0.30487364490571217,
      "grad_norm": 0.3151512145996094,
      "learning_rate": 1.575321489295943e-05,
      "loss": 0.1668,
      "step": 16885
    },
    {
      "epoch": 0.30496392433861286,
      "grad_norm": 0.4796990156173706,
      "learning_rate": 1.5750894821006094e-05,
      "loss": 0.2686,
      "step": 16890
    },
    {
      "epoch": 0.3050542037715136,
      "grad_norm": 0.44149377942085266,
      "learning_rate": 1.5748574286435083e-05,
      "loss": 0.3099,
      "step": 16895
    },
    {
      "epoch": 0.3051444832044143,
      "grad_norm": 0.4739304780960083,
      "learning_rate": 1.574625328943307e-05,
      "loss": 0.3619,
      "step": 16900
    },
    {
      "epoch": 0.305234762637315,
      "grad_norm": 0.47793883085250854,
      "learning_rate": 1.5743931830186754e-05,
      "loss": 0.3208,
      "step": 16905
    },
    {
      "epoch": 0.3053250420702157,
      "grad_norm": 0.29709920287132263,
      "learning_rate": 1.574160990888289e-05,
      "loss": 0.2187,
      "step": 16910
    },
    {
      "epoch": 0.30541532150311645,
      "grad_norm": 2.004121780395508,
      "learning_rate": 1.573928752570825e-05,
      "loss": 0.299,
      "step": 16915
    },
    {
      "epoch": 0.30550560093601714,
      "grad_norm": 0.721603274345398,
      "learning_rate": 1.573696468084966e-05,
      "loss": 0.278,
      "step": 16920
    },
    {
      "epoch": 0.3055958803689179,
      "grad_norm": 0.3206048905849457,
      "learning_rate": 1.573464137449397e-05,
      "loss": 0.27,
      "step": 16925
    },
    {
      "epoch": 0.30568615980181857,
      "grad_norm": 0.910554826259613,
      "learning_rate": 1.573231760682808e-05,
      "loss": 0.3167,
      "step": 16930
    },
    {
      "epoch": 0.3057764392347193,
      "grad_norm": 0.4243769943714142,
      "learning_rate": 1.5729993378038913e-05,
      "loss": 0.2846,
      "step": 16935
    },
    {
      "epoch": 0.30586671866762,
      "grad_norm": 0.4143887460231781,
      "learning_rate": 1.572766868831344e-05,
      "loss": 0.2467,
      "step": 16940
    },
    {
      "epoch": 0.30595699810052074,
      "grad_norm": 0.35781130194664,
      "learning_rate": 1.572534353783867e-05,
      "loss": 0.3657,
      "step": 16945
    },
    {
      "epoch": 0.3060472775334214,
      "grad_norm": 0.4417208135128021,
      "learning_rate": 1.5723017926801633e-05,
      "loss": 0.2778,
      "step": 16950
    },
    {
      "epoch": 0.30613755696632217,
      "grad_norm": 0.3955446481704712,
      "learning_rate": 1.5720691855389417e-05,
      "loss": 0.296,
      "step": 16955
    },
    {
      "epoch": 0.30622783639922285,
      "grad_norm": 0.34164705872535706,
      "learning_rate": 1.5718365323789135e-05,
      "loss": 0.2581,
      "step": 16960
    },
    {
      "epoch": 0.3063181158321236,
      "grad_norm": 0.3188674747943878,
      "learning_rate": 1.571603833218794e-05,
      "loss": 0.2536,
      "step": 16965
    },
    {
      "epoch": 0.3064083952650243,
      "grad_norm": 0.36604100465774536,
      "learning_rate": 1.571371088077302e-05,
      "loss": 0.193,
      "step": 16970
    },
    {
      "epoch": 0.306498674697925,
      "grad_norm": 0.46219056844711304,
      "learning_rate": 1.5711382969731603e-05,
      "loss": 0.3393,
      "step": 16975
    },
    {
      "epoch": 0.3065889541308257,
      "grad_norm": 0.4927115738391876,
      "learning_rate": 1.5709054599250954e-05,
      "loss": 0.2085,
      "step": 16980
    },
    {
      "epoch": 0.30667923356372645,
      "grad_norm": 0.3200385272502899,
      "learning_rate": 1.5706725769518365e-05,
      "loss": 0.2326,
      "step": 16985
    },
    {
      "epoch": 0.30676951299662714,
      "grad_norm": 0.4072023332118988,
      "learning_rate": 1.5704396480721183e-05,
      "loss": 0.2405,
      "step": 16990
    },
    {
      "epoch": 0.3068597924295279,
      "grad_norm": 0.41983669996261597,
      "learning_rate": 1.5702066733046782e-05,
      "loss": 0.4141,
      "step": 16995
    },
    {
      "epoch": 0.30695007186242856,
      "grad_norm": 0.4949229657649994,
      "learning_rate": 1.5699736526682566e-05,
      "loss": 0.3501,
      "step": 17000
    },
    {
      "epoch": 0.3070403512953293,
      "grad_norm": 0.574365496635437,
      "learning_rate": 1.569740586181599e-05,
      "loss": 0.2866,
      "step": 17005
    },
    {
      "epoch": 0.30713063072823,
      "grad_norm": 0.329974889755249,
      "learning_rate": 1.5695074738634536e-05,
      "loss": 0.314,
      "step": 17010
    },
    {
      "epoch": 0.30722091016113073,
      "grad_norm": 0.45943304896354675,
      "learning_rate": 1.569274315732573e-05,
      "loss": 0.2753,
      "step": 17015
    },
    {
      "epoch": 0.3073111895940314,
      "grad_norm": 0.5166605710983276,
      "learning_rate": 1.569041111807712e-05,
      "loss": 0.3967,
      "step": 17020
    },
    {
      "epoch": 0.30740146902693216,
      "grad_norm": 0.6126747727394104,
      "learning_rate": 1.5688078621076313e-05,
      "loss": 0.2758,
      "step": 17025
    },
    {
      "epoch": 0.30749174845983285,
      "grad_norm": 0.5386550426483154,
      "learning_rate": 1.5685745666510935e-05,
      "loss": 0.2913,
      "step": 17030
    },
    {
      "epoch": 0.3075820278927336,
      "grad_norm": 0.5282537937164307,
      "learning_rate": 1.568341225456866e-05,
      "loss": 0.3627,
      "step": 17035
    },
    {
      "epoch": 0.30767230732563433,
      "grad_norm": 1.0371299982070923,
      "learning_rate": 1.568107838543719e-05,
      "loss": 0.1561,
      "step": 17040
    },
    {
      "epoch": 0.307762586758535,
      "grad_norm": 0.54307621717453,
      "learning_rate": 1.5678744059304273e-05,
      "loss": 0.2988,
      "step": 17045
    },
    {
      "epoch": 0.30785286619143576,
      "grad_norm": 0.5720261335372925,
      "learning_rate": 1.567640927635768e-05,
      "loss": 0.2863,
      "step": 17050
    },
    {
      "epoch": 0.30794314562433645,
      "grad_norm": 0.42609718441963196,
      "learning_rate": 1.5674074036785233e-05,
      "loss": 0.216,
      "step": 17055
    },
    {
      "epoch": 0.3080334250572372,
      "grad_norm": 0.39350172877311707,
      "learning_rate": 1.5671738340774787e-05,
      "loss": 0.3556,
      "step": 17060
    },
    {
      "epoch": 0.3081237044901379,
      "grad_norm": 0.441972553730011,
      "learning_rate": 1.5669402188514228e-05,
      "loss": 0.2697,
      "step": 17065
    },
    {
      "epoch": 0.3082139839230386,
      "grad_norm": 0.5798181295394897,
      "learning_rate": 1.5667065580191482e-05,
      "loss": 0.3265,
      "step": 17070
    },
    {
      "epoch": 0.3083042633559393,
      "grad_norm": 0.4554998576641083,
      "learning_rate": 1.5664728515994514e-05,
      "loss": 0.3389,
      "step": 17075
    },
    {
      "epoch": 0.30839454278884004,
      "grad_norm": 0.3674794137477875,
      "learning_rate": 1.5662390996111322e-05,
      "loss": 0.2932,
      "step": 17080
    },
    {
      "epoch": 0.30848482222174073,
      "grad_norm": 0.6466832756996155,
      "learning_rate": 1.566005302072995e-05,
      "loss": 0.1986,
      "step": 17085
    },
    {
      "epoch": 0.30857510165464147,
      "grad_norm": 1.2485787868499756,
      "learning_rate": 1.565771459003846e-05,
      "loss": 0.2913,
      "step": 17090
    },
    {
      "epoch": 0.30866538108754216,
      "grad_norm": 1.4496780633926392,
      "learning_rate": 1.565537570422497e-05,
      "loss": 0.3014,
      "step": 17095
    },
    {
      "epoch": 0.3087556605204429,
      "grad_norm": 0.415077805519104,
      "learning_rate": 1.565303636347762e-05,
      "loss": 0.257,
      "step": 17100
    },
    {
      "epoch": 0.3088459399533436,
      "grad_norm": 0.35092994570732117,
      "learning_rate": 1.5650696567984595e-05,
      "loss": 0.2626,
      "step": 17105
    },
    {
      "epoch": 0.3089362193862443,
      "grad_norm": 0.5046998858451843,
      "learning_rate": 1.564835631793412e-05,
      "loss": 0.2361,
      "step": 17110
    },
    {
      "epoch": 0.309026498819145,
      "grad_norm": 0.46307122707366943,
      "learning_rate": 1.5646015613514444e-05,
      "loss": 0.249,
      "step": 17115
    },
    {
      "epoch": 0.30911677825204575,
      "grad_norm": 0.6530759930610657,
      "learning_rate": 1.564367445491386e-05,
      "loss": 0.2387,
      "step": 17120
    },
    {
      "epoch": 0.30920705768494644,
      "grad_norm": 0.36529412865638733,
      "learning_rate": 1.5641332842320706e-05,
      "loss": 0.272,
      "step": 17125
    },
    {
      "epoch": 0.3092973371178472,
      "grad_norm": 0.3701333999633789,
      "learning_rate": 1.5638990775923332e-05,
      "loss": 0.2612,
      "step": 17130
    },
    {
      "epoch": 0.30938761655074787,
      "grad_norm": 0.39277517795562744,
      "learning_rate": 1.5636648255910158e-05,
      "loss": 0.2861,
      "step": 17135
    },
    {
      "epoch": 0.3094778959836486,
      "grad_norm": 0.2821442782878876,
      "learning_rate": 1.5634305282469612e-05,
      "loss": 0.2338,
      "step": 17140
    },
    {
      "epoch": 0.3095681754165493,
      "grad_norm": 0.4453014135360718,
      "learning_rate": 1.563196185579017e-05,
      "loss": 0.2903,
      "step": 17145
    },
    {
      "epoch": 0.30965845484945004,
      "grad_norm": 0.4597513675689697,
      "learning_rate": 1.5629617976060345e-05,
      "loss": 0.279,
      "step": 17150
    },
    {
      "epoch": 0.3097487342823507,
      "grad_norm": 0.2701919376850128,
      "learning_rate": 1.5627273643468685e-05,
      "loss": 0.1459,
      "step": 17155
    },
    {
      "epoch": 0.30983901371525147,
      "grad_norm": 0.6496157646179199,
      "learning_rate": 1.5624928858203777e-05,
      "loss": 0.2051,
      "step": 17160
    },
    {
      "epoch": 0.30992929314815215,
      "grad_norm": 0.47751709818840027,
      "learning_rate": 1.5622583620454238e-05,
      "loss": 0.1716,
      "step": 17165
    },
    {
      "epoch": 0.3100195725810529,
      "grad_norm": 0.4866003096103668,
      "learning_rate": 1.5620237930408726e-05,
      "loss": 0.2241,
      "step": 17170
    },
    {
      "epoch": 0.3101098520139536,
      "grad_norm": 0.30454862117767334,
      "learning_rate": 1.5617891788255936e-05,
      "loss": 0.3589,
      "step": 17175
    },
    {
      "epoch": 0.3102001314468543,
      "grad_norm": 0.6282525658607483,
      "learning_rate": 1.56155451941846e-05,
      "loss": 0.3163,
      "step": 17180
    },
    {
      "epoch": 0.310290410879755,
      "grad_norm": 0.4587770700454712,
      "learning_rate": 1.5613198148383482e-05,
      "loss": 0.2203,
      "step": 17185
    },
    {
      "epoch": 0.31038069031265575,
      "grad_norm": 0.5046096444129944,
      "learning_rate": 1.561085065104138e-05,
      "loss": 0.374,
      "step": 17190
    },
    {
      "epoch": 0.31047096974555644,
      "grad_norm": 0.2771824598312378,
      "learning_rate": 1.560850270234714e-05,
      "loss": 0.1991,
      "step": 17195
    },
    {
      "epoch": 0.3105612491784572,
      "grad_norm": 0.4556228518486023,
      "learning_rate": 1.560615430248964e-05,
      "loss": 0.2416,
      "step": 17200
    },
    {
      "epoch": 0.31065152861135786,
      "grad_norm": 0.5162927508354187,
      "learning_rate": 1.560380545165778e-05,
      "loss": 0.2607,
      "step": 17205
    },
    {
      "epoch": 0.3107418080442586,
      "grad_norm": 0.3797825276851654,
      "learning_rate": 1.5601456150040522e-05,
      "loss": 0.1871,
      "step": 17210
    },
    {
      "epoch": 0.3108320874771593,
      "grad_norm": 0.3668143153190613,
      "learning_rate": 1.5599106397826836e-05,
      "loss": 0.1506,
      "step": 17215
    },
    {
      "epoch": 0.31092236691006003,
      "grad_norm": 0.2939102351665497,
      "learning_rate": 1.5596756195205755e-05,
      "loss": 0.2777,
      "step": 17220
    },
    {
      "epoch": 0.3110126463429607,
      "grad_norm": 0.34372422099113464,
      "learning_rate": 1.5594405542366327e-05,
      "loss": 0.2181,
      "step": 17225
    },
    {
      "epoch": 0.31110292577586146,
      "grad_norm": 0.8658391833305359,
      "learning_rate": 1.559205443949765e-05,
      "loss": 0.3776,
      "step": 17230
    },
    {
      "epoch": 0.31119320520876215,
      "grad_norm": 0.34222641587257385,
      "learning_rate": 1.5589702886788855e-05,
      "loss": 0.1955,
      "step": 17235
    },
    {
      "epoch": 0.3112834846416629,
      "grad_norm": 0.4252156913280487,
      "learning_rate": 1.55873508844291e-05,
      "loss": 0.2518,
      "step": 17240
    },
    {
      "epoch": 0.3113737640745636,
      "grad_norm": 1.5674724578857422,
      "learning_rate": 1.558499843260759e-05,
      "loss": 0.2681,
      "step": 17245
    },
    {
      "epoch": 0.3114640435074643,
      "grad_norm": 0.388078898191452,
      "learning_rate": 1.5582645531513563e-05,
      "loss": 0.3618,
      "step": 17250
    },
    {
      "epoch": 0.311554322940365,
      "grad_norm": 0.34436941146850586,
      "learning_rate": 1.5580292181336295e-05,
      "loss": 0.3266,
      "step": 17255
    },
    {
      "epoch": 0.31164460237326574,
      "grad_norm": 0.4869365394115448,
      "learning_rate": 1.557793838226509e-05,
      "loss": 0.2529,
      "step": 17260
    },
    {
      "epoch": 0.31173488180616643,
      "grad_norm": 0.4009968936443329,
      "learning_rate": 1.55755841344893e-05,
      "loss": 0.2321,
      "step": 17265
    },
    {
      "epoch": 0.3118251612390672,
      "grad_norm": 0.4688039720058441,
      "learning_rate": 1.5573229438198304e-05,
      "loss": 0.3298,
      "step": 17270
    },
    {
      "epoch": 0.31191544067196786,
      "grad_norm": 0.44214802980422974,
      "learning_rate": 1.5570874293581524e-05,
      "loss": 0.2518,
      "step": 17275
    },
    {
      "epoch": 0.3120057201048686,
      "grad_norm": 0.6073891520500183,
      "learning_rate": 1.556851870082841e-05,
      "loss": 0.3124,
      "step": 17280
    },
    {
      "epoch": 0.3120959995377693,
      "grad_norm": 0.5940367579460144,
      "learning_rate": 1.5566162660128452e-05,
      "loss": 0.2754,
      "step": 17285
    },
    {
      "epoch": 0.31218627897067003,
      "grad_norm": 0.4338489770889282,
      "learning_rate": 1.5563806171671177e-05,
      "loss": 0.2416,
      "step": 17290
    },
    {
      "epoch": 0.3122765584035707,
      "grad_norm": 0.4148912727832794,
      "learning_rate": 1.556144923564615e-05,
      "loss": 0.3408,
      "step": 17295
    },
    {
      "epoch": 0.31236683783647146,
      "grad_norm": 0.42692506313323975,
      "learning_rate": 1.5559091852242966e-05,
      "loss": 0.2837,
      "step": 17300
    },
    {
      "epoch": 0.31245711726937214,
      "grad_norm": 0.4609665274620056,
      "learning_rate": 1.5556734021651263e-05,
      "loss": 0.2777,
      "step": 17305
    },
    {
      "epoch": 0.3125473967022729,
      "grad_norm": 0.41917040944099426,
      "learning_rate": 1.555437574406071e-05,
      "loss": 0.2286,
      "step": 17310
    },
    {
      "epoch": 0.31263767613517357,
      "grad_norm": 0.8608391284942627,
      "learning_rate": 1.5552017019661013e-05,
      "loss": 0.2612,
      "step": 17315
    },
    {
      "epoch": 0.3127279555680743,
      "grad_norm": 0.49164366722106934,
      "learning_rate": 1.5549657848641907e-05,
      "loss": 0.3164,
      "step": 17320
    },
    {
      "epoch": 0.312818235000975,
      "grad_norm": 0.46910715103149414,
      "learning_rate": 1.5547298231193185e-05,
      "loss": 0.2725,
      "step": 17325
    },
    {
      "epoch": 0.31290851443387574,
      "grad_norm": 0.38566046953201294,
      "learning_rate": 1.554493816750465e-05,
      "loss": 0.2339,
      "step": 17330
    },
    {
      "epoch": 0.3129987938667764,
      "grad_norm": 0.38748466968536377,
      "learning_rate": 1.5542577657766152e-05,
      "loss": 0.2653,
      "step": 17335
    },
    {
      "epoch": 0.31308907329967717,
      "grad_norm": 0.3147033751010895,
      "learning_rate": 1.5540216702167587e-05,
      "loss": 0.3066,
      "step": 17340
    },
    {
      "epoch": 0.31317935273257785,
      "grad_norm": 0.3528516888618469,
      "learning_rate": 1.5537855300898862e-05,
      "loss": 0.3652,
      "step": 17345
    },
    {
      "epoch": 0.3132696321654786,
      "grad_norm": 0.43596094846725464,
      "learning_rate": 1.5535493454149947e-05,
      "loss": 0.3015,
      "step": 17350
    },
    {
      "epoch": 0.3133599115983793,
      "grad_norm": 0.6840492486953735,
      "learning_rate": 1.5533131162110827e-05,
      "loss": 0.1868,
      "step": 17355
    },
    {
      "epoch": 0.31345019103128,
      "grad_norm": 0.38821282982826233,
      "learning_rate": 1.553076842497154e-05,
      "loss": 0.196,
      "step": 17360
    },
    {
      "epoch": 0.3135404704641807,
      "grad_norm": 0.4120749831199646,
      "learning_rate": 1.5528405242922143e-05,
      "loss": 0.3502,
      "step": 17365
    },
    {
      "epoch": 0.31363074989708145,
      "grad_norm": 0.43744704127311707,
      "learning_rate": 1.552604161615274e-05,
      "loss": 0.2365,
      "step": 17370
    },
    {
      "epoch": 0.31372102932998214,
      "grad_norm": 0.31180235743522644,
      "learning_rate": 1.5523677544853464e-05,
      "loss": 0.28,
      "step": 17375
    },
    {
      "epoch": 0.3138113087628829,
      "grad_norm": 0.5090747475624084,
      "learning_rate": 1.5521313029214494e-05,
      "loss": 0.3136,
      "step": 17380
    },
    {
      "epoch": 0.31390158819578357,
      "grad_norm": 0.4170553982257843,
      "learning_rate": 1.5518948069426035e-05,
      "loss": 0.3178,
      "step": 17385
    },
    {
      "epoch": 0.3139918676286843,
      "grad_norm": 0.49591946601867676,
      "learning_rate": 1.5516582665678332e-05,
      "loss": 0.2441,
      "step": 17390
    },
    {
      "epoch": 0.314082147061585,
      "grad_norm": 1.1685659885406494,
      "learning_rate": 1.5514216818161658e-05,
      "loss": 0.3014,
      "step": 17395
    },
    {
      "epoch": 0.31417242649448573,
      "grad_norm": 0.23236806690692902,
      "learning_rate": 1.5511850527066336e-05,
      "loss": 0.3061,
      "step": 17400
    },
    {
      "epoch": 0.3142627059273864,
      "grad_norm": 0.37775471806526184,
      "learning_rate": 1.5509483792582713e-05,
      "loss": 0.2792,
      "step": 17405
    },
    {
      "epoch": 0.31435298536028716,
      "grad_norm": 0.6125426292419434,
      "learning_rate": 1.5507116614901176e-05,
      "loss": 0.2237,
      "step": 17410
    },
    {
      "epoch": 0.31444326479318785,
      "grad_norm": 0.626716673374176,
      "learning_rate": 1.5504748994212147e-05,
      "loss": 0.2518,
      "step": 17415
    },
    {
      "epoch": 0.3145335442260886,
      "grad_norm": 0.5543079972267151,
      "learning_rate": 1.550238093070609e-05,
      "loss": 0.2795,
      "step": 17420
    },
    {
      "epoch": 0.3146238236589893,
      "grad_norm": 0.35237279534339905,
      "learning_rate": 1.550001242457349e-05,
      "loss": 0.2229,
      "step": 17425
    },
    {
      "epoch": 0.31471410309189,
      "grad_norm": 0.5148335695266724,
      "learning_rate": 1.549764347600488e-05,
      "loss": 0.3217,
      "step": 17430
    },
    {
      "epoch": 0.3148043825247907,
      "grad_norm": 0.38885366916656494,
      "learning_rate": 1.5495274085190823e-05,
      "loss": 0.1891,
      "step": 17435
    },
    {
      "epoch": 0.31489466195769145,
      "grad_norm": 0.46232593059539795,
      "learning_rate": 1.549290425232192e-05,
      "loss": 0.358,
      "step": 17440
    },
    {
      "epoch": 0.3149849413905922,
      "grad_norm": 0.40540289878845215,
      "learning_rate": 1.549053397758881e-05,
      "loss": 0.2707,
      "step": 17445
    },
    {
      "epoch": 0.3150752208234929,
      "grad_norm": 0.47470027208328247,
      "learning_rate": 1.548816326118216e-05,
      "loss": 0.2484,
      "step": 17450
    },
    {
      "epoch": 0.3151655002563936,
      "grad_norm": 0.5260505080223083,
      "learning_rate": 1.5485792103292678e-05,
      "loss": 0.195,
      "step": 17455
    },
    {
      "epoch": 0.3152557796892943,
      "grad_norm": 0.4604606330394745,
      "learning_rate": 1.548342050411111e-05,
      "loss": 0.2525,
      "step": 17460
    },
    {
      "epoch": 0.31534605912219504,
      "grad_norm": 0.5449261665344238,
      "learning_rate": 1.5481048463828226e-05,
      "loss": 0.2357,
      "step": 17465
    },
    {
      "epoch": 0.31543633855509573,
      "grad_norm": 0.34216904640197754,
      "learning_rate": 1.5478675982634844e-05,
      "loss": 0.2595,
      "step": 17470
    },
    {
      "epoch": 0.31552661798799647,
      "grad_norm": 0.6009491086006165,
      "learning_rate": 1.5476303060721818e-05,
      "loss": 0.3101,
      "step": 17475
    },
    {
      "epoch": 0.31561689742089716,
      "grad_norm": 0.6484291553497314,
      "learning_rate": 1.5473929698280023e-05,
      "loss": 0.2956,
      "step": 17480
    },
    {
      "epoch": 0.3157071768537979,
      "grad_norm": 0.48652103543281555,
      "learning_rate": 1.5471555895500382e-05,
      "loss": 0.2512,
      "step": 17485
    },
    {
      "epoch": 0.3157974562866986,
      "grad_norm": 0.45876580476760864,
      "learning_rate": 1.5469181652573856e-05,
      "loss": 0.2578,
      "step": 17490
    },
    {
      "epoch": 0.3158877357195993,
      "grad_norm": 0.4625091850757599,
      "learning_rate": 1.5466806969691428e-05,
      "loss": 0.2697,
      "step": 17495
    },
    {
      "epoch": 0.3159780151525,
      "grad_norm": 2.235440492630005,
      "learning_rate": 1.5464431847044127e-05,
      "loss": 0.2677,
      "step": 17500
    },
    {
      "epoch": 0.31606829458540076,
      "grad_norm": 0.9531151652336121,
      "learning_rate": 1.5462056284823016e-05,
      "loss": 0.2773,
      "step": 17505
    },
    {
      "epoch": 0.31615857401830144,
      "grad_norm": 0.6140753030776978,
      "learning_rate": 1.545968028321919e-05,
      "loss": 0.2744,
      "step": 17510
    },
    {
      "epoch": 0.3162488534512022,
      "grad_norm": 0.3872118294239044,
      "learning_rate": 1.5457303842423777e-05,
      "loss": 0.278,
      "step": 17515
    },
    {
      "epoch": 0.31633913288410287,
      "grad_norm": 1.6485217809677124,
      "learning_rate": 1.5454926962627948e-05,
      "loss": 0.2587,
      "step": 17520
    },
    {
      "epoch": 0.3164294123170036,
      "grad_norm": 0.3472902178764343,
      "learning_rate": 1.545254964402291e-05,
      "loss": 0.2899,
      "step": 17525
    },
    {
      "epoch": 0.3165196917499043,
      "grad_norm": 0.3585287630558014,
      "learning_rate": 1.5450171886799898e-05,
      "loss": 0.2406,
      "step": 17530
    },
    {
      "epoch": 0.31660997118280504,
      "grad_norm": 0.5434662699699402,
      "learning_rate": 1.544779369115018e-05,
      "loss": 0.3442,
      "step": 17535
    },
    {
      "epoch": 0.3167002506157057,
      "grad_norm": 0.5494941473007202,
      "learning_rate": 1.5445415057265072e-05,
      "loss": 0.3096,
      "step": 17540
    },
    {
      "epoch": 0.31679053004860647,
      "grad_norm": 0.3780425190925598,
      "learning_rate": 1.5443035985335914e-05,
      "loss": 0.2801,
      "step": 17545
    },
    {
      "epoch": 0.31688080948150715,
      "grad_norm": 0.2784373164176941,
      "learning_rate": 1.5440656475554084e-05,
      "loss": 0.2165,
      "step": 17550
    },
    {
      "epoch": 0.3169710889144079,
      "grad_norm": 0.42795300483703613,
      "learning_rate": 1.5438276528111e-05,
      "loss": 0.3302,
      "step": 17555
    },
    {
      "epoch": 0.3170613683473086,
      "grad_norm": 0.4713800549507141,
      "learning_rate": 1.5435896143198107e-05,
      "loss": 0.2392,
      "step": 17560
    },
    {
      "epoch": 0.3171516477802093,
      "grad_norm": 0.4960048794746399,
      "learning_rate": 1.5433515321006895e-05,
      "loss": 0.3038,
      "step": 17565
    },
    {
      "epoch": 0.31724192721311,
      "grad_norm": 0.275836318731308,
      "learning_rate": 1.543113406172888e-05,
      "loss": 0.1655,
      "step": 17570
    },
    {
      "epoch": 0.31733220664601075,
      "grad_norm": 0.4332798719406128,
      "learning_rate": 1.5428752365555617e-05,
      "loss": 0.3069,
      "step": 17575
    },
    {
      "epoch": 0.31742248607891144,
      "grad_norm": 0.2804764211177826,
      "learning_rate": 1.5426370232678697e-05,
      "loss": 0.2446,
      "step": 17580
    },
    {
      "epoch": 0.3175127655118122,
      "grad_norm": 0.5299254059791565,
      "learning_rate": 1.5423987663289747e-05,
      "loss": 0.2971,
      "step": 17585
    },
    {
      "epoch": 0.31760304494471286,
      "grad_norm": 0.4273294508457184,
      "learning_rate": 1.542160465758042e-05,
      "loss": 0.2914,
      "step": 17590
    },
    {
      "epoch": 0.3176933243776136,
      "grad_norm": 0.35984429717063904,
      "learning_rate": 1.5419221215742423e-05,
      "loss": 0.2972,
      "step": 17595
    },
    {
      "epoch": 0.3177836038105143,
      "grad_norm": 0.5424455404281616,
      "learning_rate": 1.541683733796748e-05,
      "loss": 0.276,
      "step": 17600
    },
    {
      "epoch": 0.31787388324341503,
      "grad_norm": 0.4359292984008789,
      "learning_rate": 1.5414453024447354e-05,
      "loss": 0.3098,
      "step": 17605
    },
    {
      "epoch": 0.3179641626763157,
      "grad_norm": 0.570429265499115,
      "learning_rate": 1.5412068275373854e-05,
      "loss": 0.3422,
      "step": 17610
    },
    {
      "epoch": 0.31805444210921646,
      "grad_norm": 0.43146470189094543,
      "learning_rate": 1.540968309093881e-05,
      "loss": 0.296,
      "step": 17615
    },
    {
      "epoch": 0.31814472154211715,
      "grad_norm": 0.5638782978057861,
      "learning_rate": 1.540729747133409e-05,
      "loss": 0.3522,
      "step": 17620
    },
    {
      "epoch": 0.3182350009750179,
      "grad_norm": 0.31060877442359924,
      "learning_rate": 1.5404911416751607e-05,
      "loss": 0.2625,
      "step": 17625
    },
    {
      "epoch": 0.3183252804079186,
      "grad_norm": 0.5508368015289307,
      "learning_rate": 1.5402524927383295e-05,
      "loss": 0.3542,
      "step": 17630
    },
    {
      "epoch": 0.3184155598408193,
      "grad_norm": 0.6269552707672119,
      "learning_rate": 1.5400138003421135e-05,
      "loss": 0.2714,
      "step": 17635
    },
    {
      "epoch": 0.31850583927372,
      "grad_norm": 0.656037449836731,
      "learning_rate": 1.5397750645057135e-05,
      "loss": 0.3294,
      "step": 17640
    },
    {
      "epoch": 0.31859611870662075,
      "grad_norm": 0.45344042778015137,
      "learning_rate": 1.539536285248334e-05,
      "loss": 0.3146,
      "step": 17645
    },
    {
      "epoch": 0.31868639813952143,
      "grad_norm": 0.4088731110095978,
      "learning_rate": 1.5392974625891834e-05,
      "loss": 0.2511,
      "step": 17650
    },
    {
      "epoch": 0.3187766775724222,
      "grad_norm": 0.44984954595565796,
      "learning_rate": 1.5390585965474727e-05,
      "loss": 0.2528,
      "step": 17655
    },
    {
      "epoch": 0.31886695700532286,
      "grad_norm": 0.35342809557914734,
      "learning_rate": 1.538819687142417e-05,
      "loss": 0.2453,
      "step": 17660
    },
    {
      "epoch": 0.3189572364382236,
      "grad_norm": 0.45350322127342224,
      "learning_rate": 1.5385807343932358e-05,
      "loss": 0.2969,
      "step": 17665
    },
    {
      "epoch": 0.3190475158711243,
      "grad_norm": 0.8516509532928467,
      "learning_rate": 1.53834173831915e-05,
      "loss": 0.2054,
      "step": 17670
    },
    {
      "epoch": 0.31913779530402503,
      "grad_norm": 0.8284174203872681,
      "learning_rate": 1.5381026989393854e-05,
      "loss": 0.3415,
      "step": 17675
    },
    {
      "epoch": 0.3192280747369257,
      "grad_norm": 0.39702773094177246,
      "learning_rate": 1.537863616273171e-05,
      "loss": 0.2738,
      "step": 17680
    },
    {
      "epoch": 0.31931835416982646,
      "grad_norm": 0.32880184054374695,
      "learning_rate": 1.5376244903397397e-05,
      "loss": 0.2103,
      "step": 17685
    },
    {
      "epoch": 0.31940863360272714,
      "grad_norm": 0.3985629975795746,
      "learning_rate": 1.5373853211583267e-05,
      "loss": 0.2858,
      "step": 17690
    },
    {
      "epoch": 0.3194989130356279,
      "grad_norm": 0.397734671831131,
      "learning_rate": 1.537146108748172e-05,
      "loss": 0.3038,
      "step": 17695
    },
    {
      "epoch": 0.31958919246852857,
      "grad_norm": 0.45787420868873596,
      "learning_rate": 1.536906853128518e-05,
      "loss": 0.2162,
      "step": 17700
    },
    {
      "epoch": 0.3196794719014293,
      "grad_norm": 0.32149168848991394,
      "learning_rate": 1.5366675543186117e-05,
      "loss": 0.2645,
      "step": 17705
    },
    {
      "epoch": 0.31976975133433,
      "grad_norm": 0.5299479365348816,
      "learning_rate": 1.5364282123377022e-05,
      "loss": 0.3511,
      "step": 17710
    },
    {
      "epoch": 0.31986003076723074,
      "grad_norm": 0.4254346489906311,
      "learning_rate": 1.5361888272050435e-05,
      "loss": 0.2054,
      "step": 17715
    },
    {
      "epoch": 0.3199503102001314,
      "grad_norm": 0.6408798098564148,
      "learning_rate": 1.5359493989398923e-05,
      "loss": 0.3001,
      "step": 17720
    },
    {
      "epoch": 0.32004058963303217,
      "grad_norm": 0.3622814416885376,
      "learning_rate": 1.5357099275615085e-05,
      "loss": 0.2692,
      "step": 17725
    },
    {
      "epoch": 0.32013086906593285,
      "grad_norm": 0.7824422717094421,
      "learning_rate": 1.535470413089156e-05,
      "loss": 0.2724,
      "step": 17730
    },
    {
      "epoch": 0.3202211484988336,
      "grad_norm": 0.4633120596408844,
      "learning_rate": 1.5352308555421022e-05,
      "loss": 0.2852,
      "step": 17735
    },
    {
      "epoch": 0.3203114279317343,
      "grad_norm": 0.47430092096328735,
      "learning_rate": 1.534991254939617e-05,
      "loss": 0.3123,
      "step": 17740
    },
    {
      "epoch": 0.320401707364635,
      "grad_norm": 0.38343843817710876,
      "learning_rate": 1.5347516113009756e-05,
      "loss": 0.3564,
      "step": 17745
    },
    {
      "epoch": 0.3204919867975357,
      "grad_norm": 0.5076741576194763,
      "learning_rate": 1.534511924645455e-05,
      "loss": 0.2177,
      "step": 17750
    },
    {
      "epoch": 0.32058226623043645,
      "grad_norm": 0.35495322942733765,
      "learning_rate": 1.5342721949923363e-05,
      "loss": 0.3111,
      "step": 17755
    },
    {
      "epoch": 0.32067254566333714,
      "grad_norm": 0.5616178512573242,
      "learning_rate": 1.534032422360904e-05,
      "loss": 0.3554,
      "step": 17760
    },
    {
      "epoch": 0.3207628250962379,
      "grad_norm": 0.5561057329177856,
      "learning_rate": 1.533792606770446e-05,
      "loss": 0.4187,
      "step": 17765
    },
    {
      "epoch": 0.32085310452913857,
      "grad_norm": 0.3615930676460266,
      "learning_rate": 1.5335527482402543e-05,
      "loss": 0.225,
      "step": 17770
    },
    {
      "epoch": 0.3209433839620393,
      "grad_norm": 0.7328411936759949,
      "learning_rate": 1.5333128467896227e-05,
      "loss": 0.281,
      "step": 17775
    },
    {
      "epoch": 0.32103366339494,
      "grad_norm": 0.9044609665870667,
      "learning_rate": 1.5330729024378504e-05,
      "loss": 0.2768,
      "step": 17780
    },
    {
      "epoch": 0.32112394282784074,
      "grad_norm": 0.7998683452606201,
      "learning_rate": 1.532832915204239e-05,
      "loss": 0.2857,
      "step": 17785
    },
    {
      "epoch": 0.3212142222607414,
      "grad_norm": 0.4175611138343811,
      "learning_rate": 1.5325928851080936e-05,
      "loss": 0.2223,
      "step": 17790
    },
    {
      "epoch": 0.32130450169364216,
      "grad_norm": 0.7397657632827759,
      "learning_rate": 1.532352812168723e-05,
      "loss": 0.268,
      "step": 17795
    },
    {
      "epoch": 0.32139478112654285,
      "grad_norm": 0.30127376317977905,
      "learning_rate": 1.5321126964054394e-05,
      "loss": 0.3168,
      "step": 17800
    },
    {
      "epoch": 0.3214850605594436,
      "grad_norm": 0.2943519353866577,
      "learning_rate": 1.531872537837558e-05,
      "loss": 0.2549,
      "step": 17805
    },
    {
      "epoch": 0.3215753399923443,
      "grad_norm": 0.9851113557815552,
      "learning_rate": 1.5316323364843976e-05,
      "loss": 0.1874,
      "step": 17810
    },
    {
      "epoch": 0.321665619425245,
      "grad_norm": 0.6844465136528015,
      "learning_rate": 1.5313920923652817e-05,
      "loss": 0.2449,
      "step": 17815
    },
    {
      "epoch": 0.3217558988581457,
      "grad_norm": 0.48671990633010864,
      "learning_rate": 1.5311518054995352e-05,
      "loss": 0.341,
      "step": 17820
    },
    {
      "epoch": 0.32184617829104645,
      "grad_norm": 0.25424695014953613,
      "learning_rate": 1.530911475906488e-05,
      "loss": 0.3965,
      "step": 17825
    },
    {
      "epoch": 0.32193645772394713,
      "grad_norm": 0.36367541551589966,
      "learning_rate": 1.5306711036054727e-05,
      "loss": 0.3291,
      "step": 17830
    },
    {
      "epoch": 0.3220267371568479,
      "grad_norm": 0.5668572187423706,
      "learning_rate": 1.5304306886158254e-05,
      "loss": 0.174,
      "step": 17835
    },
    {
      "epoch": 0.3221170165897486,
      "grad_norm": 0.3800593912601471,
      "learning_rate": 1.5301902309568854e-05,
      "loss": 0.2872,
      "step": 17840
    },
    {
      "epoch": 0.3222072960226493,
      "grad_norm": 0.4047221541404724,
      "learning_rate": 1.5299497306479966e-05,
      "loss": 0.3493,
      "step": 17845
    },
    {
      "epoch": 0.32229757545555004,
      "grad_norm": 0.32829201221466064,
      "learning_rate": 1.529709187708505e-05,
      "loss": 0.2295,
      "step": 17850
    },
    {
      "epoch": 0.32238785488845073,
      "grad_norm": 0.40165066719055176,
      "learning_rate": 1.5294686021577608e-05,
      "loss": 0.247,
      "step": 17855
    },
    {
      "epoch": 0.3224781343213515,
      "grad_norm": 0.31915032863616943,
      "learning_rate": 1.5292279740151167e-05,
      "loss": 0.2794,
      "step": 17860
    },
    {
      "epoch": 0.32256841375425216,
      "grad_norm": 0.5519475340843201,
      "learning_rate": 1.5289873032999306e-05,
      "loss": 0.268,
      "step": 17865
    },
    {
      "epoch": 0.3226586931871529,
      "grad_norm": 0.49204424023628235,
      "learning_rate": 1.5287465900315616e-05,
      "loss": 0.2668,
      "step": 17870
    },
    {
      "epoch": 0.3227489726200536,
      "grad_norm": 0.3488686680793762,
      "learning_rate": 1.5285058342293738e-05,
      "loss": 0.17,
      "step": 17875
    },
    {
      "epoch": 0.32283925205295433,
      "grad_norm": 0.5201186537742615,
      "learning_rate": 1.528265035912734e-05,
      "loss": 0.2825,
      "step": 17880
    },
    {
      "epoch": 0.322929531485855,
      "grad_norm": 0.3380523920059204,
      "learning_rate": 1.5280241951010137e-05,
      "loss": 0.2673,
      "step": 17885
    },
    {
      "epoch": 0.32301981091875576,
      "grad_norm": 0.4553193151950836,
      "learning_rate": 1.5277833118135853e-05,
      "loss": 0.3188,
      "step": 17890
    },
    {
      "epoch": 0.32311009035165644,
      "grad_norm": 0.49805140495300293,
      "learning_rate": 1.5275423860698268e-05,
      "loss": 0.2471,
      "step": 17895
    },
    {
      "epoch": 0.3232003697845572,
      "grad_norm": 0.4507269859313965,
      "learning_rate": 1.5273014178891196e-05,
      "loss": 0.1797,
      "step": 17900
    },
    {
      "epoch": 0.32329064921745787,
      "grad_norm": 0.5158857107162476,
      "learning_rate": 1.5270604072908466e-05,
      "loss": 0.2785,
      "step": 17905
    },
    {
      "epoch": 0.3233809286503586,
      "grad_norm": 0.4525963366031647,
      "learning_rate": 1.526819354294396e-05,
      "loss": 0.2833,
      "step": 17910
    },
    {
      "epoch": 0.3234712080832593,
      "grad_norm": 0.46041014790534973,
      "learning_rate": 1.5265782589191588e-05,
      "loss": 0.216,
      "step": 17915
    },
    {
      "epoch": 0.32356148751616004,
      "grad_norm": 0.46751755475997925,
      "learning_rate": 1.5263371211845296e-05,
      "loss": 0.2792,
      "step": 17920
    },
    {
      "epoch": 0.3236517669490607,
      "grad_norm": 0.4012288451194763,
      "learning_rate": 1.5260959411099055e-05,
      "loss": 0.339,
      "step": 17925
    },
    {
      "epoch": 0.32374204638196147,
      "grad_norm": 0.391625314950943,
      "learning_rate": 1.525854718714688e-05,
      "loss": 0.2954,
      "step": 17930
    },
    {
      "epoch": 0.32383232581486215,
      "grad_norm": 0.6322211027145386,
      "learning_rate": 1.5256134540182817e-05,
      "loss": 0.2814,
      "step": 17935
    },
    {
      "epoch": 0.3239226052477629,
      "grad_norm": 0.4969169795513153,
      "learning_rate": 1.5253721470400947e-05,
      "loss": 0.387,
      "step": 17940
    },
    {
      "epoch": 0.3240128846806636,
      "grad_norm": 1.0259813070297241,
      "learning_rate": 1.5251307977995383e-05,
      "loss": 0.2878,
      "step": 17945
    },
    {
      "epoch": 0.3241031641135643,
      "grad_norm": 0.7180622816085815,
      "learning_rate": 1.5248894063160271e-05,
      "loss": 0.2033,
      "step": 17950
    },
    {
      "epoch": 0.324193443546465,
      "grad_norm": 0.49255871772766113,
      "learning_rate": 1.5246479726089796e-05,
      "loss": 0.3112,
      "step": 17955
    },
    {
      "epoch": 0.32428372297936575,
      "grad_norm": 0.353839248418808,
      "learning_rate": 1.5244064966978171e-05,
      "loss": 0.2683,
      "step": 17960
    },
    {
      "epoch": 0.32437400241226644,
      "grad_norm": 0.6729487776756287,
      "learning_rate": 1.5241649786019652e-05,
      "loss": 0.2065,
      "step": 17965
    },
    {
      "epoch": 0.3244642818451672,
      "grad_norm": 0.44503656029701233,
      "learning_rate": 1.5239234183408513e-05,
      "loss": 0.2631,
      "step": 17970
    },
    {
      "epoch": 0.32455456127806787,
      "grad_norm": 0.31708043813705444,
      "learning_rate": 1.523681815933908e-05,
      "loss": 0.2517,
      "step": 17975
    },
    {
      "epoch": 0.3246448407109686,
      "grad_norm": 0.4860188364982605,
      "learning_rate": 1.5234401714005702e-05,
      "loss": 0.2687,
      "step": 17980
    },
    {
      "epoch": 0.3247351201438693,
      "grad_norm": 0.4818451702594757,
      "learning_rate": 1.5231984847602764e-05,
      "loss": 0.243,
      "step": 17985
    },
    {
      "epoch": 0.32482539957677004,
      "grad_norm": 0.4228139817714691,
      "learning_rate": 1.522956756032468e-05,
      "loss": 0.1765,
      "step": 17990
    },
    {
      "epoch": 0.3249156790096707,
      "grad_norm": 0.47641831636428833,
      "learning_rate": 1.5227149852365915e-05,
      "loss": 0.2793,
      "step": 17995
    },
    {
      "epoch": 0.32500595844257146,
      "grad_norm": 0.4192675054073334,
      "learning_rate": 1.5224731723920947e-05,
      "loss": 0.2514,
      "step": 18000
    },
    {
      "epoch": 0.32509623787547215,
      "grad_norm": 0.4162924885749817,
      "learning_rate": 1.52223131751843e-05,
      "loss": 0.3272,
      "step": 18005
    },
    {
      "epoch": 0.3251865173083729,
      "grad_norm": 0.43204668164253235,
      "learning_rate": 1.5219894206350528e-05,
      "loss": 0.31,
      "step": 18010
    },
    {
      "epoch": 0.3252767967412736,
      "grad_norm": 0.4375188648700714,
      "learning_rate": 1.5217474817614217e-05,
      "loss": 0.3487,
      "step": 18015
    },
    {
      "epoch": 0.3253670761741743,
      "grad_norm": 0.746900200843811,
      "learning_rate": 1.521505500916999e-05,
      "loss": 0.252,
      "step": 18020
    },
    {
      "epoch": 0.325457355607075,
      "grad_norm": 1.7763670682907104,
      "learning_rate": 1.521263478121251e-05,
      "loss": 0.3135,
      "step": 18025
    },
    {
      "epoch": 0.32554763503997575,
      "grad_norm": 0.5207912921905518,
      "learning_rate": 1.5210214133936458e-05,
      "loss": 0.2825,
      "step": 18030
    },
    {
      "epoch": 0.32563791447287643,
      "grad_norm": 0.4232496917247772,
      "learning_rate": 1.520779306753656e-05,
      "loss": 0.3172,
      "step": 18035
    },
    {
      "epoch": 0.3257281939057772,
      "grad_norm": 0.47740769386291504,
      "learning_rate": 1.5205371582207577e-05,
      "loss": 0.3233,
      "step": 18040
    },
    {
      "epoch": 0.32581847333867786,
      "grad_norm": 0.3207866847515106,
      "learning_rate": 1.5202949678144297e-05,
      "loss": 0.2103,
      "step": 18045
    },
    {
      "epoch": 0.3259087527715786,
      "grad_norm": 0.5160874128341675,
      "learning_rate": 1.5200527355541543e-05,
      "loss": 0.3405,
      "step": 18050
    },
    {
      "epoch": 0.3259990322044793,
      "grad_norm": 0.47352078557014465,
      "learning_rate": 1.5198104614594174e-05,
      "loss": 0.2959,
      "step": 18055
    },
    {
      "epoch": 0.32608931163738003,
      "grad_norm": 0.28415295481681824,
      "learning_rate": 1.5195681455497084e-05,
      "loss": 0.3159,
      "step": 18060
    },
    {
      "epoch": 0.3261795910702807,
      "grad_norm": 0.3160787522792816,
      "learning_rate": 1.5193257878445198e-05,
      "loss": 0.2751,
      "step": 18065
    },
    {
      "epoch": 0.32626987050318146,
      "grad_norm": 0.3366694748401642,
      "learning_rate": 1.5190833883633472e-05,
      "loss": 0.2634,
      "step": 18070
    },
    {
      "epoch": 0.32636014993608214,
      "grad_norm": 0.49434515833854675,
      "learning_rate": 1.5188409471256902e-05,
      "loss": 0.2425,
      "step": 18075
    },
    {
      "epoch": 0.3264504293689829,
      "grad_norm": 0.8601022958755493,
      "learning_rate": 1.5185984641510514e-05,
      "loss": 0.309,
      "step": 18080
    },
    {
      "epoch": 0.3265407088018836,
      "grad_norm": 0.48108533024787903,
      "learning_rate": 1.5183559394589366e-05,
      "loss": 0.3005,
      "step": 18085
    },
    {
      "epoch": 0.3266309882347843,
      "grad_norm": 0.4106110632419586,
      "learning_rate": 1.5181133730688553e-05,
      "loss": 0.404,
      "step": 18090
    },
    {
      "epoch": 0.326721267667685,
      "grad_norm": 0.4805768132209778,
      "learning_rate": 1.5178707650003204e-05,
      "loss": 0.3485,
      "step": 18095
    },
    {
      "epoch": 0.32681154710058574,
      "grad_norm": 0.308956503868103,
      "learning_rate": 1.5176281152728479e-05,
      "loss": 0.2114,
      "step": 18100
    },
    {
      "epoch": 0.32690182653348643,
      "grad_norm": 0.26499325037002563,
      "learning_rate": 1.5173854239059565e-05,
      "loss": 0.2471,
      "step": 18105
    },
    {
      "epoch": 0.32699210596638717,
      "grad_norm": 0.3078134059906006,
      "learning_rate": 1.5171426909191697e-05,
      "loss": 0.2572,
      "step": 18110
    },
    {
      "epoch": 0.32708238539928786,
      "grad_norm": 0.33887648582458496,
      "learning_rate": 1.5168999163320135e-05,
      "loss": 0.2228,
      "step": 18115
    },
    {
      "epoch": 0.3271726648321886,
      "grad_norm": 0.287139356136322,
      "learning_rate": 1.5166571001640171e-05,
      "loss": 0.1838,
      "step": 18120
    },
    {
      "epoch": 0.3272629442650893,
      "grad_norm": 1.0340220928192139,
      "learning_rate": 1.5164142424347136e-05,
      "loss": 0.2525,
      "step": 18125
    },
    {
      "epoch": 0.32735322369799,
      "grad_norm": 0.3287242650985718,
      "learning_rate": 1.5161713431636389e-05,
      "loss": 0.2641,
      "step": 18130
    },
    {
      "epoch": 0.3274435031308907,
      "grad_norm": 0.4560239613056183,
      "learning_rate": 1.5159284023703326e-05,
      "loss": 0.2862,
      "step": 18135
    },
    {
      "epoch": 0.32753378256379145,
      "grad_norm": 0.5136450529098511,
      "learning_rate": 1.5156854200743376e-05,
      "loss": 0.3416,
      "step": 18140
    },
    {
      "epoch": 0.32762406199669214,
      "grad_norm": 0.7004572153091431,
      "learning_rate": 1.5154423962951998e-05,
      "loss": 0.2566,
      "step": 18145
    },
    {
      "epoch": 0.3277143414295929,
      "grad_norm": 0.5109156966209412,
      "learning_rate": 1.5151993310524692e-05,
      "loss": 0.3552,
      "step": 18150
    },
    {
      "epoch": 0.32780462086249357,
      "grad_norm": 0.3315854072570801,
      "learning_rate": 1.5149562243656977e-05,
      "loss": 0.3227,
      "step": 18155
    },
    {
      "epoch": 0.3278949002953943,
      "grad_norm": 0.4538090229034424,
      "learning_rate": 1.5147130762544424e-05,
      "loss": 0.3818,
      "step": 18160
    },
    {
      "epoch": 0.327985179728295,
      "grad_norm": 0.46064648032188416,
      "learning_rate": 1.5144698867382626e-05,
      "loss": 0.2542,
      "step": 18165
    },
    {
      "epoch": 0.32807545916119574,
      "grad_norm": 0.41564473509788513,
      "learning_rate": 1.5142266558367205e-05,
      "loss": 0.2961,
      "step": 18170
    },
    {
      "epoch": 0.3281657385940964,
      "grad_norm": 0.3369181752204895,
      "learning_rate": 1.513983383569383e-05,
      "loss": 0.2766,
      "step": 18175
    },
    {
      "epoch": 0.32825601802699717,
      "grad_norm": 0.5861334204673767,
      "learning_rate": 1.5137400699558194e-05,
      "loss": 0.2411,
      "step": 18180
    },
    {
      "epoch": 0.32834629745989785,
      "grad_norm": 0.6709774136543274,
      "learning_rate": 1.5134967150156024e-05,
      "loss": 0.3143,
      "step": 18185
    },
    {
      "epoch": 0.3284365768927986,
      "grad_norm": 0.3815455734729767,
      "learning_rate": 1.513253318768308e-05,
      "loss": 0.2594,
      "step": 18190
    },
    {
      "epoch": 0.3285268563256993,
      "grad_norm": 0.5351514220237732,
      "learning_rate": 1.513009881233516e-05,
      "loss": 0.253,
      "step": 18195
    },
    {
      "epoch": 0.3286171357586,
      "grad_norm": 0.4658789038658142,
      "learning_rate": 1.5127664024308086e-05,
      "loss": 0.1936,
      "step": 18200
    },
    {
      "epoch": 0.3287074151915007,
      "grad_norm": 0.2493811547756195,
      "learning_rate": 1.5125228823797726e-05,
      "loss": 0.1633,
      "step": 18205
    },
    {
      "epoch": 0.32879769462440145,
      "grad_norm": 0.4948270320892334,
      "learning_rate": 1.5122793210999972e-05,
      "loss": 0.2486,
      "step": 18210
    },
    {
      "epoch": 0.32888797405730213,
      "grad_norm": 0.4731934368610382,
      "learning_rate": 1.5120357186110747e-05,
      "loss": 0.2792,
      "step": 18215
    },
    {
      "epoch": 0.3289782534902029,
      "grad_norm": 0.4198252558708191,
      "learning_rate": 1.5117920749326019e-05,
      "loss": 0.281,
      "step": 18220
    },
    {
      "epoch": 0.32906853292310356,
      "grad_norm": 0.4379923641681671,
      "learning_rate": 1.5115483900841774e-05,
      "loss": 0.2259,
      "step": 18225
    },
    {
      "epoch": 0.3291588123560043,
      "grad_norm": 0.4433564841747284,
      "learning_rate": 1.5113046640854045e-05,
      "loss": 0.2752,
      "step": 18230
    },
    {
      "epoch": 0.32924909178890505,
      "grad_norm": 0.33808523416519165,
      "learning_rate": 1.5110608969558888e-05,
      "loss": 0.1777,
      "step": 18235
    },
    {
      "epoch": 0.32933937122180573,
      "grad_norm": 0.2609556317329407,
      "learning_rate": 1.51081708871524e-05,
      "loss": 0.2747,
      "step": 18240
    },
    {
      "epoch": 0.3294296506547065,
      "grad_norm": 1.2248449325561523,
      "learning_rate": 1.51057323938307e-05,
      "loss": 0.3077,
      "step": 18245
    },
    {
      "epoch": 0.32951993008760716,
      "grad_norm": 0.5488759279251099,
      "learning_rate": 1.5103293489789953e-05,
      "loss": 0.3706,
      "step": 18250
    },
    {
      "epoch": 0.3296102095205079,
      "grad_norm": 0.6112303137779236,
      "learning_rate": 1.510085417522635e-05,
      "loss": 0.2726,
      "step": 18255
    },
    {
      "epoch": 0.3297004889534086,
      "grad_norm": 0.45051974058151245,
      "learning_rate": 1.5098414450336114e-05,
      "loss": 0.1565,
      "step": 18260
    },
    {
      "epoch": 0.32979076838630933,
      "grad_norm": 0.6409130096435547,
      "learning_rate": 1.5095974315315505e-05,
      "loss": 0.2526,
      "step": 18265
    },
    {
      "epoch": 0.32988104781921,
      "grad_norm": 0.4389084279537201,
      "learning_rate": 1.5093533770360813e-05,
      "loss": 0.3013,
      "step": 18270
    },
    {
      "epoch": 0.32997132725211076,
      "grad_norm": 0.4817896783351898,
      "learning_rate": 1.5091092815668362e-05,
      "loss": 0.2014,
      "step": 18275
    },
    {
      "epoch": 0.33006160668501144,
      "grad_norm": 0.4563262462615967,
      "learning_rate": 1.5088651451434515e-05,
      "loss": 0.2679,
      "step": 18280
    },
    {
      "epoch": 0.3301518861179122,
      "grad_norm": 0.5780072212219238,
      "learning_rate": 1.5086209677855652e-05,
      "loss": 0.2654,
      "step": 18285
    },
    {
      "epoch": 0.33024216555081287,
      "grad_norm": 0.6199705600738525,
      "learning_rate": 1.50837674951282e-05,
      "loss": 0.2326,
      "step": 18290
    },
    {
      "epoch": 0.3303324449837136,
      "grad_norm": 0.9827958941459656,
      "learning_rate": 1.508132490344862e-05,
      "loss": 0.3321,
      "step": 18295
    },
    {
      "epoch": 0.3304227244166143,
      "grad_norm": 0.4238829016685486,
      "learning_rate": 1.5078881903013392e-05,
      "loss": 0.2209,
      "step": 18300
    },
    {
      "epoch": 0.33051300384951504,
      "grad_norm": 0.459552139043808,
      "learning_rate": 1.5076438494019043e-05,
      "loss": 0.3734,
      "step": 18305
    },
    {
      "epoch": 0.3306032832824157,
      "grad_norm": 0.40297162532806396,
      "learning_rate": 1.507399467666213e-05,
      "loss": 0.2856,
      "step": 18310
    },
    {
      "epoch": 0.33069356271531647,
      "grad_norm": 0.4808918237686157,
      "learning_rate": 1.5071550451139234e-05,
      "loss": 0.2648,
      "step": 18315
    },
    {
      "epoch": 0.33078384214821716,
      "grad_norm": 0.5772596597671509,
      "learning_rate": 1.5069105817646977e-05,
      "loss": 0.2349,
      "step": 18320
    },
    {
      "epoch": 0.3308741215811179,
      "grad_norm": 0.407766193151474,
      "learning_rate": 1.5066660776382014e-05,
      "loss": 0.3459,
      "step": 18325
    },
    {
      "epoch": 0.3309644010140186,
      "grad_norm": 0.46875712275505066,
      "learning_rate": 1.506421532754103e-05,
      "loss": 0.2846,
      "step": 18330
    },
    {
      "epoch": 0.3310546804469193,
      "grad_norm": 0.531537652015686,
      "learning_rate": 1.5061769471320742e-05,
      "loss": 0.1981,
      "step": 18335
    },
    {
      "epoch": 0.33114495987982,
      "grad_norm": 1.2391175031661987,
      "learning_rate": 1.5059323207917902e-05,
      "loss": 0.2979,
      "step": 18340
    },
    {
      "epoch": 0.33123523931272075,
      "grad_norm": 0.3438813388347626,
      "learning_rate": 1.5056876537529296e-05,
      "loss": 0.3106,
      "step": 18345
    },
    {
      "epoch": 0.33132551874562144,
      "grad_norm": 0.8342446088790894,
      "learning_rate": 1.5054429460351743e-05,
      "loss": 0.3556,
      "step": 18350
    },
    {
      "epoch": 0.3314157981785222,
      "grad_norm": 0.5235251784324646,
      "learning_rate": 1.5051981976582082e-05,
      "loss": 0.2515,
      "step": 18355
    },
    {
      "epoch": 0.33150607761142287,
      "grad_norm": 0.4574722647666931,
      "learning_rate": 1.5049534086417208e-05,
      "loss": 0.3817,
      "step": 18360
    },
    {
      "epoch": 0.3315963570443236,
      "grad_norm": 0.48444637656211853,
      "learning_rate": 1.5047085790054028e-05,
      "loss": 0.3425,
      "step": 18365
    },
    {
      "epoch": 0.3316866364772243,
      "grad_norm": 0.4341820478439331,
      "learning_rate": 1.5044637087689494e-05,
      "loss": 0.3034,
      "step": 18370
    },
    {
      "epoch": 0.33177691591012504,
      "grad_norm": 0.5475265383720398,
      "learning_rate": 1.5042187979520583e-05,
      "loss": 0.3048,
      "step": 18375
    },
    {
      "epoch": 0.3318671953430257,
      "grad_norm": 1.0465278625488281,
      "learning_rate": 1.5039738465744312e-05,
      "loss": 0.2356,
      "step": 18380
    },
    {
      "epoch": 0.33195747477592646,
      "grad_norm": 0.5055349469184875,
      "learning_rate": 1.5037288546557718e-05,
      "loss": 0.2453,
      "step": 18385
    },
    {
      "epoch": 0.33204775420882715,
      "grad_norm": 0.4561009109020233,
      "learning_rate": 1.503483822215789e-05,
      "loss": 0.25,
      "step": 18390
    },
    {
      "epoch": 0.3321380336417279,
      "grad_norm": 0.3330610394477844,
      "learning_rate": 1.5032387492741932e-05,
      "loss": 0.329,
      "step": 18395
    },
    {
      "epoch": 0.3322283130746286,
      "grad_norm": 0.4629840850830078,
      "learning_rate": 1.502993635850699e-05,
      "loss": 0.2658,
      "step": 18400
    },
    {
      "epoch": 0.3323185925075293,
      "grad_norm": 0.5670529007911682,
      "learning_rate": 1.5027484819650238e-05,
      "loss": 0.2284,
      "step": 18405
    },
    {
      "epoch": 0.33240887194043,
      "grad_norm": 0.4390983283519745,
      "learning_rate": 1.5025032876368885e-05,
      "loss": 0.3765,
      "step": 18410
    },
    {
      "epoch": 0.33249915137333075,
      "grad_norm": 0.3989102840423584,
      "learning_rate": 1.5022580528860175e-05,
      "loss": 0.3113,
      "step": 18415
    },
    {
      "epoch": 0.33258943080623143,
      "grad_norm": 0.36408400535583496,
      "learning_rate": 1.5020127777321376e-05,
      "loss": 0.3607,
      "step": 18420
    },
    {
      "epoch": 0.3326797102391322,
      "grad_norm": 0.43424683809280396,
      "learning_rate": 1.50176746219498e-05,
      "loss": 0.234,
      "step": 18425
    },
    {
      "epoch": 0.33276998967203286,
      "grad_norm": 0.8242179751396179,
      "learning_rate": 1.5015221062942783e-05,
      "loss": 0.1775,
      "step": 18430
    },
    {
      "epoch": 0.3328602691049336,
      "grad_norm": 0.42394378781318665,
      "learning_rate": 1.5012767100497692e-05,
      "loss": 0.2686,
      "step": 18435
    },
    {
      "epoch": 0.3329505485378343,
      "grad_norm": 0.475367933511734,
      "learning_rate": 1.5010312734811938e-05,
      "loss": 0.2521,
      "step": 18440
    },
    {
      "epoch": 0.33304082797073503,
      "grad_norm": 0.5788323283195496,
      "learning_rate": 1.5007857966082953e-05,
      "loss": 0.2721,
      "step": 18445
    },
    {
      "epoch": 0.3331311074036357,
      "grad_norm": 0.41820040345191956,
      "learning_rate": 1.5005402794508203e-05,
      "loss": 0.3111,
      "step": 18450
    },
    {
      "epoch": 0.33322138683653646,
      "grad_norm": 0.5704630017280579,
      "learning_rate": 1.500294722028519e-05,
      "loss": 0.2769,
      "step": 18455
    },
    {
      "epoch": 0.33331166626943715,
      "grad_norm": 0.3807762563228607,
      "learning_rate": 1.5000491243611453e-05,
      "loss": 0.2532,
      "step": 18460
    },
    {
      "epoch": 0.3334019457023379,
      "grad_norm": 0.6831206679344177,
      "learning_rate": 1.4998034864684547e-05,
      "loss": 0.2033,
      "step": 18465
    },
    {
      "epoch": 0.3334922251352386,
      "grad_norm": 0.427998811006546,
      "learning_rate": 1.499557808370208e-05,
      "loss": 0.2237,
      "step": 18470
    },
    {
      "epoch": 0.3335825045681393,
      "grad_norm": 0.44406625628471375,
      "learning_rate": 1.4993120900861674e-05,
      "loss": 0.2092,
      "step": 18475
    },
    {
      "epoch": 0.33367278400104,
      "grad_norm": 1.016243815422058,
      "learning_rate": 1.4990663316360998e-05,
      "loss": 0.2789,
      "step": 18480
    },
    {
      "epoch": 0.33376306343394074,
      "grad_norm": 0.7131323218345642,
      "learning_rate": 1.498820533039774e-05,
      "loss": 0.2929,
      "step": 18485
    },
    {
      "epoch": 0.33385334286684143,
      "grad_norm": 0.36451780796051025,
      "learning_rate": 1.4985746943169634e-05,
      "loss": 0.3113,
      "step": 18490
    },
    {
      "epoch": 0.33394362229974217,
      "grad_norm": 0.5136528611183167,
      "learning_rate": 1.4983288154874436e-05,
      "loss": 0.1657,
      "step": 18495
    },
    {
      "epoch": 0.33403390173264286,
      "grad_norm": 0.46431905031204224,
      "learning_rate": 1.4980828965709938e-05,
      "loss": 0.2786,
      "step": 18500
    },
    {
      "epoch": 0.3341241811655436,
      "grad_norm": 0.3372277021408081,
      "learning_rate": 1.4978369375873963e-05,
      "loss": 0.2844,
      "step": 18505
    },
    {
      "epoch": 0.3342144605984443,
      "grad_norm": 0.4561137557029724,
      "learning_rate": 1.4975909385564366e-05,
      "loss": 0.2242,
      "step": 18510
    },
    {
      "epoch": 0.334304740031345,
      "grad_norm": 0.6487745046615601,
      "learning_rate": 1.4973448994979041e-05,
      "loss": 0.2536,
      "step": 18515
    },
    {
      "epoch": 0.3343950194642457,
      "grad_norm": 0.7460372447967529,
      "learning_rate": 1.4970988204315902e-05,
      "loss": 0.2419,
      "step": 18520
    },
    {
      "epoch": 0.33448529889714645,
      "grad_norm": 0.3725984990596771,
      "learning_rate": 1.4968527013772906e-05,
      "loss": 0.2879,
      "step": 18525
    },
    {
      "epoch": 0.33457557833004714,
      "grad_norm": 0.4160342216491699,
      "learning_rate": 1.4966065423548037e-05,
      "loss": 0.3512,
      "step": 18530
    },
    {
      "epoch": 0.3346658577629479,
      "grad_norm": 0.5152134895324707,
      "learning_rate": 1.4963603433839308e-05,
      "loss": 0.3255,
      "step": 18535
    },
    {
      "epoch": 0.33475613719584857,
      "grad_norm": 0.4288351535797119,
      "learning_rate": 1.4961141044844776e-05,
      "loss": 0.253,
      "step": 18540
    },
    {
      "epoch": 0.3348464166287493,
      "grad_norm": 0.4910518229007721,
      "learning_rate": 1.4958678256762515e-05,
      "loss": 0.1859,
      "step": 18545
    },
    {
      "epoch": 0.33493669606165,
      "grad_norm": 0.5223522782325745,
      "learning_rate": 1.4956215069790645e-05,
      "loss": 0.3078,
      "step": 18550
    },
    {
      "epoch": 0.33502697549455074,
      "grad_norm": 0.4163327217102051,
      "learning_rate": 1.4953751484127304e-05,
      "loss": 0.294,
      "step": 18555
    },
    {
      "epoch": 0.3351172549274514,
      "grad_norm": 0.45471954345703125,
      "learning_rate": 1.4951287499970677e-05,
      "loss": 0.1941,
      "step": 18560
    },
    {
      "epoch": 0.33520753436035217,
      "grad_norm": 0.564900815486908,
      "learning_rate": 1.4948823117518967e-05,
      "loss": 0.2978,
      "step": 18565
    },
    {
      "epoch": 0.33529781379325285,
      "grad_norm": 0.6309841275215149,
      "learning_rate": 1.494635833697042e-05,
      "loss": 0.2194,
      "step": 18570
    },
    {
      "epoch": 0.3353880932261536,
      "grad_norm": 0.39717692136764526,
      "learning_rate": 1.4943893158523306e-05,
      "loss": 0.2625,
      "step": 18575
    },
    {
      "epoch": 0.3354783726590543,
      "grad_norm": 0.5182188749313354,
      "learning_rate": 1.4941427582375935e-05,
      "loss": 0.4234,
      "step": 18580
    },
    {
      "epoch": 0.335568652091955,
      "grad_norm": 0.43272313475608826,
      "learning_rate": 1.4938961608726642e-05,
      "loss": 0.2652,
      "step": 18585
    },
    {
      "epoch": 0.3356589315248557,
      "grad_norm": 1.8120278120040894,
      "learning_rate": 1.4936495237773798e-05,
      "loss": 0.2911,
      "step": 18590
    },
    {
      "epoch": 0.33574921095775645,
      "grad_norm": 0.6251808404922485,
      "learning_rate": 1.4934028469715805e-05,
      "loss": 0.2554,
      "step": 18595
    },
    {
      "epoch": 0.33583949039065714,
      "grad_norm": 0.3555147647857666,
      "learning_rate": 1.493156130475109e-05,
      "loss": 0.2023,
      "step": 18600
    },
    {
      "epoch": 0.3359297698235579,
      "grad_norm": 0.4704226851463318,
      "learning_rate": 1.4929093743078128e-05,
      "loss": 0.2813,
      "step": 18605
    },
    {
      "epoch": 0.33602004925645856,
      "grad_norm": 0.6650073528289795,
      "learning_rate": 1.4926625784895414e-05,
      "loss": 0.3789,
      "step": 18610
    },
    {
      "epoch": 0.3361103286893593,
      "grad_norm": 0.47177284955978394,
      "learning_rate": 1.4924157430401473e-05,
      "loss": 0.3072,
      "step": 18615
    },
    {
      "epoch": 0.33620060812226,
      "grad_norm": 0.5198081731796265,
      "learning_rate": 1.4921688679794871e-05,
      "loss": 0.2936,
      "step": 18620
    },
    {
      "epoch": 0.33629088755516073,
      "grad_norm": 0.36302539706230164,
      "learning_rate": 1.4919219533274196e-05,
      "loss": 0.1642,
      "step": 18625
    },
    {
      "epoch": 0.3363811669880615,
      "grad_norm": 0.3527725338935852,
      "learning_rate": 1.4916749991038077e-05,
      "loss": 0.1546,
      "step": 18630
    },
    {
      "epoch": 0.33647144642096216,
      "grad_norm": 1.1330320835113525,
      "learning_rate": 1.4914280053285168e-05,
      "loss": 0.2461,
      "step": 18635
    },
    {
      "epoch": 0.3365617258538629,
      "grad_norm": 0.4595145881175995,
      "learning_rate": 1.4911809720214163e-05,
      "loss": 0.207,
      "step": 18640
    },
    {
      "epoch": 0.3366520052867636,
      "grad_norm": 0.31736651062965393,
      "learning_rate": 1.4909338992023777e-05,
      "loss": 0.2515,
      "step": 18645
    },
    {
      "epoch": 0.33674228471966433,
      "grad_norm": 0.4406778812408447,
      "learning_rate": 1.4906867868912763e-05,
      "loss": 0.2084,
      "step": 18650
    },
    {
      "epoch": 0.336832564152565,
      "grad_norm": 0.3526134192943573,
      "learning_rate": 1.4904396351079903e-05,
      "loss": 0.271,
      "step": 18655
    },
    {
      "epoch": 0.33692284358546576,
      "grad_norm": 0.4426441490650177,
      "learning_rate": 1.4901924438724018e-05,
      "loss": 0.292,
      "step": 18660
    },
    {
      "epoch": 0.33701312301836645,
      "grad_norm": 0.5876598954200745,
      "learning_rate": 1.489945213204395e-05,
      "loss": 0.3156,
      "step": 18665
    },
    {
      "epoch": 0.3371034024512672,
      "grad_norm": 0.5972442030906677,
      "learning_rate": 1.4896979431238582e-05,
      "loss": 0.3115,
      "step": 18670
    },
    {
      "epoch": 0.3371936818841679,
      "grad_norm": 0.4676417112350464,
      "learning_rate": 1.4894506336506823e-05,
      "loss": 0.2342,
      "step": 18675
    },
    {
      "epoch": 0.3372839613170686,
      "grad_norm": 0.332589715719223,
      "learning_rate": 1.4892032848047615e-05,
      "loss": 0.3179,
      "step": 18680
    },
    {
      "epoch": 0.3373742407499693,
      "grad_norm": 0.46008965373039246,
      "learning_rate": 1.4889558966059934e-05,
      "loss": 0.2315,
      "step": 18685
    },
    {
      "epoch": 0.33746452018287004,
      "grad_norm": 0.29239562153816223,
      "learning_rate": 1.4887084690742783e-05,
      "loss": 0.267,
      "step": 18690
    },
    {
      "epoch": 0.33755479961577073,
      "grad_norm": 0.41400787234306335,
      "learning_rate": 1.4884610022295204e-05,
      "loss": 0.3701,
      "step": 18695
    },
    {
      "epoch": 0.33764507904867147,
      "grad_norm": 0.31794947385787964,
      "learning_rate": 1.4882134960916258e-05,
      "loss": 0.1902,
      "step": 18700
    },
    {
      "epoch": 0.33773535848157216,
      "grad_norm": 0.37944456934928894,
      "learning_rate": 1.4879659506805056e-05,
      "loss": 0.3004,
      "step": 18705
    },
    {
      "epoch": 0.3378256379144729,
      "grad_norm": 0.5071249008178711,
      "learning_rate": 1.4877183660160724e-05,
      "loss": 0.2749,
      "step": 18710
    },
    {
      "epoch": 0.3379159173473736,
      "grad_norm": 0.43339505791664124,
      "learning_rate": 1.4874707421182426e-05,
      "loss": 0.214,
      "step": 18715
    },
    {
      "epoch": 0.3380061967802743,
      "grad_norm": 0.31769007444381714,
      "learning_rate": 1.487223079006936e-05,
      "loss": 0.295,
      "step": 18720
    },
    {
      "epoch": 0.338096476213175,
      "grad_norm": 0.7044994831085205,
      "learning_rate": 1.4869753767020749e-05,
      "loss": 0.3715,
      "step": 18725
    },
    {
      "epoch": 0.33818675564607575,
      "grad_norm": 0.3921722173690796,
      "learning_rate": 1.4867276352235858e-05,
      "loss": 0.3381,
      "step": 18730
    },
    {
      "epoch": 0.33827703507897644,
      "grad_norm": 0.5172808766365051,
      "learning_rate": 1.4864798545913968e-05,
      "loss": 0.2997,
      "step": 18735
    },
    {
      "epoch": 0.3383673145118772,
      "grad_norm": 0.49731895327568054,
      "learning_rate": 1.486232034825441e-05,
      "loss": 0.2703,
      "step": 18740
    },
    {
      "epoch": 0.33845759394477787,
      "grad_norm": 0.3825044631958008,
      "learning_rate": 1.4859841759456532e-05,
      "loss": 0.3019,
      "step": 18745
    },
    {
      "epoch": 0.3385478733776786,
      "grad_norm": 0.4261285662651062,
      "learning_rate": 1.4857362779719715e-05,
      "loss": 0.27,
      "step": 18750
    },
    {
      "epoch": 0.3386381528105793,
      "grad_norm": 0.34586456418037415,
      "learning_rate": 1.4854883409243383e-05,
      "loss": 0.1818,
      "step": 18755
    },
    {
      "epoch": 0.33872843224348004,
      "grad_norm": 0.396688848733902,
      "learning_rate": 1.4852403648226976e-05,
      "loss": 0.2738,
      "step": 18760
    },
    {
      "epoch": 0.3388187116763807,
      "grad_norm": 0.6118276715278625,
      "learning_rate": 1.4849923496869979e-05,
      "loss": 0.2463,
      "step": 18765
    },
    {
      "epoch": 0.33890899110928147,
      "grad_norm": 0.3033411502838135,
      "learning_rate": 1.4847442955371896e-05,
      "loss": 0.26,
      "step": 18770
    },
    {
      "epoch": 0.33899927054218215,
      "grad_norm": 0.5740275979042053,
      "learning_rate": 1.4844962023932274e-05,
      "loss": 0.2029,
      "step": 18775
    },
    {
      "epoch": 0.3390895499750829,
      "grad_norm": 0.4284028112888336,
      "learning_rate": 1.4842480702750684e-05,
      "loss": 0.3111,
      "step": 18780
    },
    {
      "epoch": 0.3391798294079836,
      "grad_norm": 0.6427981853485107,
      "learning_rate": 1.4839998992026726e-05,
      "loss": 0.1807,
      "step": 18785
    },
    {
      "epoch": 0.3392701088408843,
      "grad_norm": 0.36368077993392944,
      "learning_rate": 1.4837516891960043e-05,
      "loss": 0.2607,
      "step": 18790
    },
    {
      "epoch": 0.339360388273785,
      "grad_norm": 0.4679321050643921,
      "learning_rate": 1.4835034402750294e-05,
      "loss": 0.2857,
      "step": 18795
    },
    {
      "epoch": 0.33945066770668575,
      "grad_norm": 0.5633551478385925,
      "learning_rate": 1.4832551524597183e-05,
      "loss": 0.3887,
      "step": 18800
    },
    {
      "epoch": 0.33954094713958644,
      "grad_norm": 0.6296786665916443,
      "learning_rate": 1.4830068257700441e-05,
      "loss": 0.3002,
      "step": 18805
    },
    {
      "epoch": 0.3396312265724872,
      "grad_norm": 0.5864987969398499,
      "learning_rate": 1.4827584602259824e-05,
      "loss": 0.2391,
      "step": 18810
    },
    {
      "epoch": 0.33972150600538786,
      "grad_norm": 0.36309221386909485,
      "learning_rate": 1.4825100558475122e-05,
      "loss": 0.3185,
      "step": 18815
    },
    {
      "epoch": 0.3398117854382886,
      "grad_norm": 0.4260045886039734,
      "learning_rate": 1.4822616126546168e-05,
      "loss": 0.2521,
      "step": 18820
    },
    {
      "epoch": 0.3399020648711893,
      "grad_norm": 0.5944346189498901,
      "learning_rate": 1.4820131306672807e-05,
      "loss": 0.3321,
      "step": 18825
    },
    {
      "epoch": 0.33999234430409003,
      "grad_norm": 0.3762878179550171,
      "learning_rate": 1.4817646099054929e-05,
      "loss": 0.2522,
      "step": 18830
    },
    {
      "epoch": 0.3400826237369907,
      "grad_norm": 0.2695557177066803,
      "learning_rate": 1.4815160503892449e-05,
      "loss": 0.2246,
      "step": 18835
    },
    {
      "epoch": 0.34017290316989146,
      "grad_norm": 0.4036068022251129,
      "learning_rate": 1.4812674521385317e-05,
      "loss": 0.2663,
      "step": 18840
    },
    {
      "epoch": 0.34026318260279215,
      "grad_norm": 0.36277270317077637,
      "learning_rate": 1.4810188151733513e-05,
      "loss": 0.2648,
      "step": 18845
    },
    {
      "epoch": 0.3403534620356929,
      "grad_norm": 0.25366660952568054,
      "learning_rate": 1.4807701395137041e-05,
      "loss": 0.2239,
      "step": 18850
    },
    {
      "epoch": 0.3404437414685936,
      "grad_norm": 0.41909897327423096,
      "learning_rate": 1.4805214251795951e-05,
      "loss": 0.0684,
      "step": 18855
    },
    {
      "epoch": 0.3405340209014943,
      "grad_norm": 0.3215011954307556,
      "learning_rate": 1.4802726721910312e-05,
      "loss": 0.3305,
      "step": 18860
    },
    {
      "epoch": 0.340624300334395,
      "grad_norm": 1.6500555276870728,
      "learning_rate": 1.4800238805680223e-05,
      "loss": 0.3719,
      "step": 18865
    },
    {
      "epoch": 0.34071457976729574,
      "grad_norm": 0.6505630016326904,
      "learning_rate": 1.4797750503305828e-05,
      "loss": 0.2377,
      "step": 18870
    },
    {
      "epoch": 0.34080485920019643,
      "grad_norm": 0.45783016085624695,
      "learning_rate": 1.4795261814987286e-05,
      "loss": 0.2104,
      "step": 18875
    },
    {
      "epoch": 0.34089513863309717,
      "grad_norm": 0.6193820834159851,
      "learning_rate": 1.4792772740924795e-05,
      "loss": 0.3273,
      "step": 18880
    },
    {
      "epoch": 0.34098541806599786,
      "grad_norm": 0.42955857515335083,
      "learning_rate": 1.4790283281318585e-05,
      "loss": 0.2735,
      "step": 18885
    },
    {
      "epoch": 0.3410756974988986,
      "grad_norm": 0.4381408989429474,
      "learning_rate": 1.4787793436368913e-05,
      "loss": 0.2838,
      "step": 18890
    },
    {
      "epoch": 0.3411659769317993,
      "grad_norm": 0.5888350605964661,
      "learning_rate": 1.4785303206276072e-05,
      "loss": 0.3188,
      "step": 18895
    },
    {
      "epoch": 0.34125625636470003,
      "grad_norm": 0.3903317451477051,
      "learning_rate": 1.478281259124038e-05,
      "loss": 0.2102,
      "step": 18900
    },
    {
      "epoch": 0.3413465357976007,
      "grad_norm": 0.4762745797634125,
      "learning_rate": 1.4780321591462186e-05,
      "loss": 0.2934,
      "step": 18905
    },
    {
      "epoch": 0.34143681523050146,
      "grad_norm": 0.4727645218372345,
      "learning_rate": 1.4777830207141878e-05,
      "loss": 0.3152,
      "step": 18910
    },
    {
      "epoch": 0.34152709466340214,
      "grad_norm": 0.30822187662124634,
      "learning_rate": 1.477533843847987e-05,
      "loss": 0.3066,
      "step": 18915
    },
    {
      "epoch": 0.3416173740963029,
      "grad_norm": 0.43814173340797424,
      "learning_rate": 1.4772846285676603e-05,
      "loss": 0.3401,
      "step": 18920
    },
    {
      "epoch": 0.34170765352920357,
      "grad_norm": 0.4088091552257538,
      "learning_rate": 1.4770353748932556e-05,
      "loss": 0.2896,
      "step": 18925
    },
    {
      "epoch": 0.3417979329621043,
      "grad_norm": 0.44749176502227783,
      "learning_rate": 1.476786082844823e-05,
      "loss": 0.2602,
      "step": 18930
    },
    {
      "epoch": 0.341888212395005,
      "grad_norm": 0.6866820454597473,
      "learning_rate": 1.476536752442417e-05,
      "loss": 0.2833,
      "step": 18935
    },
    {
      "epoch": 0.34197849182790574,
      "grad_norm": 0.6032989621162415,
      "learning_rate": 1.4762873837060939e-05,
      "loss": 0.3363,
      "step": 18940
    },
    {
      "epoch": 0.3420687712608064,
      "grad_norm": 0.31822606921195984,
      "learning_rate": 1.4760379766559136e-05,
      "loss": 0.2322,
      "step": 18945
    },
    {
      "epoch": 0.34215905069370717,
      "grad_norm": 0.46180498600006104,
      "learning_rate": 1.4757885313119394e-05,
      "loss": 0.329,
      "step": 18950
    },
    {
      "epoch": 0.34224933012660785,
      "grad_norm": 0.4387761056423187,
      "learning_rate": 1.475539047694237e-05,
      "loss": 0.194,
      "step": 18955
    },
    {
      "epoch": 0.3423396095595086,
      "grad_norm": 0.3619935214519501,
      "learning_rate": 1.4752895258228757e-05,
      "loss": 0.2013,
      "step": 18960
    },
    {
      "epoch": 0.3424298889924093,
      "grad_norm": 0.33837059140205383,
      "learning_rate": 1.4750399657179278e-05,
      "loss": 0.214,
      "step": 18965
    },
    {
      "epoch": 0.34252016842531,
      "grad_norm": 0.4133617877960205,
      "learning_rate": 1.4747903673994685e-05,
      "loss": 0.2423,
      "step": 18970
    },
    {
      "epoch": 0.3426104478582107,
      "grad_norm": 0.5128462910652161,
      "learning_rate": 1.4745407308875764e-05,
      "loss": 0.2934,
      "step": 18975
    },
    {
      "epoch": 0.34270072729111145,
      "grad_norm": 0.29368436336517334,
      "learning_rate": 1.4742910562023325e-05,
      "loss": 0.2599,
      "step": 18980
    },
    {
      "epoch": 0.34279100672401214,
      "grad_norm": 0.6440581679344177,
      "learning_rate": 1.4740413433638217e-05,
      "loss": 0.2811,
      "step": 18985
    },
    {
      "epoch": 0.3428812861569129,
      "grad_norm": 0.32762810587882996,
      "learning_rate": 1.4737915923921314e-05,
      "loss": 0.289,
      "step": 18990
    },
    {
      "epoch": 0.34297156558981357,
      "grad_norm": 0.6130682229995728,
      "learning_rate": 1.4735418033073523e-05,
      "loss": 0.2321,
      "step": 18995
    },
    {
      "epoch": 0.3430618450227143,
      "grad_norm": 0.461866557598114,
      "learning_rate": 1.4732919761295779e-05,
      "loss": 0.254,
      "step": 19000
    },
    {
      "epoch": 0.343152124455615,
      "grad_norm": 0.44246944785118103,
      "learning_rate": 1.4730421108789054e-05,
      "loss": 0.363,
      "step": 19005
    },
    {
      "epoch": 0.34324240388851573,
      "grad_norm": 0.4307284355163574,
      "learning_rate": 1.4727922075754344e-05,
      "loss": 0.288,
      "step": 19010
    },
    {
      "epoch": 0.3433326833214164,
      "grad_norm": 0.5668054223060608,
      "learning_rate": 1.4725422662392678e-05,
      "loss": 0.2427,
      "step": 19015
    },
    {
      "epoch": 0.34342296275431716,
      "grad_norm": 0.5066699981689453,
      "learning_rate": 1.4722922868905118e-05,
      "loss": 0.2555,
      "step": 19020
    },
    {
      "epoch": 0.3435132421872179,
      "grad_norm": 0.6849452257156372,
      "learning_rate": 1.4720422695492752e-05,
      "loss": 0.3359,
      "step": 19025
    },
    {
      "epoch": 0.3436035216201186,
      "grad_norm": 0.18425752222537994,
      "learning_rate": 1.4717922142356698e-05,
      "loss": 0.2555,
      "step": 19030
    },
    {
      "epoch": 0.34369380105301933,
      "grad_norm": 0.3990256190299988,
      "learning_rate": 1.4715421209698111e-05,
      "loss": 0.2023,
      "step": 19035
    },
    {
      "epoch": 0.34378408048592,
      "grad_norm": 0.5287699103355408,
      "learning_rate": 1.4712919897718174e-05,
      "loss": 0.1971,
      "step": 19040
    },
    {
      "epoch": 0.34387435991882076,
      "grad_norm": 0.4891014099121094,
      "learning_rate": 1.4710418206618093e-05,
      "loss": 0.2733,
      "step": 19045
    },
    {
      "epoch": 0.34396463935172145,
      "grad_norm": 0.7189919352531433,
      "learning_rate": 1.470791613659912e-05,
      "loss": 0.2504,
      "step": 19050
    },
    {
      "epoch": 0.3440549187846222,
      "grad_norm": 0.4761006832122803,
      "learning_rate": 1.4705413687862523e-05,
      "loss": 0.2638,
      "step": 19055
    },
    {
      "epoch": 0.3441451982175229,
      "grad_norm": 0.4221712350845337,
      "learning_rate": 1.4702910860609607e-05,
      "loss": 0.3228,
      "step": 19060
    },
    {
      "epoch": 0.3442354776504236,
      "grad_norm": 0.4744848608970642,
      "learning_rate": 1.4700407655041701e-05,
      "loss": 0.3331,
      "step": 19065
    },
    {
      "epoch": 0.3443257570833243,
      "grad_norm": 0.5731523633003235,
      "learning_rate": 1.469790407136018e-05,
      "loss": 0.3992,
      "step": 19070
    },
    {
      "epoch": 0.34441603651622504,
      "grad_norm": 0.5367938876152039,
      "learning_rate": 1.469540010976643e-05,
      "loss": 0.1992,
      "step": 19075
    },
    {
      "epoch": 0.34450631594912573,
      "grad_norm": 0.5153685212135315,
      "learning_rate": 1.4692895770461881e-05,
      "loss": 0.2705,
      "step": 19080
    },
    {
      "epoch": 0.34459659538202647,
      "grad_norm": 0.4405202269554138,
      "learning_rate": 1.469039105364799e-05,
      "loss": 0.335,
      "step": 19085
    },
    {
      "epoch": 0.34468687481492716,
      "grad_norm": 0.5037932395935059,
      "learning_rate": 1.4687885959526239e-05,
      "loss": 0.2357,
      "step": 19090
    },
    {
      "epoch": 0.3447771542478279,
      "grad_norm": 0.47482338547706604,
      "learning_rate": 1.4685380488298148e-05,
      "loss": 0.2591,
      "step": 19095
    },
    {
      "epoch": 0.3448674336807286,
      "grad_norm": 0.41249433159828186,
      "learning_rate": 1.468287464016526e-05,
      "loss": 0.2142,
      "step": 19100
    },
    {
      "epoch": 0.3449577131136293,
      "grad_norm": 0.47318199276924133,
      "learning_rate": 1.4680368415329158e-05,
      "loss": 0.2712,
      "step": 19105
    },
    {
      "epoch": 0.34504799254653,
      "grad_norm": 0.41339215636253357,
      "learning_rate": 1.4677861813991446e-05,
      "loss": 0.3777,
      "step": 19110
    },
    {
      "epoch": 0.34513827197943076,
      "grad_norm": 0.41870197653770447,
      "learning_rate": 1.4675354836353762e-05,
      "loss": 0.2222,
      "step": 19115
    },
    {
      "epoch": 0.34522855141233144,
      "grad_norm": 0.45384618639945984,
      "learning_rate": 1.4672847482617775e-05,
      "loss": 0.1345,
      "step": 19120
    },
    {
      "epoch": 0.3453188308452322,
      "grad_norm": 0.6229274868965149,
      "learning_rate": 1.4670339752985182e-05,
      "loss": 0.2492,
      "step": 19125
    },
    {
      "epoch": 0.34540911027813287,
      "grad_norm": 0.6140114665031433,
      "learning_rate": 1.4667831647657714e-05,
      "loss": 0.2459,
      "step": 19130
    },
    {
      "epoch": 0.3454993897110336,
      "grad_norm": 0.6977282762527466,
      "learning_rate": 1.4665323166837126e-05,
      "loss": 0.3348,
      "step": 19135
    },
    {
      "epoch": 0.3455896691439343,
      "grad_norm": 0.4640437066555023,
      "learning_rate": 1.4662814310725214e-05,
      "loss": 0.2192,
      "step": 19140
    },
    {
      "epoch": 0.34567994857683504,
      "grad_norm": 0.5979374051094055,
      "learning_rate": 1.4660305079523784e-05,
      "loss": 0.2042,
      "step": 19145
    },
    {
      "epoch": 0.3457702280097357,
      "grad_norm": 0.8193564414978027,
      "learning_rate": 1.4657795473434702e-05,
      "loss": 0.4243,
      "step": 19150
    },
    {
      "epoch": 0.34586050744263647,
      "grad_norm": 0.6361282467842102,
      "learning_rate": 1.4655285492659837e-05,
      "loss": 0.2487,
      "step": 19155
    },
    {
      "epoch": 0.34595078687553715,
      "grad_norm": 0.5085448622703552,
      "learning_rate": 1.46527751374011e-05,
      "loss": 0.2778,
      "step": 19160
    },
    {
      "epoch": 0.3460410663084379,
      "grad_norm": 0.3158360421657562,
      "learning_rate": 1.4650264407860432e-05,
      "loss": 0.1997,
      "step": 19165
    },
    {
      "epoch": 0.3461313457413386,
      "grad_norm": 0.3488834798336029,
      "learning_rate": 1.4647753304239807e-05,
      "loss": 0.2658,
      "step": 19170
    },
    {
      "epoch": 0.3462216251742393,
      "grad_norm": 0.3840755820274353,
      "learning_rate": 1.4645241826741218e-05,
      "loss": 0.2221,
      "step": 19175
    },
    {
      "epoch": 0.34631190460714,
      "grad_norm": 0.5721611380577087,
      "learning_rate": 1.46427299755667e-05,
      "loss": 0.3702,
      "step": 19180
    },
    {
      "epoch": 0.34640218404004075,
      "grad_norm": 0.38359248638153076,
      "learning_rate": 1.4640217750918308e-05,
      "loss": 0.2568,
      "step": 19185
    },
    {
      "epoch": 0.34649246347294144,
      "grad_norm": 0.43777433037757874,
      "learning_rate": 1.4637705152998139e-05,
      "loss": 0.3129,
      "step": 19190
    },
    {
      "epoch": 0.3465827429058422,
      "grad_norm": 0.3485426902770996,
      "learning_rate": 1.4635192182008308e-05,
      "loss": 0.2971,
      "step": 19195
    },
    {
      "epoch": 0.34667302233874286,
      "grad_norm": 0.4009533226490021,
      "learning_rate": 1.4632678838150966e-05,
      "loss": 0.2504,
      "step": 19200
    },
    {
      "epoch": 0.3467633017716436,
      "grad_norm": 0.2905276417732239,
      "learning_rate": 1.4630165121628299e-05,
      "loss": 0.2163,
      "step": 19205
    },
    {
      "epoch": 0.3468535812045443,
      "grad_norm": 0.5106023550033569,
      "learning_rate": 1.4627651032642508e-05,
      "loss": 0.314,
      "step": 19210
    },
    {
      "epoch": 0.34694386063744503,
      "grad_norm": 0.4790681004524231,
      "learning_rate": 1.462513657139584e-05,
      "loss": 0.2759,
      "step": 19215
    },
    {
      "epoch": 0.3470341400703457,
      "grad_norm": 0.39027634263038635,
      "learning_rate": 1.4622621738090563e-05,
      "loss": 0.3259,
      "step": 19220
    },
    {
      "epoch": 0.34712441950324646,
      "grad_norm": 0.7681130170822144,
      "learning_rate": 1.4620106532928975e-05,
      "loss": 0.2633,
      "step": 19225
    },
    {
      "epoch": 0.34721469893614715,
      "grad_norm": 0.5734387636184692,
      "learning_rate": 1.4617590956113408e-05,
      "loss": 0.3346,
      "step": 19230
    },
    {
      "epoch": 0.3473049783690479,
      "grad_norm": 0.3158043920993805,
      "learning_rate": 1.4615075007846222e-05,
      "loss": 0.1993,
      "step": 19235
    },
    {
      "epoch": 0.3473952578019486,
      "grad_norm": 0.3726866543292999,
      "learning_rate": 1.4612558688329808e-05,
      "loss": 0.2789,
      "step": 19240
    },
    {
      "epoch": 0.3474855372348493,
      "grad_norm": 0.8030419945716858,
      "learning_rate": 1.4610041997766581e-05,
      "loss": 0.41,
      "step": 19245
    },
    {
      "epoch": 0.34757581666775,
      "grad_norm": 0.45678412914276123,
      "learning_rate": 1.4607524936358996e-05,
      "loss": 0.2492,
      "step": 19250
    },
    {
      "epoch": 0.34766609610065075,
      "grad_norm": 0.3806954324245453,
      "learning_rate": 1.4605007504309531e-05,
      "loss": 0.2823,
      "step": 19255
    },
    {
      "epoch": 0.34775637553355143,
      "grad_norm": 0.646897554397583,
      "learning_rate": 1.4602489701820693e-05,
      "loss": 0.2263,
      "step": 19260
    },
    {
      "epoch": 0.3478466549664522,
      "grad_norm": 0.5565019845962524,
      "learning_rate": 1.4599971529095019e-05,
      "loss": 0.1936,
      "step": 19265
    },
    {
      "epoch": 0.34793693439935286,
      "grad_norm": 0.3828217685222626,
      "learning_rate": 1.4597452986335084e-05,
      "loss": 0.2337,
      "step": 19270
    },
    {
      "epoch": 0.3480272138322536,
      "grad_norm": 0.49128270149230957,
      "learning_rate": 1.459493407374348e-05,
      "loss": 0.2843,
      "step": 19275
    },
    {
      "epoch": 0.3481174932651543,
      "grad_norm": 0.40793755650520325,
      "learning_rate": 1.459241479152284e-05,
      "loss": 0.3523,
      "step": 19280
    },
    {
      "epoch": 0.34820777269805503,
      "grad_norm": 0.46196678280830383,
      "learning_rate": 1.4589895139875818e-05,
      "loss": 0.2962,
      "step": 19285
    },
    {
      "epoch": 0.3482980521309557,
      "grad_norm": 0.9009236693382263,
      "learning_rate": 1.4587375119005107e-05,
      "loss": 0.3176,
      "step": 19290
    },
    {
      "epoch": 0.34838833156385646,
      "grad_norm": 0.40262532234191895,
      "learning_rate": 1.458485472911342e-05,
      "loss": 0.3026,
      "step": 19295
    },
    {
      "epoch": 0.34847861099675714,
      "grad_norm": 0.41505348682403564,
      "learning_rate": 1.4582333970403506e-05,
      "loss": 0.2214,
      "step": 19300
    },
    {
      "epoch": 0.3485688904296579,
      "grad_norm": 0.38175857067108154,
      "learning_rate": 1.457981284307814e-05,
      "loss": 0.1901,
      "step": 19305
    },
    {
      "epoch": 0.34865916986255857,
      "grad_norm": 0.41143855452537537,
      "learning_rate": 1.4577291347340128e-05,
      "loss": 0.3221,
      "step": 19310
    },
    {
      "epoch": 0.3487494492954593,
      "grad_norm": 0.5829758048057556,
      "learning_rate": 1.4574769483392308e-05,
      "loss": 0.1806,
      "step": 19315
    },
    {
      "epoch": 0.34883972872836,
      "grad_norm": 0.4416637420654297,
      "learning_rate": 1.4572247251437545e-05,
      "loss": 0.2414,
      "step": 19320
    },
    {
      "epoch": 0.34893000816126074,
      "grad_norm": 0.3901643455028534,
      "learning_rate": 1.4569724651678736e-05,
      "loss": 0.2626,
      "step": 19325
    },
    {
      "epoch": 0.3490202875941614,
      "grad_norm": 0.6232507824897766,
      "learning_rate": 1.4567201684318798e-05,
      "loss": 0.3614,
      "step": 19330
    },
    {
      "epoch": 0.34911056702706217,
      "grad_norm": 0.48857784271240234,
      "learning_rate": 1.4564678349560697e-05,
      "loss": 0.2078,
      "step": 19335
    },
    {
      "epoch": 0.34920084645996285,
      "grad_norm": 0.4701916575431824,
      "learning_rate": 1.4562154647607409e-05,
      "loss": 0.259,
      "step": 19340
    },
    {
      "epoch": 0.3492911258928636,
      "grad_norm": 0.4478681981563568,
      "learning_rate": 1.455963057866195e-05,
      "loss": 0.229,
      "step": 19345
    },
    {
      "epoch": 0.3493814053257643,
      "grad_norm": 0.6830845475196838,
      "learning_rate": 1.4557106142927362e-05,
      "loss": 0.3134,
      "step": 19350
    },
    {
      "epoch": 0.349471684758665,
      "grad_norm": 0.6161406636238098,
      "learning_rate": 1.4554581340606721e-05,
      "loss": 0.2472,
      "step": 19355
    },
    {
      "epoch": 0.3495619641915657,
      "grad_norm": 0.711693525314331,
      "learning_rate": 1.4552056171903125e-05,
      "loss": 0.225,
      "step": 19360
    },
    {
      "epoch": 0.34965224362446645,
      "grad_norm": 0.44821977615356445,
      "learning_rate": 1.4549530637019702e-05,
      "loss": 0.4006,
      "step": 19365
    },
    {
      "epoch": 0.34974252305736714,
      "grad_norm": 0.25322824716567993,
      "learning_rate": 1.4547004736159622e-05,
      "loss": 0.2437,
      "step": 19370
    },
    {
      "epoch": 0.3498328024902679,
      "grad_norm": 0.6989191174507141,
      "learning_rate": 1.4544478469526071e-05,
      "loss": 0.3077,
      "step": 19375
    },
    {
      "epoch": 0.34992308192316857,
      "grad_norm": 0.7583245038986206,
      "learning_rate": 1.4541951837322268e-05,
      "loss": 0.31,
      "step": 19380
    },
    {
      "epoch": 0.3500133613560693,
      "grad_norm": 0.4083063006401062,
      "learning_rate": 1.4539424839751463e-05,
      "loss": 0.2243,
      "step": 19385
    },
    {
      "epoch": 0.35010364078897,
      "grad_norm": 0.5561839938163757,
      "learning_rate": 1.4536897477016936e-05,
      "loss": 0.2594,
      "step": 19390
    },
    {
      "epoch": 0.35019392022187074,
      "grad_norm": 0.29318517446517944,
      "learning_rate": 1.453436974932199e-05,
      "loss": 0.2789,
      "step": 19395
    },
    {
      "epoch": 0.3502841996547714,
      "grad_norm": 0.23433071374893188,
      "learning_rate": 1.4531841656869967e-05,
      "loss": 0.2632,
      "step": 19400
    },
    {
      "epoch": 0.35037447908767216,
      "grad_norm": 0.5063546895980835,
      "learning_rate": 1.4529313199864233e-05,
      "loss": 0.3407,
      "step": 19405
    },
    {
      "epoch": 0.35046475852057285,
      "grad_norm": 0.49939775466918945,
      "learning_rate": 1.4526784378508181e-05,
      "loss": 0.2844,
      "step": 19410
    },
    {
      "epoch": 0.3505550379534736,
      "grad_norm": 0.3330202102661133,
      "learning_rate": 1.4524255193005242e-05,
      "loss": 0.1756,
      "step": 19415
    },
    {
      "epoch": 0.35064531738637433,
      "grad_norm": 0.7029513120651245,
      "learning_rate": 1.4521725643558866e-05,
      "loss": 0.2963,
      "step": 19420
    },
    {
      "epoch": 0.350735596819275,
      "grad_norm": 0.7581319808959961,
      "learning_rate": 1.4519195730372539e-05,
      "loss": 0.1796,
      "step": 19425
    },
    {
      "epoch": 0.35082587625217576,
      "grad_norm": 1.4334921836853027,
      "learning_rate": 1.451666545364977e-05,
      "loss": 0.2531,
      "step": 19430
    },
    {
      "epoch": 0.35091615568507645,
      "grad_norm": 0.6741414070129395,
      "learning_rate": 1.4514134813594108e-05,
      "loss": 0.2778,
      "step": 19435
    },
    {
      "epoch": 0.3510064351179772,
      "grad_norm": 0.45837345719337463,
      "learning_rate": 1.4511603810409121e-05,
      "loss": 0.3039,
      "step": 19440
    },
    {
      "epoch": 0.3510967145508779,
      "grad_norm": 0.6172248721122742,
      "learning_rate": 1.4509072444298406e-05,
      "loss": 0.3513,
      "step": 19445
    },
    {
      "epoch": 0.3511869939837786,
      "grad_norm": 0.5250604152679443,
      "learning_rate": 1.45065407154656e-05,
      "loss": 0.2898,
      "step": 19450
    },
    {
      "epoch": 0.3512772734166793,
      "grad_norm": 0.4318670630455017,
      "learning_rate": 1.450400862411436e-05,
      "loss": 0.2264,
      "step": 19455
    },
    {
      "epoch": 0.35136755284958004,
      "grad_norm": 0.4514148533344269,
      "learning_rate": 1.4501476170448375e-05,
      "loss": 0.3025,
      "step": 19460
    },
    {
      "epoch": 0.35145783228248073,
      "grad_norm": 0.5048012137413025,
      "learning_rate": 1.4498943354671355e-05,
      "loss": 0.3115,
      "step": 19465
    },
    {
      "epoch": 0.3515481117153815,
      "grad_norm": 0.3340650498867035,
      "learning_rate": 1.4496410176987056e-05,
      "loss": 0.1856,
      "step": 19470
    },
    {
      "epoch": 0.35163839114828216,
      "grad_norm": 0.6395434141159058,
      "learning_rate": 1.4493876637599251e-05,
      "loss": 0.316,
      "step": 19475
    },
    {
      "epoch": 0.3517286705811829,
      "grad_norm": 0.32689008116722107,
      "learning_rate": 1.4491342736711747e-05,
      "loss": 0.2629,
      "step": 19480
    },
    {
      "epoch": 0.3518189500140836,
      "grad_norm": 0.3152740001678467,
      "learning_rate": 1.448880847452837e-05,
      "loss": 0.268,
      "step": 19485
    },
    {
      "epoch": 0.35190922944698433,
      "grad_norm": 0.4254841208457947,
      "learning_rate": 1.4486273851252989e-05,
      "loss": 0.2718,
      "step": 19490
    },
    {
      "epoch": 0.351999508879885,
      "grad_norm": 0.2890605330467224,
      "learning_rate": 1.4483738867089497e-05,
      "loss": 0.2361,
      "step": 19495
    },
    {
      "epoch": 0.35208978831278576,
      "grad_norm": 0.49806782603263855,
      "learning_rate": 1.4481203522241812e-05,
      "loss": 0.2602,
      "step": 19500
    },
    {
      "epoch": 0.35218006774568644,
      "grad_norm": 0.4263138473033905,
      "learning_rate": 1.4478667816913884e-05,
      "loss": 0.2234,
      "step": 19505
    },
    {
      "epoch": 0.3522703471785872,
      "grad_norm": 0.490191787481308,
      "learning_rate": 1.4476131751309695e-05,
      "loss": 0.2344,
      "step": 19510
    },
    {
      "epoch": 0.35236062661148787,
      "grad_norm": 0.5755016803741455,
      "learning_rate": 1.4473595325633249e-05,
      "loss": 0.3289,
      "step": 19515
    },
    {
      "epoch": 0.3524509060443886,
      "grad_norm": 0.7282212972640991,
      "learning_rate": 1.4471058540088588e-05,
      "loss": 0.3026,
      "step": 19520
    },
    {
      "epoch": 0.3525411854772893,
      "grad_norm": 0.45803165435791016,
      "learning_rate": 1.4468521394879773e-05,
      "loss": 0.3556,
      "step": 19525
    },
    {
      "epoch": 0.35263146491019004,
      "grad_norm": 0.47527822852134705,
      "learning_rate": 1.44659838902109e-05,
      "loss": 0.3038,
      "step": 19530
    },
    {
      "epoch": 0.3527217443430907,
      "grad_norm": 0.3465036153793335,
      "learning_rate": 1.4463446026286096e-05,
      "loss": 0.2434,
      "step": 19535
    },
    {
      "epoch": 0.35281202377599147,
      "grad_norm": 0.5368249416351318,
      "learning_rate": 1.4460907803309511e-05,
      "loss": 0.2178,
      "step": 19540
    },
    {
      "epoch": 0.35290230320889215,
      "grad_norm": 0.46956199407577515,
      "learning_rate": 1.4458369221485327e-05,
      "loss": 0.2635,
      "step": 19545
    },
    {
      "epoch": 0.3529925826417929,
      "grad_norm": 0.9469265937805176,
      "learning_rate": 1.4455830281017756e-05,
      "loss": 0.1715,
      "step": 19550
    },
    {
      "epoch": 0.3530828620746936,
      "grad_norm": 0.3272339403629303,
      "learning_rate": 1.4453290982111035e-05,
      "loss": 0.2596,
      "step": 19555
    },
    {
      "epoch": 0.3531731415075943,
      "grad_norm": 0.5277330875396729,
      "learning_rate": 1.4450751324969432e-05,
      "loss": 0.3983,
      "step": 19560
    },
    {
      "epoch": 0.353263420940495,
      "grad_norm": 0.4450097680091858,
      "learning_rate": 1.4448211309797247e-05,
      "loss": 0.2898,
      "step": 19565
    },
    {
      "epoch": 0.35335370037339575,
      "grad_norm": 0.5424237847328186,
      "learning_rate": 1.4445670936798806e-05,
      "loss": 0.3631,
      "step": 19570
    },
    {
      "epoch": 0.35344397980629644,
      "grad_norm": 0.48294100165367126,
      "learning_rate": 1.4443130206178456e-05,
      "loss": 0.2373,
      "step": 19575
    },
    {
      "epoch": 0.3535342592391972,
      "grad_norm": 0.6076140403747559,
      "learning_rate": 1.4440589118140586e-05,
      "loss": 0.3505,
      "step": 19580
    },
    {
      "epoch": 0.35362453867209787,
      "grad_norm": 0.35823559761047363,
      "learning_rate": 1.443804767288961e-05,
      "loss": 0.236,
      "step": 19585
    },
    {
      "epoch": 0.3537148181049986,
      "grad_norm": 0.4359067380428314,
      "learning_rate": 1.4435505870629967e-05,
      "loss": 0.3581,
      "step": 19590
    },
    {
      "epoch": 0.3538050975378993,
      "grad_norm": 0.5708020329475403,
      "learning_rate": 1.4432963711566124e-05,
      "loss": 0.2288,
      "step": 19595
    },
    {
      "epoch": 0.35389537697080004,
      "grad_norm": 1.0934189558029175,
      "learning_rate": 1.4430421195902582e-05,
      "loss": 0.2791,
      "step": 19600
    },
    {
      "epoch": 0.3539856564037007,
      "grad_norm": 0.6318325996398926,
      "learning_rate": 1.4427878323843867e-05,
      "loss": 0.2272,
      "step": 19605
    },
    {
      "epoch": 0.35407593583660146,
      "grad_norm": 0.3226698338985443,
      "learning_rate": 1.4425335095594532e-05,
      "loss": 0.2296,
      "step": 19610
    },
    {
      "epoch": 0.35416621526950215,
      "grad_norm": 0.722440242767334,
      "learning_rate": 1.442279151135917e-05,
      "loss": 0.24,
      "step": 19615
    },
    {
      "epoch": 0.3542564947024029,
      "grad_norm": 0.3280979096889496,
      "learning_rate": 1.4420247571342383e-05,
      "loss": 0.228,
      "step": 19620
    },
    {
      "epoch": 0.3543467741353036,
      "grad_norm": 0.32591092586517334,
      "learning_rate": 1.4417703275748817e-05,
      "loss": 0.1964,
      "step": 19625
    },
    {
      "epoch": 0.3544370535682043,
      "grad_norm": 0.3040795624256134,
      "learning_rate": 1.4415158624783142e-05,
      "loss": 0.1845,
      "step": 19630
    },
    {
      "epoch": 0.354527333001105,
      "grad_norm": 0.4605494737625122,
      "learning_rate": 1.441261361865006e-05,
      "loss": 0.2225,
      "step": 19635
    },
    {
      "epoch": 0.35461761243400575,
      "grad_norm": 0.4035797715187073,
      "learning_rate": 1.4410068257554295e-05,
      "loss": 0.1865,
      "step": 19640
    },
    {
      "epoch": 0.35470789186690643,
      "grad_norm": 0.5863381028175354,
      "learning_rate": 1.44075225417006e-05,
      "loss": 0.2495,
      "step": 19645
    },
    {
      "epoch": 0.3547981712998072,
      "grad_norm": 0.45320987701416016,
      "learning_rate": 1.4404976471293766e-05,
      "loss": 0.308,
      "step": 19650
    },
    {
      "epoch": 0.35488845073270786,
      "grad_norm": 0.4062706530094147,
      "learning_rate": 1.4402430046538597e-05,
      "loss": 0.2782,
      "step": 19655
    },
    {
      "epoch": 0.3549787301656086,
      "grad_norm": 0.5910449028015137,
      "learning_rate": 1.4399883267639942e-05,
      "loss": 0.2696,
      "step": 19660
    },
    {
      "epoch": 0.3550690095985093,
      "grad_norm": 0.440742552280426,
      "learning_rate": 1.4397336134802673e-05,
      "loss": 0.2436,
      "step": 19665
    },
    {
      "epoch": 0.35515928903141003,
      "grad_norm": 1.6211673021316528,
      "learning_rate": 1.4394788648231679e-05,
      "loss": 0.2452,
      "step": 19670
    },
    {
      "epoch": 0.3552495684643107,
      "grad_norm": 0.5122200846672058,
      "learning_rate": 1.4392240808131892e-05,
      "loss": 0.2505,
      "step": 19675
    },
    {
      "epoch": 0.35533984789721146,
      "grad_norm": 0.683473527431488,
      "learning_rate": 1.4389692614708266e-05,
      "loss": 0.2907,
      "step": 19680
    },
    {
      "epoch": 0.35543012733011214,
      "grad_norm": 0.4165571928024292,
      "learning_rate": 1.4387144068165788e-05,
      "loss": 0.2784,
      "step": 19685
    },
    {
      "epoch": 0.3555204067630129,
      "grad_norm": 0.35494714975357056,
      "learning_rate": 1.4384595168709465e-05,
      "loss": 0.2705,
      "step": 19690
    },
    {
      "epoch": 0.3556106861959136,
      "grad_norm": 0.4419481158256531,
      "learning_rate": 1.4382045916544339e-05,
      "loss": 0.2106,
      "step": 19695
    },
    {
      "epoch": 0.3557009656288143,
      "grad_norm": 0.5680895447731018,
      "learning_rate": 1.437949631187548e-05,
      "loss": 0.4024,
      "step": 19700
    },
    {
      "epoch": 0.355791245061715,
      "grad_norm": 0.82598876953125,
      "learning_rate": 1.4376946354907988e-05,
      "loss": 0.327,
      "step": 19705
    },
    {
      "epoch": 0.35588152449461574,
      "grad_norm": 0.4863128364086151,
      "learning_rate": 1.437439604584698e-05,
      "loss": 0.2895,
      "step": 19710
    },
    {
      "epoch": 0.35597180392751643,
      "grad_norm": 0.46011534333229065,
      "learning_rate": 1.4371845384897617e-05,
      "loss": 0.3094,
      "step": 19715
    },
    {
      "epoch": 0.35606208336041717,
      "grad_norm": 0.791948139667511,
      "learning_rate": 1.4369294372265082e-05,
      "loss": 0.1886,
      "step": 19720
    },
    {
      "epoch": 0.35615236279331786,
      "grad_norm": 0.45013314485549927,
      "learning_rate": 1.436674300815458e-05,
      "loss": 0.25,
      "step": 19725
    },
    {
      "epoch": 0.3562426422262186,
      "grad_norm": 0.2880622446537018,
      "learning_rate": 1.4364191292771352e-05,
      "loss": 0.1994,
      "step": 19730
    },
    {
      "epoch": 0.3563329216591193,
      "grad_norm": 0.338219553232193,
      "learning_rate": 1.4361639226320668e-05,
      "loss": 0.2814,
      "step": 19735
    },
    {
      "epoch": 0.35642320109202,
      "grad_norm": 0.5113375782966614,
      "learning_rate": 1.4359086809007818e-05,
      "loss": 0.2548,
      "step": 19740
    },
    {
      "epoch": 0.3565134805249207,
      "grad_norm": 0.8494113087654114,
      "learning_rate": 1.4356534041038126e-05,
      "loss": 0.2837,
      "step": 19745
    },
    {
      "epoch": 0.35660375995782145,
      "grad_norm": 0.4344501197338104,
      "learning_rate": 1.4353980922616948e-05,
      "loss": 0.3321,
      "step": 19750
    },
    {
      "epoch": 0.35669403939072214,
      "grad_norm": 0.7422098517417908,
      "learning_rate": 1.4351427453949664e-05,
      "loss": 0.302,
      "step": 19755
    },
    {
      "epoch": 0.3567843188236229,
      "grad_norm": 0.43888768553733826,
      "learning_rate": 1.4348873635241676e-05,
      "loss": 0.1837,
      "step": 19760
    },
    {
      "epoch": 0.35687459825652357,
      "grad_norm": 0.5647388100624084,
      "learning_rate": 1.4346319466698424e-05,
      "loss": 0.2307,
      "step": 19765
    },
    {
      "epoch": 0.3569648776894243,
      "grad_norm": 0.4936065375804901,
      "learning_rate": 1.4343764948525371e-05,
      "loss": 0.2358,
      "step": 19770
    },
    {
      "epoch": 0.357055157122325,
      "grad_norm": 0.43983563780784607,
      "learning_rate": 1.434121008092801e-05,
      "loss": 0.2115,
      "step": 19775
    },
    {
      "epoch": 0.35714543655522574,
      "grad_norm": 0.47324320673942566,
      "learning_rate": 1.433865486411186e-05,
      "loss": 0.2016,
      "step": 19780
    },
    {
      "epoch": 0.3572357159881264,
      "grad_norm": 0.42178916931152344,
      "learning_rate": 1.4336099298282475e-05,
      "loss": 0.258,
      "step": 19785
    },
    {
      "epoch": 0.35732599542102716,
      "grad_norm": 0.42953428626060486,
      "learning_rate": 1.4333543383645425e-05,
      "loss": 0.3026,
      "step": 19790
    },
    {
      "epoch": 0.35741627485392785,
      "grad_norm": 0.5370898842811584,
      "learning_rate": 1.4330987120406321e-05,
      "loss": 0.3165,
      "step": 19795
    },
    {
      "epoch": 0.3575065542868286,
      "grad_norm": 1.250838041305542,
      "learning_rate": 1.4328430508770789e-05,
      "loss": 0.2761,
      "step": 19800
    },
    {
      "epoch": 0.3575968337197293,
      "grad_norm": 0.37467846274375916,
      "learning_rate": 1.4325873548944498e-05,
      "loss": 0.2376,
      "step": 19805
    },
    {
      "epoch": 0.35768711315263,
      "grad_norm": 0.5512472987174988,
      "learning_rate": 1.4323316241133127e-05,
      "loss": 0.2276,
      "step": 19810
    },
    {
      "epoch": 0.35777739258553076,
      "grad_norm": 0.5403490662574768,
      "learning_rate": 1.4320758585542401e-05,
      "loss": 0.299,
      "step": 19815
    },
    {
      "epoch": 0.35786767201843145,
      "grad_norm": 0.6419376730918884,
      "learning_rate": 1.431820058237806e-05,
      "loss": 0.25,
      "step": 19820
    },
    {
      "epoch": 0.3579579514513322,
      "grad_norm": 0.4647071659564972,
      "learning_rate": 1.4315642231845876e-05,
      "loss": 0.2565,
      "step": 19825
    },
    {
      "epoch": 0.3580482308842329,
      "grad_norm": 0.49543461203575134,
      "learning_rate": 1.4313083534151657e-05,
      "loss": 0.1982,
      "step": 19830
    },
    {
      "epoch": 0.3581385103171336,
      "grad_norm": 0.5073922276496887,
      "learning_rate": 1.4310524489501225e-05,
      "loss": 0.2764,
      "step": 19835
    },
    {
      "epoch": 0.3582287897500343,
      "grad_norm": 0.6127316355705261,
      "learning_rate": 1.430796509810044e-05,
      "loss": 0.3251,
      "step": 19840
    },
    {
      "epoch": 0.35831906918293505,
      "grad_norm": 0.5740789175033569,
      "learning_rate": 1.4305405360155183e-05,
      "loss": 0.3247,
      "step": 19845
    },
    {
      "epoch": 0.35840934861583573,
      "grad_norm": 0.36514946818351746,
      "learning_rate": 1.4302845275871371e-05,
      "loss": 0.2557,
      "step": 19850
    },
    {
      "epoch": 0.3584996280487365,
      "grad_norm": 0.5043168067932129,
      "learning_rate": 1.4300284845454937e-05,
      "loss": 0.2391,
      "step": 19855
    },
    {
      "epoch": 0.35858990748163716,
      "grad_norm": 0.4452023506164551,
      "learning_rate": 1.4297724069111854e-05,
      "loss": 0.2864,
      "step": 19860
    },
    {
      "epoch": 0.3586801869145379,
      "grad_norm": 0.2947918474674225,
      "learning_rate": 1.4295162947048119e-05,
      "loss": 0.2045,
      "step": 19865
    },
    {
      "epoch": 0.3587704663474386,
      "grad_norm": 0.3564087450504303,
      "learning_rate": 1.4292601479469754e-05,
      "loss": 0.2265,
      "step": 19870
    },
    {
      "epoch": 0.35886074578033933,
      "grad_norm": 0.46823352575302124,
      "learning_rate": 1.4290039666582807e-05,
      "loss": 0.3774,
      "step": 19875
    },
    {
      "epoch": 0.35895102521324,
      "grad_norm": 0.4109562635421753,
      "learning_rate": 1.4287477508593361e-05,
      "loss": 0.2929,
      "step": 19880
    },
    {
      "epoch": 0.35904130464614076,
      "grad_norm": 0.48783037066459656,
      "learning_rate": 1.4284915005707527e-05,
      "loss": 0.2404,
      "step": 19885
    },
    {
      "epoch": 0.35913158407904144,
      "grad_norm": 0.39710333943367004,
      "learning_rate": 1.428235215813143e-05,
      "loss": 0.2284,
      "step": 19890
    },
    {
      "epoch": 0.3592218635119422,
      "grad_norm": 0.47894757986068726,
      "learning_rate": 1.427978896607124e-05,
      "loss": 0.3403,
      "step": 19895
    },
    {
      "epoch": 0.35931214294484287,
      "grad_norm": 0.4499393105506897,
      "learning_rate": 1.4277225429733146e-05,
      "loss": 0.2867,
      "step": 19900
    },
    {
      "epoch": 0.3594024223777436,
      "grad_norm": 0.36178120970726013,
      "learning_rate": 1.4274661549323363e-05,
      "loss": 0.3247,
      "step": 19905
    },
    {
      "epoch": 0.3594927018106443,
      "grad_norm": 0.5082676410675049,
      "learning_rate": 1.4272097325048135e-05,
      "loss": 0.3115,
      "step": 19910
    },
    {
      "epoch": 0.35958298124354504,
      "grad_norm": 0.4348970055580139,
      "learning_rate": 1.4269532757113743e-05,
      "loss": 0.2576,
      "step": 19915
    },
    {
      "epoch": 0.3596732606764457,
      "grad_norm": 0.34213346242904663,
      "learning_rate": 1.4266967845726482e-05,
      "loss": 0.2346,
      "step": 19920
    },
    {
      "epoch": 0.35976354010934647,
      "grad_norm": 0.5642054080963135,
      "learning_rate": 1.4264402591092678e-05,
      "loss": 0.3248,
      "step": 19925
    },
    {
      "epoch": 0.35985381954224716,
      "grad_norm": 0.3973153233528137,
      "learning_rate": 1.4261836993418695e-05,
      "loss": 0.2369,
      "step": 19930
    },
    {
      "epoch": 0.3599440989751479,
      "grad_norm": 0.5653963685035706,
      "learning_rate": 1.4259271052910912e-05,
      "loss": 0.2373,
      "step": 19935
    },
    {
      "epoch": 0.3600343784080486,
      "grad_norm": 0.4134688079357147,
      "learning_rate": 1.4256704769775742e-05,
      "loss": 0.2033,
      "step": 19940
    },
    {
      "epoch": 0.3601246578409493,
      "grad_norm": 0.5514252185821533,
      "learning_rate": 1.4254138144219623e-05,
      "loss": 0.2597,
      "step": 19945
    },
    {
      "epoch": 0.36021493727385,
      "grad_norm": 0.34145841002464294,
      "learning_rate": 1.4251571176449018e-05,
      "loss": 0.2873,
      "step": 19950
    },
    {
      "epoch": 0.36030521670675075,
      "grad_norm": 0.7172877192497253,
      "learning_rate": 1.424900386667042e-05,
      "loss": 0.3498,
      "step": 19955
    },
    {
      "epoch": 0.36039549613965144,
      "grad_norm": 0.1886661797761917,
      "learning_rate": 1.4246436215090361e-05,
      "loss": 0.3121,
      "step": 19960
    },
    {
      "epoch": 0.3604857755725522,
      "grad_norm": 0.595154881477356,
      "learning_rate": 1.4243868221915382e-05,
      "loss": 0.2444,
      "step": 19965
    },
    {
      "epoch": 0.36057605500545287,
      "grad_norm": 0.5034096240997314,
      "learning_rate": 1.4241299887352059e-05,
      "loss": 0.276,
      "step": 19970
    },
    {
      "epoch": 0.3606663344383536,
      "grad_norm": 0.4864138960838318,
      "learning_rate": 1.4238731211606995e-05,
      "loss": 0.2873,
      "step": 19975
    },
    {
      "epoch": 0.3607566138712543,
      "grad_norm": 0.5148269534111023,
      "learning_rate": 1.4236162194886827e-05,
      "loss": 0.2586,
      "step": 19980
    },
    {
      "epoch": 0.36084689330415504,
      "grad_norm": 0.455869197845459,
      "learning_rate": 1.4233592837398209e-05,
      "loss": 0.2671,
      "step": 19985
    },
    {
      "epoch": 0.3609371727370557,
      "grad_norm": 0.3940732479095459,
      "learning_rate": 1.4231023139347829e-05,
      "loss": 0.3339,
      "step": 19990
    },
    {
      "epoch": 0.36102745216995646,
      "grad_norm": 0.6617382764816284,
      "learning_rate": 1.42284531009424e-05,
      "loss": 0.3365,
      "step": 19995
    },
    {
      "epoch": 0.36111773160285715,
      "grad_norm": 0.23632988333702087,
      "learning_rate": 1.4225882722388662e-05,
      "loss": 0.2272,
      "step": 20000
    },
    {
      "epoch": 0.3612080110357579,
      "grad_norm": 0.3209882378578186,
      "learning_rate": 1.4223312003893383e-05,
      "loss": 0.2649,
      "step": 20005
    },
    {
      "epoch": 0.3612982904686586,
      "grad_norm": 0.5351271629333496,
      "learning_rate": 1.4220740945663361e-05,
      "loss": 0.2702,
      "step": 20010
    },
    {
      "epoch": 0.3613885699015593,
      "grad_norm": 1.513425350189209,
      "learning_rate": 1.4218169547905417e-05,
      "loss": 0.2621,
      "step": 20015
    },
    {
      "epoch": 0.36147884933446,
      "grad_norm": 0.35597509145736694,
      "learning_rate": 1.4215597810826405e-05,
      "loss": 0.299,
      "step": 20020
    },
    {
      "epoch": 0.36156912876736075,
      "grad_norm": 0.3152241110801697,
      "learning_rate": 1.4213025734633197e-05,
      "loss": 0.2417,
      "step": 20025
    },
    {
      "epoch": 0.36165940820026143,
      "grad_norm": 0.5848614573478699,
      "learning_rate": 1.4210453319532702e-05,
      "loss": 0.2618,
      "step": 20030
    },
    {
      "epoch": 0.3617496876331622,
      "grad_norm": 0.33715879917144775,
      "learning_rate": 1.420788056573185e-05,
      "loss": 0.2325,
      "step": 20035
    },
    {
      "epoch": 0.36183996706606286,
      "grad_norm": 0.6367013454437256,
      "learning_rate": 1.4205307473437597e-05,
      "loss": 0.2358,
      "step": 20040
    },
    {
      "epoch": 0.3619302464989636,
      "grad_norm": 0.38860565423965454,
      "learning_rate": 1.4202734042856938e-05,
      "loss": 0.3227,
      "step": 20045
    },
    {
      "epoch": 0.3620205259318643,
      "grad_norm": 0.5168018937110901,
      "learning_rate": 1.4200160274196881e-05,
      "loss": 0.3169,
      "step": 20050
    },
    {
      "epoch": 0.36211080536476503,
      "grad_norm": 0.39503562450408936,
      "learning_rate": 1.4197586167664466e-05,
      "loss": 0.2773,
      "step": 20055
    },
    {
      "epoch": 0.3622010847976657,
      "grad_norm": 0.48961323499679565,
      "learning_rate": 1.4195011723466765e-05,
      "loss": 0.238,
      "step": 20060
    },
    {
      "epoch": 0.36229136423056646,
      "grad_norm": 0.43903133273124695,
      "learning_rate": 1.4192436941810872e-05,
      "loss": 0.2567,
      "step": 20065
    },
    {
      "epoch": 0.36238164366346715,
      "grad_norm": 0.5009246468544006,
      "learning_rate": 1.4189861822903907e-05,
      "loss": 0.2474,
      "step": 20070
    },
    {
      "epoch": 0.3624719230963679,
      "grad_norm": 0.3318796455860138,
      "learning_rate": 1.4187286366953022e-05,
      "loss": 0.3398,
      "step": 20075
    },
    {
      "epoch": 0.3625622025292686,
      "grad_norm": 0.7951874136924744,
      "learning_rate": 1.4184710574165394e-05,
      "loss": 0.2387,
      "step": 20080
    },
    {
      "epoch": 0.3626524819621693,
      "grad_norm": 0.3770418167114258,
      "learning_rate": 1.4182134444748226e-05,
      "loss": 0.2669,
      "step": 20085
    },
    {
      "epoch": 0.36274276139507,
      "grad_norm": 0.32762718200683594,
      "learning_rate": 1.4179557978908748e-05,
      "loss": 0.2638,
      "step": 20090
    },
    {
      "epoch": 0.36283304082797074,
      "grad_norm": 0.637208878993988,
      "learning_rate": 1.4176981176854221e-05,
      "loss": 0.2871,
      "step": 20095
    },
    {
      "epoch": 0.36292332026087143,
      "grad_norm": 0.44785478711128235,
      "learning_rate": 1.4174404038791927e-05,
      "loss": 0.3969,
      "step": 20100
    },
    {
      "epoch": 0.36301359969377217,
      "grad_norm": 0.7451809644699097,
      "learning_rate": 1.4171826564929178e-05,
      "loss": 0.2978,
      "step": 20105
    },
    {
      "epoch": 0.36310387912667286,
      "grad_norm": 0.4416932761669159,
      "learning_rate": 1.4169248755473315e-05,
      "loss": 0.262,
      "step": 20110
    },
    {
      "epoch": 0.3631941585595736,
      "grad_norm": 0.5423164963722229,
      "learning_rate": 1.41666706106317e-05,
      "loss": 0.3489,
      "step": 20115
    },
    {
      "epoch": 0.3632844379924743,
      "grad_norm": 0.461900919675827,
      "learning_rate": 1.4164092130611735e-05,
      "loss": 0.3419,
      "step": 20120
    },
    {
      "epoch": 0.363374717425375,
      "grad_norm": 0.7312812805175781,
      "learning_rate": 1.416151331562083e-05,
      "loss": 0.325,
      "step": 20125
    },
    {
      "epoch": 0.3634649968582757,
      "grad_norm": 0.3819325566291809,
      "learning_rate": 1.4158934165866436e-05,
      "loss": 0.2519,
      "step": 20130
    },
    {
      "epoch": 0.36355527629117645,
      "grad_norm": 0.2581886053085327,
      "learning_rate": 1.4156354681556028e-05,
      "loss": 0.224,
      "step": 20135
    },
    {
      "epoch": 0.36364555572407714,
      "grad_norm": 0.44829660654067993,
      "learning_rate": 1.4153774862897102e-05,
      "loss": 0.2847,
      "step": 20140
    },
    {
      "epoch": 0.3637358351569779,
      "grad_norm": 0.4747091233730316,
      "learning_rate": 1.4151194710097194e-05,
      "loss": 0.2449,
      "step": 20145
    },
    {
      "epoch": 0.36382611458987857,
      "grad_norm": 0.9994378089904785,
      "learning_rate": 1.4148614223363852e-05,
      "loss": 0.2027,
      "step": 20150
    },
    {
      "epoch": 0.3639163940227793,
      "grad_norm": 0.3756590187549591,
      "learning_rate": 1.4146033402904657e-05,
      "loss": 0.2202,
      "step": 20155
    },
    {
      "epoch": 0.36400667345568,
      "grad_norm": 0.6926841139793396,
      "learning_rate": 1.4143452248927223e-05,
      "loss": 0.2173,
      "step": 20160
    },
    {
      "epoch": 0.36409695288858074,
      "grad_norm": 0.4116734266281128,
      "learning_rate": 1.4140870761639178e-05,
      "loss": 0.2754,
      "step": 20165
    },
    {
      "epoch": 0.3641872323214814,
      "grad_norm": 0.49817314743995667,
      "learning_rate": 1.4138288941248189e-05,
      "loss": 0.2571,
      "step": 20170
    },
    {
      "epoch": 0.36427751175438217,
      "grad_norm": 0.2756151854991913,
      "learning_rate": 1.413570678796194e-05,
      "loss": 0.2055,
      "step": 20175
    },
    {
      "epoch": 0.36436779118728285,
      "grad_norm": 1.4359185695648193,
      "learning_rate": 1.4133124301988155e-05,
      "loss": 0.379,
      "step": 20180
    },
    {
      "epoch": 0.3644580706201836,
      "grad_norm": 0.6018916368484497,
      "learning_rate": 1.4130541483534567e-05,
      "loss": 0.2789,
      "step": 20185
    },
    {
      "epoch": 0.3645483500530843,
      "grad_norm": 0.3269156217575073,
      "learning_rate": 1.4127958332808946e-05,
      "loss": 0.2321,
      "step": 20190
    },
    {
      "epoch": 0.364638629485985,
      "grad_norm": 0.4135490357875824,
      "learning_rate": 1.4125374850019095e-05,
      "loss": 0.2917,
      "step": 20195
    },
    {
      "epoch": 0.3647289089188857,
      "grad_norm": 0.23466555774211884,
      "learning_rate": 1.4122791035372831e-05,
      "loss": 0.2085,
      "step": 20200
    },
    {
      "epoch": 0.36481918835178645,
      "grad_norm": 0.800495982170105,
      "learning_rate": 1.4120206889078003e-05,
      "loss": 0.2336,
      "step": 20205
    },
    {
      "epoch": 0.36490946778468714,
      "grad_norm": 0.5827568173408508,
      "learning_rate": 1.4117622411342486e-05,
      "loss": 0.3036,
      "step": 20210
    },
    {
      "epoch": 0.3649997472175879,
      "grad_norm": 1.102327585220337,
      "learning_rate": 1.4115037602374188e-05,
      "loss": 0.257,
      "step": 20215
    },
    {
      "epoch": 0.3650900266504886,
      "grad_norm": 0.7294953465461731,
      "learning_rate": 1.411245246238103e-05,
      "loss": 0.2424,
      "step": 20220
    },
    {
      "epoch": 0.3651803060833893,
      "grad_norm": 0.4187539219856262,
      "learning_rate": 1.4109866991570975e-05,
      "loss": 0.2634,
      "step": 20225
    },
    {
      "epoch": 0.36527058551629005,
      "grad_norm": 0.40374043583869934,
      "learning_rate": 1.4107281190152e-05,
      "loss": 0.2914,
      "step": 20230
    },
    {
      "epoch": 0.36536086494919073,
      "grad_norm": 0.9304539561271667,
      "learning_rate": 1.4104695058332119e-05,
      "loss": 0.3806,
      "step": 20235
    },
    {
      "epoch": 0.3654511443820915,
      "grad_norm": 0.7214563488960266,
      "learning_rate": 1.4102108596319359e-05,
      "loss": 0.3112,
      "step": 20240
    },
    {
      "epoch": 0.36554142381499216,
      "grad_norm": 0.446776807308197,
      "learning_rate": 1.4099521804321792e-05,
      "loss": 0.2244,
      "step": 20245
    },
    {
      "epoch": 0.3656317032478929,
      "grad_norm": 0.568058431148529,
      "learning_rate": 1.40969346825475e-05,
      "loss": 0.2412,
      "step": 20250
    },
    {
      "epoch": 0.3657219826807936,
      "grad_norm": 0.7030618190765381,
      "learning_rate": 1.4094347231204602e-05,
      "loss": 0.1724,
      "step": 20255
    },
    {
      "epoch": 0.36581226211369433,
      "grad_norm": 0.3723071217536926,
      "learning_rate": 1.4091759450501236e-05,
      "loss": 0.3529,
      "step": 20260
    },
    {
      "epoch": 0.365902541546595,
      "grad_norm": 0.7579772472381592,
      "learning_rate": 1.4089171340645575e-05,
      "loss": 0.2549,
      "step": 20265
    },
    {
      "epoch": 0.36599282097949576,
      "grad_norm": 0.4400385916233063,
      "learning_rate": 1.408658290184581e-05,
      "loss": 0.2669,
      "step": 20270
    },
    {
      "epoch": 0.36608310041239644,
      "grad_norm": 0.39168331027030945,
      "learning_rate": 1.4083994134310162e-05,
      "loss": 0.2692,
      "step": 20275
    },
    {
      "epoch": 0.3661733798452972,
      "grad_norm": 0.44686418771743774,
      "learning_rate": 1.408140503824688e-05,
      "loss": 0.303,
      "step": 20280
    },
    {
      "epoch": 0.3662636592781979,
      "grad_norm": 0.48944777250289917,
      "learning_rate": 1.4078815613864239e-05,
      "loss": 0.2626,
      "step": 20285
    },
    {
      "epoch": 0.3663539387110986,
      "grad_norm": 0.4091298282146454,
      "learning_rate": 1.4076225861370531e-05,
      "loss": 0.216,
      "step": 20290
    },
    {
      "epoch": 0.3664442181439993,
      "grad_norm": 0.4132739305496216,
      "learning_rate": 1.4073635780974097e-05,
      "loss": 0.2436,
      "step": 20295
    },
    {
      "epoch": 0.36653449757690004,
      "grad_norm": 0.31656551361083984,
      "learning_rate": 1.4071045372883281e-05,
      "loss": 0.2046,
      "step": 20300
    },
    {
      "epoch": 0.36662477700980073,
      "grad_norm": 0.38056060671806335,
      "learning_rate": 1.4068454637306461e-05,
      "loss": 0.265,
      "step": 20305
    },
    {
      "epoch": 0.36671505644270147,
      "grad_norm": 0.44137445092201233,
      "learning_rate": 1.406586357445205e-05,
      "loss": 0.3173,
      "step": 20310
    },
    {
      "epoch": 0.36680533587560216,
      "grad_norm": 0.7222954034805298,
      "learning_rate": 1.4063272184528474e-05,
      "loss": 0.276,
      "step": 20315
    },
    {
      "epoch": 0.3668956153085029,
      "grad_norm": 0.521718442440033,
      "learning_rate": 1.4060680467744192e-05,
      "loss": 0.2463,
      "step": 20320
    },
    {
      "epoch": 0.3669858947414036,
      "grad_norm": 0.5072159171104431,
      "learning_rate": 1.4058088424307693e-05,
      "loss": 0.2716,
      "step": 20325
    },
    {
      "epoch": 0.3670761741743043,
      "grad_norm": 1.0182414054870605,
      "learning_rate": 1.4055496054427485e-05,
      "loss": 0.2574,
      "step": 20330
    },
    {
      "epoch": 0.367166453607205,
      "grad_norm": 0.3587333858013153,
      "learning_rate": 1.4052903358312105e-05,
      "loss": 0.2077,
      "step": 20335
    },
    {
      "epoch": 0.36725673304010575,
      "grad_norm": 0.9112486243247986,
      "learning_rate": 1.4050310336170117e-05,
      "loss": 0.2609,
      "step": 20340
    },
    {
      "epoch": 0.36734701247300644,
      "grad_norm": 0.8448390364646912,
      "learning_rate": 1.4047716988210112e-05,
      "loss": 0.3009,
      "step": 20345
    },
    {
      "epoch": 0.3674372919059072,
      "grad_norm": 0.45320257544517517,
      "learning_rate": 1.4045123314640705e-05,
      "loss": 0.2496,
      "step": 20350
    },
    {
      "epoch": 0.36752757133880787,
      "grad_norm": 0.49984174966812134,
      "learning_rate": 1.4042529315670537e-05,
      "loss": 0.2178,
      "step": 20355
    },
    {
      "epoch": 0.3676178507717086,
      "grad_norm": 0.4586024880409241,
      "learning_rate": 1.4039934991508278e-05,
      "loss": 0.2059,
      "step": 20360
    },
    {
      "epoch": 0.3677081302046093,
      "grad_norm": 0.5028537511825562,
      "learning_rate": 1.4037340342362624e-05,
      "loss": 0.3675,
      "step": 20365
    },
    {
      "epoch": 0.36779840963751004,
      "grad_norm": 0.5076274275779724,
      "learning_rate": 1.4034745368442293e-05,
      "loss": 0.2774,
      "step": 20370
    },
    {
      "epoch": 0.3678886890704107,
      "grad_norm": 0.5486329793930054,
      "learning_rate": 1.403215006995603e-05,
      "loss": 0.22,
      "step": 20375
    },
    {
      "epoch": 0.36797896850331147,
      "grad_norm": 0.6385655999183655,
      "learning_rate": 1.4029554447112612e-05,
      "loss": 0.3082,
      "step": 20380
    },
    {
      "epoch": 0.36806924793621215,
      "grad_norm": 0.46644681692123413,
      "learning_rate": 1.4026958500120837e-05,
      "loss": 0.2894,
      "step": 20385
    },
    {
      "epoch": 0.3681595273691129,
      "grad_norm": 0.4819391369819641,
      "learning_rate": 1.4024362229189528e-05,
      "loss": 0.3046,
      "step": 20390
    },
    {
      "epoch": 0.3682498068020136,
      "grad_norm": 0.5041378140449524,
      "learning_rate": 1.4021765634527537e-05,
      "loss": 0.349,
      "step": 20395
    },
    {
      "epoch": 0.3683400862349143,
      "grad_norm": 0.6669679880142212,
      "learning_rate": 1.4019168716343748e-05,
      "loss": 0.3217,
      "step": 20400
    },
    {
      "epoch": 0.368430365667815,
      "grad_norm": 0.5393589735031128,
      "learning_rate": 1.4016571474847051e-05,
      "loss": 0.1945,
      "step": 20405
    },
    {
      "epoch": 0.36852064510071575,
      "grad_norm": 0.5151278972625732,
      "learning_rate": 1.4013973910246385e-05,
      "loss": 0.3173,
      "step": 20410
    },
    {
      "epoch": 0.36861092453361644,
      "grad_norm": 0.4417298138141632,
      "learning_rate": 1.4011376022750702e-05,
      "loss": 0.304,
      "step": 20415
    },
    {
      "epoch": 0.3687012039665172,
      "grad_norm": 0.48443835973739624,
      "learning_rate": 1.4008777812568983e-05,
      "loss": 0.3338,
      "step": 20420
    },
    {
      "epoch": 0.36879148339941786,
      "grad_norm": 0.4232576787471771,
      "learning_rate": 1.4006179279910237e-05,
      "loss": 0.2347,
      "step": 20425
    },
    {
      "epoch": 0.3688817628323186,
      "grad_norm": 0.4605163335800171,
      "learning_rate": 1.4003580424983497e-05,
      "loss": 0.324,
      "step": 20430
    },
    {
      "epoch": 0.3689720422652193,
      "grad_norm": 0.3433309495449066,
      "learning_rate": 1.4000981247997816e-05,
      "loss": 0.192,
      "step": 20435
    },
    {
      "epoch": 0.36906232169812003,
      "grad_norm": 0.34693244099617004,
      "learning_rate": 1.3998381749162286e-05,
      "loss": 0.2916,
      "step": 20440
    },
    {
      "epoch": 0.3691526011310207,
      "grad_norm": 0.42569494247436523,
      "learning_rate": 1.3995781928686016e-05,
      "loss": 0.3048,
      "step": 20445
    },
    {
      "epoch": 0.36924288056392146,
      "grad_norm": 0.7065058350563049,
      "learning_rate": 1.399318178677814e-05,
      "loss": 0.3531,
      "step": 20450
    },
    {
      "epoch": 0.36933315999682215,
      "grad_norm": 0.39248576760292053,
      "learning_rate": 1.3990581323647822e-05,
      "loss": 0.3205,
      "step": 20455
    },
    {
      "epoch": 0.3694234394297229,
      "grad_norm": 0.3955816924571991,
      "learning_rate": 1.3987980539504254e-05,
      "loss": 0.3107,
      "step": 20460
    },
    {
      "epoch": 0.3695137188626236,
      "grad_norm": 0.6594250202178955,
      "learning_rate": 1.3985379434556644e-05,
      "loss": 0.2847,
      "step": 20465
    },
    {
      "epoch": 0.3696039982955243,
      "grad_norm": 0.35042914748191833,
      "learning_rate": 1.3982778009014239e-05,
      "loss": 0.2377,
      "step": 20470
    },
    {
      "epoch": 0.369694277728425,
      "grad_norm": 0.6024788618087769,
      "learning_rate": 1.3980176263086299e-05,
      "loss": 0.3413,
      "step": 20475
    },
    {
      "epoch": 0.36978455716132574,
      "grad_norm": 0.3722987771034241,
      "learning_rate": 1.3977574196982115e-05,
      "loss": 0.2523,
      "step": 20480
    },
    {
      "epoch": 0.36987483659422643,
      "grad_norm": 0.4409797489643097,
      "learning_rate": 1.3974971810911009e-05,
      "loss": 0.3234,
      "step": 20485
    },
    {
      "epoch": 0.36996511602712717,
      "grad_norm": 0.5501575469970703,
      "learning_rate": 1.397236910508232e-05,
      "loss": 0.2888,
      "step": 20490
    },
    {
      "epoch": 0.37005539546002786,
      "grad_norm": 0.5797925591468811,
      "learning_rate": 1.396976607970542e-05,
      "loss": 0.3273,
      "step": 20495
    },
    {
      "epoch": 0.3701456748929286,
      "grad_norm": 0.47177791595458984,
      "learning_rate": 1.3967162734989703e-05,
      "loss": 0.224,
      "step": 20500
    },
    {
      "epoch": 0.3702359543258293,
      "grad_norm": 0.3789895176887512,
      "learning_rate": 1.3964559071144584e-05,
      "loss": 0.2279,
      "step": 20505
    },
    {
      "epoch": 0.37032623375873003,
      "grad_norm": 0.35565948486328125,
      "learning_rate": 1.3961955088379516e-05,
      "loss": 0.1786,
      "step": 20510
    },
    {
      "epoch": 0.3704165131916307,
      "grad_norm": 0.40913644433021545,
      "learning_rate": 1.3959350786903968e-05,
      "loss": 0.1979,
      "step": 20515
    },
    {
      "epoch": 0.37050679262453146,
      "grad_norm": 0.4691525995731354,
      "learning_rate": 1.3956746166927433e-05,
      "loss": 0.2929,
      "step": 20520
    },
    {
      "epoch": 0.37059707205743214,
      "grad_norm": 0.328995019197464,
      "learning_rate": 1.3954141228659439e-05,
      "loss": 0.2487,
      "step": 20525
    },
    {
      "epoch": 0.3706873514903329,
      "grad_norm": 0.3796280324459076,
      "learning_rate": 1.3951535972309533e-05,
      "loss": 0.3008,
      "step": 20530
    },
    {
      "epoch": 0.37077763092323357,
      "grad_norm": 0.4332354962825775,
      "learning_rate": 1.3948930398087288e-05,
      "loss": 0.2755,
      "step": 20535
    },
    {
      "epoch": 0.3708679103561343,
      "grad_norm": 0.4433010220527649,
      "learning_rate": 1.3946324506202301e-05,
      "loss": 0.2256,
      "step": 20540
    },
    {
      "epoch": 0.370958189789035,
      "grad_norm": 0.38618332147598267,
      "learning_rate": 1.3943718296864205e-05,
      "loss": 0.3062,
      "step": 20545
    },
    {
      "epoch": 0.37104846922193574,
      "grad_norm": 0.48134633898735046,
      "learning_rate": 1.3941111770282644e-05,
      "loss": 0.2413,
      "step": 20550
    },
    {
      "epoch": 0.3711387486548364,
      "grad_norm": 1.5840861797332764,
      "learning_rate": 1.3938504926667295e-05,
      "loss": 0.3481,
      "step": 20555
    },
    {
      "epoch": 0.37122902808773717,
      "grad_norm": 0.6157742738723755,
      "learning_rate": 1.3935897766227862e-05,
      "loss": 0.3665,
      "step": 20560
    },
    {
      "epoch": 0.37131930752063785,
      "grad_norm": 0.40346086025238037,
      "learning_rate": 1.3933290289174069e-05,
      "loss": 0.3293,
      "step": 20565
    },
    {
      "epoch": 0.3714095869535386,
      "grad_norm": 0.41430744528770447,
      "learning_rate": 1.393068249571567e-05,
      "loss": 0.2168,
      "step": 20570
    },
    {
      "epoch": 0.3714998663864393,
      "grad_norm": 0.435720294713974,
      "learning_rate": 1.3928074386062442e-05,
      "loss": 0.2679,
      "step": 20575
    },
    {
      "epoch": 0.37159014581934,
      "grad_norm": 0.3936663568019867,
      "learning_rate": 1.392546596042419e-05,
      "loss": 0.352,
      "step": 20580
    },
    {
      "epoch": 0.3716804252522407,
      "grad_norm": 0.6342301368713379,
      "learning_rate": 1.3922857219010746e-05,
      "loss": 0.1906,
      "step": 20585
    },
    {
      "epoch": 0.37177070468514145,
      "grad_norm": 0.4912501573562622,
      "learning_rate": 1.3920248162031957e-05,
      "loss": 0.284,
      "step": 20590
    },
    {
      "epoch": 0.37186098411804214,
      "grad_norm": 0.47178614139556885,
      "learning_rate": 1.391763878969771e-05,
      "loss": 0.2427,
      "step": 20595
    },
    {
      "epoch": 0.3719512635509429,
      "grad_norm": 0.7745219469070435,
      "learning_rate": 1.3915029102217902e-05,
      "loss": 0.2535,
      "step": 20600
    },
    {
      "epoch": 0.37204154298384357,
      "grad_norm": 0.36101463437080383,
      "learning_rate": 1.3912419099802468e-05,
      "loss": 0.2445,
      "step": 20605
    },
    {
      "epoch": 0.3721318224167443,
      "grad_norm": 0.470927357673645,
      "learning_rate": 1.3909808782661368e-05,
      "loss": 0.2573,
      "step": 20610
    },
    {
      "epoch": 0.37222210184964505,
      "grad_norm": 0.39924320578575134,
      "learning_rate": 1.3907198151004572e-05,
      "loss": 0.3407,
      "step": 20615
    },
    {
      "epoch": 0.37231238128254573,
      "grad_norm": 0.6046448349952698,
      "learning_rate": 1.3904587205042096e-05,
      "loss": 0.1862,
      "step": 20620
    },
    {
      "epoch": 0.3724026607154465,
      "grad_norm": 0.4899580776691437,
      "learning_rate": 1.390197594498397e-05,
      "loss": 0.2631,
      "step": 20625
    },
    {
      "epoch": 0.37249294014834716,
      "grad_norm": 0.6295987963676453,
      "learning_rate": 1.3899364371040248e-05,
      "loss": 0.2755,
      "step": 20630
    },
    {
      "epoch": 0.3725832195812479,
      "grad_norm": 0.5494354367256165,
      "learning_rate": 1.3896752483421013e-05,
      "loss": 0.3084,
      "step": 20635
    },
    {
      "epoch": 0.3726734990141486,
      "grad_norm": 0.4012296199798584,
      "learning_rate": 1.3894140282336372e-05,
      "loss": 0.2626,
      "step": 20640
    },
    {
      "epoch": 0.37276377844704933,
      "grad_norm": 0.42580288648605347,
      "learning_rate": 1.389152776799646e-05,
      "loss": 0.3907,
      "step": 20645
    },
    {
      "epoch": 0.37285405787995,
      "grad_norm": 0.42306599020957947,
      "learning_rate": 1.3888914940611433e-05,
      "loss": 0.2946,
      "step": 20650
    },
    {
      "epoch": 0.37294433731285076,
      "grad_norm": 0.3592524230480194,
      "learning_rate": 1.3886301800391473e-05,
      "loss": 0.3285,
      "step": 20655
    },
    {
      "epoch": 0.37303461674575145,
      "grad_norm": 0.37943747639656067,
      "learning_rate": 1.3883688347546791e-05,
      "loss": 0.2716,
      "step": 20660
    },
    {
      "epoch": 0.3731248961786522,
      "grad_norm": 0.4590633809566498,
      "learning_rate": 1.3881074582287616e-05,
      "loss": 0.1602,
      "step": 20665
    },
    {
      "epoch": 0.3732151756115529,
      "grad_norm": 0.4163933992385864,
      "learning_rate": 1.3878460504824211e-05,
      "loss": 0.2677,
      "step": 20670
    },
    {
      "epoch": 0.3733054550444536,
      "grad_norm": 0.5285905003547668,
      "learning_rate": 1.387584611536686e-05,
      "loss": 0.1799,
      "step": 20675
    },
    {
      "epoch": 0.3733957344773543,
      "grad_norm": 0.4583968222141266,
      "learning_rate": 1.3873231414125868e-05,
      "loss": 0.2622,
      "step": 20680
    },
    {
      "epoch": 0.37348601391025504,
      "grad_norm": 0.4935058653354645,
      "learning_rate": 1.3870616401311565e-05,
      "loss": 0.2088,
      "step": 20685
    },
    {
      "epoch": 0.37357629334315573,
      "grad_norm": 0.5065711736679077,
      "learning_rate": 1.386800107713432e-05,
      "loss": 0.2332,
      "step": 20690
    },
    {
      "epoch": 0.37366657277605647,
      "grad_norm": 0.43563568592071533,
      "learning_rate": 1.3865385441804508e-05,
      "loss": 0.249,
      "step": 20695
    },
    {
      "epoch": 0.37375685220895716,
      "grad_norm": 0.46323660016059875,
      "learning_rate": 1.3862769495532542e-05,
      "loss": 0.1794,
      "step": 20700
    },
    {
      "epoch": 0.3738471316418579,
      "grad_norm": 0.6760948300361633,
      "learning_rate": 1.3860153238528855e-05,
      "loss": 0.3137,
      "step": 20705
    },
    {
      "epoch": 0.3739374110747586,
      "grad_norm": 0.4988843500614166,
      "learning_rate": 1.3857536671003908e-05,
      "loss": 0.3945,
      "step": 20710
    },
    {
      "epoch": 0.3740276905076593,
      "grad_norm": 0.5700904726982117,
      "learning_rate": 1.3854919793168183e-05,
      "loss": 0.2762,
      "step": 20715
    },
    {
      "epoch": 0.37411796994056,
      "grad_norm": 0.8820415139198303,
      "learning_rate": 1.3852302605232186e-05,
      "loss": 0.1863,
      "step": 20720
    },
    {
      "epoch": 0.37420824937346076,
      "grad_norm": 0.35778066515922546,
      "learning_rate": 1.3849685107406454e-05,
      "loss": 0.306,
      "step": 20725
    },
    {
      "epoch": 0.37429852880636144,
      "grad_norm": 0.3076165020465851,
      "learning_rate": 1.3847067299901547e-05,
      "loss": 0.1948,
      "step": 20730
    },
    {
      "epoch": 0.3743888082392622,
      "grad_norm": 0.5232268571853638,
      "learning_rate": 1.3844449182928044e-05,
      "loss": 0.1947,
      "step": 20735
    },
    {
      "epoch": 0.37447908767216287,
      "grad_norm": 0.5433857440948486,
      "learning_rate": 1.3841830756696557e-05,
      "loss": 0.2965,
      "step": 20740
    },
    {
      "epoch": 0.3745693671050636,
      "grad_norm": 0.35156840085983276,
      "learning_rate": 1.3839212021417718e-05,
      "loss": 0.1864,
      "step": 20745
    },
    {
      "epoch": 0.3746596465379643,
      "grad_norm": 0.36232057213783264,
      "learning_rate": 1.3836592977302188e-05,
      "loss": 0.1613,
      "step": 20750
    },
    {
      "epoch": 0.37474992597086504,
      "grad_norm": 0.33974605798721313,
      "learning_rate": 1.3833973624560644e-05,
      "loss": 0.2699,
      "step": 20755
    },
    {
      "epoch": 0.3748402054037657,
      "grad_norm": 0.2400882989168167,
      "learning_rate": 1.38313539634038e-05,
      "loss": 0.3049,
      "step": 20760
    },
    {
      "epoch": 0.37493048483666647,
      "grad_norm": 0.3881498873233795,
      "learning_rate": 1.3828733994042387e-05,
      "loss": 0.1705,
      "step": 20765
    },
    {
      "epoch": 0.37502076426956715,
      "grad_norm": 0.6005303263664246,
      "learning_rate": 1.3826113716687162e-05,
      "loss": 0.2377,
      "step": 20770
    },
    {
      "epoch": 0.3751110437024679,
      "grad_norm": 0.8644828200340271,
      "learning_rate": 1.3823493131548906e-05,
      "loss": 0.3372,
      "step": 20775
    },
    {
      "epoch": 0.3752013231353686,
      "grad_norm": 0.37639057636260986,
      "learning_rate": 1.3820872238838426e-05,
      "loss": 0.3125,
      "step": 20780
    },
    {
      "epoch": 0.3752916025682693,
      "grad_norm": 0.33708035945892334,
      "learning_rate": 1.3818251038766555e-05,
      "loss": 0.3017,
      "step": 20785
    },
    {
      "epoch": 0.37538188200117,
      "grad_norm": 0.28124114871025085,
      "learning_rate": 1.381562953154415e-05,
      "loss": 0.2504,
      "step": 20790
    },
    {
      "epoch": 0.37547216143407075,
      "grad_norm": 0.5806324481964111,
      "learning_rate": 1.3813007717382092e-05,
      "loss": 0.307,
      "step": 20795
    },
    {
      "epoch": 0.37556244086697144,
      "grad_norm": 0.43579456210136414,
      "learning_rate": 1.3810385596491282e-05,
      "loss": 0.2074,
      "step": 20800
    },
    {
      "epoch": 0.3756527202998722,
      "grad_norm": 0.34230563044548035,
      "learning_rate": 1.380776316908266e-05,
      "loss": 0.2473,
      "step": 20805
    },
    {
      "epoch": 0.37574299973277286,
      "grad_norm": 0.5098487734794617,
      "learning_rate": 1.3805140435367177e-05,
      "loss": 0.2555,
      "step": 20810
    },
    {
      "epoch": 0.3758332791656736,
      "grad_norm": 0.6360251307487488,
      "learning_rate": 1.380251739555581e-05,
      "loss": 0.3145,
      "step": 20815
    },
    {
      "epoch": 0.3759235585985743,
      "grad_norm": 0.3359304666519165,
      "learning_rate": 1.3799894049859563e-05,
      "loss": 0.3457,
      "step": 20820
    },
    {
      "epoch": 0.37601383803147503,
      "grad_norm": 0.7483426928520203,
      "learning_rate": 1.3797270398489471e-05,
      "loss": 0.3247,
      "step": 20825
    },
    {
      "epoch": 0.3761041174643757,
      "grad_norm": 0.31878402829170227,
      "learning_rate": 1.3794646441656583e-05,
      "loss": 0.1573,
      "step": 20830
    },
    {
      "epoch": 0.37619439689727646,
      "grad_norm": 0.3818129897117615,
      "learning_rate": 1.3792022179571977e-05,
      "loss": 0.2322,
      "step": 20835
    },
    {
      "epoch": 0.37628467633017715,
      "grad_norm": 0.4125143587589264,
      "learning_rate": 1.378939761244676e-05,
      "loss": 0.2426,
      "step": 20840
    },
    {
      "epoch": 0.3763749557630779,
      "grad_norm": 0.37870392203330994,
      "learning_rate": 1.3786772740492055e-05,
      "loss": 0.2716,
      "step": 20845
    },
    {
      "epoch": 0.3764652351959786,
      "grad_norm": 0.7333554625511169,
      "learning_rate": 1.378414756391902e-05,
      "loss": 0.2721,
      "step": 20850
    },
    {
      "epoch": 0.3765555146288793,
      "grad_norm": 0.4476829171180725,
      "learning_rate": 1.378152208293882e-05,
      "loss": 0.1462,
      "step": 20855
    },
    {
      "epoch": 0.37664579406178,
      "grad_norm": 0.33928850293159485,
      "learning_rate": 1.377889629776267e-05,
      "loss": 0.2591,
      "step": 20860
    },
    {
      "epoch": 0.37673607349468075,
      "grad_norm": 0.4483809769153595,
      "learning_rate": 1.377627020860178e-05,
      "loss": 0.2472,
      "step": 20865
    },
    {
      "epoch": 0.37682635292758143,
      "grad_norm": 0.3207505941390991,
      "learning_rate": 1.3773643815667413e-05,
      "loss": 0.257,
      "step": 20870
    },
    {
      "epoch": 0.3769166323604822,
      "grad_norm": 0.47099733352661133,
      "learning_rate": 1.3771017119170838e-05,
      "loss": 0.1816,
      "step": 20875
    },
    {
      "epoch": 0.37700691179338286,
      "grad_norm": 0.4887351989746094,
      "learning_rate": 1.3768390119323352e-05,
      "loss": 0.2619,
      "step": 20880
    },
    {
      "epoch": 0.3770971912262836,
      "grad_norm": 0.42729565501213074,
      "learning_rate": 1.3765762816336279e-05,
      "loss": 0.2753,
      "step": 20885
    },
    {
      "epoch": 0.3771874706591843,
      "grad_norm": 1.1441235542297363,
      "learning_rate": 1.3763135210420971e-05,
      "loss": 0.3516,
      "step": 20890
    },
    {
      "epoch": 0.37727775009208503,
      "grad_norm": 0.18297132849693298,
      "learning_rate": 1.3760507301788792e-05,
      "loss": 0.3361,
      "step": 20895
    },
    {
      "epoch": 0.3773680295249857,
      "grad_norm": 0.43833106756210327,
      "learning_rate": 1.3757879090651143e-05,
      "loss": 0.3026,
      "step": 20900
    },
    {
      "epoch": 0.37745830895788646,
      "grad_norm": 0.43935343623161316,
      "learning_rate": 1.3755250577219443e-05,
      "loss": 0.2527,
      "step": 20905
    },
    {
      "epoch": 0.37754858839078714,
      "grad_norm": 0.7618777751922607,
      "learning_rate": 1.375262176170514e-05,
      "loss": 0.2651,
      "step": 20910
    },
    {
      "epoch": 0.3776388678236879,
      "grad_norm": 0.434602290391922,
      "learning_rate": 1.3749992644319701e-05,
      "loss": 0.2037,
      "step": 20915
    },
    {
      "epoch": 0.37772914725658857,
      "grad_norm": 0.8464885354042053,
      "learning_rate": 1.3747363225274615e-05,
      "loss": 0.266,
      "step": 20920
    },
    {
      "epoch": 0.3778194266894893,
      "grad_norm": 0.42587873339653015,
      "learning_rate": 1.3744733504781405e-05,
      "loss": 0.2134,
      "step": 20925
    },
    {
      "epoch": 0.37790970612239,
      "grad_norm": 0.3875392973423004,
      "learning_rate": 1.3742103483051612e-05,
      "loss": 0.2555,
      "step": 20930
    },
    {
      "epoch": 0.37799998555529074,
      "grad_norm": 0.4170796573162079,
      "learning_rate": 1.3739473160296798e-05,
      "loss": 0.2634,
      "step": 20935
    },
    {
      "epoch": 0.3780902649881914,
      "grad_norm": 0.411149799823761,
      "learning_rate": 1.3736842536728563e-05,
      "loss": 0.239,
      "step": 20940
    },
    {
      "epoch": 0.37818054442109217,
      "grad_norm": 0.4498842656612396,
      "learning_rate": 1.3734211612558513e-05,
      "loss": 0.2336,
      "step": 20945
    },
    {
      "epoch": 0.37827082385399285,
      "grad_norm": 0.382417231798172,
      "learning_rate": 1.373158038799829e-05,
      "loss": 0.3019,
      "step": 20950
    },
    {
      "epoch": 0.3783611032868936,
      "grad_norm": 0.3414664566516876,
      "learning_rate": 1.3728948863259555e-05,
      "loss": 0.3018,
      "step": 20955
    },
    {
      "epoch": 0.3784513827197943,
      "grad_norm": 0.5627580285072327,
      "learning_rate": 1.3726317038553996e-05,
      "loss": 0.2927,
      "step": 20960
    },
    {
      "epoch": 0.378541662152695,
      "grad_norm": 0.36109137535095215,
      "learning_rate": 1.372368491409332e-05,
      "loss": 0.2494,
      "step": 20965
    },
    {
      "epoch": 0.3786319415855957,
      "grad_norm": 0.2758963406085968,
      "learning_rate": 1.3721052490089272e-05,
      "loss": 0.2305,
      "step": 20970
    },
    {
      "epoch": 0.37872222101849645,
      "grad_norm": 0.37376853823661804,
      "learning_rate": 1.3718419766753603e-05,
      "loss": 0.2622,
      "step": 20975
    },
    {
      "epoch": 0.37881250045139714,
      "grad_norm": 0.47796663641929626,
      "learning_rate": 1.3715786744298103e-05,
      "loss": 0.335,
      "step": 20980
    },
    {
      "epoch": 0.3789027798842979,
      "grad_norm": 0.4772331118583679,
      "learning_rate": 1.3713153422934571e-05,
      "loss": 0.3328,
      "step": 20985
    },
    {
      "epoch": 0.37899305931719857,
      "grad_norm": 0.29484912753105164,
      "learning_rate": 1.3710519802874847e-05,
      "loss": 0.2899,
      "step": 20990
    },
    {
      "epoch": 0.3790833387500993,
      "grad_norm": 0.3672797977924347,
      "learning_rate": 1.3707885884330781e-05,
      "loss": 0.3458,
      "step": 20995
    },
    {
      "epoch": 0.379173618183,
      "grad_norm": 0.33409571647644043,
      "learning_rate": 1.3705251667514255e-05,
      "loss": 0.2574,
      "step": 21000
    },
    {
      "epoch": 0.37926389761590074,
      "grad_norm": 0.35732537508010864,
      "learning_rate": 1.370261715263717e-05,
      "loss": 0.2169,
      "step": 21005
    },
    {
      "epoch": 0.3793541770488015,
      "grad_norm": 0.4787987768650055,
      "learning_rate": 1.3699982339911459e-05,
      "loss": 0.2929,
      "step": 21010
    },
    {
      "epoch": 0.37944445648170216,
      "grad_norm": 0.566339910030365,
      "learning_rate": 1.369734722954907e-05,
      "loss": 0.3447,
      "step": 21015
    },
    {
      "epoch": 0.3795347359146029,
      "grad_norm": 0.5799692869186401,
      "learning_rate": 1.3694711821761976e-05,
      "loss": 0.2302,
      "step": 21020
    },
    {
      "epoch": 0.3796250153475036,
      "grad_norm": 0.5670090913772583,
      "learning_rate": 1.369207611676218e-05,
      "loss": 0.2486,
      "step": 21025
    },
    {
      "epoch": 0.37971529478040433,
      "grad_norm": 0.4892391264438629,
      "learning_rate": 1.3689440114761708e-05,
      "loss": 0.3051,
      "step": 21030
    },
    {
      "epoch": 0.379805574213305,
      "grad_norm": 0.6479849815368652,
      "learning_rate": 1.3686803815972599e-05,
      "loss": 0.2842,
      "step": 21035
    },
    {
      "epoch": 0.37989585364620576,
      "grad_norm": 0.37288862466812134,
      "learning_rate": 1.3684167220606929e-05,
      "loss": 0.2737,
      "step": 21040
    },
    {
      "epoch": 0.37998613307910645,
      "grad_norm": 0.4827922284603119,
      "learning_rate": 1.3681530328876798e-05,
      "loss": 0.2171,
      "step": 21045
    },
    {
      "epoch": 0.3800764125120072,
      "grad_norm": 0.49152135848999023,
      "learning_rate": 1.3678893140994314e-05,
      "loss": 0.3088,
      "step": 21050
    },
    {
      "epoch": 0.3801666919449079,
      "grad_norm": 0.40357691049575806,
      "learning_rate": 1.3676255657171627e-05,
      "loss": 0.3664,
      "step": 21055
    },
    {
      "epoch": 0.3802569713778086,
      "grad_norm": 0.5374104976654053,
      "learning_rate": 1.3673617877620902e-05,
      "loss": 0.3357,
      "step": 21060
    },
    {
      "epoch": 0.3803472508107093,
      "grad_norm": 0.5262294411659241,
      "learning_rate": 1.3670979802554325e-05,
      "loss": 0.1575,
      "step": 21065
    },
    {
      "epoch": 0.38043753024361004,
      "grad_norm": 0.48867297172546387,
      "learning_rate": 1.3668341432184116e-05,
      "loss": 0.2814,
      "step": 21070
    },
    {
      "epoch": 0.38052780967651073,
      "grad_norm": 0.2917868196964264,
      "learning_rate": 1.366570276672251e-05,
      "loss": 0.1978,
      "step": 21075
    },
    {
      "epoch": 0.3806180891094115,
      "grad_norm": 0.4150635004043579,
      "learning_rate": 1.366306380638177e-05,
      "loss": 0.3177,
      "step": 21080
    },
    {
      "epoch": 0.38070836854231216,
      "grad_norm": 0.45227259397506714,
      "learning_rate": 1.3660424551374175e-05,
      "loss": 0.2428,
      "step": 21085
    },
    {
      "epoch": 0.3807986479752129,
      "grad_norm": 0.43083256483078003,
      "learning_rate": 1.3657785001912045e-05,
      "loss": 0.349,
      "step": 21090
    },
    {
      "epoch": 0.3808889274081136,
      "grad_norm": 0.36342352628707886,
      "learning_rate": 1.3655145158207703e-05,
      "loss": 0.2375,
      "step": 21095
    },
    {
      "epoch": 0.38097920684101433,
      "grad_norm": 0.9517899751663208,
      "learning_rate": 1.3652505020473509e-05,
      "loss": 0.2719,
      "step": 21100
    },
    {
      "epoch": 0.381069486273915,
      "grad_norm": 0.6163972616195679,
      "learning_rate": 1.3649864588921841e-05,
      "loss": 0.3468,
      "step": 21105
    },
    {
      "epoch": 0.38115976570681576,
      "grad_norm": 0.6210305094718933,
      "learning_rate": 1.3647223863765107e-05,
      "loss": 0.2118,
      "step": 21110
    },
    {
      "epoch": 0.38125004513971644,
      "grad_norm": 0.5331239700317383,
      "learning_rate": 1.364458284521573e-05,
      "loss": 0.3312,
      "step": 21115
    },
    {
      "epoch": 0.3813403245726172,
      "grad_norm": 0.7067062854766846,
      "learning_rate": 1.3641941533486158e-05,
      "loss": 0.3288,
      "step": 21120
    },
    {
      "epoch": 0.38143060400551787,
      "grad_norm": 0.4058865010738373,
      "learning_rate": 1.3639299928788875e-05,
      "loss": 0.2042,
      "step": 21125
    },
    {
      "epoch": 0.3815208834384186,
      "grad_norm": 0.3381866216659546,
      "learning_rate": 1.363665803133637e-05,
      "loss": 0.2507,
      "step": 21130
    },
    {
      "epoch": 0.3816111628713193,
      "grad_norm": 0.5158727765083313,
      "learning_rate": 1.3634015841341168e-05,
      "loss": 0.2603,
      "step": 21135
    },
    {
      "epoch": 0.38170144230422004,
      "grad_norm": 0.6629533171653748,
      "learning_rate": 1.3631373359015813e-05,
      "loss": 0.2761,
      "step": 21140
    },
    {
      "epoch": 0.3817917217371207,
      "grad_norm": 0.6049644947052002,
      "learning_rate": 1.3628730584572874e-05,
      "loss": 0.2861,
      "step": 21145
    },
    {
      "epoch": 0.38188200117002147,
      "grad_norm": 0.4999704658985138,
      "learning_rate": 1.362608751822494e-05,
      "loss": 0.3413,
      "step": 21150
    },
    {
      "epoch": 0.38197228060292215,
      "grad_norm": 0.3376184403896332,
      "learning_rate": 1.3623444160184633e-05,
      "loss": 0.2189,
      "step": 21155
    },
    {
      "epoch": 0.3820625600358229,
      "grad_norm": 0.4643104672431946,
      "learning_rate": 1.3620800510664588e-05,
      "loss": 0.2927,
      "step": 21160
    },
    {
      "epoch": 0.3821528394687236,
      "grad_norm": 0.37722495198249817,
      "learning_rate": 1.3618156569877466e-05,
      "loss": 0.2584,
      "step": 21165
    },
    {
      "epoch": 0.3822431189016243,
      "grad_norm": 0.4606938660144806,
      "learning_rate": 1.3615512338035957e-05,
      "loss": 0.304,
      "step": 21170
    },
    {
      "epoch": 0.382333398334525,
      "grad_norm": 0.31286484003067017,
      "learning_rate": 1.3612867815352766e-05,
      "loss": 0.1905,
      "step": 21175
    },
    {
      "epoch": 0.38242367776742575,
      "grad_norm": 0.37826356291770935,
      "learning_rate": 1.3610223002040628e-05,
      "loss": 0.1855,
      "step": 21180
    },
    {
      "epoch": 0.38251395720032644,
      "grad_norm": 0.43756335973739624,
      "learning_rate": 1.3607577898312297e-05,
      "loss": 0.229,
      "step": 21185
    },
    {
      "epoch": 0.3826042366332272,
      "grad_norm": 0.4016486406326294,
      "learning_rate": 1.3604932504380554e-05,
      "loss": 0.3214,
      "step": 21190
    },
    {
      "epoch": 0.38269451606612787,
      "grad_norm": 0.523820698261261,
      "learning_rate": 1.3602286820458204e-05,
      "loss": 0.3178,
      "step": 21195
    },
    {
      "epoch": 0.3827847954990286,
      "grad_norm": 0.43063297867774963,
      "learning_rate": 1.3599640846758067e-05,
      "loss": 0.2159,
      "step": 21200
    },
    {
      "epoch": 0.3828750749319293,
      "grad_norm": 0.5800107717514038,
      "learning_rate": 1.3596994583492997e-05,
      "loss": 0.2275,
      "step": 21205
    },
    {
      "epoch": 0.38296535436483004,
      "grad_norm": 0.4514652490615845,
      "learning_rate": 1.3594348030875866e-05,
      "loss": 0.2793,
      "step": 21210
    },
    {
      "epoch": 0.3830556337977307,
      "grad_norm": 0.4241120517253876,
      "learning_rate": 1.359170118911957e-05,
      "loss": 0.3655,
      "step": 21215
    },
    {
      "epoch": 0.38314591323063146,
      "grad_norm": 0.5384905934333801,
      "learning_rate": 1.3589054058437026e-05,
      "loss": 0.2727,
      "step": 21220
    },
    {
      "epoch": 0.38323619266353215,
      "grad_norm": 0.3744039535522461,
      "learning_rate": 1.3586406639041179e-05,
      "loss": 0.2432,
      "step": 21225
    },
    {
      "epoch": 0.3833264720964329,
      "grad_norm": 0.4343605935573578,
      "learning_rate": 1.3583758931144995e-05,
      "loss": 0.2756,
      "step": 21230
    },
    {
      "epoch": 0.3834167515293336,
      "grad_norm": 0.6409178972244263,
      "learning_rate": 1.3581110934961459e-05,
      "loss": 0.3335,
      "step": 21235
    },
    {
      "epoch": 0.3835070309622343,
      "grad_norm": 0.3212273418903351,
      "learning_rate": 1.3578462650703585e-05,
      "loss": 0.215,
      "step": 21240
    },
    {
      "epoch": 0.383597310395135,
      "grad_norm": 0.3993001878261566,
      "learning_rate": 1.3575814078584411e-05,
      "loss": 0.2568,
      "step": 21245
    },
    {
      "epoch": 0.38368758982803575,
      "grad_norm": 0.8827431797981262,
      "learning_rate": 1.3573165218816991e-05,
      "loss": 0.2671,
      "step": 21250
    },
    {
      "epoch": 0.38377786926093643,
      "grad_norm": 0.3880627751350403,
      "learning_rate": 1.357051607161441e-05,
      "loss": 0.2351,
      "step": 21255
    },
    {
      "epoch": 0.3838681486938372,
      "grad_norm": 0.45617276430130005,
      "learning_rate": 1.356786663718977e-05,
      "loss": 0.2768,
      "step": 21260
    },
    {
      "epoch": 0.38395842812673786,
      "grad_norm": 0.4872366487979889,
      "learning_rate": 1.3565216915756199e-05,
      "loss": 0.2729,
      "step": 21265
    },
    {
      "epoch": 0.3840487075596386,
      "grad_norm": 0.638798177242279,
      "learning_rate": 1.3562566907526851e-05,
      "loss": 0.3035,
      "step": 21270
    },
    {
      "epoch": 0.3841389869925393,
      "grad_norm": 0.7527413964271545,
      "learning_rate": 1.3559916612714898e-05,
      "loss": 0.3511,
      "step": 21275
    },
    {
      "epoch": 0.38422926642544003,
      "grad_norm": 0.5562458634376526,
      "learning_rate": 1.3557266031533536e-05,
      "loss": 0.2414,
      "step": 21280
    },
    {
      "epoch": 0.3843195458583407,
      "grad_norm": 0.4700249135494232,
      "learning_rate": 1.3554615164195983e-05,
      "loss": 0.2109,
      "step": 21285
    },
    {
      "epoch": 0.38440982529124146,
      "grad_norm": 0.31953826546669006,
      "learning_rate": 1.3551964010915488e-05,
      "loss": 0.2183,
      "step": 21290
    },
    {
      "epoch": 0.38450010472414214,
      "grad_norm": 0.4933184087276459,
      "learning_rate": 1.3549312571905313e-05,
      "loss": 0.2185,
      "step": 21295
    },
    {
      "epoch": 0.3845903841570429,
      "grad_norm": 0.5202569961547852,
      "learning_rate": 1.3546660847378746e-05,
      "loss": 0.2111,
      "step": 21300
    },
    {
      "epoch": 0.38468066358994357,
      "grad_norm": 0.5522760152816772,
      "learning_rate": 1.35440088375491e-05,
      "loss": 0.2552,
      "step": 21305
    },
    {
      "epoch": 0.3847709430228443,
      "grad_norm": 0.4002067744731903,
      "learning_rate": 1.3541356542629713e-05,
      "loss": 0.2786,
      "step": 21310
    },
    {
      "epoch": 0.384861222455745,
      "grad_norm": 0.44965609908103943,
      "learning_rate": 1.353870396283394e-05,
      "loss": 0.2857,
      "step": 21315
    },
    {
      "epoch": 0.38495150188864574,
      "grad_norm": 0.4091430604457855,
      "learning_rate": 1.3536051098375162e-05,
      "loss": 0.3082,
      "step": 21320
    },
    {
      "epoch": 0.38504178132154643,
      "grad_norm": 0.4339289963245392,
      "learning_rate": 1.3533397949466785e-05,
      "loss": 0.2586,
      "step": 21325
    },
    {
      "epoch": 0.38513206075444717,
      "grad_norm": 0.5966063737869263,
      "learning_rate": 1.3530744516322227e-05,
      "loss": 0.3589,
      "step": 21330
    },
    {
      "epoch": 0.38522234018734786,
      "grad_norm": 0.5281093716621399,
      "learning_rate": 1.3528090799154946e-05,
      "loss": 0.3192,
      "step": 21335
    },
    {
      "epoch": 0.3853126196202486,
      "grad_norm": 1.1246488094329834,
      "learning_rate": 1.3525436798178416e-05,
      "loss": 0.2516,
      "step": 21340
    },
    {
      "epoch": 0.3854028990531493,
      "grad_norm": 0.5170313715934753,
      "learning_rate": 1.3522782513606123e-05,
      "loss": 0.258,
      "step": 21345
    },
    {
      "epoch": 0.38549317848605,
      "grad_norm": 0.3201058804988861,
      "learning_rate": 1.352012794565159e-05,
      "loss": 0.3076,
      "step": 21350
    },
    {
      "epoch": 0.3855834579189507,
      "grad_norm": 0.5544672608375549,
      "learning_rate": 1.3517473094528359e-05,
      "loss": 0.1814,
      "step": 21355
    },
    {
      "epoch": 0.38567373735185145,
      "grad_norm": 0.32615146040916443,
      "learning_rate": 1.3514817960449994e-05,
      "loss": 0.2321,
      "step": 21360
    },
    {
      "epoch": 0.38576401678475214,
      "grad_norm": 0.4158816337585449,
      "learning_rate": 1.3512162543630076e-05,
      "loss": 0.3257,
      "step": 21365
    },
    {
      "epoch": 0.3858542962176529,
      "grad_norm": 0.4445611536502838,
      "learning_rate": 1.3509506844282221e-05,
      "loss": 0.2883,
      "step": 21370
    },
    {
      "epoch": 0.38594457565055357,
      "grad_norm": 0.5737717151641846,
      "learning_rate": 1.3506850862620055e-05,
      "loss": 0.2249,
      "step": 21375
    },
    {
      "epoch": 0.3860348550834543,
      "grad_norm": 0.5599545240402222,
      "learning_rate": 1.3504194598857234e-05,
      "loss": 0.2635,
      "step": 21380
    },
    {
      "epoch": 0.386125134516355,
      "grad_norm": 0.47273439168930054,
      "learning_rate": 1.3501538053207435e-05,
      "loss": 0.2185,
      "step": 21385
    },
    {
      "epoch": 0.38621541394925574,
      "grad_norm": 0.3157688081264496,
      "learning_rate": 1.349888122588436e-05,
      "loss": 0.2757,
      "step": 21390
    },
    {
      "epoch": 0.3863056933821564,
      "grad_norm": 0.38830289244651794,
      "learning_rate": 1.3496224117101728e-05,
      "loss": 0.2529,
      "step": 21395
    },
    {
      "epoch": 0.38639597281505716,
      "grad_norm": 0.26463016867637634,
      "learning_rate": 1.3493566727073284e-05,
      "loss": 0.2809,
      "step": 21400
    },
    {
      "epoch": 0.3864862522479579,
      "grad_norm": 0.2879667580127716,
      "learning_rate": 1.3490909056012803e-05,
      "loss": 0.2566,
      "step": 21405
    },
    {
      "epoch": 0.3865765316808586,
      "grad_norm": 0.4215259850025177,
      "learning_rate": 1.3488251104134065e-05,
      "loss": 0.287,
      "step": 21410
    },
    {
      "epoch": 0.38666681111375933,
      "grad_norm": 0.40119993686676025,
      "learning_rate": 1.3485592871650889e-05,
      "loss": 0.2952,
      "step": 21415
    },
    {
      "epoch": 0.38675709054666,
      "grad_norm": 0.3995286226272583,
      "learning_rate": 1.348293435877711e-05,
      "loss": 0.2095,
      "step": 21420
    },
    {
      "epoch": 0.38684736997956076,
      "grad_norm": 0.46701765060424805,
      "learning_rate": 1.3480275565726585e-05,
      "loss": 0.286,
      "step": 21425
    },
    {
      "epoch": 0.38693764941246145,
      "grad_norm": 1.2765496969223022,
      "learning_rate": 1.3477616492713195e-05,
      "loss": 0.2907,
      "step": 21430
    },
    {
      "epoch": 0.3870279288453622,
      "grad_norm": 0.43694746494293213,
      "learning_rate": 1.3474957139950844e-05,
      "loss": 0.2878,
      "step": 21435
    },
    {
      "epoch": 0.3871182082782629,
      "grad_norm": 0.41188398003578186,
      "learning_rate": 1.3472297507653454e-05,
      "loss": 0.2602,
      "step": 21440
    },
    {
      "epoch": 0.3872084877111636,
      "grad_norm": 0.6653173565864563,
      "learning_rate": 1.3469637596034978e-05,
      "loss": 0.2713,
      "step": 21445
    },
    {
      "epoch": 0.3872987671440643,
      "grad_norm": 0.49647408723831177,
      "learning_rate": 1.346697740530938e-05,
      "loss": 0.2557,
      "step": 21450
    },
    {
      "epoch": 0.38738904657696505,
      "grad_norm": 0.3037748336791992,
      "learning_rate": 1.346431693569066e-05,
      "loss": 0.1898,
      "step": 21455
    },
    {
      "epoch": 0.38747932600986573,
      "grad_norm": 0.4379635155200958,
      "learning_rate": 1.346165618739283e-05,
      "loss": 0.2372,
      "step": 21460
    },
    {
      "epoch": 0.3875696054427665,
      "grad_norm": 0.5771574974060059,
      "learning_rate": 1.3458995160629928e-05,
      "loss": 0.2403,
      "step": 21465
    },
    {
      "epoch": 0.38765988487566716,
      "grad_norm": 0.7019838690757751,
      "learning_rate": 1.3456333855616017e-05,
      "loss": 0.2481,
      "step": 21470
    },
    {
      "epoch": 0.3877501643085679,
      "grad_norm": 0.4674120247364044,
      "learning_rate": 1.3453672272565175e-05,
      "loss": 0.2677,
      "step": 21475
    },
    {
      "epoch": 0.3878404437414686,
      "grad_norm": 0.8151453733444214,
      "learning_rate": 1.345101041169151e-05,
      "loss": 0.189,
      "step": 21480
    },
    {
      "epoch": 0.38793072317436933,
      "grad_norm": 0.35866454243659973,
      "learning_rate": 1.3448348273209147e-05,
      "loss": 0.1593,
      "step": 21485
    },
    {
      "epoch": 0.38802100260727,
      "grad_norm": 0.5386594533920288,
      "learning_rate": 1.344568585733224e-05,
      "loss": 0.2609,
      "step": 21490
    },
    {
      "epoch": 0.38811128204017076,
      "grad_norm": 0.41673269867897034,
      "learning_rate": 1.3443023164274958e-05,
      "loss": 0.2291,
      "step": 21495
    },
    {
      "epoch": 0.38820156147307144,
      "grad_norm": 0.3212771415710449,
      "learning_rate": 1.3440360194251494e-05,
      "loss": 0.2891,
      "step": 21500
    },
    {
      "epoch": 0.3882918409059722,
      "grad_norm": 0.5306001901626587,
      "learning_rate": 1.3437696947476068e-05,
      "loss": 0.2982,
      "step": 21505
    },
    {
      "epoch": 0.38838212033887287,
      "grad_norm": 0.3461056351661682,
      "learning_rate": 1.3435033424162917e-05,
      "loss": 0.2977,
      "step": 21510
    },
    {
      "epoch": 0.3884723997717736,
      "grad_norm": 0.4729425609111786,
      "learning_rate": 1.34323696245263e-05,
      "loss": 0.361,
      "step": 21515
    },
    {
      "epoch": 0.3885626792046743,
      "grad_norm": 0.3501926064491272,
      "learning_rate": 1.3429705548780505e-05,
      "loss": 0.2411,
      "step": 21520
    },
    {
      "epoch": 0.38865295863757504,
      "grad_norm": 0.3689830005168915,
      "learning_rate": 1.3427041197139835e-05,
      "loss": 0.2079,
      "step": 21525
    },
    {
      "epoch": 0.3887432380704757,
      "grad_norm": 0.3276408016681671,
      "learning_rate": 1.3424376569818614e-05,
      "loss": 0.3374,
      "step": 21530
    },
    {
      "epoch": 0.38883351750337647,
      "grad_norm": 0.4310067594051361,
      "learning_rate": 1.34217116670312e-05,
      "loss": 0.273,
      "step": 21535
    },
    {
      "epoch": 0.38892379693627716,
      "grad_norm": 0.3747933506965637,
      "learning_rate": 1.3419046488991955e-05,
      "loss": 0.1641,
      "step": 21540
    },
    {
      "epoch": 0.3890140763691779,
      "grad_norm": 0.3694609999656677,
      "learning_rate": 1.3416381035915282e-05,
      "loss": 0.2563,
      "step": 21545
    },
    {
      "epoch": 0.3891043558020786,
      "grad_norm": 0.3853660821914673,
      "learning_rate": 1.3413715308015596e-05,
      "loss": 0.3523,
      "step": 21550
    },
    {
      "epoch": 0.3891946352349793,
      "grad_norm": 0.4884442090988159,
      "learning_rate": 1.3411049305507332e-05,
      "loss": 0.2402,
      "step": 21555
    },
    {
      "epoch": 0.38928491466788,
      "grad_norm": 0.5782903432846069,
      "learning_rate": 1.3408383028604952e-05,
      "loss": 0.2228,
      "step": 21560
    },
    {
      "epoch": 0.38937519410078075,
      "grad_norm": 0.5393293499946594,
      "learning_rate": 1.3405716477522938e-05,
      "loss": 0.2738,
      "step": 21565
    },
    {
      "epoch": 0.38946547353368144,
      "grad_norm": 0.505789577960968,
      "learning_rate": 1.3403049652475797e-05,
      "loss": 0.2972,
      "step": 21570
    },
    {
      "epoch": 0.3895557529665822,
      "grad_norm": 0.6107163429260254,
      "learning_rate": 1.3400382553678053e-05,
      "loss": 0.3708,
      "step": 21575
    },
    {
      "epoch": 0.38964603239948287,
      "grad_norm": 0.41639187932014465,
      "learning_rate": 1.3397715181344255e-05,
      "loss": 0.336,
      "step": 21580
    },
    {
      "epoch": 0.3897363118323836,
      "grad_norm": 0.5116377472877502,
      "learning_rate": 1.3395047535688977e-05,
      "loss": 0.1877,
      "step": 21585
    },
    {
      "epoch": 0.3898265912652843,
      "grad_norm": 0.5431540012359619,
      "learning_rate": 1.3392379616926808e-05,
      "loss": 0.2468,
      "step": 21590
    },
    {
      "epoch": 0.38991687069818504,
      "grad_norm": 0.562221884727478,
      "learning_rate": 1.3389711425272363e-05,
      "loss": 0.2512,
      "step": 21595
    },
    {
      "epoch": 0.3900071501310857,
      "grad_norm": 0.4357643127441406,
      "learning_rate": 1.338704296094028e-05,
      "loss": 0.1748,
      "step": 21600
    },
    {
      "epoch": 0.39009742956398646,
      "grad_norm": 0.5541381239891052,
      "learning_rate": 1.3384374224145219e-05,
      "loss": 0.2372,
      "step": 21605
    },
    {
      "epoch": 0.39018770899688715,
      "grad_norm": 0.5287481546401978,
      "learning_rate": 1.3381705215101857e-05,
      "loss": 0.3041,
      "step": 21610
    },
    {
      "epoch": 0.3902779884297879,
      "grad_norm": 0.5050212144851685,
      "learning_rate": 1.3379035934024894e-05,
      "loss": 0.239,
      "step": 21615
    },
    {
      "epoch": 0.3903682678626886,
      "grad_norm": 0.6532121896743774,
      "learning_rate": 1.337636638112906e-05,
      "loss": 0.2447,
      "step": 21620
    },
    {
      "epoch": 0.3904585472955893,
      "grad_norm": 0.4641731381416321,
      "learning_rate": 1.3373696556629102e-05,
      "loss": 0.2577,
      "step": 21625
    },
    {
      "epoch": 0.39054882672849,
      "grad_norm": 0.36768922209739685,
      "learning_rate": 1.3371026460739783e-05,
      "loss": 0.2899,
      "step": 21630
    },
    {
      "epoch": 0.39063910616139075,
      "grad_norm": 0.4381278455257416,
      "learning_rate": 1.3368356093675896e-05,
      "loss": 0.2455,
      "step": 21635
    },
    {
      "epoch": 0.39072938559429143,
      "grad_norm": 0.4196430444717407,
      "learning_rate": 1.336568545565225e-05,
      "loss": 0.2358,
      "step": 21640
    },
    {
      "epoch": 0.3908196650271922,
      "grad_norm": 0.6201593279838562,
      "learning_rate": 1.3363014546883679e-05,
      "loss": 0.2729,
      "step": 21645
    },
    {
      "epoch": 0.39090994446009286,
      "grad_norm": 0.6143485903739929,
      "learning_rate": 1.3360343367585042e-05,
      "loss": 0.2572,
      "step": 21650
    },
    {
      "epoch": 0.3910002238929936,
      "grad_norm": 0.2882515788078308,
      "learning_rate": 1.3357671917971213e-05,
      "loss": 0.184,
      "step": 21655
    },
    {
      "epoch": 0.3910905033258943,
      "grad_norm": 0.43455737829208374,
      "learning_rate": 1.335500019825709e-05,
      "loss": 0.3488,
      "step": 21660
    },
    {
      "epoch": 0.39118078275879503,
      "grad_norm": 0.35219860076904297,
      "learning_rate": 1.3352328208657594e-05,
      "loss": 0.2462,
      "step": 21665
    },
    {
      "epoch": 0.3912710621916957,
      "grad_norm": 0.4262147545814514,
      "learning_rate": 1.3349655949387667e-05,
      "loss": 0.3275,
      "step": 21670
    },
    {
      "epoch": 0.39136134162459646,
      "grad_norm": 0.345892995595932,
      "learning_rate": 1.3346983420662275e-05,
      "loss": 0.219,
      "step": 21675
    },
    {
      "epoch": 0.39145162105749715,
      "grad_norm": 0.5044654607772827,
      "learning_rate": 1.33443106226964e-05,
      "loss": 0.2579,
      "step": 21680
    },
    {
      "epoch": 0.3915419004903979,
      "grad_norm": 0.5730448365211487,
      "learning_rate": 1.3341637555705051e-05,
      "loss": 0.3406,
      "step": 21685
    },
    {
      "epoch": 0.3916321799232986,
      "grad_norm": 0.5646150708198547,
      "learning_rate": 1.333896421990326e-05,
      "loss": 0.3259,
      "step": 21690
    },
    {
      "epoch": 0.3917224593561993,
      "grad_norm": 0.5319784879684448,
      "learning_rate": 1.3336290615506072e-05,
      "loss": 0.2592,
      "step": 21695
    },
    {
      "epoch": 0.3918127387891,
      "grad_norm": 0.5225646495819092,
      "learning_rate": 1.3333616742728559e-05,
      "loss": 0.2272,
      "step": 21700
    },
    {
      "epoch": 0.39190301822200074,
      "grad_norm": 0.6103857159614563,
      "learning_rate": 1.3330942601785819e-05,
      "loss": 0.309,
      "step": 21705
    },
    {
      "epoch": 0.39199329765490143,
      "grad_norm": 0.5389057397842407,
      "learning_rate": 1.3328268192892963e-05,
      "loss": 0.2542,
      "step": 21710
    },
    {
      "epoch": 0.39208357708780217,
      "grad_norm": 0.21751132607460022,
      "learning_rate": 1.3325593516265132e-05,
      "loss": 0.1974,
      "step": 21715
    },
    {
      "epoch": 0.39217385652070286,
      "grad_norm": 0.3460856080055237,
      "learning_rate": 1.3322918572117483e-05,
      "loss": 0.2763,
      "step": 21720
    },
    {
      "epoch": 0.3922641359536036,
      "grad_norm": 0.45296546816825867,
      "learning_rate": 1.3320243360665193e-05,
      "loss": 0.2324,
      "step": 21725
    },
    {
      "epoch": 0.3923544153865043,
      "grad_norm": 0.6587622761726379,
      "learning_rate": 1.3317567882123463e-05,
      "loss": 0.2041,
      "step": 21730
    },
    {
      "epoch": 0.392444694819405,
      "grad_norm": 0.3234507441520691,
      "learning_rate": 1.331489213670752e-05,
      "loss": 0.2638,
      "step": 21735
    },
    {
      "epoch": 0.3925349742523057,
      "grad_norm": 0.3528299331665039,
      "learning_rate": 1.3312216124632609e-05,
      "loss": 0.2844,
      "step": 21740
    },
    {
      "epoch": 0.39262525368520645,
      "grad_norm": 0.5674415826797485,
      "learning_rate": 1.3309539846113988e-05,
      "loss": 0.277,
      "step": 21745
    },
    {
      "epoch": 0.39271553311810714,
      "grad_norm": 0.2994835674762726,
      "learning_rate": 1.330686330136695e-05,
      "loss": 0.2854,
      "step": 21750
    },
    {
      "epoch": 0.3928058125510079,
      "grad_norm": 0.7018741369247437,
      "learning_rate": 1.3304186490606805e-05,
      "loss": 0.2942,
      "step": 21755
    },
    {
      "epoch": 0.39289609198390857,
      "grad_norm": 0.4478186368942261,
      "learning_rate": 1.330150941404888e-05,
      "loss": 0.2202,
      "step": 21760
    },
    {
      "epoch": 0.3929863714168093,
      "grad_norm": 0.455462783575058,
      "learning_rate": 1.3298832071908525e-05,
      "loss": 0.2077,
      "step": 21765
    },
    {
      "epoch": 0.39307665084971,
      "grad_norm": 0.5022167563438416,
      "learning_rate": 1.3296154464401117e-05,
      "loss": 0.2687,
      "step": 21770
    },
    {
      "epoch": 0.39316693028261074,
      "grad_norm": 0.5095340609550476,
      "learning_rate": 1.3293476591742047e-05,
      "loss": 0.3387,
      "step": 21775
    },
    {
      "epoch": 0.3932572097155114,
      "grad_norm": 0.4713091254234314,
      "learning_rate": 1.3290798454146731e-05,
      "loss": 0.2839,
      "step": 21780
    },
    {
      "epoch": 0.39334748914841217,
      "grad_norm": 0.5228136777877808,
      "learning_rate": 1.3288120051830607e-05,
      "loss": 0.2061,
      "step": 21785
    },
    {
      "epoch": 0.39343776858131285,
      "grad_norm": 0.5140119194984436,
      "learning_rate": 1.328544138500913e-05,
      "loss": 0.2326,
      "step": 21790
    },
    {
      "epoch": 0.3935280480142136,
      "grad_norm": 0.4276304543018341,
      "learning_rate": 1.3282762453897779e-05,
      "loss": 0.1985,
      "step": 21795
    },
    {
      "epoch": 0.39361832744711434,
      "grad_norm": 0.2913047969341278,
      "learning_rate": 1.328008325871206e-05,
      "loss": 0.2378,
      "step": 21800
    },
    {
      "epoch": 0.393708606880015,
      "grad_norm": 0.5427494049072266,
      "learning_rate": 1.3277403799667493e-05,
      "loss": 0.2491,
      "step": 21805
    },
    {
      "epoch": 0.39379888631291576,
      "grad_norm": 0.47361496090888977,
      "learning_rate": 1.3274724076979612e-05,
      "loss": 0.2625,
      "step": 21810
    },
    {
      "epoch": 0.39388916574581645,
      "grad_norm": 0.42510926723480225,
      "learning_rate": 1.3272044090863995e-05,
      "loss": 0.2812,
      "step": 21815
    },
    {
      "epoch": 0.3939794451787172,
      "grad_norm": 0.5460023880004883,
      "learning_rate": 1.3269363841536218e-05,
      "loss": 0.3689,
      "step": 21820
    },
    {
      "epoch": 0.3940697246116179,
      "grad_norm": 0.5921125411987305,
      "learning_rate": 1.3266683329211894e-05,
      "loss": 0.2266,
      "step": 21825
    },
    {
      "epoch": 0.3941600040445186,
      "grad_norm": 0.6031335592269897,
      "learning_rate": 1.3264002554106642e-05,
      "loss": 0.3288,
      "step": 21830
    },
    {
      "epoch": 0.3942502834774193,
      "grad_norm": 0.4571972191333771,
      "learning_rate": 1.326132151643612e-05,
      "loss": 0.244,
      "step": 21835
    },
    {
      "epoch": 0.39434056291032005,
      "grad_norm": 0.3885684609413147,
      "learning_rate": 1.3258640216415995e-05,
      "loss": 0.272,
      "step": 21840
    },
    {
      "epoch": 0.39443084234322073,
      "grad_norm": 0.788741409778595,
      "learning_rate": 1.3255958654261952e-05,
      "loss": 0.3086,
      "step": 21845
    },
    {
      "epoch": 0.3945211217761215,
      "grad_norm": 0.5432313680648804,
      "learning_rate": 1.3253276830189712e-05,
      "loss": 0.2451,
      "step": 21850
    },
    {
      "epoch": 0.39461140120902216,
      "grad_norm": 0.27433672547340393,
      "learning_rate": 1.3250594744415006e-05,
      "loss": 0.284,
      "step": 21855
    },
    {
      "epoch": 0.3947016806419229,
      "grad_norm": 0.4588322639465332,
      "learning_rate": 1.3247912397153586e-05,
      "loss": 0.279,
      "step": 21860
    },
    {
      "epoch": 0.3947919600748236,
      "grad_norm": 0.35396280884742737,
      "learning_rate": 1.3245229788621226e-05,
      "loss": 0.3236,
      "step": 21865
    },
    {
      "epoch": 0.39488223950772433,
      "grad_norm": 0.4001164138317108,
      "learning_rate": 1.3242546919033726e-05,
      "loss": 0.1426,
      "step": 21870
    },
    {
      "epoch": 0.394972518940625,
      "grad_norm": 2.4019925594329834,
      "learning_rate": 1.3239863788606902e-05,
      "loss": 0.1902,
      "step": 21875
    },
    {
      "epoch": 0.39506279837352576,
      "grad_norm": 0.35852307081222534,
      "learning_rate": 1.3237180397556592e-05,
      "loss": 0.2625,
      "step": 21880
    },
    {
      "epoch": 0.39515307780642644,
      "grad_norm": 0.6176601052284241,
      "learning_rate": 1.3234496746098657e-05,
      "loss": 0.2788,
      "step": 21885
    },
    {
      "epoch": 0.3952433572393272,
      "grad_norm": 0.3619178533554077,
      "learning_rate": 1.3231812834448976e-05,
      "loss": 0.2386,
      "step": 21890
    },
    {
      "epoch": 0.3953336366722279,
      "grad_norm": 0.30432456731796265,
      "learning_rate": 1.3229128662823447e-05,
      "loss": 0.2922,
      "step": 21895
    },
    {
      "epoch": 0.3954239161051286,
      "grad_norm": 0.5979883670806885,
      "learning_rate": 1.3226444231437998e-05,
      "loss": 0.2852,
      "step": 21900
    },
    {
      "epoch": 0.3955141955380293,
      "grad_norm": 0.3860207498073578,
      "learning_rate": 1.3223759540508568e-05,
      "loss": 0.2429,
      "step": 21905
    },
    {
      "epoch": 0.39560447497093004,
      "grad_norm": 0.5804364681243896,
      "learning_rate": 1.3221074590251121e-05,
      "loss": 0.2835,
      "step": 21910
    },
    {
      "epoch": 0.39569475440383073,
      "grad_norm": 0.7440268993377686,
      "learning_rate": 1.3218389380881645e-05,
      "loss": 0.3498,
      "step": 21915
    },
    {
      "epoch": 0.39578503383673147,
      "grad_norm": 0.5781992673873901,
      "learning_rate": 1.3215703912616143e-05,
      "loss": 0.3228,
      "step": 21920
    },
    {
      "epoch": 0.39587531326963216,
      "grad_norm": 0.6474190950393677,
      "learning_rate": 1.3213018185670638e-05,
      "loss": 0.3353,
      "step": 21925
    },
    {
      "epoch": 0.3959655927025329,
      "grad_norm": 0.28923219442367554,
      "learning_rate": 1.3210332200261183e-05,
      "loss": 0.3499,
      "step": 21930
    },
    {
      "epoch": 0.3960558721354336,
      "grad_norm": 0.452716588973999,
      "learning_rate": 1.3207645956603845e-05,
      "loss": 0.2217,
      "step": 21935
    },
    {
      "epoch": 0.3961461515683343,
      "grad_norm": 0.4790239632129669,
      "learning_rate": 1.3204959454914712e-05,
      "loss": 0.3062,
      "step": 21940
    },
    {
      "epoch": 0.396236431001235,
      "grad_norm": 0.5556036233901978,
      "learning_rate": 1.320227269540989e-05,
      "loss": 0.286,
      "step": 21945
    },
    {
      "epoch": 0.39632671043413575,
      "grad_norm": 0.6638554930686951,
      "learning_rate": 1.3199585678305516e-05,
      "loss": 0.2642,
      "step": 21950
    },
    {
      "epoch": 0.39641698986703644,
      "grad_norm": 0.6628899574279785,
      "learning_rate": 1.3196898403817735e-05,
      "loss": 0.2448,
      "step": 21955
    },
    {
      "epoch": 0.3965072692999372,
      "grad_norm": 0.35188084840774536,
      "learning_rate": 1.3194210872162723e-05,
      "loss": 0.334,
      "step": 21960
    },
    {
      "epoch": 0.39659754873283787,
      "grad_norm": 0.33103373646736145,
      "learning_rate": 1.319152308355667e-05,
      "loss": 0.2284,
      "step": 21965
    },
    {
      "epoch": 0.3966878281657386,
      "grad_norm": 0.4334578514099121,
      "learning_rate": 1.3188835038215788e-05,
      "loss": 0.2292,
      "step": 21970
    },
    {
      "epoch": 0.3967781075986393,
      "grad_norm": 0.6531385779380798,
      "learning_rate": 1.3186146736356314e-05,
      "loss": 0.2756,
      "step": 21975
    },
    {
      "epoch": 0.39686838703154004,
      "grad_norm": 0.3080330491065979,
      "learning_rate": 1.3183458178194499e-05,
      "loss": 0.2535,
      "step": 21980
    },
    {
      "epoch": 0.3969586664644407,
      "grad_norm": 2.054600238800049,
      "learning_rate": 1.3180769363946621e-05,
      "loss": 0.2555,
      "step": 21985
    },
    {
      "epoch": 0.39704894589734147,
      "grad_norm": 0.43283167481422424,
      "learning_rate": 1.3178080293828974e-05,
      "loss": 0.291,
      "step": 21990
    },
    {
      "epoch": 0.39713922533024215,
      "grad_norm": 0.371512234210968,
      "learning_rate": 1.3175390968057872e-05,
      "loss": 0.3182,
      "step": 21995
    },
    {
      "epoch": 0.3972295047631429,
      "grad_norm": 0.5806812644004822,
      "learning_rate": 1.3172701386849659e-05,
      "loss": 0.2578,
      "step": 22000
    },
    {
      "epoch": 0.3973197841960436,
      "grad_norm": 0.40547439455986023,
      "learning_rate": 1.3170011550420682e-05,
      "loss": 0.2778,
      "step": 22005
    },
    {
      "epoch": 0.3974100636289443,
      "grad_norm": 0.5131949186325073,
      "learning_rate": 1.3167321458987326e-05,
      "loss": 0.1756,
      "step": 22010
    },
    {
      "epoch": 0.397500343061845,
      "grad_norm": 0.5438373684883118,
      "learning_rate": 1.3164631112765987e-05,
      "loss": 0.2903,
      "step": 22015
    },
    {
      "epoch": 0.39759062249474575,
      "grad_norm": 0.4344537556171417,
      "learning_rate": 1.3161940511973084e-05,
      "loss": 0.3439,
      "step": 22020
    },
    {
      "epoch": 0.39768090192764644,
      "grad_norm": 0.4997982680797577,
      "learning_rate": 1.3159249656825057e-05,
      "loss": 0.345,
      "step": 22025
    },
    {
      "epoch": 0.3977711813605472,
      "grad_norm": 0.36932867765426636,
      "learning_rate": 1.3156558547538362e-05,
      "loss": 0.212,
      "step": 22030
    },
    {
      "epoch": 0.39786146079344786,
      "grad_norm": 0.6555669903755188,
      "learning_rate": 1.3153867184329488e-05,
      "loss": 0.2374,
      "step": 22035
    },
    {
      "epoch": 0.3979517402263486,
      "grad_norm": 3.40368390083313,
      "learning_rate": 1.3151175567414927e-05,
      "loss": 0.2516,
      "step": 22040
    },
    {
      "epoch": 0.3980420196592493,
      "grad_norm": 0.6993756890296936,
      "learning_rate": 1.3148483697011201e-05,
      "loss": 0.3079,
      "step": 22045
    },
    {
      "epoch": 0.39813229909215003,
      "grad_norm": 0.5923683047294617,
      "learning_rate": 1.3145791573334854e-05,
      "loss": 0.2734,
      "step": 22050
    },
    {
      "epoch": 0.3982225785250507,
      "grad_norm": 0.4434371888637543,
      "learning_rate": 1.314309919660245e-05,
      "loss": 0.2839,
      "step": 22055
    },
    {
      "epoch": 0.39831285795795146,
      "grad_norm": 0.5689956545829773,
      "learning_rate": 1.3140406567030565e-05,
      "loss": 0.2707,
      "step": 22060
    },
    {
      "epoch": 0.39840313739085215,
      "grad_norm": 0.354709267616272,
      "learning_rate": 1.3137713684835806e-05,
      "loss": 0.2433,
      "step": 22065
    },
    {
      "epoch": 0.3984934168237529,
      "grad_norm": 0.44803231954574585,
      "learning_rate": 1.3135020550234797e-05,
      "loss": 0.2332,
      "step": 22070
    },
    {
      "epoch": 0.3985836962566536,
      "grad_norm": 0.2759532630443573,
      "learning_rate": 1.3132327163444174e-05,
      "loss": 0.181,
      "step": 22075
    },
    {
      "epoch": 0.3986739756895543,
      "grad_norm": 0.36494264006614685,
      "learning_rate": 1.3129633524680604e-05,
      "loss": 0.2335,
      "step": 22080
    },
    {
      "epoch": 0.398764255122455,
      "grad_norm": 0.37547898292541504,
      "learning_rate": 1.3126939634160775e-05,
      "loss": 0.2587,
      "step": 22085
    },
    {
      "epoch": 0.39885453455535574,
      "grad_norm": 0.4002203643321991,
      "learning_rate": 1.3124245492101389e-05,
      "loss": 0.2481,
      "step": 22090
    },
    {
      "epoch": 0.39894481398825643,
      "grad_norm": 0.4837264120578766,
      "learning_rate": 1.3121551098719165e-05,
      "loss": 0.3074,
      "step": 22095
    },
    {
      "epoch": 0.39903509342115717,
      "grad_norm": 0.38516107201576233,
      "learning_rate": 1.3118856454230853e-05,
      "loss": 0.1741,
      "step": 22100
    },
    {
      "epoch": 0.39912537285405786,
      "grad_norm": 0.43117663264274597,
      "learning_rate": 1.3116161558853216e-05,
      "loss": 0.2468,
      "step": 22105
    },
    {
      "epoch": 0.3992156522869586,
      "grad_norm": 0.8077203035354614,
      "learning_rate": 1.3113466412803036e-05,
      "loss": 0.2171,
      "step": 22110
    },
    {
      "epoch": 0.3993059317198593,
      "grad_norm": 0.5902336835861206,
      "learning_rate": 1.311077101629712e-05,
      "loss": 0.2886,
      "step": 22115
    },
    {
      "epoch": 0.39939621115276003,
      "grad_norm": 0.927211582660675,
      "learning_rate": 1.3108075369552295e-05,
      "loss": 0.317,
      "step": 22120
    },
    {
      "epoch": 0.3994864905856607,
      "grad_norm": 0.27386292815208435,
      "learning_rate": 1.3105379472785405e-05,
      "loss": 0.2825,
      "step": 22125
    },
    {
      "epoch": 0.39957677001856146,
      "grad_norm": 0.4687485992908478,
      "learning_rate": 1.3102683326213312e-05,
      "loss": 0.3169,
      "step": 22130
    },
    {
      "epoch": 0.39966704945146214,
      "grad_norm": 0.5701487064361572,
      "learning_rate": 1.3099986930052906e-05,
      "loss": 0.1922,
      "step": 22135
    },
    {
      "epoch": 0.3997573288843629,
      "grad_norm": 0.5942817330360413,
      "learning_rate": 1.3097290284521091e-05,
      "loss": 0.2372,
      "step": 22140
    },
    {
      "epoch": 0.39984760831726357,
      "grad_norm": 0.6339251399040222,
      "learning_rate": 1.3094593389834789e-05,
      "loss": 0.2858,
      "step": 22145
    },
    {
      "epoch": 0.3999378877501643,
      "grad_norm": 0.30864328145980835,
      "learning_rate": 1.3091896246210953e-05,
      "loss": 0.3315,
      "step": 22150
    },
    {
      "epoch": 0.400028167183065,
      "grad_norm": 0.7753767967224121,
      "learning_rate": 1.3089198853866543e-05,
      "loss": 0.2209,
      "step": 22155
    },
    {
      "epoch": 0.40011844661596574,
      "grad_norm": 0.4362318217754364,
      "learning_rate": 1.3086501213018546e-05,
      "loss": 0.2329,
      "step": 22160
    },
    {
      "epoch": 0.4002087260488664,
      "grad_norm": 0.4449676275253296,
      "learning_rate": 1.3083803323883966e-05,
      "loss": 0.2642,
      "step": 22165
    },
    {
      "epoch": 0.40029900548176717,
      "grad_norm": 0.5464814901351929,
      "learning_rate": 1.308110518667983e-05,
      "loss": 0.3336,
      "step": 22170
    },
    {
      "epoch": 0.40038928491466785,
      "grad_norm": 0.43090885877609253,
      "learning_rate": 1.3078406801623182e-05,
      "loss": 0.1789,
      "step": 22175
    },
    {
      "epoch": 0.4004795643475686,
      "grad_norm": 0.5409093499183655,
      "learning_rate": 1.3075708168931093e-05,
      "loss": 0.2445,
      "step": 22180
    },
    {
      "epoch": 0.4005698437804693,
      "grad_norm": 0.4051618278026581,
      "learning_rate": 1.3073009288820643e-05,
      "loss": 0.2874,
      "step": 22185
    },
    {
      "epoch": 0.40066012321337,
      "grad_norm": 0.3605162799358368,
      "learning_rate": 1.3070310161508939e-05,
      "loss": 0.2565,
      "step": 22190
    },
    {
      "epoch": 0.40075040264627076,
      "grad_norm": 0.31601181626319885,
      "learning_rate": 1.3067610787213103e-05,
      "loss": 0.2436,
      "step": 22195
    },
    {
      "epoch": 0.40084068207917145,
      "grad_norm": 0.4679415225982666,
      "learning_rate": 1.3064911166150284e-05,
      "loss": 0.3179,
      "step": 22200
    },
    {
      "epoch": 0.4009309615120722,
      "grad_norm": 0.23421575129032135,
      "learning_rate": 1.3062211298537648e-05,
      "loss": 0.2115,
      "step": 22205
    },
    {
      "epoch": 0.4010212409449729,
      "grad_norm": 0.3195156455039978,
      "learning_rate": 1.3059511184592375e-05,
      "loss": 0.202,
      "step": 22210
    },
    {
      "epoch": 0.4011115203778736,
      "grad_norm": 0.31775689125061035,
      "learning_rate": 1.3056810824531674e-05,
      "loss": 0.1927,
      "step": 22215
    },
    {
      "epoch": 0.4012017998107743,
      "grad_norm": 0.42000657320022583,
      "learning_rate": 1.3054110218572766e-05,
      "loss": 0.2534,
      "step": 22220
    },
    {
      "epoch": 0.40129207924367505,
      "grad_norm": 0.3960062861442566,
      "learning_rate": 1.3051409366932894e-05,
      "loss": 0.3371,
      "step": 22225
    },
    {
      "epoch": 0.40138235867657573,
      "grad_norm": 0.39160043001174927,
      "learning_rate": 1.304870826982933e-05,
      "loss": 0.2168,
      "step": 22230
    },
    {
      "epoch": 0.4014726381094765,
      "grad_norm": 0.3743045926094055,
      "learning_rate": 1.3046006927479349e-05,
      "loss": 0.2778,
      "step": 22235
    },
    {
      "epoch": 0.40156291754237716,
      "grad_norm": 0.31039419770240784,
      "learning_rate": 1.3043305340100259e-05,
      "loss": 0.2947,
      "step": 22240
    },
    {
      "epoch": 0.4016531969752779,
      "grad_norm": 0.4929141402244568,
      "learning_rate": 1.304060350790938e-05,
      "loss": 0.3441,
      "step": 22245
    },
    {
      "epoch": 0.4017434764081786,
      "grad_norm": 0.4496571719646454,
      "learning_rate": 1.3037901431124053e-05,
      "loss": 0.2276,
      "step": 22250
    },
    {
      "epoch": 0.40183375584107933,
      "grad_norm": 0.44919899106025696,
      "learning_rate": 1.3035199109961649e-05,
      "loss": 0.3278,
      "step": 22255
    },
    {
      "epoch": 0.40192403527398,
      "grad_norm": 0.3038470149040222,
      "learning_rate": 1.3032496544639539e-05,
      "loss": 0.3122,
      "step": 22260
    },
    {
      "epoch": 0.40201431470688076,
      "grad_norm": 0.368513822555542,
      "learning_rate": 1.3029793735375136e-05,
      "loss": 0.2556,
      "step": 22265
    },
    {
      "epoch": 0.40210459413978145,
      "grad_norm": 0.5456978678703308,
      "learning_rate": 1.3027090682385852e-05,
      "loss": 0.3737,
      "step": 22270
    },
    {
      "epoch": 0.4021948735726822,
      "grad_norm": 0.38461658358573914,
      "learning_rate": 1.3024387385889132e-05,
      "loss": 0.2524,
      "step": 22275
    },
    {
      "epoch": 0.4022851530055829,
      "grad_norm": 0.5025389790534973,
      "learning_rate": 1.3021683846102437e-05,
      "loss": 0.3344,
      "step": 22280
    },
    {
      "epoch": 0.4023754324384836,
      "grad_norm": 0.8047762513160706,
      "learning_rate": 1.3018980063243248e-05,
      "loss": 0.2436,
      "step": 22285
    },
    {
      "epoch": 0.4024657118713843,
      "grad_norm": 0.515868067741394,
      "learning_rate": 1.301627603752906e-05,
      "loss": 0.23,
      "step": 22290
    },
    {
      "epoch": 0.40255599130428504,
      "grad_norm": 0.3612858057022095,
      "learning_rate": 1.3013571769177395e-05,
      "loss": 0.2775,
      "step": 22295
    },
    {
      "epoch": 0.40264627073718573,
      "grad_norm": 0.5300875902175903,
      "learning_rate": 1.3010867258405796e-05,
      "loss": 0.3596,
      "step": 22300
    },
    {
      "epoch": 0.40273655017008647,
      "grad_norm": 1.0323123931884766,
      "learning_rate": 1.3008162505431813e-05,
      "loss": 0.3625,
      "step": 22305
    },
    {
      "epoch": 0.40282682960298716,
      "grad_norm": 0.526929497718811,
      "learning_rate": 1.300545751047303e-05,
      "loss": 0.2416,
      "step": 22310
    },
    {
      "epoch": 0.4029171090358879,
      "grad_norm": 0.5251116752624512,
      "learning_rate": 1.3002752273747041e-05,
      "loss": 0.313,
      "step": 22315
    },
    {
      "epoch": 0.4030073884687886,
      "grad_norm": 0.3955122232437134,
      "learning_rate": 1.3000046795471467e-05,
      "loss": 0.2874,
      "step": 22320
    },
    {
      "epoch": 0.4030976679016893,
      "grad_norm": 0.4317103326320648,
      "learning_rate": 1.2997341075863934e-05,
      "loss": 0.2103,
      "step": 22325
    },
    {
      "epoch": 0.40318794733459,
      "grad_norm": 0.6640254259109497,
      "learning_rate": 1.299463511514211e-05,
      "loss": 0.326,
      "step": 22330
    },
    {
      "epoch": 0.40327822676749076,
      "grad_norm": 0.4563729763031006,
      "learning_rate": 1.2991928913523663e-05,
      "loss": 0.221,
      "step": 22335
    },
    {
      "epoch": 0.40336850620039144,
      "grad_norm": 0.35100239515304565,
      "learning_rate": 1.2989222471226288e-05,
      "loss": 0.2703,
      "step": 22340
    },
    {
      "epoch": 0.4034587856332922,
      "grad_norm": 0.3896128237247467,
      "learning_rate": 1.29865157884677e-05,
      "loss": 0.3096,
      "step": 22345
    },
    {
      "epoch": 0.40354906506619287,
      "grad_norm": 1.338800072669983,
      "learning_rate": 1.2983808865465629e-05,
      "loss": 0.2735,
      "step": 22350
    },
    {
      "epoch": 0.4036393444990936,
      "grad_norm": 0.6762623190879822,
      "learning_rate": 1.2981101702437831e-05,
      "loss": 0.2059,
      "step": 22355
    },
    {
      "epoch": 0.4037296239319943,
      "grad_norm": 0.44800934195518494,
      "learning_rate": 1.2978394299602073e-05,
      "loss": 0.225,
      "step": 22360
    },
    {
      "epoch": 0.40381990336489504,
      "grad_norm": 0.2908240556716919,
      "learning_rate": 1.297568665717615e-05,
      "loss": 0.2992,
      "step": 22365
    },
    {
      "epoch": 0.4039101827977957,
      "grad_norm": 0.4273534119129181,
      "learning_rate": 1.297297877537787e-05,
      "loss": 0.3031,
      "step": 22370
    },
    {
      "epoch": 0.40400046223069647,
      "grad_norm": 0.5294716954231262,
      "learning_rate": 1.2970270654425063e-05,
      "loss": 0.2404,
      "step": 22375
    },
    {
      "epoch": 0.40409074166359715,
      "grad_norm": 0.3720170557498932,
      "learning_rate": 1.296756229453558e-05,
      "loss": 0.2095,
      "step": 22380
    },
    {
      "epoch": 0.4041810210964979,
      "grad_norm": 0.9239777326583862,
      "learning_rate": 1.2964853695927287e-05,
      "loss": 0.2013,
      "step": 22385
    },
    {
      "epoch": 0.4042713005293986,
      "grad_norm": 0.13262271881103516,
      "learning_rate": 1.2962144858818067e-05,
      "loss": 0.208,
      "step": 22390
    },
    {
      "epoch": 0.4043615799622993,
      "grad_norm": 0.3081349730491638,
      "learning_rate": 1.2959435783425834e-05,
      "loss": 0.3017,
      "step": 22395
    },
    {
      "epoch": 0.4044518593952,
      "grad_norm": 0.36518728733062744,
      "learning_rate": 1.2956726469968506e-05,
      "loss": 0.277,
      "step": 22400
    },
    {
      "epoch": 0.40454213882810075,
      "grad_norm": 0.3570597767829895,
      "learning_rate": 1.2954016918664034e-05,
      "loss": 0.1976,
      "step": 22405
    },
    {
      "epoch": 0.40463241826100144,
      "grad_norm": 0.5864139795303345,
      "learning_rate": 1.2951307129730375e-05,
      "loss": 0.2879,
      "step": 22410
    },
    {
      "epoch": 0.4047226976939022,
      "grad_norm": 0.27137792110443115,
      "learning_rate": 1.2948597103385518e-05,
      "loss": 0.1695,
      "step": 22415
    },
    {
      "epoch": 0.40481297712680286,
      "grad_norm": 0.894551694393158,
      "learning_rate": 1.2945886839847464e-05,
      "loss": 0.3315,
      "step": 22420
    },
    {
      "epoch": 0.4049032565597036,
      "grad_norm": 0.7413949966430664,
      "learning_rate": 1.294317633933423e-05,
      "loss": 0.2742,
      "step": 22425
    },
    {
      "epoch": 0.4049935359926043,
      "grad_norm": 0.4280196726322174,
      "learning_rate": 1.2940465602063862e-05,
      "loss": 0.2159,
      "step": 22430
    },
    {
      "epoch": 0.40508381542550503,
      "grad_norm": 0.5879983901977539,
      "learning_rate": 1.293775462825441e-05,
      "loss": 0.2604,
      "step": 22435
    },
    {
      "epoch": 0.4051740948584057,
      "grad_norm": 0.46284785866737366,
      "learning_rate": 1.293504341812396e-05,
      "loss": 0.216,
      "step": 22440
    },
    {
      "epoch": 0.40526437429130646,
      "grad_norm": 0.2746809720993042,
      "learning_rate": 1.2932331971890608e-05,
      "loss": 0.2496,
      "step": 22445
    },
    {
      "epoch": 0.40535465372420715,
      "grad_norm": 0.30468782782554626,
      "learning_rate": 1.2929620289772469e-05,
      "loss": 0.2481,
      "step": 22450
    },
    {
      "epoch": 0.4054449331571079,
      "grad_norm": 0.5176340937614441,
      "learning_rate": 1.2926908371987675e-05,
      "loss": 0.2774,
      "step": 22455
    },
    {
      "epoch": 0.4055352125900086,
      "grad_norm": 0.4324280619621277,
      "learning_rate": 1.2924196218754387e-05,
      "loss": 0.1958,
      "step": 22460
    },
    {
      "epoch": 0.4056254920229093,
      "grad_norm": 0.47290143370628357,
      "learning_rate": 1.292148383029077e-05,
      "loss": 0.2765,
      "step": 22465
    },
    {
      "epoch": 0.40571577145581,
      "grad_norm": 0.4480784237384796,
      "learning_rate": 1.2918771206815023e-05,
      "loss": 0.3096,
      "step": 22470
    },
    {
      "epoch": 0.40580605088871075,
      "grad_norm": 0.5606126189231873,
      "learning_rate": 1.2916058348545352e-05,
      "loss": 0.268,
      "step": 22475
    },
    {
      "epoch": 0.40589633032161143,
      "grad_norm": 0.35950228571891785,
      "learning_rate": 1.2913345255699992e-05,
      "loss": 0.3666,
      "step": 22480
    },
    {
      "epoch": 0.4059866097545122,
      "grad_norm": 0.4312303364276886,
      "learning_rate": 1.2910631928497187e-05,
      "loss": 0.2431,
      "step": 22485
    },
    {
      "epoch": 0.40607688918741286,
      "grad_norm": 0.41166362166404724,
      "learning_rate": 1.2907918367155203e-05,
      "loss": 0.2415,
      "step": 22490
    },
    {
      "epoch": 0.4061671686203136,
      "grad_norm": 0.5689197778701782,
      "learning_rate": 1.2905204571892332e-05,
      "loss": 0.2023,
      "step": 22495
    },
    {
      "epoch": 0.4062574480532143,
      "grad_norm": 0.4476293623447418,
      "learning_rate": 1.2902490542926875e-05,
      "loss": 0.1702,
      "step": 22500
    },
    {
      "epoch": 0.40634772748611503,
      "grad_norm": 0.2847082316875458,
      "learning_rate": 1.2899776280477158e-05,
      "loss": 0.3206,
      "step": 22505
    },
    {
      "epoch": 0.4064380069190157,
      "grad_norm": 0.3037201464176178,
      "learning_rate": 1.289706178476152e-05,
      "loss": 0.1943,
      "step": 22510
    },
    {
      "epoch": 0.40652828635191646,
      "grad_norm": 0.5437815189361572,
      "learning_rate": 1.2894347055998329e-05,
      "loss": 0.2397,
      "step": 22515
    },
    {
      "epoch": 0.40661856578481714,
      "grad_norm": 0.461439847946167,
      "learning_rate": 1.2891632094405962e-05,
      "loss": 0.3052,
      "step": 22520
    },
    {
      "epoch": 0.4067088452177179,
      "grad_norm": 0.25330743193626404,
      "learning_rate": 1.288891690020281e-05,
      "loss": 0.2289,
      "step": 22525
    },
    {
      "epoch": 0.40679912465061857,
      "grad_norm": 0.6357237696647644,
      "learning_rate": 1.2886201473607305e-05,
      "loss": 0.2533,
      "step": 22530
    },
    {
      "epoch": 0.4068894040835193,
      "grad_norm": 0.3013076186180115,
      "learning_rate": 1.2883485814837872e-05,
      "loss": 0.2569,
      "step": 22535
    },
    {
      "epoch": 0.40697968351642,
      "grad_norm": 0.45817866921424866,
      "learning_rate": 1.288076992411297e-05,
      "loss": 0.2713,
      "step": 22540
    },
    {
      "epoch": 0.40706996294932074,
      "grad_norm": 0.367135226726532,
      "learning_rate": 1.2878053801651076e-05,
      "loss": 0.2432,
      "step": 22545
    },
    {
      "epoch": 0.4071602423822214,
      "grad_norm": 0.4528060853481293,
      "learning_rate": 1.2875337447670677e-05,
      "loss": 0.2921,
      "step": 22550
    },
    {
      "epoch": 0.40725052181512217,
      "grad_norm": 0.47508856654167175,
      "learning_rate": 1.2872620862390286e-05,
      "loss": 0.2638,
      "step": 22555
    },
    {
      "epoch": 0.40734080124802285,
      "grad_norm": 0.6407237648963928,
      "learning_rate": 1.2869904046028434e-05,
      "loss": 0.2509,
      "step": 22560
    },
    {
      "epoch": 0.4074310806809236,
      "grad_norm": 0.3994412124156952,
      "learning_rate": 1.2867186998803666e-05,
      "loss": 0.2549,
      "step": 22565
    },
    {
      "epoch": 0.4075213601138243,
      "grad_norm": 0.42670902609825134,
      "learning_rate": 1.2864469720934553e-05,
      "loss": 0.3734,
      "step": 22570
    },
    {
      "epoch": 0.407611639546725,
      "grad_norm": 0.42069903016090393,
      "learning_rate": 1.2861752212639673e-05,
      "loss": 0.2171,
      "step": 22575
    },
    {
      "epoch": 0.4077019189796257,
      "grad_norm": 0.5737390518188477,
      "learning_rate": 1.2859034474137639e-05,
      "loss": 0.3261,
      "step": 22580
    },
    {
      "epoch": 0.40779219841252645,
      "grad_norm": 0.4968356490135193,
      "learning_rate": 1.2856316505647066e-05,
      "loss": 0.2422,
      "step": 22585
    },
    {
      "epoch": 0.4078824778454272,
      "grad_norm": 0.5124558806419373,
      "learning_rate": 1.28535983073866e-05,
      "loss": 0.2636,
      "step": 22590
    },
    {
      "epoch": 0.4079727572783279,
      "grad_norm": 0.586444616317749,
      "learning_rate": 1.2850879879574896e-05,
      "loss": 0.23,
      "step": 22595
    },
    {
      "epoch": 0.4080630367112286,
      "grad_norm": 0.5707995891571045,
      "learning_rate": 1.2848161222430635e-05,
      "loss": 0.2049,
      "step": 22600
    },
    {
      "epoch": 0.4081533161441293,
      "grad_norm": 0.32396647334098816,
      "learning_rate": 1.2845442336172512e-05,
      "loss": 0.254,
      "step": 22605
    },
    {
      "epoch": 0.40824359557703005,
      "grad_norm": 0.4279758036136627,
      "learning_rate": 1.284272322101924e-05,
      "loss": 0.2778,
      "step": 22610
    },
    {
      "epoch": 0.40833387500993074,
      "grad_norm": 0.4514820873737335,
      "learning_rate": 1.2840003877189554e-05,
      "loss": 0.1986,
      "step": 22615
    },
    {
      "epoch": 0.4084241544428315,
      "grad_norm": 0.44896379113197327,
      "learning_rate": 1.2837284304902202e-05,
      "loss": 0.2626,
      "step": 22620
    },
    {
      "epoch": 0.40851443387573216,
      "grad_norm": 0.3960501551628113,
      "learning_rate": 1.2834564504375959e-05,
      "loss": 0.2557,
      "step": 22625
    },
    {
      "epoch": 0.4086047133086329,
      "grad_norm": 0.32725077867507935,
      "learning_rate": 1.2831844475829613e-05,
      "loss": 0.2286,
      "step": 22630
    },
    {
      "epoch": 0.4086949927415336,
      "grad_norm": 0.6208406090736389,
      "learning_rate": 1.2829124219481967e-05,
      "loss": 0.2369,
      "step": 22635
    },
    {
      "epoch": 0.40878527217443433,
      "grad_norm": 0.5370245575904846,
      "learning_rate": 1.2826403735551844e-05,
      "loss": 0.2311,
      "step": 22640
    },
    {
      "epoch": 0.408875551607335,
      "grad_norm": 0.5427417159080505,
      "learning_rate": 1.2823683024258096e-05,
      "loss": 0.276,
      "step": 22645
    },
    {
      "epoch": 0.40896583104023576,
      "grad_norm": 0.49803993105888367,
      "learning_rate": 1.2820962085819577e-05,
      "loss": 0.1651,
      "step": 22650
    },
    {
      "epoch": 0.40905611047313645,
      "grad_norm": 0.7014377117156982,
      "learning_rate": 1.2818240920455166e-05,
      "loss": 0.2302,
      "step": 22655
    },
    {
      "epoch": 0.4091463899060372,
      "grad_norm": 0.31781819462776184,
      "learning_rate": 1.2815519528383767e-05,
      "loss": 0.2744,
      "step": 22660
    },
    {
      "epoch": 0.4092366693389379,
      "grad_norm": 0.4156563878059387,
      "learning_rate": 1.2812797909824291e-05,
      "loss": 0.2408,
      "step": 22665
    },
    {
      "epoch": 0.4093269487718386,
      "grad_norm": 0.4748717248439789,
      "learning_rate": 1.2810076064995673e-05,
      "loss": 0.2519,
      "step": 22670
    },
    {
      "epoch": 0.4094172282047393,
      "grad_norm": 0.38562139868736267,
      "learning_rate": 1.2807353994116866e-05,
      "loss": 0.2516,
      "step": 22675
    },
    {
      "epoch": 0.40950750763764004,
      "grad_norm": 0.5304756164550781,
      "learning_rate": 1.2804631697406845e-05,
      "loss": 0.2908,
      "step": 22680
    },
    {
      "epoch": 0.40959778707054073,
      "grad_norm": 0.48928162455558777,
      "learning_rate": 1.2801909175084595e-05,
      "loss": 0.3051,
      "step": 22685
    },
    {
      "epoch": 0.4096880665034415,
      "grad_norm": 0.5467295050621033,
      "learning_rate": 1.2799186427369121e-05,
      "loss": 0.2783,
      "step": 22690
    },
    {
      "epoch": 0.40977834593634216,
      "grad_norm": 1.0147714614868164,
      "learning_rate": 1.2796463454479454e-05,
      "loss": 0.2478,
      "step": 22695
    },
    {
      "epoch": 0.4098686253692429,
      "grad_norm": 0.624610424041748,
      "learning_rate": 1.2793740256634632e-05,
      "loss": 0.4032,
      "step": 22700
    },
    {
      "epoch": 0.4099589048021436,
      "grad_norm": 0.3689691424369812,
      "learning_rate": 1.2791016834053723e-05,
      "loss": 0.2545,
      "step": 22705
    },
    {
      "epoch": 0.41004918423504433,
      "grad_norm": 0.45096608996391296,
      "learning_rate": 1.2788293186955798e-05,
      "loss": 0.2484,
      "step": 22710
    },
    {
      "epoch": 0.410139463667945,
      "grad_norm": 0.5789922475814819,
      "learning_rate": 1.2785569315559963e-05,
      "loss": 0.2958,
      "step": 22715
    },
    {
      "epoch": 0.41022974310084576,
      "grad_norm": 0.3758523166179657,
      "learning_rate": 1.2782845220085325e-05,
      "loss": 0.2646,
      "step": 22720
    },
    {
      "epoch": 0.41032002253374644,
      "grad_norm": 0.41890549659729004,
      "learning_rate": 1.2780120900751026e-05,
      "loss": 0.2976,
      "step": 22725
    },
    {
      "epoch": 0.4104103019666472,
      "grad_norm": 0.6090922355651855,
      "learning_rate": 1.277739635777621e-05,
      "loss": 0.2118,
      "step": 22730
    },
    {
      "epoch": 0.41050058139954787,
      "grad_norm": 0.5390254259109497,
      "learning_rate": 1.2774671591380054e-05,
      "loss": 0.3203,
      "step": 22735
    },
    {
      "epoch": 0.4105908608324486,
      "grad_norm": 0.355477511882782,
      "learning_rate": 1.277194660178174e-05,
      "loss": 0.2058,
      "step": 22740
    },
    {
      "epoch": 0.4106811402653493,
      "grad_norm": 0.3480255603790283,
      "learning_rate": 1.2769221389200478e-05,
      "loss": 0.258,
      "step": 22745
    },
    {
      "epoch": 0.41077141969825004,
      "grad_norm": 0.3804841637611389,
      "learning_rate": 1.276649595385549e-05,
      "loss": 0.2201,
      "step": 22750
    },
    {
      "epoch": 0.4108616991311507,
      "grad_norm": 0.5269613862037659,
      "learning_rate": 1.2763770295966013e-05,
      "loss": 0.3018,
      "step": 22755
    },
    {
      "epoch": 0.41095197856405147,
      "grad_norm": 0.523177444934845,
      "learning_rate": 1.2761044415751313e-05,
      "loss": 0.2714,
      "step": 22760
    },
    {
      "epoch": 0.41104225799695215,
      "grad_norm": 0.3963806927204132,
      "learning_rate": 1.2758318313430665e-05,
      "loss": 0.2408,
      "step": 22765
    },
    {
      "epoch": 0.4111325374298529,
      "grad_norm": 0.5341278314590454,
      "learning_rate": 1.2755591989223361e-05,
      "loss": 0.1856,
      "step": 22770
    },
    {
      "epoch": 0.4112228168627536,
      "grad_norm": 0.735899031162262,
      "learning_rate": 1.2752865443348719e-05,
      "loss": 0.3275,
      "step": 22775
    },
    {
      "epoch": 0.4113130962956543,
      "grad_norm": 0.23954825103282928,
      "learning_rate": 1.2750138676026068e-05,
      "loss": 0.3269,
      "step": 22780
    },
    {
      "epoch": 0.411403375728555,
      "grad_norm": 0.5179336667060852,
      "learning_rate": 1.2747411687474753e-05,
      "loss": 0.2438,
      "step": 22785
    },
    {
      "epoch": 0.41149365516145575,
      "grad_norm": 0.3703598082065582,
      "learning_rate": 1.2744684477914144e-05,
      "loss": 0.3266,
      "step": 22790
    },
    {
      "epoch": 0.41158393459435644,
      "grad_norm": 0.427204430103302,
      "learning_rate": 1.2741957047563627e-05,
      "loss": 0.2964,
      "step": 22795
    },
    {
      "epoch": 0.4116742140272572,
      "grad_norm": 0.5042241215705872,
      "learning_rate": 1.2739229396642601e-05,
      "loss": 0.2434,
      "step": 22800
    },
    {
      "epoch": 0.41176449346015787,
      "grad_norm": 0.44526392221450806,
      "learning_rate": 1.2736501525370483e-05,
      "loss": 0.2303,
      "step": 22805
    },
    {
      "epoch": 0.4118547728930586,
      "grad_norm": 0.41565486788749695,
      "learning_rate": 1.2733773433966717e-05,
      "loss": 0.2661,
      "step": 22810
    },
    {
      "epoch": 0.4119450523259593,
      "grad_norm": 0.8846193552017212,
      "learning_rate": 1.2731045122650754e-05,
      "loss": 0.2412,
      "step": 22815
    },
    {
      "epoch": 0.41203533175886004,
      "grad_norm": 0.537651002407074,
      "learning_rate": 1.2728316591642065e-05,
      "loss": 0.2317,
      "step": 22820
    },
    {
      "epoch": 0.4121256111917607,
      "grad_norm": 0.41685473918914795,
      "learning_rate": 1.2725587841160146e-05,
      "loss": 0.2167,
      "step": 22825
    },
    {
      "epoch": 0.41221589062466146,
      "grad_norm": 0.43952709436416626,
      "learning_rate": 1.27228588714245e-05,
      "loss": 0.3201,
      "step": 22830
    },
    {
      "epoch": 0.41230617005756215,
      "grad_norm": 0.4234485328197479,
      "learning_rate": 1.2720129682654659e-05,
      "loss": 0.2127,
      "step": 22835
    },
    {
      "epoch": 0.4123964494904629,
      "grad_norm": 0.5349525809288025,
      "learning_rate": 1.2717400275070154e-05,
      "loss": 0.2614,
      "step": 22840
    },
    {
      "epoch": 0.4124867289233636,
      "grad_norm": 0.6713441014289856,
      "learning_rate": 1.2714670648890562e-05,
      "loss": 0.3333,
      "step": 22845
    },
    {
      "epoch": 0.4125770083562643,
      "grad_norm": 0.3737652897834778,
      "learning_rate": 1.2711940804335452e-05,
      "loss": 0.1947,
      "step": 22850
    },
    {
      "epoch": 0.412667287789165,
      "grad_norm": 0.4805316627025604,
      "learning_rate": 1.2709210741624418e-05,
      "loss": 0.2479,
      "step": 22855
    },
    {
      "epoch": 0.41275756722206575,
      "grad_norm": 0.5672625303268433,
      "learning_rate": 1.2706480460977081e-05,
      "loss": 0.2142,
      "step": 22860
    },
    {
      "epoch": 0.41284784665496643,
      "grad_norm": 0.4623817801475525,
      "learning_rate": 1.270374996261307e-05,
      "loss": 0.2617,
      "step": 22865
    },
    {
      "epoch": 0.4129381260878672,
      "grad_norm": 0.37226539850234985,
      "learning_rate": 1.2701019246752034e-05,
      "loss": 0.2183,
      "step": 22870
    },
    {
      "epoch": 0.41302840552076786,
      "grad_norm": 0.889231264591217,
      "learning_rate": 1.2698288313613634e-05,
      "loss": 0.2961,
      "step": 22875
    },
    {
      "epoch": 0.4131186849536686,
      "grad_norm": 0.506869912147522,
      "learning_rate": 1.2695557163417558e-05,
      "loss": 0.2368,
      "step": 22880
    },
    {
      "epoch": 0.4132089643865693,
      "grad_norm": 0.7621840238571167,
      "learning_rate": 1.2692825796383508e-05,
      "loss": 0.2583,
      "step": 22885
    },
    {
      "epoch": 0.41329924381947003,
      "grad_norm": 0.5102425217628479,
      "learning_rate": 1.2690094212731202e-05,
      "loss": 0.3368,
      "step": 22890
    },
    {
      "epoch": 0.4133895232523707,
      "grad_norm": 0.3810027837753296,
      "learning_rate": 1.2687362412680375e-05,
      "loss": 0.1905,
      "step": 22895
    },
    {
      "epoch": 0.41347980268527146,
      "grad_norm": 0.4110488295555115,
      "learning_rate": 1.2684630396450782e-05,
      "loss": 0.3392,
      "step": 22900
    },
    {
      "epoch": 0.41357008211817214,
      "grad_norm": 0.3828012943267822,
      "learning_rate": 1.268189816426219e-05,
      "loss": 0.1826,
      "step": 22905
    },
    {
      "epoch": 0.4136603615510729,
      "grad_norm": 0.5053306221961975,
      "learning_rate": 1.2679165716334391e-05,
      "loss": 0.3074,
      "step": 22910
    },
    {
      "epoch": 0.41375064098397357,
      "grad_norm": 0.3866608440876007,
      "learning_rate": 1.2676433052887192e-05,
      "loss": 0.3494,
      "step": 22915
    },
    {
      "epoch": 0.4138409204168743,
      "grad_norm": 0.5869969129562378,
      "learning_rate": 1.2673700174140408e-05,
      "loss": 0.3472,
      "step": 22920
    },
    {
      "epoch": 0.413931199849775,
      "grad_norm": 0.6221544742584229,
      "learning_rate": 1.267096708031389e-05,
      "loss": 0.2017,
      "step": 22925
    },
    {
      "epoch": 0.41402147928267574,
      "grad_norm": 0.5329904556274414,
      "learning_rate": 1.2668233771627488e-05,
      "loss": 0.3864,
      "step": 22930
    },
    {
      "epoch": 0.41411175871557643,
      "grad_norm": 0.6042413711547852,
      "learning_rate": 1.2665500248301079e-05,
      "loss": 0.3168,
      "step": 22935
    },
    {
      "epoch": 0.41420203814847717,
      "grad_norm": 0.43554195761680603,
      "learning_rate": 1.2662766510554554e-05,
      "loss": 0.3326,
      "step": 22940
    },
    {
      "epoch": 0.41429231758137786,
      "grad_norm": 0.4658290445804596,
      "learning_rate": 1.2660032558607827e-05,
      "loss": 0.2667,
      "step": 22945
    },
    {
      "epoch": 0.4143825970142786,
      "grad_norm": 0.514781653881073,
      "learning_rate": 1.265729839268082e-05,
      "loss": 0.3871,
      "step": 22950
    },
    {
      "epoch": 0.4144728764471793,
      "grad_norm": 0.47646093368530273,
      "learning_rate": 1.2654564012993472e-05,
      "loss": 0.2737,
      "step": 22955
    },
    {
      "epoch": 0.41456315588008,
      "grad_norm": 0.49809929728507996,
      "learning_rate": 1.2651829419765755e-05,
      "loss": 0.3645,
      "step": 22960
    },
    {
      "epoch": 0.4146534353129807,
      "grad_norm": 0.35647469758987427,
      "learning_rate": 1.2649094613217643e-05,
      "loss": 0.2278,
      "step": 22965
    },
    {
      "epoch": 0.41474371474588145,
      "grad_norm": 1.0289150476455688,
      "learning_rate": 1.2646359593569128e-05,
      "loss": 0.2856,
      "step": 22970
    },
    {
      "epoch": 0.41483399417878214,
      "grad_norm": 0.35534340143203735,
      "learning_rate": 1.2643624361040222e-05,
      "loss": 0.1699,
      "step": 22975
    },
    {
      "epoch": 0.4149242736116829,
      "grad_norm": 1.2184815406799316,
      "learning_rate": 1.2640888915850964e-05,
      "loss": 0.2914,
      "step": 22980
    },
    {
      "epoch": 0.4150145530445836,
      "grad_norm": 0.30880922079086304,
      "learning_rate": 1.2638153258221389e-05,
      "loss": 0.2778,
      "step": 22985
    },
    {
      "epoch": 0.4151048324774843,
      "grad_norm": 0.45844656229019165,
      "learning_rate": 1.2635417388371565e-05,
      "loss": 0.3023,
      "step": 22990
    },
    {
      "epoch": 0.41519511191038505,
      "grad_norm": 0.40415072441101074,
      "learning_rate": 1.2632681306521577e-05,
      "loss": 0.2527,
      "step": 22995
    },
    {
      "epoch": 0.41528539134328574,
      "grad_norm": 0.37578916549682617,
      "learning_rate": 1.2629945012891515e-05,
      "loss": 0.2063,
      "step": 23000
    },
    {
      "epoch": 0.4153756707761865,
      "grad_norm": 0.448086678981781,
      "learning_rate": 1.2627208507701501e-05,
      "loss": 0.31,
      "step": 23005
    },
    {
      "epoch": 0.41546595020908716,
      "grad_norm": 0.4715213179588318,
      "learning_rate": 1.2624471791171665e-05,
      "loss": 0.2169,
      "step": 23010
    },
    {
      "epoch": 0.4155562296419879,
      "grad_norm": 0.7206951975822449,
      "learning_rate": 1.2621734863522152e-05,
      "loss": 0.3105,
      "step": 23015
    },
    {
      "epoch": 0.4156465090748886,
      "grad_norm": 0.8258635997772217,
      "learning_rate": 1.2618997724973132e-05,
      "loss": 0.2576,
      "step": 23020
    },
    {
      "epoch": 0.41573678850778933,
      "grad_norm": 0.3769485652446747,
      "learning_rate": 1.2616260375744788e-05,
      "loss": 0.1799,
      "step": 23025
    },
    {
      "epoch": 0.41582706794069,
      "grad_norm": 0.514301598072052,
      "learning_rate": 1.2613522816057319e-05,
      "loss": 0.3868,
      "step": 23030
    },
    {
      "epoch": 0.41591734737359076,
      "grad_norm": 0.456122487783432,
      "learning_rate": 1.2610785046130943e-05,
      "loss": 0.2225,
      "step": 23035
    },
    {
      "epoch": 0.41600762680649145,
      "grad_norm": 0.6266130805015564,
      "learning_rate": 1.260804706618589e-05,
      "loss": 0.226,
      "step": 23040
    },
    {
      "epoch": 0.4160979062393922,
      "grad_norm": 0.33325594663619995,
      "learning_rate": 1.2605308876442412e-05,
      "loss": 0.2569,
      "step": 23045
    },
    {
      "epoch": 0.4161881856722929,
      "grad_norm": 0.5136666297912598,
      "learning_rate": 1.2602570477120779e-05,
      "loss": 0.1929,
      "step": 23050
    },
    {
      "epoch": 0.4162784651051936,
      "grad_norm": 0.35206708312034607,
      "learning_rate": 1.2599831868441272e-05,
      "loss": 0.1797,
      "step": 23055
    },
    {
      "epoch": 0.4163687445380943,
      "grad_norm": 0.38802480697631836,
      "learning_rate": 1.2597093050624196e-05,
      "loss": 0.3006,
      "step": 23060
    },
    {
      "epoch": 0.41645902397099505,
      "grad_norm": 0.575931966304779,
      "learning_rate": 1.2594354023889865e-05,
      "loss": 0.2287,
      "step": 23065
    },
    {
      "epoch": 0.41654930340389573,
      "grad_norm": 1.2482432126998901,
      "learning_rate": 1.259161478845862e-05,
      "loss": 0.2879,
      "step": 23070
    },
    {
      "epoch": 0.4166395828367965,
      "grad_norm": 0.28200772404670715,
      "learning_rate": 1.2588875344550805e-05,
      "loss": 0.2682,
      "step": 23075
    },
    {
      "epoch": 0.41672986226969716,
      "grad_norm": 0.40231940150260925,
      "learning_rate": 1.258613569238679e-05,
      "loss": 0.2936,
      "step": 23080
    },
    {
      "epoch": 0.4168201417025979,
      "grad_norm": 0.43480750918388367,
      "learning_rate": 1.258339583218696e-05,
      "loss": 0.3142,
      "step": 23085
    },
    {
      "epoch": 0.4169104211354986,
      "grad_norm": 0.4709104895591736,
      "learning_rate": 1.2580655764171724e-05,
      "loss": 0.2487,
      "step": 23090
    },
    {
      "epoch": 0.41700070056839933,
      "grad_norm": 0.27902746200561523,
      "learning_rate": 1.2577915488561492e-05,
      "loss": 0.2756,
      "step": 23095
    },
    {
      "epoch": 0.4170909800013,
      "grad_norm": 0.5836136341094971,
      "learning_rate": 1.25751750055767e-05,
      "loss": 0.2459,
      "step": 23100
    },
    {
      "epoch": 0.41718125943420076,
      "grad_norm": 0.3742866814136505,
      "learning_rate": 1.2572434315437805e-05,
      "loss": 0.244,
      "step": 23105
    },
    {
      "epoch": 0.41727153886710144,
      "grad_norm": 0.48316583037376404,
      "learning_rate": 1.2569693418365272e-05,
      "loss": 0.2519,
      "step": 23110
    },
    {
      "epoch": 0.4173618183000022,
      "grad_norm": 0.3660636246204376,
      "learning_rate": 1.2566952314579589e-05,
      "loss": 0.2304,
      "step": 23115
    },
    {
      "epoch": 0.41745209773290287,
      "grad_norm": 0.6036543846130371,
      "learning_rate": 1.256421100430125e-05,
      "loss": 0.3532,
      "step": 23120
    },
    {
      "epoch": 0.4175423771658036,
      "grad_norm": 0.3633591830730438,
      "learning_rate": 1.2561469487750784e-05,
      "loss": 0.2204,
      "step": 23125
    },
    {
      "epoch": 0.4176326565987043,
      "grad_norm": 0.30054494738578796,
      "learning_rate": 1.2558727765148722e-05,
      "loss": 0.2039,
      "step": 23130
    },
    {
      "epoch": 0.41772293603160504,
      "grad_norm": 0.3902128338813782,
      "learning_rate": 1.2555985836715612e-05,
      "loss": 0.2077,
      "step": 23135
    },
    {
      "epoch": 0.4178132154645057,
      "grad_norm": 0.3770866394042969,
      "learning_rate": 1.2553243702672027e-05,
      "loss": 0.2709,
      "step": 23140
    },
    {
      "epoch": 0.41790349489740647,
      "grad_norm": 0.3882823586463928,
      "learning_rate": 1.255050136323855e-05,
      "loss": 0.1715,
      "step": 23145
    },
    {
      "epoch": 0.41799377433030716,
      "grad_norm": 0.46126997470855713,
      "learning_rate": 1.2547758818635782e-05,
      "loss": 0.2966,
      "step": 23150
    },
    {
      "epoch": 0.4180840537632079,
      "grad_norm": 0.2830755412578583,
      "learning_rate": 1.2545016069084339e-05,
      "loss": 0.2856,
      "step": 23155
    },
    {
      "epoch": 0.4181743331961086,
      "grad_norm": 0.4069587290287018,
      "learning_rate": 1.2542273114804862e-05,
      "loss": 0.2747,
      "step": 23160
    },
    {
      "epoch": 0.4182646126290093,
      "grad_norm": 0.593215823173523,
      "learning_rate": 1.2539529956017997e-05,
      "loss": 0.3418,
      "step": 23165
    },
    {
      "epoch": 0.41835489206191,
      "grad_norm": 0.5367265939712524,
      "learning_rate": 1.2536786592944405e-05,
      "loss": 0.2829,
      "step": 23170
    },
    {
      "epoch": 0.41844517149481075,
      "grad_norm": 0.44455626606941223,
      "learning_rate": 1.253404302580478e-05,
      "loss": 0.1921,
      "step": 23175
    },
    {
      "epoch": 0.41853545092771144,
      "grad_norm": 0.49821528792381287,
      "learning_rate": 1.253129925481982e-05,
      "loss": 0.334,
      "step": 23180
    },
    {
      "epoch": 0.4186257303606122,
      "grad_norm": 0.4233429729938507,
      "learning_rate": 1.2528555280210233e-05,
      "loss": 0.2826,
      "step": 23185
    },
    {
      "epoch": 0.41871600979351287,
      "grad_norm": 0.5704105496406555,
      "learning_rate": 1.2525811102196763e-05,
      "loss": 0.2453,
      "step": 23190
    },
    {
      "epoch": 0.4188062892264136,
      "grad_norm": 0.46656692028045654,
      "learning_rate": 1.2523066721000154e-05,
      "loss": 0.3139,
      "step": 23195
    },
    {
      "epoch": 0.4188965686593143,
      "grad_norm": 0.423334538936615,
      "learning_rate": 1.2520322136841169e-05,
      "loss": 0.2353,
      "step": 23200
    },
    {
      "epoch": 0.41898684809221504,
      "grad_norm": 0.47647279500961304,
      "learning_rate": 1.2517577349940595e-05,
      "loss": 0.2704,
      "step": 23205
    },
    {
      "epoch": 0.4190771275251157,
      "grad_norm": 0.5600226521492004,
      "learning_rate": 1.2514832360519227e-05,
      "loss": 0.2792,
      "step": 23210
    },
    {
      "epoch": 0.41916740695801646,
      "grad_norm": 0.43187180161476135,
      "learning_rate": 1.2512087168797882e-05,
      "loss": 0.2976,
      "step": 23215
    },
    {
      "epoch": 0.41925768639091715,
      "grad_norm": 0.6845264434814453,
      "learning_rate": 1.2509341774997385e-05,
      "loss": 0.2715,
      "step": 23220
    },
    {
      "epoch": 0.4193479658238179,
      "grad_norm": 0.702416181564331,
      "learning_rate": 1.250659617933859e-05,
      "loss": 0.21,
      "step": 23225
    },
    {
      "epoch": 0.4194382452567186,
      "grad_norm": 0.45580577850341797,
      "learning_rate": 1.2503850382042355e-05,
      "loss": 0.2385,
      "step": 23230
    },
    {
      "epoch": 0.4195285246896193,
      "grad_norm": 0.5321224927902222,
      "learning_rate": 1.250110438332956e-05,
      "loss": 0.2274,
      "step": 23235
    },
    {
      "epoch": 0.41961880412252,
      "grad_norm": 0.49888378381729126,
      "learning_rate": 1.2498358183421106e-05,
      "loss": 0.3107,
      "step": 23240
    },
    {
      "epoch": 0.41970908355542075,
      "grad_norm": 0.4273422956466675,
      "learning_rate": 1.2495611782537902e-05,
      "loss": 0.1971,
      "step": 23245
    },
    {
      "epoch": 0.41979936298832143,
      "grad_norm": 0.6001827716827393,
      "learning_rate": 1.2492865180900874e-05,
      "loss": 0.2154,
      "step": 23250
    },
    {
      "epoch": 0.4198896424212222,
      "grad_norm": 0.3414769172668457,
      "learning_rate": 1.2490118378730964e-05,
      "loss": 0.2462,
      "step": 23255
    },
    {
      "epoch": 0.41997992185412286,
      "grad_norm": 0.5230898857116699,
      "learning_rate": 1.2487371376249137e-05,
      "loss": 0.1892,
      "step": 23260
    },
    {
      "epoch": 0.4200702012870236,
      "grad_norm": 0.4658863842487335,
      "learning_rate": 1.248462417367637e-05,
      "loss": 0.2641,
      "step": 23265
    },
    {
      "epoch": 0.4201604807199243,
      "grad_norm": 0.6205943822860718,
      "learning_rate": 1.248187677123365e-05,
      "loss": 0.3032,
      "step": 23270
    },
    {
      "epoch": 0.42025076015282503,
      "grad_norm": 0.4406639635562897,
      "learning_rate": 1.247912916914199e-05,
      "loss": 0.2362,
      "step": 23275
    },
    {
      "epoch": 0.4203410395857257,
      "grad_norm": 0.44867652654647827,
      "learning_rate": 1.2476381367622414e-05,
      "loss": 0.2107,
      "step": 23280
    },
    {
      "epoch": 0.42043131901862646,
      "grad_norm": 0.341111958026886,
      "learning_rate": 1.2473633366895959e-05,
      "loss": 0.2036,
      "step": 23285
    },
    {
      "epoch": 0.42052159845152715,
      "grad_norm": 0.4511333703994751,
      "learning_rate": 1.2470885167183685e-05,
      "loss": 0.2541,
      "step": 23290
    },
    {
      "epoch": 0.4206118778844279,
      "grad_norm": 0.3271433711051941,
      "learning_rate": 1.2468136768706669e-05,
      "loss": 0.2803,
      "step": 23295
    },
    {
      "epoch": 0.4207021573173286,
      "grad_norm": 0.5134062767028809,
      "learning_rate": 1.246538817168599e-05,
      "loss": 0.3147,
      "step": 23300
    },
    {
      "epoch": 0.4207924367502293,
      "grad_norm": 0.9824153780937195,
      "learning_rate": 1.246263937634276e-05,
      "loss": 0.3014,
      "step": 23305
    },
    {
      "epoch": 0.42088271618313,
      "grad_norm": 1.0824674367904663,
      "learning_rate": 1.2459890382898098e-05,
      "loss": 0.2731,
      "step": 23310
    },
    {
      "epoch": 0.42097299561603074,
      "grad_norm": 0.33683472871780396,
      "learning_rate": 1.245714119157314e-05,
      "loss": 0.3048,
      "step": 23315
    },
    {
      "epoch": 0.42106327504893143,
      "grad_norm": 0.5700289607048035,
      "learning_rate": 1.2454391802589036e-05,
      "loss": 0.2221,
      "step": 23320
    },
    {
      "epoch": 0.42115355448183217,
      "grad_norm": 0.25581130385398865,
      "learning_rate": 1.245164221616696e-05,
      "loss": 0.216,
      "step": 23325
    },
    {
      "epoch": 0.42124383391473286,
      "grad_norm": 0.4467751979827881,
      "learning_rate": 1.244889243252809e-05,
      "loss": 0.336,
      "step": 23330
    },
    {
      "epoch": 0.4213341133476336,
      "grad_norm": 0.22876004874706268,
      "learning_rate": 1.2446142451893628e-05,
      "loss": 0.1749,
      "step": 23335
    },
    {
      "epoch": 0.4214243927805343,
      "grad_norm": 0.7102181911468506,
      "learning_rate": 1.2443392274484794e-05,
      "loss": 0.298,
      "step": 23340
    },
    {
      "epoch": 0.421514672213435,
      "grad_norm": 0.45810016989707947,
      "learning_rate": 1.2440641900522818e-05,
      "loss": 0.2286,
      "step": 23345
    },
    {
      "epoch": 0.4216049516463357,
      "grad_norm": 0.3493264615535736,
      "learning_rate": 1.2437891330228942e-05,
      "loss": 0.3413,
      "step": 23350
    },
    {
      "epoch": 0.42169523107923645,
      "grad_norm": 0.262546181678772,
      "learning_rate": 1.2435140563824435e-05,
      "loss": 0.2425,
      "step": 23355
    },
    {
      "epoch": 0.42178551051213714,
      "grad_norm": 0.39406266808509827,
      "learning_rate": 1.2432389601530576e-05,
      "loss": 0.2303,
      "step": 23360
    },
    {
      "epoch": 0.4218757899450379,
      "grad_norm": 0.5727577209472656,
      "learning_rate": 1.2429638443568657e-05,
      "loss": 0.3111,
      "step": 23365
    },
    {
      "epoch": 0.42196606937793857,
      "grad_norm": 0.6123302578926086,
      "learning_rate": 1.2426887090159992e-05,
      "loss": 0.2073,
      "step": 23370
    },
    {
      "epoch": 0.4220563488108393,
      "grad_norm": 0.5969008207321167,
      "learning_rate": 1.2424135541525906e-05,
      "loss": 0.2418,
      "step": 23375
    },
    {
      "epoch": 0.42214662824374,
      "grad_norm": 0.36988839507102966,
      "learning_rate": 1.2421383797887738e-05,
      "loss": 0.2219,
      "step": 23380
    },
    {
      "epoch": 0.42223690767664074,
      "grad_norm": 0.4886693060398102,
      "learning_rate": 1.241863185946685e-05,
      "loss": 0.2436,
      "step": 23385
    },
    {
      "epoch": 0.4223271871095415,
      "grad_norm": 0.334306001663208,
      "learning_rate": 1.2415879726484614e-05,
      "loss": 0.2905,
      "step": 23390
    },
    {
      "epoch": 0.42241746654244217,
      "grad_norm": 0.4661378264427185,
      "learning_rate": 1.2413127399162422e-05,
      "loss": 0.3199,
      "step": 23395
    },
    {
      "epoch": 0.4225077459753429,
      "grad_norm": 0.2870534062385559,
      "learning_rate": 1.2410374877721672e-05,
      "loss": 0.2479,
      "step": 23400
    },
    {
      "epoch": 0.4225980254082436,
      "grad_norm": 0.3354700803756714,
      "learning_rate": 1.240762216238379e-05,
      "loss": 0.3226,
      "step": 23405
    },
    {
      "epoch": 0.42268830484114434,
      "grad_norm": 0.8332720398902893,
      "learning_rate": 1.2404869253370212e-05,
      "loss": 0.167,
      "step": 23410
    },
    {
      "epoch": 0.422778584274045,
      "grad_norm": 0.4037450850009918,
      "learning_rate": 1.2402116150902384e-05,
      "loss": 0.2679,
      "step": 23415
    },
    {
      "epoch": 0.42286886370694576,
      "grad_norm": 0.5018265843391418,
      "learning_rate": 1.2399362855201778e-05,
      "loss": 0.2422,
      "step": 23420
    },
    {
      "epoch": 0.42295914313984645,
      "grad_norm": 0.3457806408405304,
      "learning_rate": 1.2396609366489876e-05,
      "loss": 0.2515,
      "step": 23425
    },
    {
      "epoch": 0.4230494225727472,
      "grad_norm": 0.4327681064605713,
      "learning_rate": 1.2393855684988176e-05,
      "loss": 0.2149,
      "step": 23430
    },
    {
      "epoch": 0.4231397020056479,
      "grad_norm": 0.49335646629333496,
      "learning_rate": 1.239110181091819e-05,
      "loss": 0.3475,
      "step": 23435
    },
    {
      "epoch": 0.4232299814385486,
      "grad_norm": 0.4823317229747772,
      "learning_rate": 1.238834774450145e-05,
      "loss": 0.358,
      "step": 23440
    },
    {
      "epoch": 0.4233202608714493,
      "grad_norm": 0.38144388794898987,
      "learning_rate": 1.2385593485959497e-05,
      "loss": 0.256,
      "step": 23445
    },
    {
      "epoch": 0.42341054030435005,
      "grad_norm": 0.5280448198318481,
      "learning_rate": 1.2382839035513894e-05,
      "loss": 0.2755,
      "step": 23450
    },
    {
      "epoch": 0.42350081973725073,
      "grad_norm": 0.3799925446510315,
      "learning_rate": 1.2380084393386214e-05,
      "loss": 0.1969,
      "step": 23455
    },
    {
      "epoch": 0.4235910991701515,
      "grad_norm": 0.6687923073768616,
      "learning_rate": 1.2377329559798053e-05,
      "loss": 0.2413,
      "step": 23460
    },
    {
      "epoch": 0.42368137860305216,
      "grad_norm": 0.3646118640899658,
      "learning_rate": 1.237457453497101e-05,
      "loss": 0.1871,
      "step": 23465
    },
    {
      "epoch": 0.4237716580359529,
      "grad_norm": 0.4773862957954407,
      "learning_rate": 1.237181931912671e-05,
      "loss": 0.2827,
      "step": 23470
    },
    {
      "epoch": 0.4238619374688536,
      "grad_norm": 0.4353986382484436,
      "learning_rate": 1.2369063912486793e-05,
      "loss": 0.2613,
      "step": 23475
    },
    {
      "epoch": 0.42395221690175433,
      "grad_norm": 0.9043532013893127,
      "learning_rate": 1.2366308315272911e-05,
      "loss": 0.2748,
      "step": 23480
    },
    {
      "epoch": 0.424042496334655,
      "grad_norm": 0.6638239622116089,
      "learning_rate": 1.2363552527706724e-05,
      "loss": 0.2582,
      "step": 23485
    },
    {
      "epoch": 0.42413277576755576,
      "grad_norm": 0.3441218435764313,
      "learning_rate": 1.2360796550009924e-05,
      "loss": 0.2524,
      "step": 23490
    },
    {
      "epoch": 0.42422305520045644,
      "grad_norm": 0.5626394748687744,
      "learning_rate": 1.2358040382404207e-05,
      "loss": 0.3088,
      "step": 23495
    },
    {
      "epoch": 0.4243133346333572,
      "grad_norm": 0.40816786885261536,
      "learning_rate": 1.2355284025111282e-05,
      "loss": 0.2522,
      "step": 23500
    },
    {
      "epoch": 0.4244036140662579,
      "grad_norm": 0.6815304756164551,
      "learning_rate": 1.2352527478352886e-05,
      "loss": 0.3174,
      "step": 23505
    },
    {
      "epoch": 0.4244938934991586,
      "grad_norm": 0.4805319309234619,
      "learning_rate": 1.2349770742350759e-05,
      "loss": 0.2346,
      "step": 23510
    },
    {
      "epoch": 0.4245841729320593,
      "grad_norm": 0.4371818006038666,
      "learning_rate": 1.234701381732666e-05,
      "loss": 0.3627,
      "step": 23515
    },
    {
      "epoch": 0.42467445236496004,
      "grad_norm": 0.4816952645778656,
      "learning_rate": 1.234425670350236e-05,
      "loss": 0.3147,
      "step": 23520
    },
    {
      "epoch": 0.42476473179786073,
      "grad_norm": 0.25366294384002686,
      "learning_rate": 1.2341499401099656e-05,
      "loss": 0.1932,
      "step": 23525
    },
    {
      "epoch": 0.42485501123076147,
      "grad_norm": 0.5283558964729309,
      "learning_rate": 1.233874191034035e-05,
      "loss": 0.2441,
      "step": 23530
    },
    {
      "epoch": 0.42494529066366216,
      "grad_norm": 0.4625545144081116,
      "learning_rate": 1.233598423144626e-05,
      "loss": 0.1845,
      "step": 23535
    },
    {
      "epoch": 0.4250355700965629,
      "grad_norm": 0.5480676889419556,
      "learning_rate": 1.2333226364639226e-05,
      "loss": 0.2443,
      "step": 23540
    },
    {
      "epoch": 0.4251258495294636,
      "grad_norm": 0.40750113129615784,
      "learning_rate": 1.2330468310141092e-05,
      "loss": 0.2311,
      "step": 23545
    },
    {
      "epoch": 0.4252161289623643,
      "grad_norm": 0.30785858631134033,
      "learning_rate": 1.2327710068173726e-05,
      "loss": 0.2228,
      "step": 23550
    },
    {
      "epoch": 0.425306408395265,
      "grad_norm": 0.3887018859386444,
      "learning_rate": 1.2324951638959011e-05,
      "loss": 0.1793,
      "step": 23555
    },
    {
      "epoch": 0.42539668782816575,
      "grad_norm": 0.4880499243736267,
      "learning_rate": 1.232219302271884e-05,
      "loss": 0.1978,
      "step": 23560
    },
    {
      "epoch": 0.42548696726106644,
      "grad_norm": 0.42758166790008545,
      "learning_rate": 1.231943421967512e-05,
      "loss": 0.2948,
      "step": 23565
    },
    {
      "epoch": 0.4255772466939672,
      "grad_norm": 0.4794660806655884,
      "learning_rate": 1.2316675230049788e-05,
      "loss": 0.2399,
      "step": 23570
    },
    {
      "epoch": 0.42566752612686787,
      "grad_norm": 0.32832571864128113,
      "learning_rate": 1.2313916054064774e-05,
      "loss": 0.286,
      "step": 23575
    },
    {
      "epoch": 0.4257578055597686,
      "grad_norm": 0.3535519540309906,
      "learning_rate": 1.231115669194204e-05,
      "loss": 0.2398,
      "step": 23580
    },
    {
      "epoch": 0.4258480849926693,
      "grad_norm": 0.49113744497299194,
      "learning_rate": 1.230839714390355e-05,
      "loss": 0.3338,
      "step": 23585
    },
    {
      "epoch": 0.42593836442557004,
      "grad_norm": 0.28829047083854675,
      "learning_rate": 1.2305637410171296e-05,
      "loss": 0.274,
      "step": 23590
    },
    {
      "epoch": 0.4260286438584707,
      "grad_norm": 0.244572713971138,
      "learning_rate": 1.2302877490967275e-05,
      "loss": 0.1644,
      "step": 23595
    },
    {
      "epoch": 0.42611892329137147,
      "grad_norm": 0.3454359173774719,
      "learning_rate": 1.2300117386513501e-05,
      "loss": 0.178,
      "step": 23600
    },
    {
      "epoch": 0.42620920272427215,
      "grad_norm": 0.3768734931945801,
      "learning_rate": 1.229735709703201e-05,
      "loss": 0.3869,
      "step": 23605
    },
    {
      "epoch": 0.4262994821571729,
      "grad_norm": 0.4586016535758972,
      "learning_rate": 1.2294596622744845e-05,
      "loss": 0.2703,
      "step": 23610
    },
    {
      "epoch": 0.4263897615900736,
      "grad_norm": 0.4032464623451233,
      "learning_rate": 1.2291835963874063e-05,
      "loss": 0.273,
      "step": 23615
    },
    {
      "epoch": 0.4264800410229743,
      "grad_norm": 0.4911908209323883,
      "learning_rate": 1.228907512064174e-05,
      "loss": 0.2433,
      "step": 23620
    },
    {
      "epoch": 0.426570320455875,
      "grad_norm": 0.3181964159011841,
      "learning_rate": 1.228631409326997e-05,
      "loss": 0.2955,
      "step": 23625
    },
    {
      "epoch": 0.42666059988877575,
      "grad_norm": 0.35814782977104187,
      "learning_rate": 1.2283552881980852e-05,
      "loss": 0.2569,
      "step": 23630
    },
    {
      "epoch": 0.42675087932167644,
      "grad_norm": 0.6058424115180969,
      "learning_rate": 1.2280791486996509e-05,
      "loss": 0.251,
      "step": 23635
    },
    {
      "epoch": 0.4268411587545772,
      "grad_norm": 0.5719448328018188,
      "learning_rate": 1.2278029908539073e-05,
      "loss": 0.2149,
      "step": 23640
    },
    {
      "epoch": 0.42693143818747786,
      "grad_norm": 0.3337777554988861,
      "learning_rate": 1.2275268146830691e-05,
      "loss": 0.275,
      "step": 23645
    },
    {
      "epoch": 0.4270217176203786,
      "grad_norm": 0.47124147415161133,
      "learning_rate": 1.2272506202093533e-05,
      "loss": 0.3303,
      "step": 23650
    },
    {
      "epoch": 0.4271119970532793,
      "grad_norm": 0.4274919331073761,
      "learning_rate": 1.2269744074549773e-05,
      "loss": 0.2398,
      "step": 23655
    },
    {
      "epoch": 0.42720227648618003,
      "grad_norm": 0.3180335462093353,
      "learning_rate": 1.2266981764421603e-05,
      "loss": 0.1748,
      "step": 23660
    },
    {
      "epoch": 0.4272925559190807,
      "grad_norm": 0.753650426864624,
      "learning_rate": 1.2264219271931234e-05,
      "loss": 0.2856,
      "step": 23665
    },
    {
      "epoch": 0.42738283535198146,
      "grad_norm": 0.622544527053833,
      "learning_rate": 1.2261456597300888e-05,
      "loss": 0.2688,
      "step": 23670
    },
    {
      "epoch": 0.42747311478488215,
      "grad_norm": 0.42950546741485596,
      "learning_rate": 1.22586937407528e-05,
      "loss": 0.3049,
      "step": 23675
    },
    {
      "epoch": 0.4275633942177829,
      "grad_norm": 0.632057249546051,
      "learning_rate": 1.2255930702509224e-05,
      "loss": 0.3121,
      "step": 23680
    },
    {
      "epoch": 0.4276536736506836,
      "grad_norm": 0.3873615562915802,
      "learning_rate": 1.2253167482792423e-05,
      "loss": 0.3521,
      "step": 23685
    },
    {
      "epoch": 0.4277439530835843,
      "grad_norm": 0.4572433829307556,
      "learning_rate": 1.2250404081824683e-05,
      "loss": 0.1936,
      "step": 23690
    },
    {
      "epoch": 0.427834232516485,
      "grad_norm": 0.5588839054107666,
      "learning_rate": 1.2247640499828295e-05,
      "loss": 0.2833,
      "step": 23695
    },
    {
      "epoch": 0.42792451194938574,
      "grad_norm": 0.5518315434455872,
      "learning_rate": 1.2244876737025568e-05,
      "loss": 0.3176,
      "step": 23700
    },
    {
      "epoch": 0.42801479138228643,
      "grad_norm": 0.5185930728912354,
      "learning_rate": 1.2242112793638833e-05,
      "loss": 0.2754,
      "step": 23705
    },
    {
      "epoch": 0.42810507081518717,
      "grad_norm": 0.3396175503730774,
      "learning_rate": 1.2239348669890426e-05,
      "loss": 0.1739,
      "step": 23710
    },
    {
      "epoch": 0.42819535024808786,
      "grad_norm": 0.3102880120277405,
      "learning_rate": 1.22365843660027e-05,
      "loss": 0.152,
      "step": 23715
    },
    {
      "epoch": 0.4282856296809886,
      "grad_norm": 0.307441383600235,
      "learning_rate": 1.2233819882198021e-05,
      "loss": 0.2695,
      "step": 23720
    },
    {
      "epoch": 0.4283759091138893,
      "grad_norm": 0.5969069004058838,
      "learning_rate": 1.2231055218698775e-05,
      "loss": 0.301,
      "step": 23725
    },
    {
      "epoch": 0.42846618854679,
      "grad_norm": 0.27882397174835205,
      "learning_rate": 1.2228290375727358e-05,
      "loss": 0.2568,
      "step": 23730
    },
    {
      "epoch": 0.4285564679796907,
      "grad_norm": 0.4679926037788391,
      "learning_rate": 1.2225525353506182e-05,
      "loss": 0.2583,
      "step": 23735
    },
    {
      "epoch": 0.42864674741259146,
      "grad_norm": 0.4088680148124695,
      "learning_rate": 1.2222760152257674e-05,
      "loss": 0.2584,
      "step": 23740
    },
    {
      "epoch": 0.42873702684549214,
      "grad_norm": 0.8026046752929688,
      "learning_rate": 1.221999477220427e-05,
      "loss": 0.2621,
      "step": 23745
    },
    {
      "epoch": 0.4288273062783929,
      "grad_norm": 0.5004242062568665,
      "learning_rate": 1.2217229213568427e-05,
      "loss": 0.3945,
      "step": 23750
    },
    {
      "epoch": 0.42891758571129357,
      "grad_norm": 0.5434352159500122,
      "learning_rate": 1.2214463476572615e-05,
      "loss": 0.2763,
      "step": 23755
    },
    {
      "epoch": 0.4290078651441943,
      "grad_norm": 0.5118503570556641,
      "learning_rate": 1.2211697561439318e-05,
      "loss": 0.2636,
      "step": 23760
    },
    {
      "epoch": 0.429098144577095,
      "grad_norm": 0.5239500403404236,
      "learning_rate": 1.2208931468391031e-05,
      "loss": 0.1781,
      "step": 23765
    },
    {
      "epoch": 0.42918842400999574,
      "grad_norm": 0.4253959357738495,
      "learning_rate": 1.220616519765027e-05,
      "loss": 0.3,
      "step": 23770
    },
    {
      "epoch": 0.4292787034428964,
      "grad_norm": 0.5123077630996704,
      "learning_rate": 1.220339874943956e-05,
      "loss": 0.2837,
      "step": 23775
    },
    {
      "epoch": 0.42936898287579717,
      "grad_norm": 0.6357778310775757,
      "learning_rate": 1.220063212398144e-05,
      "loss": 0.1701,
      "step": 23780
    },
    {
      "epoch": 0.4294592623086979,
      "grad_norm": 0.4320623278617859,
      "learning_rate": 1.2197865321498463e-05,
      "loss": 0.3303,
      "step": 23785
    },
    {
      "epoch": 0.4295495417415986,
      "grad_norm": 0.4319918155670166,
      "learning_rate": 1.2195098342213205e-05,
      "loss": 0.307,
      "step": 23790
    },
    {
      "epoch": 0.42963982117449934,
      "grad_norm": 0.4354119896888733,
      "learning_rate": 1.2192331186348243e-05,
      "loss": 0.2688,
      "step": 23795
    },
    {
      "epoch": 0.4297301006074,
      "grad_norm": 0.44331490993499756,
      "learning_rate": 1.2189563854126177e-05,
      "loss": 0.2869,
      "step": 23800
    },
    {
      "epoch": 0.42982038004030076,
      "grad_norm": 0.5300556421279907,
      "learning_rate": 1.2186796345769619e-05,
      "loss": 0.1792,
      "step": 23805
    },
    {
      "epoch": 0.42991065947320145,
      "grad_norm": 0.44784700870513916,
      "learning_rate": 1.2184028661501197e-05,
      "loss": 0.3702,
      "step": 23810
    },
    {
      "epoch": 0.4300009389061022,
      "grad_norm": 0.5087341666221619,
      "learning_rate": 1.2181260801543545e-05,
      "loss": 0.342,
      "step": 23815
    },
    {
      "epoch": 0.4300912183390029,
      "grad_norm": 0.6763350963592529,
      "learning_rate": 1.217849276611932e-05,
      "loss": 0.3196,
      "step": 23820
    },
    {
      "epoch": 0.4301814977719036,
      "grad_norm": 0.30958354473114014,
      "learning_rate": 1.2175724555451195e-05,
      "loss": 0.2174,
      "step": 23825
    },
    {
      "epoch": 0.4302717772048043,
      "grad_norm": 0.5187954306602478,
      "learning_rate": 1.2172956169761845e-05,
      "loss": 0.2851,
      "step": 23830
    },
    {
      "epoch": 0.43036205663770505,
      "grad_norm": 0.8124983906745911,
      "learning_rate": 1.217018760927397e-05,
      "loss": 0.2497,
      "step": 23835
    },
    {
      "epoch": 0.43045233607060573,
      "grad_norm": 0.3612107038497925,
      "learning_rate": 1.2167418874210285e-05,
      "loss": 0.2518,
      "step": 23840
    },
    {
      "epoch": 0.4305426155035065,
      "grad_norm": 0.6983647346496582,
      "learning_rate": 1.2164649964793506e-05,
      "loss": 0.2057,
      "step": 23845
    },
    {
      "epoch": 0.43063289493640716,
      "grad_norm": 0.4899793267250061,
      "learning_rate": 1.2161880881246378e-05,
      "loss": 0.3078,
      "step": 23850
    },
    {
      "epoch": 0.4307231743693079,
      "grad_norm": 0.33246493339538574,
      "learning_rate": 1.2159111623791651e-05,
      "loss": 0.1778,
      "step": 23855
    },
    {
      "epoch": 0.4308134538022086,
      "grad_norm": 0.46771249175071716,
      "learning_rate": 1.2156342192652093e-05,
      "loss": 0.2582,
      "step": 23860
    },
    {
      "epoch": 0.43090373323510933,
      "grad_norm": 0.6646603345870972,
      "learning_rate": 1.2153572588050479e-05,
      "loss": 0.3399,
      "step": 23865
    },
    {
      "epoch": 0.43099401266801,
      "grad_norm": 0.3225894272327423,
      "learning_rate": 1.2150802810209614e-05,
      "loss": 0.2621,
      "step": 23870
    },
    {
      "epoch": 0.43108429210091076,
      "grad_norm": 0.44844329357147217,
      "learning_rate": 1.2148032859352298e-05,
      "loss": 0.2642,
      "step": 23875
    },
    {
      "epoch": 0.43117457153381145,
      "grad_norm": 0.39339956641197205,
      "learning_rate": 1.2145262735701357e-05,
      "loss": 0.2938,
      "step": 23880
    },
    {
      "epoch": 0.4312648509667122,
      "grad_norm": 0.4607665240764618,
      "learning_rate": 1.2142492439479625e-05,
      "loss": 0.3199,
      "step": 23885
    },
    {
      "epoch": 0.4313551303996129,
      "grad_norm": 0.4481928050518036,
      "learning_rate": 1.2139721970909956e-05,
      "loss": 0.2893,
      "step": 23890
    },
    {
      "epoch": 0.4314454098325136,
      "grad_norm": 0.538311243057251,
      "learning_rate": 1.2136951330215212e-05,
      "loss": 0.2104,
      "step": 23895
    },
    {
      "epoch": 0.4315356892654143,
      "grad_norm": 0.35558781027793884,
      "learning_rate": 1.213418051761827e-05,
      "loss": 0.2451,
      "step": 23900
    },
    {
      "epoch": 0.43162596869831504,
      "grad_norm": 0.5374397039413452,
      "learning_rate": 1.2131409533342022e-05,
      "loss": 0.1912,
      "step": 23905
    },
    {
      "epoch": 0.43171624813121573,
      "grad_norm": 0.7439382672309875,
      "learning_rate": 1.2128638377609373e-05,
      "loss": 0.2741,
      "step": 23910
    },
    {
      "epoch": 0.43180652756411647,
      "grad_norm": 0.46327850222587585,
      "learning_rate": 1.2125867050643243e-05,
      "loss": 0.3571,
      "step": 23915
    },
    {
      "epoch": 0.43189680699701716,
      "grad_norm": 0.4672674238681793,
      "learning_rate": 1.2123095552666566e-05,
      "loss": 0.3456,
      "step": 23920
    },
    {
      "epoch": 0.4319870864299179,
      "grad_norm": 0.45635950565338135,
      "learning_rate": 1.212032388390229e-05,
      "loss": 0.2549,
      "step": 23925
    },
    {
      "epoch": 0.4320773658628186,
      "grad_norm": 0.49297451972961426,
      "learning_rate": 1.211755204457337e-05,
      "loss": 0.2797,
      "step": 23930
    },
    {
      "epoch": 0.4321676452957193,
      "grad_norm": 0.45111674070358276,
      "learning_rate": 1.2114780034902787e-05,
      "loss": 0.2937,
      "step": 23935
    },
    {
      "epoch": 0.43225792472862,
      "grad_norm": 0.4889915883541107,
      "learning_rate": 1.2112007855113526e-05,
      "loss": 0.3148,
      "step": 23940
    },
    {
      "epoch": 0.43234820416152075,
      "grad_norm": 0.29992303252220154,
      "learning_rate": 1.210923550542859e-05,
      "loss": 0.3375,
      "step": 23945
    },
    {
      "epoch": 0.43243848359442144,
      "grad_norm": 0.6777974367141724,
      "learning_rate": 1.2106462986070988e-05,
      "loss": 0.285,
      "step": 23950
    },
    {
      "epoch": 0.4325287630273222,
      "grad_norm": 0.5728838443756104,
      "learning_rate": 1.210369029726376e-05,
      "loss": 0.1933,
      "step": 23955
    },
    {
      "epoch": 0.43261904246022287,
      "grad_norm": 0.5338889360427856,
      "learning_rate": 1.210091743922994e-05,
      "loss": 0.2527,
      "step": 23960
    },
    {
      "epoch": 0.4327093218931236,
      "grad_norm": 0.43478772044181824,
      "learning_rate": 1.2098144412192587e-05,
      "loss": 0.3012,
      "step": 23965
    },
    {
      "epoch": 0.4327996013260243,
      "grad_norm": 0.3326312303543091,
      "learning_rate": 1.2095371216374771e-05,
      "loss": 0.293,
      "step": 23970
    },
    {
      "epoch": 0.43288988075892504,
      "grad_norm": 0.4028029143810272,
      "learning_rate": 1.2092597851999578e-05,
      "loss": 0.2374,
      "step": 23975
    },
    {
      "epoch": 0.4329801601918257,
      "grad_norm": 0.45053476095199585,
      "learning_rate": 1.2089824319290097e-05,
      "loss": 0.2518,
      "step": 23980
    },
    {
      "epoch": 0.43307043962472647,
      "grad_norm": 0.5274935364723206,
      "learning_rate": 1.208705061846945e-05,
      "loss": 0.198,
      "step": 23985
    },
    {
      "epoch": 0.43316071905762715,
      "grad_norm": 0.5276974439620972,
      "learning_rate": 1.2084276749760756e-05,
      "loss": 0.2876,
      "step": 23990
    },
    {
      "epoch": 0.4332509984905279,
      "grad_norm": 0.49531668424606323,
      "learning_rate": 1.2081502713387145e-05,
      "loss": 0.3453,
      "step": 23995
    },
    {
      "epoch": 0.4333412779234286,
      "grad_norm": 0.5129486918449402,
      "learning_rate": 1.207872850957178e-05,
      "loss": 0.2143,
      "step": 24000
    },
    {
      "epoch": 0.4334315573563293,
      "grad_norm": 0.8107872009277344,
      "learning_rate": 1.2075954138537817e-05,
      "loss": 0.202,
      "step": 24005
    },
    {
      "epoch": 0.43352183678923,
      "grad_norm": 0.5041257739067078,
      "learning_rate": 1.207317960050844e-05,
      "loss": 0.2312,
      "step": 24010
    },
    {
      "epoch": 0.43361211622213075,
      "grad_norm": 0.4606269896030426,
      "learning_rate": 1.2070404895706835e-05,
      "loss": 0.3045,
      "step": 24015
    },
    {
      "epoch": 0.43370239565503144,
      "grad_norm": 0.42314013838768005,
      "learning_rate": 1.2067630024356212e-05,
      "loss": 0.28,
      "step": 24020
    },
    {
      "epoch": 0.4337926750879322,
      "grad_norm": 0.5237497091293335,
      "learning_rate": 1.2064854986679784e-05,
      "loss": 0.3149,
      "step": 24025
    },
    {
      "epoch": 0.43388295452083286,
      "grad_norm": 0.47119367122650146,
      "learning_rate": 1.2062079782900785e-05,
      "loss": 0.2723,
      "step": 24030
    },
    {
      "epoch": 0.4339732339537336,
      "grad_norm": 0.4915752112865448,
      "learning_rate": 1.2059304413242463e-05,
      "loss": 0.3197,
      "step": 24035
    },
    {
      "epoch": 0.4340635133866343,
      "grad_norm": 0.43169498443603516,
      "learning_rate": 1.2056528877928073e-05,
      "loss": 0.2272,
      "step": 24040
    },
    {
      "epoch": 0.43415379281953503,
      "grad_norm": 0.25861960649490356,
      "learning_rate": 1.2053753177180888e-05,
      "loss": 0.2338,
      "step": 24045
    },
    {
      "epoch": 0.4342440722524357,
      "grad_norm": 0.6282423138618469,
      "learning_rate": 1.2050977311224188e-05,
      "loss": 0.3242,
      "step": 24050
    },
    {
      "epoch": 0.43433435168533646,
      "grad_norm": 0.4926910400390625,
      "learning_rate": 1.2048201280281278e-05,
      "loss": 0.2658,
      "step": 24055
    },
    {
      "epoch": 0.43442463111823715,
      "grad_norm": 0.42777496576309204,
      "learning_rate": 1.2045425084575466e-05,
      "loss": 0.283,
      "step": 24060
    },
    {
      "epoch": 0.4345149105511379,
      "grad_norm": 0.5594484210014343,
      "learning_rate": 1.2042648724330078e-05,
      "loss": 0.3006,
      "step": 24065
    },
    {
      "epoch": 0.4346051899840386,
      "grad_norm": 0.4370863437652588,
      "learning_rate": 1.2039872199768451e-05,
      "loss": 0.2146,
      "step": 24070
    },
    {
      "epoch": 0.4346954694169393,
      "grad_norm": 0.44897207617759705,
      "learning_rate": 1.203709551111394e-05,
      "loss": 0.2844,
      "step": 24075
    },
    {
      "epoch": 0.43478574884984,
      "grad_norm": 0.5095186829566956,
      "learning_rate": 1.2034318658589902e-05,
      "loss": 0.3773,
      "step": 24080
    },
    {
      "epoch": 0.43487602828274075,
      "grad_norm": 0.45647507905960083,
      "learning_rate": 1.2031541642419723e-05,
      "loss": 0.3153,
      "step": 24085
    },
    {
      "epoch": 0.43496630771564143,
      "grad_norm": 0.4166284203529358,
      "learning_rate": 1.2028764462826789e-05,
      "loss": 0.1815,
      "step": 24090
    },
    {
      "epoch": 0.4350565871485422,
      "grad_norm": 0.2727082371711731,
      "learning_rate": 1.2025987120034502e-05,
      "loss": 0.2251,
      "step": 24095
    },
    {
      "epoch": 0.43514686658144286,
      "grad_norm": 0.2521149218082428,
      "learning_rate": 1.2023209614266282e-05,
      "loss": 0.239,
      "step": 24100
    },
    {
      "epoch": 0.4352371460143436,
      "grad_norm": 0.3187403380870819,
      "learning_rate": 1.2020431945745561e-05,
      "loss": 0.2084,
      "step": 24105
    },
    {
      "epoch": 0.4353274254472443,
      "grad_norm": 0.797163724899292,
      "learning_rate": 1.2017654114695779e-05,
      "loss": 0.2989,
      "step": 24110
    },
    {
      "epoch": 0.43541770488014503,
      "grad_norm": 1.0744043588638306,
      "learning_rate": 1.2014876121340389e-05,
      "loss": 0.2556,
      "step": 24115
    },
    {
      "epoch": 0.4355079843130457,
      "grad_norm": 0.7450100779533386,
      "learning_rate": 1.201209796590287e-05,
      "loss": 0.2484,
      "step": 24120
    },
    {
      "epoch": 0.43559826374594646,
      "grad_norm": 0.5865199565887451,
      "learning_rate": 1.2009319648606696e-05,
      "loss": 0.3165,
      "step": 24125
    },
    {
      "epoch": 0.43568854317884714,
      "grad_norm": 0.3864075243473053,
      "learning_rate": 1.2006541169675365e-05,
      "loss": 0.251,
      "step": 24130
    },
    {
      "epoch": 0.4357788226117479,
      "grad_norm": 0.5079840421676636,
      "learning_rate": 1.2003762529332388e-05,
      "loss": 0.198,
      "step": 24135
    },
    {
      "epoch": 0.43586910204464857,
      "grad_norm": 0.4628409445285797,
      "learning_rate": 1.2000983727801285e-05,
      "loss": 0.3116,
      "step": 24140
    },
    {
      "epoch": 0.4359593814775493,
      "grad_norm": 0.41381558775901794,
      "learning_rate": 1.1998204765305585e-05,
      "loss": 0.2505,
      "step": 24145
    },
    {
      "epoch": 0.43604966091045,
      "grad_norm": 1.1286911964416504,
      "learning_rate": 1.1995425642068845e-05,
      "loss": 0.2429,
      "step": 24150
    },
    {
      "epoch": 0.43613994034335074,
      "grad_norm": 0.700363278388977,
      "learning_rate": 1.1992646358314619e-05,
      "loss": 0.1841,
      "step": 24155
    },
    {
      "epoch": 0.4362302197762514,
      "grad_norm": 0.538574755191803,
      "learning_rate": 1.1989866914266481e-05,
      "loss": 0.2661,
      "step": 24160
    },
    {
      "epoch": 0.43632049920915217,
      "grad_norm": 0.3559228479862213,
      "learning_rate": 1.1987087310148016e-05,
      "loss": 0.2594,
      "step": 24165
    },
    {
      "epoch": 0.43641077864205285,
      "grad_norm": 0.5627803802490234,
      "learning_rate": 1.1984307546182827e-05,
      "loss": 0.2349,
      "step": 24170
    },
    {
      "epoch": 0.4365010580749536,
      "grad_norm": 0.5101236701011658,
      "learning_rate": 1.1981527622594525e-05,
      "loss": 0.2979,
      "step": 24175
    },
    {
      "epoch": 0.43659133750785434,
      "grad_norm": 0.5389246344566345,
      "learning_rate": 1.1978747539606728e-05,
      "loss": 0.2205,
      "step": 24180
    },
    {
      "epoch": 0.436681616940755,
      "grad_norm": 0.4691467583179474,
      "learning_rate": 1.197596729744308e-05,
      "loss": 0.2294,
      "step": 24185
    },
    {
      "epoch": 0.43677189637365577,
      "grad_norm": 0.3519728481769562,
      "learning_rate": 1.1973186896327232e-05,
      "loss": 0.255,
      "step": 24190
    },
    {
      "epoch": 0.43686217580655645,
      "grad_norm": 0.5833948850631714,
      "learning_rate": 1.197040633648284e-05,
      "loss": 0.3013,
      "step": 24195
    },
    {
      "epoch": 0.4369524552394572,
      "grad_norm": 0.7892794013023376,
      "learning_rate": 1.1967625618133586e-05,
      "loss": 0.3072,
      "step": 24200
    },
    {
      "epoch": 0.4370427346723579,
      "grad_norm": 2.0248167514801025,
      "learning_rate": 1.1964844741503159e-05,
      "loss": 0.2601,
      "step": 24205
    },
    {
      "epoch": 0.4371330141052586,
      "grad_norm": 0.3766515254974365,
      "learning_rate": 1.1962063706815256e-05,
      "loss": 0.2675,
      "step": 24210
    },
    {
      "epoch": 0.4372232935381593,
      "grad_norm": 0.4443298280239105,
      "learning_rate": 1.1959282514293596e-05,
      "loss": 0.3409,
      "step": 24215
    },
    {
      "epoch": 0.43731357297106005,
      "grad_norm": 0.40835708379745483,
      "learning_rate": 1.1956501164161903e-05,
      "loss": 0.1445,
      "step": 24220
    },
    {
      "epoch": 0.43740385240396074,
      "grad_norm": 0.5901013016700745,
      "learning_rate": 1.1953719656643914e-05,
      "loss": 0.2464,
      "step": 24225
    },
    {
      "epoch": 0.4374941318368615,
      "grad_norm": 0.5692929625511169,
      "learning_rate": 1.1950937991963385e-05,
      "loss": 0.2389,
      "step": 24230
    },
    {
      "epoch": 0.43758441126976216,
      "grad_norm": 0.4530095160007477,
      "learning_rate": 1.1948156170344079e-05,
      "loss": 0.2317,
      "step": 24235
    },
    {
      "epoch": 0.4376746907026629,
      "grad_norm": 0.46489042043685913,
      "learning_rate": 1.1945374192009773e-05,
      "loss": 0.2053,
      "step": 24240
    },
    {
      "epoch": 0.4377649701355636,
      "grad_norm": 0.4560989439487457,
      "learning_rate": 1.1942592057184256e-05,
      "loss": 0.2311,
      "step": 24245
    },
    {
      "epoch": 0.43785524956846433,
      "grad_norm": 0.6620902419090271,
      "learning_rate": 1.1939809766091334e-05,
      "loss": 0.2191,
      "step": 24250
    },
    {
      "epoch": 0.437945529001365,
      "grad_norm": 0.48353275656700134,
      "learning_rate": 1.1937027318954823e-05,
      "loss": 0.2172,
      "step": 24255
    },
    {
      "epoch": 0.43803580843426576,
      "grad_norm": 0.2793586254119873,
      "learning_rate": 1.1934244715998547e-05,
      "loss": 0.223,
      "step": 24260
    },
    {
      "epoch": 0.43812608786716645,
      "grad_norm": 0.4108912944793701,
      "learning_rate": 1.1931461957446342e-05,
      "loss": 0.3581,
      "step": 24265
    },
    {
      "epoch": 0.4382163673000672,
      "grad_norm": 0.2431604564189911,
      "learning_rate": 1.1928679043522071e-05,
      "loss": 0.1932,
      "step": 24270
    },
    {
      "epoch": 0.4383066467329679,
      "grad_norm": 0.4234777092933655,
      "learning_rate": 1.1925895974449588e-05,
      "loss": 0.2708,
      "step": 24275
    },
    {
      "epoch": 0.4383969261658686,
      "grad_norm": 0.7244442105293274,
      "learning_rate": 1.192311275045278e-05,
      "loss": 0.2523,
      "step": 24280
    },
    {
      "epoch": 0.4384872055987693,
      "grad_norm": 0.505651593208313,
      "learning_rate": 1.1920329371755535e-05,
      "loss": 0.2314,
      "step": 24285
    },
    {
      "epoch": 0.43857748503167004,
      "grad_norm": 0.4934924840927124,
      "learning_rate": 1.1917545838581754e-05,
      "loss": 0.2486,
      "step": 24290
    },
    {
      "epoch": 0.43866776446457073,
      "grad_norm": 0.39593183994293213,
      "learning_rate": 1.1914762151155348e-05,
      "loss": 0.2177,
      "step": 24295
    },
    {
      "epoch": 0.4387580438974715,
      "grad_norm": 0.700186014175415,
      "learning_rate": 1.1911978309700252e-05,
      "loss": 0.3491,
      "step": 24300
    },
    {
      "epoch": 0.43884832333037216,
      "grad_norm": 0.4597525894641876,
      "learning_rate": 1.1909194314440402e-05,
      "loss": 0.2484,
      "step": 24305
    },
    {
      "epoch": 0.4389386027632729,
      "grad_norm": 1.573953628540039,
      "learning_rate": 1.1906410165599748e-05,
      "loss": 0.2956,
      "step": 24310
    },
    {
      "epoch": 0.4390288821961736,
      "grad_norm": 0.4366190731525421,
      "learning_rate": 1.1903625863402261e-05,
      "loss": 0.1852,
      "step": 24315
    },
    {
      "epoch": 0.43911916162907433,
      "grad_norm": 0.2975316643714905,
      "learning_rate": 1.1900841408071912e-05,
      "loss": 0.3055,
      "step": 24320
    },
    {
      "epoch": 0.439209441061975,
      "grad_norm": 0.5020982027053833,
      "learning_rate": 1.1898056799832694e-05,
      "loss": 0.2907,
      "step": 24325
    },
    {
      "epoch": 0.43929972049487576,
      "grad_norm": 0.4672057032585144,
      "learning_rate": 1.1895272038908604e-05,
      "loss": 0.2599,
      "step": 24330
    },
    {
      "epoch": 0.43938999992777644,
      "grad_norm": 0.33109766244888306,
      "learning_rate": 1.189248712552366e-05,
      "loss": 0.2419,
      "step": 24335
    },
    {
      "epoch": 0.4394802793606772,
      "grad_norm": 0.2925727963447571,
      "learning_rate": 1.1889702059901887e-05,
      "loss": 0.1978,
      "step": 24340
    },
    {
      "epoch": 0.43957055879357787,
      "grad_norm": 0.3118140697479248,
      "learning_rate": 1.188691684226732e-05,
      "loss": 0.1405,
      "step": 24345
    },
    {
      "epoch": 0.4396608382264786,
      "grad_norm": 0.37156930565834045,
      "learning_rate": 1.1884131472844013e-05,
      "loss": 0.1736,
      "step": 24350
    },
    {
      "epoch": 0.4397511176593793,
      "grad_norm": 1.8306384086608887,
      "learning_rate": 1.1881345951856031e-05,
      "loss": 0.32,
      "step": 24355
    },
    {
      "epoch": 0.43984139709228004,
      "grad_norm": 0.772351086139679,
      "learning_rate": 1.1878560279527445e-05,
      "loss": 0.2236,
      "step": 24360
    },
    {
      "epoch": 0.4399316765251807,
      "grad_norm": 0.4637385904788971,
      "learning_rate": 1.1875774456082342e-05,
      "loss": 0.182,
      "step": 24365
    },
    {
      "epoch": 0.44002195595808147,
      "grad_norm": 0.4528012275695801,
      "learning_rate": 1.1872988481744822e-05,
      "loss": 0.2313,
      "step": 24370
    },
    {
      "epoch": 0.44011223539098215,
      "grad_norm": 0.42925751209259033,
      "learning_rate": 1.1870202356738996e-05,
      "loss": 0.3206,
      "step": 24375
    },
    {
      "epoch": 0.4402025148238829,
      "grad_norm": 0.611354649066925,
      "learning_rate": 1.186741608128899e-05,
      "loss": 0.2631,
      "step": 24380
    },
    {
      "epoch": 0.4402927942567836,
      "grad_norm": 0.2863406240940094,
      "learning_rate": 1.1864629655618938e-05,
      "loss": 0.2927,
      "step": 24385
    },
    {
      "epoch": 0.4403830736896843,
      "grad_norm": 0.5054121613502502,
      "learning_rate": 1.1861843079952986e-05,
      "loss": 0.2239,
      "step": 24390
    },
    {
      "epoch": 0.440473353122585,
      "grad_norm": 0.44482725858688354,
      "learning_rate": 1.1859056354515292e-05,
      "loss": 0.2104,
      "step": 24395
    },
    {
      "epoch": 0.44056363255548575,
      "grad_norm": 0.47482234239578247,
      "learning_rate": 1.1856269479530036e-05,
      "loss": 0.3047,
      "step": 24400
    },
    {
      "epoch": 0.44065391198838644,
      "grad_norm": 0.6331995725631714,
      "learning_rate": 1.1853482455221395e-05,
      "loss": 0.1987,
      "step": 24405
    },
    {
      "epoch": 0.4407441914212872,
      "grad_norm": 0.45549681782722473,
      "learning_rate": 1.1850695281813565e-05,
      "loss": 0.2732,
      "step": 24410
    },
    {
      "epoch": 0.44083447085418787,
      "grad_norm": 0.2735913097858429,
      "learning_rate": 1.1847907959530755e-05,
      "loss": 0.2056,
      "step": 24415
    },
    {
      "epoch": 0.4409247502870886,
      "grad_norm": 0.45871978998184204,
      "learning_rate": 1.1845120488597189e-05,
      "loss": 0.2659,
      "step": 24420
    },
    {
      "epoch": 0.4410150297199893,
      "grad_norm": 0.40871867537498474,
      "learning_rate": 1.1842332869237094e-05,
      "loss": 0.2257,
      "step": 24425
    },
    {
      "epoch": 0.44110530915289003,
      "grad_norm": 0.5864661931991577,
      "learning_rate": 1.1839545101674712e-05,
      "loss": 0.2137,
      "step": 24430
    },
    {
      "epoch": 0.4411955885857907,
      "grad_norm": 0.4320935904979706,
      "learning_rate": 1.1836757186134302e-05,
      "loss": 0.2865,
      "step": 24435
    },
    {
      "epoch": 0.44128586801869146,
      "grad_norm": 0.6176201105117798,
      "learning_rate": 1.1833969122840134e-05,
      "loss": 0.3079,
      "step": 24440
    },
    {
      "epoch": 0.44137614745159215,
      "grad_norm": 0.6133599877357483,
      "learning_rate": 1.183118091201648e-05,
      "loss": 0.336,
      "step": 24445
    },
    {
      "epoch": 0.4414664268844929,
      "grad_norm": 0.3526920974254608,
      "learning_rate": 1.182839255388764e-05,
      "loss": 0.255,
      "step": 24450
    },
    {
      "epoch": 0.4415567063173936,
      "grad_norm": 0.3885346055030823,
      "learning_rate": 1.1825604048677913e-05,
      "loss": 0.1731,
      "step": 24455
    },
    {
      "epoch": 0.4416469857502943,
      "grad_norm": 0.4212827682495117,
      "learning_rate": 1.182281539661161e-05,
      "loss": 0.2904,
      "step": 24460
    },
    {
      "epoch": 0.441737265183195,
      "grad_norm": 0.2621988356113434,
      "learning_rate": 1.1820026597913061e-05,
      "loss": 0.2868,
      "step": 24465
    },
    {
      "epoch": 0.44182754461609575,
      "grad_norm": 0.537579357624054,
      "learning_rate": 1.1817237652806607e-05,
      "loss": 0.2877,
      "step": 24470
    },
    {
      "epoch": 0.44191782404899643,
      "grad_norm": 0.4155954420566559,
      "learning_rate": 1.1814448561516595e-05,
      "loss": 0.2729,
      "step": 24475
    },
    {
      "epoch": 0.4420081034818972,
      "grad_norm": 0.6439969539642334,
      "learning_rate": 1.181165932426739e-05,
      "loss": 0.2617,
      "step": 24480
    },
    {
      "epoch": 0.44209838291479786,
      "grad_norm": 0.5035080313682556,
      "learning_rate": 1.1808869941283367e-05,
      "loss": 0.2067,
      "step": 24485
    },
    {
      "epoch": 0.4421886623476986,
      "grad_norm": 0.5220370888710022,
      "learning_rate": 1.1806080412788905e-05,
      "loss": 0.2631,
      "step": 24490
    },
    {
      "epoch": 0.4422789417805993,
      "grad_norm": 0.3431147336959839,
      "learning_rate": 1.1803290739008406e-05,
      "loss": 0.2998,
      "step": 24495
    },
    {
      "epoch": 0.44236922121350003,
      "grad_norm": 0.411819189786911,
      "learning_rate": 1.180050092016628e-05,
      "loss": 0.2614,
      "step": 24500
    },
    {
      "epoch": 0.4424595006464007,
      "grad_norm": 0.3154963552951813,
      "learning_rate": 1.1797710956486946e-05,
      "loss": 0.239,
      "step": 24505
    },
    {
      "epoch": 0.44254978007930146,
      "grad_norm": 0.39507150650024414,
      "learning_rate": 1.179492084819483e-05,
      "loss": 0.2851,
      "step": 24510
    },
    {
      "epoch": 0.44264005951220214,
      "grad_norm": 0.47061803936958313,
      "learning_rate": 1.1792130595514388e-05,
      "loss": 0.2528,
      "step": 24515
    },
    {
      "epoch": 0.4427303389451029,
      "grad_norm": 0.5809831619262695,
      "learning_rate": 1.178934019867007e-05,
      "loss": 0.2567,
      "step": 24520
    },
    {
      "epoch": 0.44282061837800357,
      "grad_norm": 0.34608837962150574,
      "learning_rate": 1.1786549657886342e-05,
      "loss": 0.2107,
      "step": 24525
    },
    {
      "epoch": 0.4429108978109043,
      "grad_norm": 0.39510759711265564,
      "learning_rate": 1.1783758973387682e-05,
      "loss": 0.3024,
      "step": 24530
    },
    {
      "epoch": 0.443001177243805,
      "grad_norm": 0.4356182813644409,
      "learning_rate": 1.1780968145398581e-05,
      "loss": 0.2404,
      "step": 24535
    },
    {
      "epoch": 0.44309145667670574,
      "grad_norm": 0.5873695611953735,
      "learning_rate": 1.1778177174143546e-05,
      "loss": 0.2397,
      "step": 24540
    },
    {
      "epoch": 0.44318173610960643,
      "grad_norm": 2.7690255641937256,
      "learning_rate": 1.1775386059847083e-05,
      "loss": 0.3104,
      "step": 24545
    },
    {
      "epoch": 0.44327201554250717,
      "grad_norm": 0.33572685718536377,
      "learning_rate": 1.177259480273372e-05,
      "loss": 0.2598,
      "step": 24550
    },
    {
      "epoch": 0.44336229497540786,
      "grad_norm": 0.37144502997398376,
      "learning_rate": 1.1769803403027992e-05,
      "loss": 0.2386,
      "step": 24555
    },
    {
      "epoch": 0.4434525744083086,
      "grad_norm": 0.2875464856624603,
      "learning_rate": 1.176701186095445e-05,
      "loss": 0.2305,
      "step": 24560
    },
    {
      "epoch": 0.4435428538412093,
      "grad_norm": 0.4528631567955017,
      "learning_rate": 1.1764220176737649e-05,
      "loss": 0.2791,
      "step": 24565
    },
    {
      "epoch": 0.44363313327411,
      "grad_norm": 0.6390991806983948,
      "learning_rate": 1.1761428350602164e-05,
      "loss": 0.381,
      "step": 24570
    },
    {
      "epoch": 0.44372341270701077,
      "grad_norm": 0.38894256949424744,
      "learning_rate": 1.1758636382772572e-05,
      "loss": 0.2269,
      "step": 24575
    },
    {
      "epoch": 0.44381369213991145,
      "grad_norm": 0.3657304048538208,
      "learning_rate": 1.1755844273473473e-05,
      "loss": 0.275,
      "step": 24580
    },
    {
      "epoch": 0.4439039715728122,
      "grad_norm": 0.4715575873851776,
      "learning_rate": 1.1753052022929468e-05,
      "loss": 0.2281,
      "step": 24585
    },
    {
      "epoch": 0.4439942510057129,
      "grad_norm": 0.322095662355423,
      "learning_rate": 1.1750259631365173e-05,
      "loss": 0.2573,
      "step": 24590
    },
    {
      "epoch": 0.4440845304386136,
      "grad_norm": 0.304795503616333,
      "learning_rate": 1.1747467099005214e-05,
      "loss": 0.3018,
      "step": 24595
    },
    {
      "epoch": 0.4441748098715143,
      "grad_norm": 0.6157793998718262,
      "learning_rate": 1.1744674426074234e-05,
      "loss": 0.2722,
      "step": 24600
    },
    {
      "epoch": 0.44426508930441505,
      "grad_norm": 0.53531813621521,
      "learning_rate": 1.1741881612796883e-05,
      "loss": 0.2626,
      "step": 24605
    },
    {
      "epoch": 0.44435536873731574,
      "grad_norm": 0.42279717326164246,
      "learning_rate": 1.1739088659397817e-05,
      "loss": 0.2222,
      "step": 24610
    },
    {
      "epoch": 0.4444456481702165,
      "grad_norm": 0.4403485655784607,
      "learning_rate": 1.1736295566101717e-05,
      "loss": 0.3001,
      "step": 24615
    },
    {
      "epoch": 0.44453592760311716,
      "grad_norm": 0.3616626262664795,
      "learning_rate": 1.1733502333133261e-05,
      "loss": 0.2753,
      "step": 24620
    },
    {
      "epoch": 0.4446262070360179,
      "grad_norm": 0.6336793303489685,
      "learning_rate": 1.1730708960717147e-05,
      "loss": 0.264,
      "step": 24625
    },
    {
      "epoch": 0.4447164864689186,
      "grad_norm": 0.38618341088294983,
      "learning_rate": 1.1727915449078077e-05,
      "loss": 0.2094,
      "step": 24630
    },
    {
      "epoch": 0.44480676590181933,
      "grad_norm": 0.45772358775138855,
      "learning_rate": 1.172512179844078e-05,
      "loss": 0.2328,
      "step": 24635
    },
    {
      "epoch": 0.44489704533472,
      "grad_norm": 0.34782519936561584,
      "learning_rate": 1.1722328009029969e-05,
      "loss": 0.171,
      "step": 24640
    },
    {
      "epoch": 0.44498732476762076,
      "grad_norm": 0.37528547644615173,
      "learning_rate": 1.1719534081070396e-05,
      "loss": 0.2841,
      "step": 24645
    },
    {
      "epoch": 0.44507760420052145,
      "grad_norm": 0.3782618045806885,
      "learning_rate": 1.1716740014786809e-05,
      "loss": 0.2331,
      "step": 24650
    },
    {
      "epoch": 0.4451678836334222,
      "grad_norm": 0.5546488761901855,
      "learning_rate": 1.1713945810403968e-05,
      "loss": 0.2844,
      "step": 24655
    },
    {
      "epoch": 0.4452581630663229,
      "grad_norm": 0.375745564699173,
      "learning_rate": 1.1711151468146648e-05,
      "loss": 0.2949,
      "step": 24660
    },
    {
      "epoch": 0.4453484424992236,
      "grad_norm": 0.24210117757320404,
      "learning_rate": 1.1708356988239636e-05,
      "loss": 0.1624,
      "step": 24665
    },
    {
      "epoch": 0.4454387219321243,
      "grad_norm": 0.30786609649658203,
      "learning_rate": 1.1705562370907724e-05,
      "loss": 0.2435,
      "step": 24670
    },
    {
      "epoch": 0.44552900136502505,
      "grad_norm": 0.3296472132205963,
      "learning_rate": 1.1702767616375717e-05,
      "loss": 0.2309,
      "step": 24675
    },
    {
      "epoch": 0.44561928079792573,
      "grad_norm": 0.29616475105285645,
      "learning_rate": 1.1699972724868438e-05,
      "loss": 0.246,
      "step": 24680
    },
    {
      "epoch": 0.4457095602308265,
      "grad_norm": 0.5503436923027039,
      "learning_rate": 1.1697177696610713e-05,
      "loss": 0.2321,
      "step": 24685
    },
    {
      "epoch": 0.44579983966372716,
      "grad_norm": 0.4225671589374542,
      "learning_rate": 1.1694382531827385e-05,
      "loss": 0.1839,
      "step": 24690
    },
    {
      "epoch": 0.4458901190966279,
      "grad_norm": 0.5932523608207703,
      "learning_rate": 1.1691587230743297e-05,
      "loss": 0.2267,
      "step": 24695
    },
    {
      "epoch": 0.4459803985295286,
      "grad_norm": 0.4628986716270447,
      "learning_rate": 1.1688791793583318e-05,
      "loss": 0.2749,
      "step": 24700
    },
    {
      "epoch": 0.44607067796242933,
      "grad_norm": 0.4341941177845001,
      "learning_rate": 1.1685996220572318e-05,
      "loss": 0.2234,
      "step": 24705
    },
    {
      "epoch": 0.44616095739533,
      "grad_norm": 0.3851144015789032,
      "learning_rate": 1.1683200511935177e-05,
      "loss": 0.2791,
      "step": 24710
    },
    {
      "epoch": 0.44625123682823076,
      "grad_norm": 0.8717007637023926,
      "learning_rate": 1.1680404667896797e-05,
      "loss": 0.3386,
      "step": 24715
    },
    {
      "epoch": 0.44634151626113144,
      "grad_norm": 0.46784061193466187,
      "learning_rate": 1.1677608688682077e-05,
      "loss": 0.2856,
      "step": 24720
    },
    {
      "epoch": 0.4464317956940322,
      "grad_norm": 0.22613851726055145,
      "learning_rate": 1.1674812574515938e-05,
      "loss": 0.2416,
      "step": 24725
    },
    {
      "epoch": 0.44652207512693287,
      "grad_norm": 0.30826452374458313,
      "learning_rate": 1.16720163256233e-05,
      "loss": 0.2006,
      "step": 24730
    },
    {
      "epoch": 0.4466123545598336,
      "grad_norm": 0.4111037254333496,
      "learning_rate": 1.166921994222911e-05,
      "loss": 0.307,
      "step": 24735
    },
    {
      "epoch": 0.4467026339927343,
      "grad_norm": 0.3378446698188782,
      "learning_rate": 1.1666423424558309e-05,
      "loss": 0.1922,
      "step": 24740
    },
    {
      "epoch": 0.44679291342563504,
      "grad_norm": 0.5012000203132629,
      "learning_rate": 1.166362677283586e-05,
      "loss": 0.2858,
      "step": 24745
    },
    {
      "epoch": 0.4468831928585357,
      "grad_norm": 0.5589067935943604,
      "learning_rate": 1.1660829987286734e-05,
      "loss": 0.2197,
      "step": 24750
    },
    {
      "epoch": 0.44697347229143647,
      "grad_norm": 0.7075532078742981,
      "learning_rate": 1.1658033068135912e-05,
      "loss": 0.4324,
      "step": 24755
    },
    {
      "epoch": 0.44706375172433716,
      "grad_norm": 0.5003488659858704,
      "learning_rate": 1.165523601560838e-05,
      "loss": 0.281,
      "step": 24760
    },
    {
      "epoch": 0.4471540311572379,
      "grad_norm": 0.3863239288330078,
      "learning_rate": 1.1652438829929149e-05,
      "loss": 0.2743,
      "step": 24765
    },
    {
      "epoch": 0.4472443105901386,
      "grad_norm": 0.1344427764415741,
      "learning_rate": 1.164964151132323e-05,
      "loss": 0.2354,
      "step": 24770
    },
    {
      "epoch": 0.4473345900230393,
      "grad_norm": 0.6510555744171143,
      "learning_rate": 1.164684406001564e-05,
      "loss": 0.2332,
      "step": 24775
    },
    {
      "epoch": 0.44742486945594,
      "grad_norm": 0.45478320121765137,
      "learning_rate": 1.1644046476231423e-05,
      "loss": 0.2003,
      "step": 24780
    },
    {
      "epoch": 0.44751514888884075,
      "grad_norm": 0.5587338805198669,
      "learning_rate": 1.164124876019562e-05,
      "loss": 0.2565,
      "step": 24785
    },
    {
      "epoch": 0.44760542832174144,
      "grad_norm": 0.4031519591808319,
      "learning_rate": 1.1638450912133288e-05,
      "loss": 0.2197,
      "step": 24790
    },
    {
      "epoch": 0.4476957077546422,
      "grad_norm": 0.5361907482147217,
      "learning_rate": 1.163565293226949e-05,
      "loss": 0.2296,
      "step": 24795
    },
    {
      "epoch": 0.44778598718754287,
      "grad_norm": 0.41891780495643616,
      "learning_rate": 1.163285482082931e-05,
      "loss": 0.2556,
      "step": 24800
    },
    {
      "epoch": 0.4478762666204436,
      "grad_norm": 0.4059182107448578,
      "learning_rate": 1.163005657803783e-05,
      "loss": 0.3062,
      "step": 24805
    },
    {
      "epoch": 0.4479665460533443,
      "grad_norm": 0.6129568219184875,
      "learning_rate": 1.1627258204120146e-05,
      "loss": 0.3074,
      "step": 24810
    },
    {
      "epoch": 0.44805682548624504,
      "grad_norm": 0.7538573145866394,
      "learning_rate": 1.1624459699301375e-05,
      "loss": 0.1785,
      "step": 24815
    },
    {
      "epoch": 0.4481471049191457,
      "grad_norm": 0.40244433283805847,
      "learning_rate": 1.1621661063806635e-05,
      "loss": 0.2526,
      "step": 24820
    },
    {
      "epoch": 0.44823738435204646,
      "grad_norm": 0.6283896565437317,
      "learning_rate": 1.1618862297861046e-05,
      "loss": 0.2274,
      "step": 24825
    },
    {
      "epoch": 0.44832766378494715,
      "grad_norm": 0.34380289912223816,
      "learning_rate": 1.161606340168976e-05,
      "loss": 0.2386,
      "step": 24830
    },
    {
      "epoch": 0.4484179432178479,
      "grad_norm": 0.5499282479286194,
      "learning_rate": 1.1613264375517923e-05,
      "loss": 0.2102,
      "step": 24835
    },
    {
      "epoch": 0.4485082226507486,
      "grad_norm": 0.23248335719108582,
      "learning_rate": 1.1610465219570695e-05,
      "loss": 0.2723,
      "step": 24840
    },
    {
      "epoch": 0.4485985020836493,
      "grad_norm": 0.347027063369751,
      "learning_rate": 1.1607665934073248e-05,
      "loss": 0.2351,
      "step": 24845
    },
    {
      "epoch": 0.44868878151655,
      "grad_norm": 0.5538097620010376,
      "learning_rate": 1.1604866519250768e-05,
      "loss": 0.3567,
      "step": 24850
    },
    {
      "epoch": 0.44877906094945075,
      "grad_norm": 0.292410284280777,
      "learning_rate": 1.1602066975328446e-05,
      "loss": 0.2196,
      "step": 24855
    },
    {
      "epoch": 0.44886934038235143,
      "grad_norm": 0.7871230840682983,
      "learning_rate": 1.159926730253148e-05,
      "loss": 0.2483,
      "step": 24860
    },
    {
      "epoch": 0.4489596198152522,
      "grad_norm": 0.3869844675064087,
      "learning_rate": 1.159646750108509e-05,
      "loss": 0.2416,
      "step": 24865
    },
    {
      "epoch": 0.44904989924815286,
      "grad_norm": 0.5626687407493591,
      "learning_rate": 1.1593667571214498e-05,
      "loss": 0.2799,
      "step": 24870
    },
    {
      "epoch": 0.4491401786810536,
      "grad_norm": 0.6964197158813477,
      "learning_rate": 1.1590867513144934e-05,
      "loss": 0.1481,
      "step": 24875
    },
    {
      "epoch": 0.4492304581139543,
      "grad_norm": 0.3524242341518402,
      "learning_rate": 1.1588067327101649e-05,
      "loss": 0.2354,
      "step": 24880
    },
    {
      "epoch": 0.44932073754685503,
      "grad_norm": 0.5947927832603455,
      "learning_rate": 1.1585267013309892e-05,
      "loss": 0.2916,
      "step": 24885
    },
    {
      "epoch": 0.4494110169797557,
      "grad_norm": 0.5788018107414246,
      "learning_rate": 1.1582466571994925e-05,
      "loss": 0.2421,
      "step": 24890
    },
    {
      "epoch": 0.44950129641265646,
      "grad_norm": 0.662961483001709,
      "learning_rate": 1.1579666003382035e-05,
      "loss": 0.258,
      "step": 24895
    },
    {
      "epoch": 0.44959157584555715,
      "grad_norm": 0.44949132204055786,
      "learning_rate": 1.1576865307696499e-05,
      "loss": 0.2572,
      "step": 24900
    },
    {
      "epoch": 0.4496818552784579,
      "grad_norm": 1.094267725944519,
      "learning_rate": 1.1574064485163613e-05,
      "loss": 0.3229,
      "step": 24905
    },
    {
      "epoch": 0.4497721347113586,
      "grad_norm": 0.39188432693481445,
      "learning_rate": 1.157126353600868e-05,
      "loss": 0.2184,
      "step": 24910
    },
    {
      "epoch": 0.4498624141442593,
      "grad_norm": 0.34874260425567627,
      "learning_rate": 1.1568462460457028e-05,
      "loss": 0.2069,
      "step": 24915
    },
    {
      "epoch": 0.44995269357716,
      "grad_norm": 0.6210698485374451,
      "learning_rate": 1.156566125873397e-05,
      "loss": 0.3669,
      "step": 24920
    },
    {
      "epoch": 0.45004297301006074,
      "grad_norm": 0.4312707781791687,
      "learning_rate": 1.1562859931064845e-05,
      "loss": 0.1418,
      "step": 24925
    },
    {
      "epoch": 0.45013325244296143,
      "grad_norm": 0.3792107403278351,
      "learning_rate": 1.1560058477675004e-05,
      "loss": 0.3266,
      "step": 24930
    },
    {
      "epoch": 0.45022353187586217,
      "grad_norm": 0.37607336044311523,
      "learning_rate": 1.1557256898789801e-05,
      "loss": 0.2038,
      "step": 24935
    },
    {
      "epoch": 0.45031381130876286,
      "grad_norm": 0.49079403281211853,
      "learning_rate": 1.1554455194634601e-05,
      "loss": 0.2078,
      "step": 24940
    },
    {
      "epoch": 0.4504040907416636,
      "grad_norm": 0.3289194703102112,
      "learning_rate": 1.1551653365434787e-05,
      "loss": 0.2301,
      "step": 24945
    },
    {
      "epoch": 0.4504943701745643,
      "grad_norm": 0.4723817706108093,
      "learning_rate": 1.1548851411415739e-05,
      "loss": 0.1823,
      "step": 24950
    },
    {
      "epoch": 0.450584649607465,
      "grad_norm": 0.3931238055229187,
      "learning_rate": 1.1546049332802855e-05,
      "loss": 0.2415,
      "step": 24955
    },
    {
      "epoch": 0.4506749290403657,
      "grad_norm": 0.2613414525985718,
      "learning_rate": 1.1543247129821545e-05,
      "loss": 0.2462,
      "step": 24960
    },
    {
      "epoch": 0.45076520847326645,
      "grad_norm": 0.5624940991401672,
      "learning_rate": 1.1540444802697221e-05,
      "loss": 0.3012,
      "step": 24965
    },
    {
      "epoch": 0.4508554879061672,
      "grad_norm": 0.35730329155921936,
      "learning_rate": 1.1537642351655316e-05,
      "loss": 0.2074,
      "step": 24970
    },
    {
      "epoch": 0.4509457673390679,
      "grad_norm": 0.5674639940261841,
      "learning_rate": 1.153483977692126e-05,
      "loss": 0.3158,
      "step": 24975
    },
    {
      "epoch": 0.4510360467719686,
      "grad_norm": 0.676696240901947,
      "learning_rate": 1.1532037078720508e-05,
      "loss": 0.1583,
      "step": 24980
    },
    {
      "epoch": 0.4511263262048693,
      "grad_norm": 0.5902599096298218,
      "learning_rate": 1.152923425727851e-05,
      "loss": 0.2561,
      "step": 24985
    },
    {
      "epoch": 0.45121660563777005,
      "grad_norm": 0.6040018200874329,
      "learning_rate": 1.1526431312820731e-05,
      "loss": 0.2519,
      "step": 24990
    },
    {
      "epoch": 0.45130688507067074,
      "grad_norm": 0.3789654076099396,
      "learning_rate": 1.1523628245572655e-05,
      "loss": 0.2538,
      "step": 24995
    },
    {
      "epoch": 0.4513971645035715,
      "grad_norm": 0.5607237219810486,
      "learning_rate": 1.1520825055759765e-05,
      "loss": 0.4231,
      "step": 25000
    },
    {
      "epoch": 0.45148744393647217,
      "grad_norm": 0.45001712441444397,
      "learning_rate": 1.1518021743607558e-05,
      "loss": 0.2767,
      "step": 25005
    },
    {
      "epoch": 0.4515777233693729,
      "grad_norm": 0.3102862536907196,
      "learning_rate": 1.1515218309341535e-05,
      "loss": 0.2226,
      "step": 25010
    },
    {
      "epoch": 0.4516680028022736,
      "grad_norm": 0.5020765662193298,
      "learning_rate": 1.1512414753187218e-05,
      "loss": 0.2048,
      "step": 25015
    },
    {
      "epoch": 0.45175828223517434,
      "grad_norm": 0.4714259207248688,
      "learning_rate": 1.1509611075370131e-05,
      "loss": 0.2712,
      "step": 25020
    },
    {
      "epoch": 0.451848561668075,
      "grad_norm": 0.9056620001792908,
      "learning_rate": 1.1506807276115806e-05,
      "loss": 0.3684,
      "step": 25025
    },
    {
      "epoch": 0.45193884110097576,
      "grad_norm": 0.6726963520050049,
      "learning_rate": 1.1504003355649794e-05,
      "loss": 0.2887,
      "step": 25030
    },
    {
      "epoch": 0.45202912053387645,
      "grad_norm": 0.5326406359672546,
      "learning_rate": 1.1501199314197644e-05,
      "loss": 0.2195,
      "step": 25035
    },
    {
      "epoch": 0.4521193999667772,
      "grad_norm": 0.5724204182624817,
      "learning_rate": 1.1498395151984925e-05,
      "loss": 0.3536,
      "step": 25040
    },
    {
      "epoch": 0.4522096793996779,
      "grad_norm": 0.4841223657131195,
      "learning_rate": 1.1495590869237213e-05,
      "loss": 0.1531,
      "step": 25045
    },
    {
      "epoch": 0.4522999588325786,
      "grad_norm": 0.3502753674983978,
      "learning_rate": 1.1492786466180086e-05,
      "loss": 0.2277,
      "step": 25050
    },
    {
      "epoch": 0.4523902382654793,
      "grad_norm": 0.4281364381313324,
      "learning_rate": 1.1489981943039141e-05,
      "loss": 0.2408,
      "step": 25055
    },
    {
      "epoch": 0.45248051769838005,
      "grad_norm": 0.5470902323722839,
      "learning_rate": 1.1487177300039984e-05,
      "loss": 0.2491,
      "step": 25060
    },
    {
      "epoch": 0.45257079713128073,
      "grad_norm": 0.4884530305862427,
      "learning_rate": 1.1484372537408229e-05,
      "loss": 0.1792,
      "step": 25065
    },
    {
      "epoch": 0.4526610765641815,
      "grad_norm": 0.481823205947876,
      "learning_rate": 1.1481567655369491e-05,
      "loss": 0.1847,
      "step": 25070
    },
    {
      "epoch": 0.45275135599708216,
      "grad_norm": 0.41728538274765015,
      "learning_rate": 1.1478762654149409e-05,
      "loss": 0.2356,
      "step": 25075
    },
    {
      "epoch": 0.4528416354299829,
      "grad_norm": 0.5025848746299744,
      "learning_rate": 1.1475957533973623e-05,
      "loss": 0.2256,
      "step": 25080
    },
    {
      "epoch": 0.4529319148628836,
      "grad_norm": 0.5028762817382812,
      "learning_rate": 1.1473152295067787e-05,
      "loss": 0.2719,
      "step": 25085
    },
    {
      "epoch": 0.45302219429578433,
      "grad_norm": 0.5323634147644043,
      "learning_rate": 1.1470346937657555e-05,
      "loss": 0.2936,
      "step": 25090
    },
    {
      "epoch": 0.453112473728685,
      "grad_norm": 0.4943752586841583,
      "learning_rate": 1.1467541461968607e-05,
      "loss": 0.2228,
      "step": 25095
    },
    {
      "epoch": 0.45320275316158576,
      "grad_norm": 0.6112527251243591,
      "learning_rate": 1.1464735868226618e-05,
      "loss": 0.2062,
      "step": 25100
    },
    {
      "epoch": 0.45329303259448644,
      "grad_norm": 0.4090018570423126,
      "learning_rate": 1.1461930156657273e-05,
      "loss": 0.3034,
      "step": 25105
    },
    {
      "epoch": 0.4533833120273872,
      "grad_norm": 0.6104006767272949,
      "learning_rate": 1.1459124327486279e-05,
      "loss": 0.2675,
      "step": 25110
    },
    {
      "epoch": 0.4534735914602879,
      "grad_norm": 0.33331215381622314,
      "learning_rate": 1.1456318380939344e-05,
      "loss": 0.2653,
      "step": 25115
    },
    {
      "epoch": 0.4535638708931886,
      "grad_norm": 0.38586562871932983,
      "learning_rate": 1.145351231724218e-05,
      "loss": 0.274,
      "step": 25120
    },
    {
      "epoch": 0.4536541503260893,
      "grad_norm": 0.41210177540779114,
      "learning_rate": 1.1450706136620523e-05,
      "loss": 0.2445,
      "step": 25125
    },
    {
      "epoch": 0.45374442975899004,
      "grad_norm": 0.35308703780174255,
      "learning_rate": 1.14478998393001e-05,
      "loss": 0.1937,
      "step": 25130
    },
    {
      "epoch": 0.45383470919189073,
      "grad_norm": 0.42387494444847107,
      "learning_rate": 1.1445093425506666e-05,
      "loss": 0.3267,
      "step": 25135
    },
    {
      "epoch": 0.45392498862479147,
      "grad_norm": 0.4690084755420685,
      "learning_rate": 1.1442286895465968e-05,
      "loss": 0.2214,
      "step": 25140
    },
    {
      "epoch": 0.45401526805769216,
      "grad_norm": 0.6119925379753113,
      "learning_rate": 1.143948024940378e-05,
      "loss": 0.2009,
      "step": 25145
    },
    {
      "epoch": 0.4541055474905929,
      "grad_norm": 0.2987114489078522,
      "learning_rate": 1.143667348754587e-05,
      "loss": 0.1875,
      "step": 25150
    },
    {
      "epoch": 0.4541958269234936,
      "grad_norm": 0.6101985573768616,
      "learning_rate": 1.1433866610118023e-05,
      "loss": 0.2181,
      "step": 25155
    },
    {
      "epoch": 0.4542861063563943,
      "grad_norm": 0.7137722373008728,
      "learning_rate": 1.1431059617346034e-05,
      "loss": 0.2883,
      "step": 25160
    },
    {
      "epoch": 0.454376385789295,
      "grad_norm": 0.2992022931575775,
      "learning_rate": 1.1428252509455704e-05,
      "loss": 0.2287,
      "step": 25165
    },
    {
      "epoch": 0.45446666522219575,
      "grad_norm": 0.3725275993347168,
      "learning_rate": 1.1425445286672843e-05,
      "loss": 0.2557,
      "step": 25170
    },
    {
      "epoch": 0.45455694465509644,
      "grad_norm": 0.534310519695282,
      "learning_rate": 1.1422637949223272e-05,
      "loss": 0.1826,
      "step": 25175
    },
    {
      "epoch": 0.4546472240879972,
      "grad_norm": 0.31912413239479065,
      "learning_rate": 1.1419830497332822e-05,
      "loss": 0.2428,
      "step": 25180
    },
    {
      "epoch": 0.45473750352089787,
      "grad_norm": 0.5752959251403809,
      "learning_rate": 1.1417022931227332e-05,
      "loss": 0.2639,
      "step": 25185
    },
    {
      "epoch": 0.4548277829537986,
      "grad_norm": 0.352507084608078,
      "learning_rate": 1.1414215251132652e-05,
      "loss": 0.3151,
      "step": 25190
    },
    {
      "epoch": 0.4549180623866993,
      "grad_norm": 0.6433412432670593,
      "learning_rate": 1.1411407457274634e-05,
      "loss": 0.2357,
      "step": 25195
    },
    {
      "epoch": 0.45500834181960004,
      "grad_norm": 0.5227245688438416,
      "learning_rate": 1.1408599549879151e-05,
      "loss": 0.2158,
      "step": 25200
    },
    {
      "epoch": 0.4550986212525007,
      "grad_norm": 0.8377803564071655,
      "learning_rate": 1.1405791529172072e-05,
      "loss": 0.2124,
      "step": 25205
    },
    {
      "epoch": 0.45518890068540147,
      "grad_norm": 0.5074353218078613,
      "learning_rate": 1.140298339537929e-05,
      "loss": 0.1699,
      "step": 25210
    },
    {
      "epoch": 0.45527918011830215,
      "grad_norm": 0.3799591064453125,
      "learning_rate": 1.1400175148726693e-05,
      "loss": 0.3373,
      "step": 25215
    },
    {
      "epoch": 0.4553694595512029,
      "grad_norm": 0.6017751097679138,
      "learning_rate": 1.1397366789440183e-05,
      "loss": 0.2376,
      "step": 25220
    },
    {
      "epoch": 0.4554597389841036,
      "grad_norm": 0.5219336748123169,
      "learning_rate": 1.1394558317745677e-05,
      "loss": 0.2721,
      "step": 25225
    },
    {
      "epoch": 0.4555500184170043,
      "grad_norm": 0.578639030456543,
      "learning_rate": 1.1391749733869094e-05,
      "loss": 0.2706,
      "step": 25230
    },
    {
      "epoch": 0.455640297849905,
      "grad_norm": 0.4614681601524353,
      "learning_rate": 1.1388941038036363e-05,
      "loss": 0.2798,
      "step": 25235
    },
    {
      "epoch": 0.45573057728280575,
      "grad_norm": 0.2860408425331116,
      "learning_rate": 1.1386132230473421e-05,
      "loss": 0.2544,
      "step": 25240
    },
    {
      "epoch": 0.45582085671570644,
      "grad_norm": 0.3257483243942261,
      "learning_rate": 1.1383323311406223e-05,
      "loss": 0.2625,
      "step": 25245
    },
    {
      "epoch": 0.4559111361486072,
      "grad_norm": 0.29570409655570984,
      "learning_rate": 1.1380514281060723e-05,
      "loss": 0.2186,
      "step": 25250
    },
    {
      "epoch": 0.45600141558150786,
      "grad_norm": 0.5505267381668091,
      "learning_rate": 1.1377705139662882e-05,
      "loss": 0.2637,
      "step": 25255
    },
    {
      "epoch": 0.4560916950144086,
      "grad_norm": 0.4645271897315979,
      "learning_rate": 1.1374895887438684e-05,
      "loss": 0.2102,
      "step": 25260
    },
    {
      "epoch": 0.4561819744473093,
      "grad_norm": 0.4300084710121155,
      "learning_rate": 1.1372086524614107e-05,
      "loss": 0.2516,
      "step": 25265
    },
    {
      "epoch": 0.45627225388021003,
      "grad_norm": 0.46449944376945496,
      "learning_rate": 1.1369277051415146e-05,
      "loss": 0.2904,
      "step": 25270
    },
    {
      "epoch": 0.4563625333131107,
      "grad_norm": 0.4134475588798523,
      "learning_rate": 1.13664674680678e-05,
      "loss": 0.2882,
      "step": 25275
    },
    {
      "epoch": 0.45645281274601146,
      "grad_norm": 0.4197487533092499,
      "learning_rate": 1.1363657774798084e-05,
      "loss": 0.1875,
      "step": 25280
    },
    {
      "epoch": 0.45654309217891215,
      "grad_norm": 0.34673747420310974,
      "learning_rate": 1.1360847971832012e-05,
      "loss": 0.2741,
      "step": 25285
    },
    {
      "epoch": 0.4566333716118129,
      "grad_norm": 0.35914045572280884,
      "learning_rate": 1.1358038059395615e-05,
      "loss": 0.2237,
      "step": 25290
    },
    {
      "epoch": 0.4567236510447136,
      "grad_norm": 0.9550446271896362,
      "learning_rate": 1.1355228037714932e-05,
      "loss": 0.2301,
      "step": 25295
    },
    {
      "epoch": 0.4568139304776143,
      "grad_norm": 0.716317892074585,
      "learning_rate": 1.1352417907016007e-05,
      "loss": 0.2322,
      "step": 25300
    },
    {
      "epoch": 0.456904209910515,
      "grad_norm": 0.45155155658721924,
      "learning_rate": 1.1349607667524892e-05,
      "loss": 0.2183,
      "step": 25305
    },
    {
      "epoch": 0.45699448934341574,
      "grad_norm": 0.392500102519989,
      "learning_rate": 1.1346797319467652e-05,
      "loss": 0.3804,
      "step": 25310
    },
    {
      "epoch": 0.45708476877631643,
      "grad_norm": 0.6335352063179016,
      "learning_rate": 1.1343986863070363e-05,
      "loss": 0.2859,
      "step": 25315
    },
    {
      "epoch": 0.45717504820921717,
      "grad_norm": 0.3605762720108032,
      "learning_rate": 1.13411762985591e-05,
      "loss": 0.2202,
      "step": 25320
    },
    {
      "epoch": 0.45726532764211786,
      "grad_norm": 0.37601563334465027,
      "learning_rate": 1.1338365626159957e-05,
      "loss": 0.2028,
      "step": 25325
    },
    {
      "epoch": 0.4573556070750186,
      "grad_norm": 0.3913663327693939,
      "learning_rate": 1.1335554846099029e-05,
      "loss": 0.2659,
      "step": 25330
    },
    {
      "epoch": 0.4574458865079193,
      "grad_norm": 0.3613426983356476,
      "learning_rate": 1.1332743958602426e-05,
      "loss": 0.2835,
      "step": 25335
    },
    {
      "epoch": 0.45753616594082,
      "grad_norm": 0.40373799204826355,
      "learning_rate": 1.1329932963896257e-05,
      "loss": 0.2213,
      "step": 25340
    },
    {
      "epoch": 0.4576264453737207,
      "grad_norm": 0.5222960114479065,
      "learning_rate": 1.1327121862206656e-05,
      "loss": 0.1844,
      "step": 25345
    },
    {
      "epoch": 0.45771672480662146,
      "grad_norm": 0.48120495676994324,
      "learning_rate": 1.1324310653759748e-05,
      "loss": 0.2273,
      "step": 25350
    },
    {
      "epoch": 0.45780700423952214,
      "grad_norm": 0.5859454274177551,
      "learning_rate": 1.1321499338781675e-05,
      "loss": 0.2183,
      "step": 25355
    },
    {
      "epoch": 0.4578972836724229,
      "grad_norm": 0.4917299449443817,
      "learning_rate": 1.131868791749859e-05,
      "loss": 0.2939,
      "step": 25360
    },
    {
      "epoch": 0.4579875631053236,
      "grad_norm": 0.6798126697540283,
      "learning_rate": 1.131587639013665e-05,
      "loss": 0.3164,
      "step": 25365
    },
    {
      "epoch": 0.4580778425382243,
      "grad_norm": 0.6465598940849304,
      "learning_rate": 1.1313064756922023e-05,
      "loss": 0.2533,
      "step": 25370
    },
    {
      "epoch": 0.45816812197112505,
      "grad_norm": 0.25099602341651917,
      "learning_rate": 1.1310253018080881e-05,
      "loss": 0.2755,
      "step": 25375
    },
    {
      "epoch": 0.45825840140402574,
      "grad_norm": 0.823212742805481,
      "learning_rate": 1.1307441173839413e-05,
      "loss": 0.2083,
      "step": 25380
    },
    {
      "epoch": 0.4583486808369265,
      "grad_norm": 0.43957242369651794,
      "learning_rate": 1.1304629224423805e-05,
      "loss": 0.2222,
      "step": 25385
    },
    {
      "epoch": 0.45843896026982717,
      "grad_norm": 0.3509773910045624,
      "learning_rate": 1.1301817170060262e-05,
      "loss": 0.2608,
      "step": 25390
    },
    {
      "epoch": 0.4585292397027279,
      "grad_norm": 0.43266743421554565,
      "learning_rate": 1.1299005010974994e-05,
      "loss": 0.247,
      "step": 25395
    },
    {
      "epoch": 0.4586195191356286,
      "grad_norm": 0.3913414776325226,
      "learning_rate": 1.1296192747394219e-05,
      "loss": 0.1879,
      "step": 25400
    },
    {
      "epoch": 0.45870979856852934,
      "grad_norm": 0.3474053144454956,
      "learning_rate": 1.1293380379544159e-05,
      "loss": 0.2345,
      "step": 25405
    },
    {
      "epoch": 0.45880007800143,
      "grad_norm": 0.5937274098396301,
      "learning_rate": 1.1290567907651053e-05,
      "loss": 0.2455,
      "step": 25410
    },
    {
      "epoch": 0.45889035743433076,
      "grad_norm": 0.5838733911514282,
      "learning_rate": 1.1287755331941142e-05,
      "loss": 0.1668,
      "step": 25415
    },
    {
      "epoch": 0.45898063686723145,
      "grad_norm": 0.4663761258125305,
      "learning_rate": 1.1284942652640674e-05,
      "loss": 0.2531,
      "step": 25420
    },
    {
      "epoch": 0.4590709163001322,
      "grad_norm": 0.7597901225090027,
      "learning_rate": 1.1282129869975915e-05,
      "loss": 0.3999,
      "step": 25425
    },
    {
      "epoch": 0.4591611957330329,
      "grad_norm": 0.694631814956665,
      "learning_rate": 1.1279316984173132e-05,
      "loss": 0.2582,
      "step": 25430
    },
    {
      "epoch": 0.4592514751659336,
      "grad_norm": 1.2105693817138672,
      "learning_rate": 1.1276503995458596e-05,
      "loss": 0.3124,
      "step": 25435
    },
    {
      "epoch": 0.4593417545988343,
      "grad_norm": 0.5668565630912781,
      "learning_rate": 1.1273690904058595e-05,
      "loss": 0.2417,
      "step": 25440
    },
    {
      "epoch": 0.45943203403173505,
      "grad_norm": 0.5524459481239319,
      "learning_rate": 1.1270877710199421e-05,
      "loss": 0.238,
      "step": 25445
    },
    {
      "epoch": 0.45952231346463573,
      "grad_norm": 0.5903627872467041,
      "learning_rate": 1.1268064414107379e-05,
      "loss": 0.2152,
      "step": 25450
    },
    {
      "epoch": 0.4596125928975365,
      "grad_norm": 0.7859571576118469,
      "learning_rate": 1.126525101600877e-05,
      "loss": 0.3142,
      "step": 25455
    },
    {
      "epoch": 0.45970287233043716,
      "grad_norm": 0.3708322048187256,
      "learning_rate": 1.1262437516129918e-05,
      "loss": 0.3153,
      "step": 25460
    },
    {
      "epoch": 0.4597931517633379,
      "grad_norm": 0.5059872269630432,
      "learning_rate": 1.1259623914697149e-05,
      "loss": 0.2025,
      "step": 25465
    },
    {
      "epoch": 0.4598834311962386,
      "grad_norm": 0.365987092256546,
      "learning_rate": 1.125681021193679e-05,
      "loss": 0.226,
      "step": 25470
    },
    {
      "epoch": 0.45997371062913933,
      "grad_norm": 0.37558290362358093,
      "learning_rate": 1.1253996408075192e-05,
      "loss": 0.1914,
      "step": 25475
    },
    {
      "epoch": 0.46006399006204,
      "grad_norm": 0.45481544733047485,
      "learning_rate": 1.1251182503338698e-05,
      "loss": 0.2769,
      "step": 25480
    },
    {
      "epoch": 0.46015426949494076,
      "grad_norm": 0.6855883598327637,
      "learning_rate": 1.1248368497953666e-05,
      "loss": 0.322,
      "step": 25485
    },
    {
      "epoch": 0.46024454892784145,
      "grad_norm": 0.3675699830055237,
      "learning_rate": 1.1245554392146467e-05,
      "loss": 0.316,
      "step": 25490
    },
    {
      "epoch": 0.4603348283607422,
      "grad_norm": 0.46729376912117004,
      "learning_rate": 1.1242740186143475e-05,
      "loss": 0.275,
      "step": 25495
    },
    {
      "epoch": 0.4604251077936429,
      "grad_norm": 0.3551467955112457,
      "learning_rate": 1.123992588017107e-05,
      "loss": 0.3124,
      "step": 25500
    },
    {
      "epoch": 0.4605153872265436,
      "grad_norm": 0.5993281006813049,
      "learning_rate": 1.123711147445564e-05,
      "loss": 0.3097,
      "step": 25505
    },
    {
      "epoch": 0.4606056666594443,
      "grad_norm": 0.322431743144989,
      "learning_rate": 1.123429696922359e-05,
      "loss": 0.2159,
      "step": 25510
    },
    {
      "epoch": 0.46069594609234504,
      "grad_norm": 0.5621485114097595,
      "learning_rate": 1.1231482364701323e-05,
      "loss": 0.3048,
      "step": 25515
    },
    {
      "epoch": 0.46078622552524573,
      "grad_norm": 0.34407296776771545,
      "learning_rate": 1.1228667661115249e-05,
      "loss": 0.2101,
      "step": 25520
    },
    {
      "epoch": 0.46087650495814647,
      "grad_norm": 0.48472723364830017,
      "learning_rate": 1.12258528586918e-05,
      "loss": 0.1564,
      "step": 25525
    },
    {
      "epoch": 0.46096678439104716,
      "grad_norm": 0.41901537775993347,
      "learning_rate": 1.1223037957657399e-05,
      "loss": 0.1618,
      "step": 25530
    },
    {
      "epoch": 0.4610570638239479,
      "grad_norm": 0.3657739460468292,
      "learning_rate": 1.122022295823849e-05,
      "loss": 0.208,
      "step": 25535
    },
    {
      "epoch": 0.4611473432568486,
      "grad_norm": 0.443582683801651,
      "learning_rate": 1.1217407860661509e-05,
      "loss": 0.2967,
      "step": 25540
    },
    {
      "epoch": 0.4612376226897493,
      "grad_norm": 0.8349863290786743,
      "learning_rate": 1.1214592665152924e-05,
      "loss": 0.2257,
      "step": 25545
    },
    {
      "epoch": 0.46132790212265,
      "grad_norm": 0.5645694732666016,
      "learning_rate": 1.1211777371939186e-05,
      "loss": 0.2244,
      "step": 25550
    },
    {
      "epoch": 0.46141818155555075,
      "grad_norm": 0.254520982503891,
      "learning_rate": 1.1208961981246768e-05,
      "loss": 0.1787,
      "step": 25555
    },
    {
      "epoch": 0.46150846098845144,
      "grad_norm": 0.3871762752532959,
      "learning_rate": 1.1206146493302155e-05,
      "loss": 0.321,
      "step": 25560
    },
    {
      "epoch": 0.4615987404213522,
      "grad_norm": 0.39559558033943176,
      "learning_rate": 1.1203330908331822e-05,
      "loss": 0.2453,
      "step": 25565
    },
    {
      "epoch": 0.46168901985425287,
      "grad_norm": 0.3969613015651703,
      "learning_rate": 1.1200515226562264e-05,
      "loss": 0.2542,
      "step": 25570
    },
    {
      "epoch": 0.4617792992871536,
      "grad_norm": 0.4680093228816986,
      "learning_rate": 1.1197699448219988e-05,
      "loss": 0.3573,
      "step": 25575
    },
    {
      "epoch": 0.4618695787200543,
      "grad_norm": 0.38695165514945984,
      "learning_rate": 1.11948835735315e-05,
      "loss": 0.2162,
      "step": 25580
    },
    {
      "epoch": 0.46195985815295504,
      "grad_norm": 0.4600764513015747,
      "learning_rate": 1.1192067602723311e-05,
      "loss": 0.2679,
      "step": 25585
    },
    {
      "epoch": 0.4620501375858557,
      "grad_norm": 0.5053609013557434,
      "learning_rate": 1.1189251536021955e-05,
      "loss": 0.3082,
      "step": 25590
    },
    {
      "epoch": 0.46214041701875647,
      "grad_norm": 0.5011573433876038,
      "learning_rate": 1.1186435373653959e-05,
      "loss": 0.2691,
      "step": 25595
    },
    {
      "epoch": 0.46223069645165715,
      "grad_norm": 0.42619091272354126,
      "learning_rate": 1.1183619115845863e-05,
      "loss": 0.1645,
      "step": 25600
    },
    {
      "epoch": 0.4623209758845579,
      "grad_norm": 0.8814185857772827,
      "learning_rate": 1.1180802762824215e-05,
      "loss": 0.2909,
      "step": 25605
    },
    {
      "epoch": 0.4624112553174586,
      "grad_norm": 0.4841145873069763,
      "learning_rate": 1.117798631481557e-05,
      "loss": 0.2913,
      "step": 25610
    },
    {
      "epoch": 0.4625015347503593,
      "grad_norm": 0.4144599437713623,
      "learning_rate": 1.117516977204649e-05,
      "loss": 0.2421,
      "step": 25615
    },
    {
      "epoch": 0.46259181418326,
      "grad_norm": 0.4766666293144226,
      "learning_rate": 1.1172353134743546e-05,
      "loss": 0.2619,
      "step": 25620
    },
    {
      "epoch": 0.46268209361616075,
      "grad_norm": 0.4696897566318512,
      "learning_rate": 1.1169536403133319e-05,
      "loss": 0.3143,
      "step": 25625
    },
    {
      "epoch": 0.46277237304906144,
      "grad_norm": 0.8463373184204102,
      "learning_rate": 1.1166719577442389e-05,
      "loss": 0.2457,
      "step": 25630
    },
    {
      "epoch": 0.4628626524819622,
      "grad_norm": 0.3645748496055603,
      "learning_rate": 1.1163902657897355e-05,
      "loss": 0.2786,
      "step": 25635
    },
    {
      "epoch": 0.46295293191486286,
      "grad_norm": 0.3574855625629425,
      "learning_rate": 1.116108564472481e-05,
      "loss": 0.2294,
      "step": 25640
    },
    {
      "epoch": 0.4630432113477636,
      "grad_norm": 0.3958921730518341,
      "learning_rate": 1.1158268538151369e-05,
      "loss": 0.2407,
      "step": 25645
    },
    {
      "epoch": 0.4631334907806643,
      "grad_norm": 0.4588448405265808,
      "learning_rate": 1.1155451338403645e-05,
      "loss": 0.2109,
      "step": 25650
    },
    {
      "epoch": 0.46322377021356503,
      "grad_norm": 0.5760933756828308,
      "learning_rate": 1.1152634045708263e-05,
      "loss": 0.1958,
      "step": 25655
    },
    {
      "epoch": 0.4633140496464657,
      "grad_norm": 0.5492554903030396,
      "learning_rate": 1.1149816660291852e-05,
      "loss": 0.2414,
      "step": 25660
    },
    {
      "epoch": 0.46340432907936646,
      "grad_norm": 0.5135606527328491,
      "learning_rate": 1.1146999182381051e-05,
      "loss": 0.2112,
      "step": 25665
    },
    {
      "epoch": 0.46349460851226715,
      "grad_norm": 0.2507036328315735,
      "learning_rate": 1.1144181612202502e-05,
      "loss": 0.2593,
      "step": 25670
    },
    {
      "epoch": 0.4635848879451679,
      "grad_norm": 0.672838568687439,
      "learning_rate": 1.1141363949982867e-05,
      "loss": 0.18,
      "step": 25675
    },
    {
      "epoch": 0.4636751673780686,
      "grad_norm": 0.29975202679634094,
      "learning_rate": 1.1138546195948798e-05,
      "loss": 0.3004,
      "step": 25680
    },
    {
      "epoch": 0.4637654468109693,
      "grad_norm": 0.44783079624176025,
      "learning_rate": 1.1135728350326965e-05,
      "loss": 0.2838,
      "step": 25685
    },
    {
      "epoch": 0.46385572624387,
      "grad_norm": 0.5177004933357239,
      "learning_rate": 1.1132910413344042e-05,
      "loss": 0.296,
      "step": 25690
    },
    {
      "epoch": 0.46394600567677075,
      "grad_norm": 0.47807779908180237,
      "learning_rate": 1.1130092385226717e-05,
      "loss": 0.2883,
      "step": 25695
    },
    {
      "epoch": 0.46403628510967143,
      "grad_norm": 0.3265976011753082,
      "learning_rate": 1.1127274266201676e-05,
      "loss": 0.2491,
      "step": 25700
    },
    {
      "epoch": 0.4641265645425722,
      "grad_norm": 0.5433142781257629,
      "learning_rate": 1.1124456056495611e-05,
      "loss": 0.2103,
      "step": 25705
    },
    {
      "epoch": 0.46421684397547286,
      "grad_norm": 0.36991173028945923,
      "learning_rate": 1.1121637756335238e-05,
      "loss": 0.2209,
      "step": 25710
    },
    {
      "epoch": 0.4643071234083736,
      "grad_norm": 0.6328502893447876,
      "learning_rate": 1.111881936594726e-05,
      "loss": 0.3442,
      "step": 25715
    },
    {
      "epoch": 0.4643974028412743,
      "grad_norm": 0.36517852544784546,
      "learning_rate": 1.1116000885558398e-05,
      "loss": 0.1797,
      "step": 25720
    },
    {
      "epoch": 0.46448768227417503,
      "grad_norm": 0.4057893455028534,
      "learning_rate": 1.1113182315395379e-05,
      "loss": 0.3115,
      "step": 25725
    },
    {
      "epoch": 0.4645779617070757,
      "grad_norm": 0.6015501618385315,
      "learning_rate": 1.111036365568494e-05,
      "loss": 0.3159,
      "step": 25730
    },
    {
      "epoch": 0.46466824113997646,
      "grad_norm": 0.3591579496860504,
      "learning_rate": 1.1107544906653811e-05,
      "loss": 0.2598,
      "step": 25735
    },
    {
      "epoch": 0.46475852057287714,
      "grad_norm": 0.5037890672683716,
      "learning_rate": 1.1104726068528755e-05,
      "loss": 0.2334,
      "step": 25740
    },
    {
      "epoch": 0.4648488000057779,
      "grad_norm": 0.31298428773880005,
      "learning_rate": 1.1101907141536514e-05,
      "loss": 0.2427,
      "step": 25745
    },
    {
      "epoch": 0.46493907943867857,
      "grad_norm": 0.2975746989250183,
      "learning_rate": 1.1099088125903855e-05,
      "loss": 0.298,
      "step": 25750
    },
    {
      "epoch": 0.4650293588715793,
      "grad_norm": 0.4254186749458313,
      "learning_rate": 1.109626902185755e-05,
      "loss": 0.2685,
      "step": 25755
    },
    {
      "epoch": 0.46511963830448005,
      "grad_norm": 0.4576852023601532,
      "learning_rate": 1.1093449829624372e-05,
      "loss": 0.2352,
      "step": 25760
    },
    {
      "epoch": 0.46520991773738074,
      "grad_norm": 0.2610947787761688,
      "learning_rate": 1.1090630549431105e-05,
      "loss": 0.1378,
      "step": 25765
    },
    {
      "epoch": 0.4653001971702815,
      "grad_norm": 0.35531389713287354,
      "learning_rate": 1.1087811181504538e-05,
      "loss": 0.1956,
      "step": 25770
    },
    {
      "epoch": 0.46539047660318217,
      "grad_norm": 0.52121901512146,
      "learning_rate": 1.1084991726071472e-05,
      "loss": 0.293,
      "step": 25775
    },
    {
      "epoch": 0.4654807560360829,
      "grad_norm": 0.5938030481338501,
      "learning_rate": 1.1082172183358713e-05,
      "loss": 0.2746,
      "step": 25780
    },
    {
      "epoch": 0.4655710354689836,
      "grad_norm": 0.5315929055213928,
      "learning_rate": 1.1079352553593067e-05,
      "loss": 0.3005,
      "step": 25785
    },
    {
      "epoch": 0.46566131490188434,
      "grad_norm": 0.40332263708114624,
      "learning_rate": 1.1076532837001355e-05,
      "loss": 0.2105,
      "step": 25790
    },
    {
      "epoch": 0.465751594334785,
      "grad_norm": 0.4936860203742981,
      "learning_rate": 1.1073713033810405e-05,
      "loss": 0.2534,
      "step": 25795
    },
    {
      "epoch": 0.46584187376768577,
      "grad_norm": 0.5080345869064331,
      "learning_rate": 1.1070893144247046e-05,
      "loss": 0.2169,
      "step": 25800
    },
    {
      "epoch": 0.46593215320058645,
      "grad_norm": 0.46547484397888184,
      "learning_rate": 1.1068073168538123e-05,
      "loss": 0.2345,
      "step": 25805
    },
    {
      "epoch": 0.4660224326334872,
      "grad_norm": 0.4039542078971863,
      "learning_rate": 1.1065253106910476e-05,
      "loss": 0.2727,
      "step": 25810
    },
    {
      "epoch": 0.4661127120663879,
      "grad_norm": 0.5644627213478088,
      "learning_rate": 1.1062432959590965e-05,
      "loss": 0.2923,
      "step": 25815
    },
    {
      "epoch": 0.4662029914992886,
      "grad_norm": 0.3832181692123413,
      "learning_rate": 1.1059612726806444e-05,
      "loss": 0.2747,
      "step": 25820
    },
    {
      "epoch": 0.4662932709321893,
      "grad_norm": 0.24436330795288086,
      "learning_rate": 1.1056792408783786e-05,
      "loss": 0.2788,
      "step": 25825
    },
    {
      "epoch": 0.46638355036509005,
      "grad_norm": 0.39775246381759644,
      "learning_rate": 1.1053972005749861e-05,
      "loss": 0.1851,
      "step": 25830
    },
    {
      "epoch": 0.46647382979799074,
      "grad_norm": 0.7041085958480835,
      "learning_rate": 1.1051151517931551e-05,
      "loss": 0.32,
      "step": 25835
    },
    {
      "epoch": 0.4665641092308915,
      "grad_norm": 0.49672818183898926,
      "learning_rate": 1.1048330945555745e-05,
      "loss": 0.2239,
      "step": 25840
    },
    {
      "epoch": 0.46665438866379216,
      "grad_norm": 0.5358633399009705,
      "learning_rate": 1.1045510288849336e-05,
      "loss": 0.3883,
      "step": 25845
    },
    {
      "epoch": 0.4667446680966929,
      "grad_norm": 0.41424885392189026,
      "learning_rate": 1.1042689548039224e-05,
      "loss": 0.2624,
      "step": 25850
    },
    {
      "epoch": 0.4668349475295936,
      "grad_norm": 0.5248550772666931,
      "learning_rate": 1.103986872335232e-05,
      "loss": 0.3169,
      "step": 25855
    },
    {
      "epoch": 0.46692522696249433,
      "grad_norm": 0.4089259207248688,
      "learning_rate": 1.103704781501554e-05,
      "loss": 0.3263,
      "step": 25860
    },
    {
      "epoch": 0.467015506395395,
      "grad_norm": 0.6299268007278442,
      "learning_rate": 1.1034226823255802e-05,
      "loss": 0.252,
      "step": 25865
    },
    {
      "epoch": 0.46710578582829576,
      "grad_norm": 0.5844053030014038,
      "learning_rate": 1.1031405748300035e-05,
      "loss": 0.3051,
      "step": 25870
    },
    {
      "epoch": 0.46719606526119645,
      "grad_norm": 0.472973108291626,
      "learning_rate": 1.1028584590375175e-05,
      "loss": 0.2391,
      "step": 25875
    },
    {
      "epoch": 0.4672863446940972,
      "grad_norm": 0.5440942049026489,
      "learning_rate": 1.1025763349708163e-05,
      "loss": 0.3175,
      "step": 25880
    },
    {
      "epoch": 0.4673766241269979,
      "grad_norm": 0.5540207028388977,
      "learning_rate": 1.1022942026525947e-05,
      "loss": 0.2434,
      "step": 25885
    },
    {
      "epoch": 0.4674669035598986,
      "grad_norm": 0.5777327418327332,
      "learning_rate": 1.1020120621055483e-05,
      "loss": 0.2079,
      "step": 25890
    },
    {
      "epoch": 0.4675571829927993,
      "grad_norm": 0.658270001411438,
      "learning_rate": 1.1017299133523734e-05,
      "loss": 0.2382,
      "step": 25895
    },
    {
      "epoch": 0.46764746242570004,
      "grad_norm": 0.5810391902923584,
      "learning_rate": 1.1014477564157661e-05,
      "loss": 0.262,
      "step": 25900
    },
    {
      "epoch": 0.46773774185860073,
      "grad_norm": 0.3897687494754791,
      "learning_rate": 1.1011655913184249e-05,
      "loss": 0.2611,
      "step": 25905
    },
    {
      "epoch": 0.4678280212915015,
      "grad_norm": 0.45382043719291687,
      "learning_rate": 1.1008834180830472e-05,
      "loss": 0.3405,
      "step": 25910
    },
    {
      "epoch": 0.46791830072440216,
      "grad_norm": 0.3488757908344269,
      "learning_rate": 1.1006012367323322e-05,
      "loss": 0.2745,
      "step": 25915
    },
    {
      "epoch": 0.4680085801573029,
      "grad_norm": 0.35227203369140625,
      "learning_rate": 1.1003190472889788e-05,
      "loss": 0.3258,
      "step": 25920
    },
    {
      "epoch": 0.4680988595902036,
      "grad_norm": 0.4962708652019501,
      "learning_rate": 1.100036849775688e-05,
      "loss": 0.2832,
      "step": 25925
    },
    {
      "epoch": 0.46818913902310433,
      "grad_norm": 0.6424579620361328,
      "learning_rate": 1.0997546442151597e-05,
      "loss": 0.2182,
      "step": 25930
    },
    {
      "epoch": 0.468279418456005,
      "grad_norm": 0.35105547308921814,
      "learning_rate": 1.0994724306300953e-05,
      "loss": 0.2243,
      "step": 25935
    },
    {
      "epoch": 0.46836969788890576,
      "grad_norm": 0.6380845904350281,
      "learning_rate": 1.0991902090431971e-05,
      "loss": 0.1955,
      "step": 25940
    },
    {
      "epoch": 0.46845997732180644,
      "grad_norm": 0.31674501299858093,
      "learning_rate": 1.0989079794771682e-05,
      "loss": 0.2868,
      "step": 25945
    },
    {
      "epoch": 0.4685502567547072,
      "grad_norm": 0.6792436242103577,
      "learning_rate": 1.098625741954711e-05,
      "loss": 0.2341,
      "step": 25950
    },
    {
      "epoch": 0.46864053618760787,
      "grad_norm": 0.5717369318008423,
      "learning_rate": 1.0983434964985303e-05,
      "loss": 0.1819,
      "step": 25955
    },
    {
      "epoch": 0.4687308156205086,
      "grad_norm": 0.49002474546432495,
      "learning_rate": 1.0980612431313302e-05,
      "loss": 0.2725,
      "step": 25960
    },
    {
      "epoch": 0.4688210950534093,
      "grad_norm": 0.6652733683586121,
      "learning_rate": 1.0977789818758157e-05,
      "loss": 0.3086,
      "step": 25965
    },
    {
      "epoch": 0.46891137448631004,
      "grad_norm": 0.5847036838531494,
      "learning_rate": 1.0974967127546937e-05,
      "loss": 0.3057,
      "step": 25970
    },
    {
      "epoch": 0.4690016539192107,
      "grad_norm": 0.5464393496513367,
      "learning_rate": 1.0972144357906696e-05,
      "loss": 0.3747,
      "step": 25975
    },
    {
      "epoch": 0.46909193335211147,
      "grad_norm": 0.6326963305473328,
      "learning_rate": 1.0969321510064512e-05,
      "loss": 0.2874,
      "step": 25980
    },
    {
      "epoch": 0.46918221278501215,
      "grad_norm": 0.4930112659931183,
      "learning_rate": 1.0966498584247455e-05,
      "loss": 0.2897,
      "step": 25985
    },
    {
      "epoch": 0.4692724922179129,
      "grad_norm": 0.3699823319911957,
      "learning_rate": 1.096367558068262e-05,
      "loss": 0.1979,
      "step": 25990
    },
    {
      "epoch": 0.4693627716508136,
      "grad_norm": 0.5999264717102051,
      "learning_rate": 1.096085249959709e-05,
      "loss": 0.2248,
      "step": 25995
    },
    {
      "epoch": 0.4694530510837143,
      "grad_norm": 0.3253791332244873,
      "learning_rate": 1.095802934121796e-05,
      "loss": 0.2934,
      "step": 26000
    },
    {
      "epoch": 0.469543330516615,
      "grad_norm": 0.3865527808666229,
      "learning_rate": 1.0955206105772336e-05,
      "loss": 0.2156,
      "step": 26005
    },
    {
      "epoch": 0.46963360994951575,
      "grad_norm": 1.0743937492370605,
      "learning_rate": 1.0952382793487327e-05,
      "loss": 0.2454,
      "step": 26010
    },
    {
      "epoch": 0.46972388938241644,
      "grad_norm": 0.5202724933624268,
      "learning_rate": 1.0949559404590046e-05,
      "loss": 0.2771,
      "step": 26015
    },
    {
      "epoch": 0.4698141688153172,
      "grad_norm": 0.4546166956424713,
      "learning_rate": 1.0946735939307616e-05,
      "loss": 0.2314,
      "step": 26020
    },
    {
      "epoch": 0.46990444824821787,
      "grad_norm": 0.4739367365837097,
      "learning_rate": 1.0943912397867162e-05,
      "loss": 0.2837,
      "step": 26025
    },
    {
      "epoch": 0.4699947276811186,
      "grad_norm": 0.6224332451820374,
      "learning_rate": 1.0941088780495819e-05,
      "loss": 0.2555,
      "step": 26030
    },
    {
      "epoch": 0.4700850071140193,
      "grad_norm": 0.7361112833023071,
      "learning_rate": 1.0938265087420725e-05,
      "loss": 0.3031,
      "step": 26035
    },
    {
      "epoch": 0.47017528654692003,
      "grad_norm": 0.3367628753185272,
      "learning_rate": 1.0935441318869027e-05,
      "loss": 0.2023,
      "step": 26040
    },
    {
      "epoch": 0.4702655659798207,
      "grad_norm": 0.378570556640625,
      "learning_rate": 1.0932617475067877e-05,
      "loss": 0.3861,
      "step": 26045
    },
    {
      "epoch": 0.47035584541272146,
      "grad_norm": 0.5609474182128906,
      "learning_rate": 1.092979355624443e-05,
      "loss": 0.22,
      "step": 26050
    },
    {
      "epoch": 0.47044612484562215,
      "grad_norm": 0.5063912868499756,
      "learning_rate": 1.0926969562625852e-05,
      "loss": 0.279,
      "step": 26055
    },
    {
      "epoch": 0.4705364042785229,
      "grad_norm": 0.21222810447216034,
      "learning_rate": 1.0924145494439316e-05,
      "loss": 0.3078,
      "step": 26060
    },
    {
      "epoch": 0.4706266837114236,
      "grad_norm": 0.4927471876144409,
      "learning_rate": 1.0921321351911989e-05,
      "loss": 0.1854,
      "step": 26065
    },
    {
      "epoch": 0.4707169631443243,
      "grad_norm": 0.5247211456298828,
      "learning_rate": 1.091849713527106e-05,
      "loss": 0.3319,
      "step": 26070
    },
    {
      "epoch": 0.470807242577225,
      "grad_norm": 0.3175283372402191,
      "learning_rate": 1.0915672844743717e-05,
      "loss": 0.2239,
      "step": 26075
    },
    {
      "epoch": 0.47089752201012575,
      "grad_norm": 0.4371960163116455,
      "learning_rate": 1.091284848055715e-05,
      "loss": 0.3051,
      "step": 26080
    },
    {
      "epoch": 0.47098780144302643,
      "grad_norm": 0.5870938897132874,
      "learning_rate": 1.0910024042938556e-05,
      "loss": 0.2086,
      "step": 26085
    },
    {
      "epoch": 0.4710780808759272,
      "grad_norm": 0.6371119618415833,
      "learning_rate": 1.0907199532115149e-05,
      "loss": 0.2639,
      "step": 26090
    },
    {
      "epoch": 0.47116836030882786,
      "grad_norm": 0.5096482634544373,
      "learning_rate": 1.0904374948314134e-05,
      "loss": 0.2345,
      "step": 26095
    },
    {
      "epoch": 0.4712586397417286,
      "grad_norm": 0.48447567224502563,
      "learning_rate": 1.0901550291762729e-05,
      "loss": 0.3136,
      "step": 26100
    },
    {
      "epoch": 0.4713489191746293,
      "grad_norm": 0.49430206418037415,
      "learning_rate": 1.0898725562688157e-05,
      "loss": 0.1687,
      "step": 26105
    },
    {
      "epoch": 0.47143919860753003,
      "grad_norm": 0.36696097254753113,
      "learning_rate": 1.0895900761317653e-05,
      "loss": 0.3424,
      "step": 26110
    },
    {
      "epoch": 0.4715294780404307,
      "grad_norm": 0.56471848487854,
      "learning_rate": 1.0893075887878442e-05,
      "loss": 0.2464,
      "step": 26115
    },
    {
      "epoch": 0.47161975747333146,
      "grad_norm": 1.4889129400253296,
      "learning_rate": 1.089025094259777e-05,
      "loss": 0.1967,
      "step": 26120
    },
    {
      "epoch": 0.47171003690623214,
      "grad_norm": 0.4604998826980591,
      "learning_rate": 1.0887425925702884e-05,
      "loss": 0.2182,
      "step": 26125
    },
    {
      "epoch": 0.4718003163391329,
      "grad_norm": 0.5376400351524353,
      "learning_rate": 1.0884600837421032e-05,
      "loss": 0.3413,
      "step": 26130
    },
    {
      "epoch": 0.47189059577203357,
      "grad_norm": 0.5955811738967896,
      "learning_rate": 1.0881775677979474e-05,
      "loss": 0.2632,
      "step": 26135
    },
    {
      "epoch": 0.4719808752049343,
      "grad_norm": 0.7458768486976624,
      "learning_rate": 1.0878950447605476e-05,
      "loss": 0.2911,
      "step": 26140
    },
    {
      "epoch": 0.472071154637835,
      "grad_norm": 0.6885311007499695,
      "learning_rate": 1.0876125146526304e-05,
      "loss": 0.1883,
      "step": 26145
    },
    {
      "epoch": 0.47216143407073574,
      "grad_norm": 0.5044176578521729,
      "learning_rate": 1.0873299774969233e-05,
      "loss": 0.3279,
      "step": 26150
    },
    {
      "epoch": 0.4722517135036365,
      "grad_norm": 0.39769864082336426,
      "learning_rate": 1.0870474333161547e-05,
      "loss": 0.2401,
      "step": 26155
    },
    {
      "epoch": 0.47234199293653717,
      "grad_norm": 0.2982105016708374,
      "learning_rate": 1.0867648821330529e-05,
      "loss": 0.2673,
      "step": 26160
    },
    {
      "epoch": 0.4724322723694379,
      "grad_norm": 0.4600881040096283,
      "learning_rate": 1.0864823239703471e-05,
      "loss": 0.3246,
      "step": 26165
    },
    {
      "epoch": 0.4725225518023386,
      "grad_norm": 0.513886570930481,
      "learning_rate": 1.086199758850767e-05,
      "loss": 0.2409,
      "step": 26170
    },
    {
      "epoch": 0.47261283123523934,
      "grad_norm": 0.39647242426872253,
      "learning_rate": 1.0859171867970434e-05,
      "loss": 0.2802,
      "step": 26175
    },
    {
      "epoch": 0.47270311066814,
      "grad_norm": 0.45733892917633057,
      "learning_rate": 1.0856346078319067e-05,
      "loss": 0.2692,
      "step": 26180
    },
    {
      "epoch": 0.47279339010104077,
      "grad_norm": 2.128892660140991,
      "learning_rate": 1.0853520219780882e-05,
      "loss": 0.3395,
      "step": 26185
    },
    {
      "epoch": 0.47288366953394145,
      "grad_norm": 0.3948085606098175,
      "learning_rate": 1.0850694292583201e-05,
      "loss": 0.2767,
      "step": 26190
    },
    {
      "epoch": 0.4729739489668422,
      "grad_norm": 0.3208751678466797,
      "learning_rate": 1.0847868296953352e-05,
      "loss": 0.3812,
      "step": 26195
    },
    {
      "epoch": 0.4730642283997429,
      "grad_norm": 0.47710272669792175,
      "learning_rate": 1.0845042233118663e-05,
      "loss": 0.3216,
      "step": 26200
    },
    {
      "epoch": 0.4731545078326436,
      "grad_norm": 0.4054561257362366,
      "learning_rate": 1.0842216101306468e-05,
      "loss": 0.2704,
      "step": 26205
    },
    {
      "epoch": 0.4732447872655443,
      "grad_norm": 0.5142204761505127,
      "learning_rate": 1.0839389901744113e-05,
      "loss": 0.2548,
      "step": 26210
    },
    {
      "epoch": 0.47333506669844505,
      "grad_norm": 0.5620430707931519,
      "learning_rate": 1.083656363465894e-05,
      "loss": 0.2676,
      "step": 26215
    },
    {
      "epoch": 0.47342534613134574,
      "grad_norm": 0.4242805242538452,
      "learning_rate": 1.0833737300278312e-05,
      "loss": 0.2053,
      "step": 26220
    },
    {
      "epoch": 0.4735156255642465,
      "grad_norm": 0.787665843963623,
      "learning_rate": 1.0830910898829573e-05,
      "loss": 0.3023,
      "step": 26225
    },
    {
      "epoch": 0.47360590499714716,
      "grad_norm": 0.36403128504753113,
      "learning_rate": 1.0828084430540096e-05,
      "loss": 0.2033,
      "step": 26230
    },
    {
      "epoch": 0.4736961844300479,
      "grad_norm": 0.35800427198410034,
      "learning_rate": 1.0825257895637246e-05,
      "loss": 0.2328,
      "step": 26235
    },
    {
      "epoch": 0.4737864638629486,
      "grad_norm": 0.5503618121147156,
      "learning_rate": 1.0822431294348403e-05,
      "loss": 0.2239,
      "step": 26240
    },
    {
      "epoch": 0.47387674329584933,
      "grad_norm": 0.5361478328704834,
      "learning_rate": 1.081960462690094e-05,
      "loss": 0.3129,
      "step": 26245
    },
    {
      "epoch": 0.47396702272875,
      "grad_norm": 0.9362848997116089,
      "learning_rate": 1.081677789352224e-05,
      "loss": 0.2872,
      "step": 26250
    },
    {
      "epoch": 0.47405730216165076,
      "grad_norm": 0.3955424129962921,
      "learning_rate": 1.08139510944397e-05,
      "loss": 0.2665,
      "step": 26255
    },
    {
      "epoch": 0.47414758159455145,
      "grad_norm": 0.4825882315635681,
      "learning_rate": 1.0811124229880715e-05,
      "loss": 0.1996,
      "step": 26260
    },
    {
      "epoch": 0.4742378610274522,
      "grad_norm": 0.5206092000007629,
      "learning_rate": 1.0808297300072677e-05,
      "loss": 0.2343,
      "step": 26265
    },
    {
      "epoch": 0.4743281404603529,
      "grad_norm": 0.4103998839855194,
      "learning_rate": 1.0805470305243001e-05,
      "loss": 0.2462,
      "step": 26270
    },
    {
      "epoch": 0.4744184198932536,
      "grad_norm": 0.31804096698760986,
      "learning_rate": 1.0802643245619097e-05,
      "loss": 0.2299,
      "step": 26275
    },
    {
      "epoch": 0.4745086993261543,
      "grad_norm": 0.35483410954475403,
      "learning_rate": 1.079981612142838e-05,
      "loss": 0.2021,
      "step": 26280
    },
    {
      "epoch": 0.47459897875905505,
      "grad_norm": 0.6194055676460266,
      "learning_rate": 1.0796988932898267e-05,
      "loss": 0.3255,
      "step": 26285
    },
    {
      "epoch": 0.47468925819195573,
      "grad_norm": 0.44059252738952637,
      "learning_rate": 1.0794161680256194e-05,
      "loss": 0.2525,
      "step": 26290
    },
    {
      "epoch": 0.4747795376248565,
      "grad_norm": 0.34184208512306213,
      "learning_rate": 1.0791334363729586e-05,
      "loss": 0.2726,
      "step": 26295
    },
    {
      "epoch": 0.47486981705775716,
      "grad_norm": 0.48157361149787903,
      "learning_rate": 1.0788506983545881e-05,
      "loss": 0.1915,
      "step": 26300
    },
    {
      "epoch": 0.4749600964906579,
      "grad_norm": 0.503015398979187,
      "learning_rate": 1.0785679539932526e-05,
      "loss": 0.2278,
      "step": 26305
    },
    {
      "epoch": 0.4750503759235586,
      "grad_norm": 0.37844520807266235,
      "learning_rate": 1.0782852033116959e-05,
      "loss": 0.1732,
      "step": 26310
    },
    {
      "epoch": 0.47514065535645933,
      "grad_norm": 0.49388542771339417,
      "learning_rate": 1.078002446332664e-05,
      "loss": 0.3277,
      "step": 26315
    },
    {
      "epoch": 0.47523093478936,
      "grad_norm": 0.29744821786880493,
      "learning_rate": 1.0777196830789025e-05,
      "loss": 0.2413,
      "step": 26320
    },
    {
      "epoch": 0.47532121422226076,
      "grad_norm": 0.604690670967102,
      "learning_rate": 1.0774369135731575e-05,
      "loss": 0.2685,
      "step": 26325
    },
    {
      "epoch": 0.47541149365516144,
      "grad_norm": 0.32783612608909607,
      "learning_rate": 1.0771541378381758e-05,
      "loss": 0.2013,
      "step": 26330
    },
    {
      "epoch": 0.4755017730880622,
      "grad_norm": 0.29162389039993286,
      "learning_rate": 1.0768713558967048e-05,
      "loss": 0.2443,
      "step": 26335
    },
    {
      "epoch": 0.47559205252096287,
      "grad_norm": 0.38048410415649414,
      "learning_rate": 1.076588567771492e-05,
      "loss": 0.2531,
      "step": 26340
    },
    {
      "epoch": 0.4756823319538636,
      "grad_norm": 0.5431281924247742,
      "learning_rate": 1.0763057734852857e-05,
      "loss": 0.2447,
      "step": 26345
    },
    {
      "epoch": 0.4757726113867643,
      "grad_norm": 0.40416088700294495,
      "learning_rate": 1.0760229730608348e-05,
      "loss": 0.2327,
      "step": 26350
    },
    {
      "epoch": 0.47586289081966504,
      "grad_norm": 0.3202001452445984,
      "learning_rate": 1.0757401665208886e-05,
      "loss": 0.1971,
      "step": 26355
    },
    {
      "epoch": 0.4759531702525657,
      "grad_norm": 0.2053840309381485,
      "learning_rate": 1.0754573538881968e-05,
      "loss": 0.2461,
      "step": 26360
    },
    {
      "epoch": 0.47604344968546647,
      "grad_norm": 1.687205195426941,
      "learning_rate": 1.075174535185509e-05,
      "loss": 0.2447,
      "step": 26365
    },
    {
      "epoch": 0.47613372911836715,
      "grad_norm": 0.4893333911895752,
      "learning_rate": 1.0748917104355766e-05,
      "loss": 0.3474,
      "step": 26370
    },
    {
      "epoch": 0.4762240085512679,
      "grad_norm": 0.6483590602874756,
      "learning_rate": 1.074608879661151e-05,
      "loss": 0.1697,
      "step": 26375
    },
    {
      "epoch": 0.4763142879841686,
      "grad_norm": 0.29549115896224976,
      "learning_rate": 1.0743260428849832e-05,
      "loss": 0.1897,
      "step": 26380
    },
    {
      "epoch": 0.4764045674170693,
      "grad_norm": 0.5002672076225281,
      "learning_rate": 1.0740432001298258e-05,
      "loss": 0.2914,
      "step": 26385
    },
    {
      "epoch": 0.47649484684997,
      "grad_norm": 0.5598426461219788,
      "learning_rate": 1.0737603514184312e-05,
      "loss": 0.2492,
      "step": 26390
    },
    {
      "epoch": 0.47658512628287075,
      "grad_norm": 1.129087209701538,
      "learning_rate": 1.0734774967735523e-05,
      "loss": 0.2702,
      "step": 26395
    },
    {
      "epoch": 0.47667540571577144,
      "grad_norm": 0.6787183880805969,
      "learning_rate": 1.0731946362179434e-05,
      "loss": 0.2756,
      "step": 26400
    },
    {
      "epoch": 0.4767656851486722,
      "grad_norm": 0.6386555433273315,
      "learning_rate": 1.072911769774358e-05,
      "loss": 0.245,
      "step": 26405
    },
    {
      "epoch": 0.47685596458157287,
      "grad_norm": 0.30959513783454895,
      "learning_rate": 1.072628897465551e-05,
      "loss": 0.2177,
      "step": 26410
    },
    {
      "epoch": 0.4769462440144736,
      "grad_norm": 0.36446917057037354,
      "learning_rate": 1.0723460193142772e-05,
      "loss": 0.3273,
      "step": 26415
    },
    {
      "epoch": 0.4770365234473743,
      "grad_norm": 0.5002758502960205,
      "learning_rate": 1.072063135343292e-05,
      "loss": 0.2759,
      "step": 26420
    },
    {
      "epoch": 0.47712680288027504,
      "grad_norm": 0.3369194269180298,
      "learning_rate": 1.0717802455753519e-05,
      "loss": 0.191,
      "step": 26425
    },
    {
      "epoch": 0.4772170823131757,
      "grad_norm": 0.5441814661026001,
      "learning_rate": 1.0714973500332124e-05,
      "loss": 0.2409,
      "step": 26430
    },
    {
      "epoch": 0.47730736174607646,
      "grad_norm": 1.0081517696380615,
      "learning_rate": 1.0712144487396313e-05,
      "loss": 0.2947,
      "step": 26435
    },
    {
      "epoch": 0.47739764117897715,
      "grad_norm": 0.4605311155319214,
      "learning_rate": 1.0709315417173656e-05,
      "loss": 0.2141,
      "step": 26440
    },
    {
      "epoch": 0.4774879206118779,
      "grad_norm": 0.4445546567440033,
      "learning_rate": 1.0706486289891729e-05,
      "loss": 0.2481,
      "step": 26445
    },
    {
      "epoch": 0.4775782000447786,
      "grad_norm": 0.4483776092529297,
      "learning_rate": 1.0703657105778116e-05,
      "loss": 0.2655,
      "step": 26450
    },
    {
      "epoch": 0.4776684794776793,
      "grad_norm": 0.6984508037567139,
      "learning_rate": 1.0700827865060409e-05,
      "loss": 0.29,
      "step": 26455
    },
    {
      "epoch": 0.47775875891058,
      "grad_norm": 0.36635181307792664,
      "learning_rate": 1.0697998567966192e-05,
      "loss": 0.2816,
      "step": 26460
    },
    {
      "epoch": 0.47784903834348075,
      "grad_norm": 0.4521954357624054,
      "learning_rate": 1.0695169214723065e-05,
      "loss": 0.2911,
      "step": 26465
    },
    {
      "epoch": 0.47793931777638143,
      "grad_norm": 0.3105730712413788,
      "learning_rate": 1.0692339805558632e-05,
      "loss": 0.3114,
      "step": 26470
    },
    {
      "epoch": 0.4780295972092822,
      "grad_norm": 0.34403181076049805,
      "learning_rate": 1.0689510340700493e-05,
      "loss": 0.2183,
      "step": 26475
    },
    {
      "epoch": 0.47811987664218286,
      "grad_norm": 0.3706817328929901,
      "learning_rate": 1.068668082037626e-05,
      "loss": 0.3542,
      "step": 26480
    },
    {
      "epoch": 0.4782101560750836,
      "grad_norm": 0.7294043302536011,
      "learning_rate": 1.068385124481355e-05,
      "loss": 0.2699,
      "step": 26485
    },
    {
      "epoch": 0.4783004355079843,
      "grad_norm": 0.5159411430358887,
      "learning_rate": 1.068102161423998e-05,
      "loss": 0.3034,
      "step": 26490
    },
    {
      "epoch": 0.47839071494088503,
      "grad_norm": 0.32585665583610535,
      "learning_rate": 1.0678191928883168e-05,
      "loss": 0.1609,
      "step": 26495
    },
    {
      "epoch": 0.4784809943737857,
      "grad_norm": 0.45307326316833496,
      "learning_rate": 1.067536218897075e-05,
      "loss": 0.2491,
      "step": 26500
    },
    {
      "epoch": 0.47857127380668646,
      "grad_norm": 0.7151533365249634,
      "learning_rate": 1.0672532394730354e-05,
      "loss": 0.2381,
      "step": 26505
    },
    {
      "epoch": 0.47866155323958715,
      "grad_norm": 0.3486301004886627,
      "learning_rate": 1.0669702546389617e-05,
      "loss": 0.247,
      "step": 26510
    },
    {
      "epoch": 0.4787518326724879,
      "grad_norm": 0.8058130741119385,
      "learning_rate": 1.0666872644176176e-05,
      "loss": 0.2803,
      "step": 26515
    },
    {
      "epoch": 0.4788421121053886,
      "grad_norm": 0.3897343873977661,
      "learning_rate": 1.0664042688317682e-05,
      "loss": 0.1999,
      "step": 26520
    },
    {
      "epoch": 0.4789323915382893,
      "grad_norm": 0.49622663855552673,
      "learning_rate": 1.0661212679041784e-05,
      "loss": 0.1602,
      "step": 26525
    },
    {
      "epoch": 0.47902267097119,
      "grad_norm": 0.7422386407852173,
      "learning_rate": 1.0658382616576128e-05,
      "loss": 0.2099,
      "step": 26530
    },
    {
      "epoch": 0.47911295040409074,
      "grad_norm": 0.42769497632980347,
      "learning_rate": 1.065555250114838e-05,
      "loss": 0.2941,
      "step": 26535
    },
    {
      "epoch": 0.47920322983699143,
      "grad_norm": 0.5021001696586609,
      "learning_rate": 1.0652722332986201e-05,
      "loss": 0.2319,
      "step": 26540
    },
    {
      "epoch": 0.47929350926989217,
      "grad_norm": 0.4900134205818176,
      "learning_rate": 1.0649892112317258e-05,
      "loss": 0.2659,
      "step": 26545
    },
    {
      "epoch": 0.47938378870279286,
      "grad_norm": 0.5937042832374573,
      "learning_rate": 1.0647061839369213e-05,
      "loss": 0.3298,
      "step": 26550
    },
    {
      "epoch": 0.4794740681356936,
      "grad_norm": 0.40473803877830505,
      "learning_rate": 1.0644231514369751e-05,
      "loss": 0.13,
      "step": 26555
    },
    {
      "epoch": 0.47956434756859434,
      "grad_norm": 0.38396719098091125,
      "learning_rate": 1.064140113754655e-05,
      "loss": 0.227,
      "step": 26560
    },
    {
      "epoch": 0.479654627001495,
      "grad_norm": 0.1891910433769226,
      "learning_rate": 1.0638570709127287e-05,
      "loss": 0.2141,
      "step": 26565
    },
    {
      "epoch": 0.47974490643439577,
      "grad_norm": 0.31764113903045654,
      "learning_rate": 1.0635740229339657e-05,
      "loss": 0.2441,
      "step": 26570
    },
    {
      "epoch": 0.47983518586729645,
      "grad_norm": 0.4471171796321869,
      "learning_rate": 1.063290969841135e-05,
      "loss": 0.291,
      "step": 26575
    },
    {
      "epoch": 0.4799254653001972,
      "grad_norm": 0.25125494599342346,
      "learning_rate": 1.063007911657005e-05,
      "loss": 0.2535,
      "step": 26580
    },
    {
      "epoch": 0.4800157447330979,
      "grad_norm": 0.4169851541519165,
      "learning_rate": 1.0627248484043473e-05,
      "loss": 0.333,
      "step": 26585
    },
    {
      "epoch": 0.4801060241659986,
      "grad_norm": 0.5211341381072998,
      "learning_rate": 1.0624417801059317e-05,
      "loss": 0.2166,
      "step": 26590
    },
    {
      "epoch": 0.4801963035988993,
      "grad_norm": 0.5397112965583801,
      "learning_rate": 1.0621587067845285e-05,
      "loss": 0.2862,
      "step": 26595
    },
    {
      "epoch": 0.48028658303180005,
      "grad_norm": 0.9622629284858704,
      "learning_rate": 1.0618756284629095e-05,
      "loss": 0.2888,
      "step": 26600
    },
    {
      "epoch": 0.48037686246470074,
      "grad_norm": 0.5056719779968262,
      "learning_rate": 1.061592545163846e-05,
      "loss": 0.216,
      "step": 26605
    },
    {
      "epoch": 0.4804671418976015,
      "grad_norm": 0.27500560879707336,
      "learning_rate": 1.06130945691011e-05,
      "loss": 0.2227,
      "step": 26610
    },
    {
      "epoch": 0.48055742133050217,
      "grad_norm": 0.5323085784912109,
      "learning_rate": 1.061026363724474e-05,
      "loss": 0.2357,
      "step": 26615
    },
    {
      "epoch": 0.4806477007634029,
      "grad_norm": 0.6595057845115662,
      "learning_rate": 1.0607432656297108e-05,
      "loss": 0.2965,
      "step": 26620
    },
    {
      "epoch": 0.4807379801963036,
      "grad_norm": 0.4934602975845337,
      "learning_rate": 1.0604601626485935e-05,
      "loss": 0.2645,
      "step": 26625
    },
    {
      "epoch": 0.48082825962920434,
      "grad_norm": 0.42322760820388794,
      "learning_rate": 1.0601770548038957e-05,
      "loss": 0.2174,
      "step": 26630
    },
    {
      "epoch": 0.480918539062105,
      "grad_norm": 0.35548311471939087,
      "learning_rate": 1.0598939421183914e-05,
      "loss": 0.2364,
      "step": 26635
    },
    {
      "epoch": 0.48100881849500576,
      "grad_norm": 0.21498507261276245,
      "learning_rate": 1.0596108246148549e-05,
      "loss": 0.2376,
      "step": 26640
    },
    {
      "epoch": 0.48109909792790645,
      "grad_norm": 0.22194571793079376,
      "learning_rate": 1.059327702316061e-05,
      "loss": 0.1536,
      "step": 26645
    },
    {
      "epoch": 0.4811893773608072,
      "grad_norm": 0.3227093815803528,
      "learning_rate": 1.0590445752447847e-05,
      "loss": 0.3052,
      "step": 26650
    },
    {
      "epoch": 0.4812796567937079,
      "grad_norm": 0.45720019936561584,
      "learning_rate": 1.0587614434238017e-05,
      "loss": 0.2323,
      "step": 26655
    },
    {
      "epoch": 0.4813699362266086,
      "grad_norm": 0.6177743077278137,
      "learning_rate": 1.0584783068758878e-05,
      "loss": 0.1444,
      "step": 26660
    },
    {
      "epoch": 0.4814602156595093,
      "grad_norm": 0.6074416637420654,
      "learning_rate": 1.0581951656238194e-05,
      "loss": 0.3255,
      "step": 26665
    },
    {
      "epoch": 0.48155049509241005,
      "grad_norm": 0.9755043983459473,
      "learning_rate": 1.057912019690373e-05,
      "loss": 0.3404,
      "step": 26670
    },
    {
      "epoch": 0.48164077452531073,
      "grad_norm": 0.7641077637672424,
      "learning_rate": 1.0576288690983255e-05,
      "loss": 0.2589,
      "step": 26675
    },
    {
      "epoch": 0.4817310539582115,
      "grad_norm": 0.35916680097579956,
      "learning_rate": 1.0573457138704546e-05,
      "loss": 0.2822,
      "step": 26680
    },
    {
      "epoch": 0.48182133339111216,
      "grad_norm": 0.3647823631763458,
      "learning_rate": 1.057062554029538e-05,
      "loss": 0.1295,
      "step": 26685
    },
    {
      "epoch": 0.4819116128240129,
      "grad_norm": 0.4745776057243347,
      "learning_rate": 1.0567793895983538e-05,
      "loss": 0.2181,
      "step": 26690
    },
    {
      "epoch": 0.4820018922569136,
      "grad_norm": 0.5288276076316833,
      "learning_rate": 1.0564962205996802e-05,
      "loss": 0.277,
      "step": 26695
    },
    {
      "epoch": 0.48209217168981433,
      "grad_norm": 0.49253758788108826,
      "learning_rate": 1.0562130470562968e-05,
      "loss": 0.2034,
      "step": 26700
    },
    {
      "epoch": 0.482182451122715,
      "grad_norm": 0.669064998626709,
      "learning_rate": 1.0559298689909825e-05,
      "loss": 0.238,
      "step": 26705
    },
    {
      "epoch": 0.48227273055561576,
      "grad_norm": 0.6141138672828674,
      "learning_rate": 1.0556466864265167e-05,
      "loss": 0.2832,
      "step": 26710
    },
    {
      "epoch": 0.48236300998851644,
      "grad_norm": 0.3418901264667511,
      "learning_rate": 1.0553634993856796e-05,
      "loss": 0.2015,
      "step": 26715
    },
    {
      "epoch": 0.4824532894214172,
      "grad_norm": 0.5083296895027161,
      "learning_rate": 1.0550803078912517e-05,
      "loss": 0.1523,
      "step": 26720
    },
    {
      "epoch": 0.4825435688543179,
      "grad_norm": 0.4708094894886017,
      "learning_rate": 1.0547971119660135e-05,
      "loss": 0.2269,
      "step": 26725
    },
    {
      "epoch": 0.4826338482872186,
      "grad_norm": 0.500835657119751,
      "learning_rate": 1.0545139116327456e-05,
      "loss": 0.2771,
      "step": 26730
    },
    {
      "epoch": 0.4827241277201193,
      "grad_norm": 0.4330381751060486,
      "learning_rate": 1.0542307069142303e-05,
      "loss": 0.246,
      "step": 26735
    },
    {
      "epoch": 0.48281440715302004,
      "grad_norm": 0.4174950122833252,
      "learning_rate": 1.0539474978332489e-05,
      "loss": 0.2575,
      "step": 26740
    },
    {
      "epoch": 0.48290468658592073,
      "grad_norm": 0.4970250129699707,
      "learning_rate": 1.0536642844125834e-05,
      "loss": 0.1805,
      "step": 26745
    },
    {
      "epoch": 0.48299496601882147,
      "grad_norm": 0.5641002655029297,
      "learning_rate": 1.0533810666750163e-05,
      "loss": 0.2918,
      "step": 26750
    },
    {
      "epoch": 0.48308524545172216,
      "grad_norm": 0.4091084897518158,
      "learning_rate": 1.0530978446433309e-05,
      "loss": 0.2336,
      "step": 26755
    },
    {
      "epoch": 0.4831755248846229,
      "grad_norm": 0.3855745494365692,
      "learning_rate": 1.0528146183403097e-05,
      "loss": 0.2463,
      "step": 26760
    },
    {
      "epoch": 0.4832658043175236,
      "grad_norm": 0.48720014095306396,
      "learning_rate": 1.0525313877887364e-05,
      "loss": 0.2497,
      "step": 26765
    },
    {
      "epoch": 0.4833560837504243,
      "grad_norm": 0.22547881305217743,
      "learning_rate": 1.052248153011395e-05,
      "loss": 0.2767,
      "step": 26770
    },
    {
      "epoch": 0.483446363183325,
      "grad_norm": 0.9942919611930847,
      "learning_rate": 1.0519649140310695e-05,
      "loss": 0.26,
      "step": 26775
    },
    {
      "epoch": 0.48353664261622575,
      "grad_norm": 0.4433083236217499,
      "learning_rate": 1.0516816708705444e-05,
      "loss": 0.3044,
      "step": 26780
    },
    {
      "epoch": 0.48362692204912644,
      "grad_norm": 0.5108844637870789,
      "learning_rate": 1.0513984235526048e-05,
      "loss": 0.3148,
      "step": 26785
    },
    {
      "epoch": 0.4837172014820272,
      "grad_norm": 0.5388258695602417,
      "learning_rate": 1.0511151721000357e-05,
      "loss": 0.3072,
      "step": 26790
    },
    {
      "epoch": 0.48380748091492787,
      "grad_norm": 0.4262354373931885,
      "learning_rate": 1.0508319165356225e-05,
      "loss": 0.2556,
      "step": 26795
    },
    {
      "epoch": 0.4838977603478286,
      "grad_norm": 0.3868439197540283,
      "learning_rate": 1.0505486568821514e-05,
      "loss": 0.237,
      "step": 26800
    },
    {
      "epoch": 0.4839880397807293,
      "grad_norm": 0.47711315751075745,
      "learning_rate": 1.0502653931624083e-05,
      "loss": 0.3017,
      "step": 26805
    },
    {
      "epoch": 0.48407831921363004,
      "grad_norm": 0.5847601294517517,
      "learning_rate": 1.0499821253991797e-05,
      "loss": 0.3238,
      "step": 26810
    },
    {
      "epoch": 0.4841685986465307,
      "grad_norm": 0.3355996012687683,
      "learning_rate": 1.0496988536152526e-05,
      "loss": 0.2366,
      "step": 26815
    },
    {
      "epoch": 0.48425887807943147,
      "grad_norm": 0.4323943555355072,
      "learning_rate": 1.049415577833414e-05,
      "loss": 0.2569,
      "step": 26820
    },
    {
      "epoch": 0.48434915751233215,
      "grad_norm": 0.6366611123085022,
      "learning_rate": 1.0491322980764517e-05,
      "loss": 0.302,
      "step": 26825
    },
    {
      "epoch": 0.4844394369452329,
      "grad_norm": 0.78076171875,
      "learning_rate": 1.048849014367153e-05,
      "loss": 0.2917,
      "step": 26830
    },
    {
      "epoch": 0.4845297163781336,
      "grad_norm": 0.4806523025035858,
      "learning_rate": 1.0485657267283062e-05,
      "loss": 0.2046,
      "step": 26835
    },
    {
      "epoch": 0.4846199958110343,
      "grad_norm": 0.6299446821212769,
      "learning_rate": 1.0482824351827001e-05,
      "loss": 0.2716,
      "step": 26840
    },
    {
      "epoch": 0.484710275243935,
      "grad_norm": 0.469390332698822,
      "learning_rate": 1.047999139753123e-05,
      "loss": 0.1242,
      "step": 26845
    },
    {
      "epoch": 0.48480055467683575,
      "grad_norm": 0.6657130718231201,
      "learning_rate": 1.0477158404623641e-05,
      "loss": 0.2257,
      "step": 26850
    },
    {
      "epoch": 0.48489083410973643,
      "grad_norm": 0.371386855840683,
      "learning_rate": 1.0474325373332129e-05,
      "loss": 0.2508,
      "step": 26855
    },
    {
      "epoch": 0.4849811135426372,
      "grad_norm": 0.6997236609458923,
      "learning_rate": 1.0471492303884585e-05,
      "loss": 0.1855,
      "step": 26860
    },
    {
      "epoch": 0.48507139297553786,
      "grad_norm": 0.4666514992713928,
      "learning_rate": 1.0468659196508916e-05,
      "loss": 0.3018,
      "step": 26865
    },
    {
      "epoch": 0.4851616724084386,
      "grad_norm": 0.37955641746520996,
      "learning_rate": 1.0465826051433023e-05,
      "loss": 0.2873,
      "step": 26870
    },
    {
      "epoch": 0.4852519518413393,
      "grad_norm": 0.36574721336364746,
      "learning_rate": 1.0462992868884808e-05,
      "loss": 0.2312,
      "step": 26875
    },
    {
      "epoch": 0.48534223127424003,
      "grad_norm": 0.440373033285141,
      "learning_rate": 1.0460159649092186e-05,
      "loss": 0.3481,
      "step": 26880
    },
    {
      "epoch": 0.4854325107071407,
      "grad_norm": 0.4811852276325226,
      "learning_rate": 1.0457326392283066e-05,
      "loss": 0.2859,
      "step": 26885
    },
    {
      "epoch": 0.48552279014004146,
      "grad_norm": 0.6694021224975586,
      "learning_rate": 1.0454493098685361e-05,
      "loss": 0.2588,
      "step": 26890
    },
    {
      "epoch": 0.48561306957294215,
      "grad_norm": 0.5687317848205566,
      "learning_rate": 1.045165976852699e-05,
      "loss": 0.2771,
      "step": 26895
    },
    {
      "epoch": 0.4857033490058429,
      "grad_norm": 0.4221053123474121,
      "learning_rate": 1.0448826402035876e-05,
      "loss": 0.2494,
      "step": 26900
    },
    {
      "epoch": 0.4857936284387436,
      "grad_norm": 0.496270090341568,
      "learning_rate": 1.0445992999439942e-05,
      "loss": 0.3064,
      "step": 26905
    },
    {
      "epoch": 0.4858839078716443,
      "grad_norm": 0.30873286724090576,
      "learning_rate": 1.0443159560967112e-05,
      "loss": 0.198,
      "step": 26910
    },
    {
      "epoch": 0.485974187304545,
      "grad_norm": 0.4473693072795868,
      "learning_rate": 1.0440326086845317e-05,
      "loss": 0.2304,
      "step": 26915
    },
    {
      "epoch": 0.48606446673744574,
      "grad_norm": 0.3942261040210724,
      "learning_rate": 1.0437492577302492e-05,
      "loss": 0.2258,
      "step": 26920
    },
    {
      "epoch": 0.48615474617034643,
      "grad_norm": 0.438599169254303,
      "learning_rate": 1.043465903256657e-05,
      "loss": 0.1977,
      "step": 26925
    },
    {
      "epoch": 0.48624502560324717,
      "grad_norm": 0.47521236538887024,
      "learning_rate": 1.0431825452865487e-05,
      "loss": 0.2347,
      "step": 26930
    },
    {
      "epoch": 0.48633530503614786,
      "grad_norm": 0.4236838221549988,
      "learning_rate": 1.0428991838427187e-05,
      "loss": 0.2296,
      "step": 26935
    },
    {
      "epoch": 0.4864255844690486,
      "grad_norm": 0.5367502570152283,
      "learning_rate": 1.0426158189479614e-05,
      "loss": 0.2377,
      "step": 26940
    },
    {
      "epoch": 0.4865158639019493,
      "grad_norm": 0.46861499547958374,
      "learning_rate": 1.0423324506250715e-05,
      "loss": 0.2945,
      "step": 26945
    },
    {
      "epoch": 0.48660614333485,
      "grad_norm": 0.32173192501068115,
      "learning_rate": 1.0420490788968434e-05,
      "loss": 0.2429,
      "step": 26950
    },
    {
      "epoch": 0.48669642276775077,
      "grad_norm": 0.5245801210403442,
      "learning_rate": 1.0417657037860727e-05,
      "loss": 0.3033,
      "step": 26955
    },
    {
      "epoch": 0.48678670220065146,
      "grad_norm": 0.40964144468307495,
      "learning_rate": 1.0414823253155546e-05,
      "loss": 0.1634,
      "step": 26960
    },
    {
      "epoch": 0.4868769816335522,
      "grad_norm": 0.4763515591621399,
      "learning_rate": 1.0411989435080851e-05,
      "loss": 0.2826,
      "step": 26965
    },
    {
      "epoch": 0.4869672610664529,
      "grad_norm": 0.547917366027832,
      "learning_rate": 1.0409155583864604e-05,
      "loss": 0.2297,
      "step": 26970
    },
    {
      "epoch": 0.4870575404993536,
      "grad_norm": 0.24356386065483093,
      "learning_rate": 1.0406321699734761e-05,
      "loss": 0.2226,
      "step": 26975
    },
    {
      "epoch": 0.4871478199322543,
      "grad_norm": 0.8425018191337585,
      "learning_rate": 1.0403487782919295e-05,
      "loss": 0.2663,
      "step": 26980
    },
    {
      "epoch": 0.48723809936515505,
      "grad_norm": 0.35435715317726135,
      "learning_rate": 1.0400653833646171e-05,
      "loss": 0.27,
      "step": 26985
    },
    {
      "epoch": 0.48732837879805574,
      "grad_norm": 0.5056458711624146,
      "learning_rate": 1.0397819852143359e-05,
      "loss": 0.215,
      "step": 26990
    },
    {
      "epoch": 0.4874186582309565,
      "grad_norm": 0.5763304829597473,
      "learning_rate": 1.0394985838638827e-05,
      "loss": 0.2151,
      "step": 26995
    },
    {
      "epoch": 0.48750893766385717,
      "grad_norm": 0.5899175405502319,
      "learning_rate": 1.0392151793360562e-05,
      "loss": 0.2686,
      "step": 27000
    },
    {
      "epoch": 0.4875992170967579,
      "grad_norm": 0.5203468799591064,
      "learning_rate": 1.0389317716536536e-05,
      "loss": 0.2753,
      "step": 27005
    },
    {
      "epoch": 0.4876894965296586,
      "grad_norm": 0.5390440821647644,
      "learning_rate": 1.0386483608394726e-05,
      "loss": 0.2311,
      "step": 27010
    },
    {
      "epoch": 0.48777977596255934,
      "grad_norm": 0.39601603150367737,
      "learning_rate": 1.0383649469163124e-05,
      "loss": 0.2765,
      "step": 27015
    },
    {
      "epoch": 0.48787005539546,
      "grad_norm": 0.28642338514328003,
      "learning_rate": 1.0380815299069709e-05,
      "loss": 0.2079,
      "step": 27020
    },
    {
      "epoch": 0.48796033482836076,
      "grad_norm": 0.42527803778648376,
      "learning_rate": 1.0377981098342473e-05,
      "loss": 0.2355,
      "step": 27025
    },
    {
      "epoch": 0.48805061426126145,
      "grad_norm": 0.41193994879722595,
      "learning_rate": 1.0375146867209408e-05,
      "loss": 0.214,
      "step": 27030
    },
    {
      "epoch": 0.4881408936941622,
      "grad_norm": 0.7089897394180298,
      "learning_rate": 1.0372312605898505e-05,
      "loss": 0.1538,
      "step": 27035
    },
    {
      "epoch": 0.4882311731270629,
      "grad_norm": 0.4400389492511749,
      "learning_rate": 1.0369478314637755e-05,
      "loss": 0.2489,
      "step": 27040
    },
    {
      "epoch": 0.4883214525599636,
      "grad_norm": 0.5277612805366516,
      "learning_rate": 1.0366643993655163e-05,
      "loss": 0.2024,
      "step": 27045
    },
    {
      "epoch": 0.4884117319928643,
      "grad_norm": 0.41362714767456055,
      "learning_rate": 1.0363809643178731e-05,
      "loss": 0.2166,
      "step": 27050
    },
    {
      "epoch": 0.48850201142576505,
      "grad_norm": 0.8280554413795471,
      "learning_rate": 1.0360975263436455e-05,
      "loss": 0.2197,
      "step": 27055
    },
    {
      "epoch": 0.48859229085866573,
      "grad_norm": 0.44255056977272034,
      "learning_rate": 1.0358140854656341e-05,
      "loss": 0.3865,
      "step": 27060
    },
    {
      "epoch": 0.4886825702915665,
      "grad_norm": 0.5058305859565735,
      "learning_rate": 1.0355306417066404e-05,
      "loss": 0.2712,
      "step": 27065
    },
    {
      "epoch": 0.48877284972446716,
      "grad_norm": 0.3069395422935486,
      "learning_rate": 1.0352471950894649e-05,
      "loss": 0.1268,
      "step": 27070
    },
    {
      "epoch": 0.4888631291573679,
      "grad_norm": 0.36307913064956665,
      "learning_rate": 1.0349637456369086e-05,
      "loss": 0.1912,
      "step": 27075
    },
    {
      "epoch": 0.4889534085902686,
      "grad_norm": 0.5039262771606445,
      "learning_rate": 1.0346802933717733e-05,
      "loss": 0.1844,
      "step": 27080
    },
    {
      "epoch": 0.48904368802316933,
      "grad_norm": 0.7958729267120361,
      "learning_rate": 1.0343968383168605e-05,
      "loss": 0.2095,
      "step": 27085
    },
    {
      "epoch": 0.48913396745607,
      "grad_norm": 0.4576447308063507,
      "learning_rate": 1.0341133804949723e-05,
      "loss": 0.2588,
      "step": 27090
    },
    {
      "epoch": 0.48922424688897076,
      "grad_norm": 0.3759396970272064,
      "learning_rate": 1.0338299199289103e-05,
      "loss": 0.2037,
      "step": 27095
    },
    {
      "epoch": 0.48931452632187145,
      "grad_norm": 0.4838416576385498,
      "learning_rate": 1.0335464566414775e-05,
      "loss": 0.2597,
      "step": 27100
    },
    {
      "epoch": 0.4894048057547722,
      "grad_norm": 0.28845927119255066,
      "learning_rate": 1.0332629906554762e-05,
      "loss": 0.2132,
      "step": 27105
    },
    {
      "epoch": 0.4894950851876729,
      "grad_norm": 0.4330255091190338,
      "learning_rate": 1.0329795219937089e-05,
      "loss": 0.2401,
      "step": 27110
    },
    {
      "epoch": 0.4895853646205736,
      "grad_norm": 0.5459669828414917,
      "learning_rate": 1.0326960506789789e-05,
      "loss": 0.3111,
      "step": 27115
    },
    {
      "epoch": 0.4896756440534743,
      "grad_norm": 1.0174330472946167,
      "learning_rate": 1.0324125767340896e-05,
      "loss": 0.2445,
      "step": 27120
    },
    {
      "epoch": 0.48976592348637504,
      "grad_norm": 0.327204167842865,
      "learning_rate": 1.032129100181844e-05,
      "loss": 0.2275,
      "step": 27125
    },
    {
      "epoch": 0.48985620291927573,
      "grad_norm": 0.5526273250579834,
      "learning_rate": 1.0318456210450461e-05,
      "loss": 0.2523,
      "step": 27130
    },
    {
      "epoch": 0.48994648235217647,
      "grad_norm": 0.5036475658416748,
      "learning_rate": 1.0315621393464992e-05,
      "loss": 0.2304,
      "step": 27135
    },
    {
      "epoch": 0.49003676178507716,
      "grad_norm": 0.37863689661026,
      "learning_rate": 1.0312786551090077e-05,
      "loss": 0.2368,
      "step": 27140
    },
    {
      "epoch": 0.4901270412179779,
      "grad_norm": 0.600937008857727,
      "learning_rate": 1.0309951683553758e-05,
      "loss": 0.3067,
      "step": 27145
    },
    {
      "epoch": 0.4902173206508786,
      "grad_norm": 0.6421152353286743,
      "learning_rate": 1.0307116791084082e-05,
      "loss": 0.2312,
      "step": 27150
    },
    {
      "epoch": 0.4903076000837793,
      "grad_norm": 0.6270744800567627,
      "learning_rate": 1.0304281873909093e-05,
      "loss": 0.2451,
      "step": 27155
    },
    {
      "epoch": 0.49039787951668,
      "grad_norm": 0.29649534821510315,
      "learning_rate": 1.0301446932256835e-05,
      "loss": 0.276,
      "step": 27160
    },
    {
      "epoch": 0.49048815894958075,
      "grad_norm": 0.5644583702087402,
      "learning_rate": 1.0298611966355368e-05,
      "loss": 0.2725,
      "step": 27165
    },
    {
      "epoch": 0.49057843838248144,
      "grad_norm": 0.5755888223648071,
      "learning_rate": 1.0295776976432739e-05,
      "loss": 0.2189,
      "step": 27170
    },
    {
      "epoch": 0.4906687178153822,
      "grad_norm": 0.3351040184497833,
      "learning_rate": 1.0292941962717e-05,
      "loss": 0.1996,
      "step": 27175
    },
    {
      "epoch": 0.49075899724828287,
      "grad_norm": 0.34694960713386536,
      "learning_rate": 1.0290106925436212e-05,
      "loss": 0.2004,
      "step": 27180
    },
    {
      "epoch": 0.4908492766811836,
      "grad_norm": 0.5113522410392761,
      "learning_rate": 1.0287271864818431e-05,
      "loss": 0.2258,
      "step": 27185
    },
    {
      "epoch": 0.4909395561140843,
      "grad_norm": 0.3528189957141876,
      "learning_rate": 1.0284436781091718e-05,
      "loss": 0.3493,
      "step": 27190
    },
    {
      "epoch": 0.49102983554698504,
      "grad_norm": 0.3386617600917816,
      "learning_rate": 1.028160167448413e-05,
      "loss": 0.2776,
      "step": 27195
    },
    {
      "epoch": 0.4911201149798857,
      "grad_norm": 0.4666965901851654,
      "learning_rate": 1.0278766545223737e-05,
      "loss": 0.2439,
      "step": 27200
    },
    {
      "epoch": 0.49121039441278647,
      "grad_norm": 0.4585404396057129,
      "learning_rate": 1.0275931393538605e-05,
      "loss": 0.2297,
      "step": 27205
    },
    {
      "epoch": 0.49130067384568715,
      "grad_norm": 0.4618760645389557,
      "learning_rate": 1.0273096219656794e-05,
      "loss": 0.234,
      "step": 27210
    },
    {
      "epoch": 0.4913909532785879,
      "grad_norm": 0.3211655616760254,
      "learning_rate": 1.0270261023806383e-05,
      "loss": 0.2691,
      "step": 27215
    },
    {
      "epoch": 0.4914812327114886,
      "grad_norm": 0.3691023290157318,
      "learning_rate": 1.0267425806215437e-05,
      "loss": 0.2486,
      "step": 27220
    },
    {
      "epoch": 0.4915715121443893,
      "grad_norm": 0.24695229530334473,
      "learning_rate": 1.0264590567112022e-05,
      "loss": 0.3352,
      "step": 27225
    },
    {
      "epoch": 0.49166179157729,
      "grad_norm": 0.4610646069049835,
      "learning_rate": 1.0261755306724226e-05,
      "loss": 0.2407,
      "step": 27230
    },
    {
      "epoch": 0.49175207101019075,
      "grad_norm": 0.5048497319221497,
      "learning_rate": 1.0258920025280119e-05,
      "loss": 0.2619,
      "step": 27235
    },
    {
      "epoch": 0.49184235044309144,
      "grad_norm": 0.4628535509109497,
      "learning_rate": 1.0256084723007775e-05,
      "loss": 0.2581,
      "step": 27240
    },
    {
      "epoch": 0.4919326298759922,
      "grad_norm": 0.3776392340660095,
      "learning_rate": 1.0253249400135279e-05,
      "loss": 0.2212,
      "step": 27245
    },
    {
      "epoch": 0.49202290930889286,
      "grad_norm": 0.36695045232772827,
      "learning_rate": 1.0250414056890711e-05,
      "loss": 0.3446,
      "step": 27250
    },
    {
      "epoch": 0.4921131887417936,
      "grad_norm": 0.47604474425315857,
      "learning_rate": 1.0247578693502152e-05,
      "loss": 0.2258,
      "step": 27255
    },
    {
      "epoch": 0.4922034681746943,
      "grad_norm": 0.510190486907959,
      "learning_rate": 1.0244743310197685e-05,
      "loss": 0.2477,
      "step": 27260
    },
    {
      "epoch": 0.49229374760759503,
      "grad_norm": 0.5116159319877625,
      "learning_rate": 1.0241907907205402e-05,
      "loss": 0.2524,
      "step": 27265
    },
    {
      "epoch": 0.4923840270404957,
      "grad_norm": 0.39241963624954224,
      "learning_rate": 1.0239072484753385e-05,
      "loss": 0.2668,
      "step": 27270
    },
    {
      "epoch": 0.49247430647339646,
      "grad_norm": 0.3967352509498596,
      "learning_rate": 1.0236237043069721e-05,
      "loss": 0.1858,
      "step": 27275
    },
    {
      "epoch": 0.49256458590629715,
      "grad_norm": 0.5310468077659607,
      "learning_rate": 1.023340158238251e-05,
      "loss": 0.2226,
      "step": 27280
    },
    {
      "epoch": 0.4926548653391979,
      "grad_norm": 0.6683197021484375,
      "learning_rate": 1.0230566102919836e-05,
      "loss": 0.2233,
      "step": 27285
    },
    {
      "epoch": 0.4927451447720986,
      "grad_norm": 0.5171959400177002,
      "learning_rate": 1.0227730604909798e-05,
      "loss": 0.2645,
      "step": 27290
    },
    {
      "epoch": 0.4928354242049993,
      "grad_norm": 0.35554239153862,
      "learning_rate": 1.0224895088580484e-05,
      "loss": 0.225,
      "step": 27295
    },
    {
      "epoch": 0.4929257036379,
      "grad_norm": 0.5169106721878052,
      "learning_rate": 1.022205955416e-05,
      "loss": 0.2111,
      "step": 27300
    },
    {
      "epoch": 0.49301598307080075,
      "grad_norm": 0.5154178142547607,
      "learning_rate": 1.0219224001876438e-05,
      "loss": 0.3203,
      "step": 27305
    },
    {
      "epoch": 0.49310626250370143,
      "grad_norm": 0.4990963637828827,
      "learning_rate": 1.0216388431957897e-05,
      "loss": 0.2666,
      "step": 27310
    },
    {
      "epoch": 0.4931965419366022,
      "grad_norm": 0.5841171741485596,
      "learning_rate": 1.0213552844632483e-05,
      "loss": 0.2205,
      "step": 27315
    },
    {
      "epoch": 0.49328682136950286,
      "grad_norm": 0.27949053049087524,
      "learning_rate": 1.0210717240128294e-05,
      "loss": 0.1742,
      "step": 27320
    },
    {
      "epoch": 0.4933771008024036,
      "grad_norm": 0.55351322889328,
      "learning_rate": 1.0207881618673435e-05,
      "loss": 0.2336,
      "step": 27325
    },
    {
      "epoch": 0.4934673802353043,
      "grad_norm": 0.5123849511146545,
      "learning_rate": 1.0205045980496008e-05,
      "loss": 0.2447,
      "step": 27330
    },
    {
      "epoch": 0.49355765966820503,
      "grad_norm": 0.5589601397514343,
      "learning_rate": 1.0202210325824126e-05,
      "loss": 0.2753,
      "step": 27335
    },
    {
      "epoch": 0.4936479391011057,
      "grad_norm": 0.7552227973937988,
      "learning_rate": 1.0199374654885895e-05,
      "loss": 0.2253,
      "step": 27340
    },
    {
      "epoch": 0.49373821853400646,
      "grad_norm": 0.46851953864097595,
      "learning_rate": 1.019653896790942e-05,
      "loss": 0.2223,
      "step": 27345
    },
    {
      "epoch": 0.4938284979669072,
      "grad_norm": 0.3763858675956726,
      "learning_rate": 1.0193703265122814e-05,
      "loss": 0.2983,
      "step": 27350
    },
    {
      "epoch": 0.4939187773998079,
      "grad_norm": 0.5535316467285156,
      "learning_rate": 1.0190867546754191e-05,
      "loss": 0.254,
      "step": 27355
    },
    {
      "epoch": 0.4940090568327086,
      "grad_norm": 0.407514750957489,
      "learning_rate": 1.0188031813031658e-05,
      "loss": 0.194,
      "step": 27360
    },
    {
      "epoch": 0.4940993362656093,
      "grad_norm": 0.6036298274993896,
      "learning_rate": 1.0185196064183337e-05,
      "loss": 0.2785,
      "step": 27365
    },
    {
      "epoch": 0.49418961569851005,
      "grad_norm": 0.2379753738641739,
      "learning_rate": 1.0182360300437338e-05,
      "loss": 0.173,
      "step": 27370
    },
    {
      "epoch": 0.49427989513141074,
      "grad_norm": 0.6542347073554993,
      "learning_rate": 1.0179524522021772e-05,
      "loss": 0.2448,
      "step": 27375
    },
    {
      "epoch": 0.4943701745643115,
      "grad_norm": 0.2842325270175934,
      "learning_rate": 1.0176688729164771e-05,
      "loss": 0.2481,
      "step": 27380
    },
    {
      "epoch": 0.49446045399721217,
      "grad_norm": 0.3398384153842926,
      "learning_rate": 1.0173852922094447e-05,
      "loss": 0.2973,
      "step": 27385
    },
    {
      "epoch": 0.4945507334301129,
      "grad_norm": 0.46946772933006287,
      "learning_rate": 1.0171017101038916e-05,
      "loss": 0.2458,
      "step": 27390
    },
    {
      "epoch": 0.4946410128630136,
      "grad_norm": 0.6143315434455872,
      "learning_rate": 1.0168181266226303e-05,
      "loss": 0.3659,
      "step": 27395
    },
    {
      "epoch": 0.49473129229591434,
      "grad_norm": 0.2996760308742523,
      "learning_rate": 1.0165345417884733e-05,
      "loss": 0.1668,
      "step": 27400
    },
    {
      "epoch": 0.494821571728815,
      "grad_norm": 0.8874998092651367,
      "learning_rate": 1.0162509556242321e-05,
      "loss": 0.3254,
      "step": 27405
    },
    {
      "epoch": 0.49491185116171577,
      "grad_norm": 0.5014644265174866,
      "learning_rate": 1.01596736815272e-05,
      "loss": 0.3182,
      "step": 27410
    },
    {
      "epoch": 0.49500213059461645,
      "grad_norm": 0.33739808201789856,
      "learning_rate": 1.0156837793967494e-05,
      "loss": 0.2717,
      "step": 27415
    },
    {
      "epoch": 0.4950924100275172,
      "grad_norm": 0.38993531465530396,
      "learning_rate": 1.0154001893791325e-05,
      "loss": 0.2172,
      "step": 27420
    },
    {
      "epoch": 0.4951826894604179,
      "grad_norm": 0.444154292345047,
      "learning_rate": 1.0151165981226823e-05,
      "loss": 0.3275,
      "step": 27425
    },
    {
      "epoch": 0.4952729688933186,
      "grad_norm": 0.5773775577545166,
      "learning_rate": 1.0148330056502118e-05,
      "loss": 0.2471,
      "step": 27430
    },
    {
      "epoch": 0.4953632483262193,
      "grad_norm": 0.7944090962409973,
      "learning_rate": 1.014549411984534e-05,
      "loss": 0.2043,
      "step": 27435
    },
    {
      "epoch": 0.49545352775912005,
      "grad_norm": 0.41809704899787903,
      "learning_rate": 1.0142658171484614e-05,
      "loss": 0.2755,
      "step": 27440
    },
    {
      "epoch": 0.49554380719202074,
      "grad_norm": 0.4073884189128876,
      "learning_rate": 1.0139822211648078e-05,
      "loss": 0.1245,
      "step": 27445
    },
    {
      "epoch": 0.4956340866249215,
      "grad_norm": 0.5428813695907593,
      "learning_rate": 1.0136986240563862e-05,
      "loss": 0.3305,
      "step": 27450
    },
    {
      "epoch": 0.49572436605782216,
      "grad_norm": 0.27504950761795044,
      "learning_rate": 1.0134150258460098e-05,
      "loss": 0.2412,
      "step": 27455
    },
    {
      "epoch": 0.4958146454907229,
      "grad_norm": 0.2455800175666809,
      "learning_rate": 1.013131426556492e-05,
      "loss": 0.3058,
      "step": 27460
    },
    {
      "epoch": 0.4959049249236236,
      "grad_norm": 0.3766688108444214,
      "learning_rate": 1.0128478262106468e-05,
      "loss": 0.28,
      "step": 27465
    },
    {
      "epoch": 0.49599520435652433,
      "grad_norm": 0.5050990581512451,
      "learning_rate": 1.012564224831287e-05,
      "loss": 0.2476,
      "step": 27470
    },
    {
      "epoch": 0.496085483789425,
      "grad_norm": 0.4720533788204193,
      "learning_rate": 1.0122806224412267e-05,
      "loss": 0.2766,
      "step": 27475
    },
    {
      "epoch": 0.49617576322232576,
      "grad_norm": 0.27817681431770325,
      "learning_rate": 1.0119970190632797e-05,
      "loss": 0.308,
      "step": 27480
    },
    {
      "epoch": 0.49626604265522645,
      "grad_norm": 0.3395104706287384,
      "learning_rate": 1.0117134147202598e-05,
      "loss": 0.253,
      "step": 27485
    },
    {
      "epoch": 0.4963563220881272,
      "grad_norm": 0.28128769993782043,
      "learning_rate": 1.0114298094349808e-05,
      "loss": 0.2615,
      "step": 27490
    },
    {
      "epoch": 0.4964466015210279,
      "grad_norm": 0.35323697328567505,
      "learning_rate": 1.0111462032302568e-05,
      "loss": 0.2483,
      "step": 27495
    },
    {
      "epoch": 0.4965368809539286,
      "grad_norm": 0.5828583836555481,
      "learning_rate": 1.0108625961289018e-05,
      "loss": 0.2116,
      "step": 27500
    },
    {
      "epoch": 0.4966271603868293,
      "grad_norm": 0.48245134949684143,
      "learning_rate": 1.0105789881537296e-05,
      "loss": 0.2822,
      "step": 27505
    },
    {
      "epoch": 0.49671743981973004,
      "grad_norm": 0.30727630853652954,
      "learning_rate": 1.0102953793275552e-05,
      "loss": 0.2791,
      "step": 27510
    },
    {
      "epoch": 0.49680771925263073,
      "grad_norm": 0.3550146222114563,
      "learning_rate": 1.0100117696731921e-05,
      "loss": 0.2403,
      "step": 27515
    },
    {
      "epoch": 0.4968979986855315,
      "grad_norm": 0.4729647934436798,
      "learning_rate": 1.0097281592134552e-05,
      "loss": 0.2758,
      "step": 27520
    },
    {
      "epoch": 0.49698827811843216,
      "grad_norm": 0.3293568789958954,
      "learning_rate": 1.0094445479711581e-05,
      "loss": 0.3106,
      "step": 27525
    },
    {
      "epoch": 0.4970785575513329,
      "grad_norm": 0.4167075455188751,
      "learning_rate": 1.0091609359691162e-05,
      "loss": 0.281,
      "step": 27530
    },
    {
      "epoch": 0.4971688369842336,
      "grad_norm": 0.5646471977233887,
      "learning_rate": 1.0088773232301434e-05,
      "loss": 0.3009,
      "step": 27535
    },
    {
      "epoch": 0.49725911641713433,
      "grad_norm": 0.2843344211578369,
      "learning_rate": 1.0085937097770544e-05,
      "loss": 0.2627,
      "step": 27540
    },
    {
      "epoch": 0.497349395850035,
      "grad_norm": 0.4866458475589752,
      "learning_rate": 1.0083100956326642e-05,
      "loss": 0.2308,
      "step": 27545
    },
    {
      "epoch": 0.49743967528293576,
      "grad_norm": 0.5150853395462036,
      "learning_rate": 1.0080264808197873e-05,
      "loss": 0.3069,
      "step": 27550
    },
    {
      "epoch": 0.49752995471583644,
      "grad_norm": 0.3755788505077362,
      "learning_rate": 1.007742865361238e-05,
      "loss": 0.117,
      "step": 27555
    },
    {
      "epoch": 0.4976202341487372,
      "grad_norm": 0.41021981835365295,
      "learning_rate": 1.0074592492798317e-05,
      "loss": 0.2048,
      "step": 27560
    },
    {
      "epoch": 0.49771051358163787,
      "grad_norm": 0.3073005974292755,
      "learning_rate": 1.007175632598383e-05,
      "loss": 0.1914,
      "step": 27565
    },
    {
      "epoch": 0.4978007930145386,
      "grad_norm": 0.4285506308078766,
      "learning_rate": 1.006892015339707e-05,
      "loss": 0.1857,
      "step": 27570
    },
    {
      "epoch": 0.4978910724474393,
      "grad_norm": 0.5439529418945312,
      "learning_rate": 1.006608397526618e-05,
      "loss": 0.1857,
      "step": 27575
    },
    {
      "epoch": 0.49798135188034004,
      "grad_norm": 0.6931535005569458,
      "learning_rate": 1.0063247791819319e-05,
      "loss": 0.2686,
      "step": 27580
    },
    {
      "epoch": 0.4980716313132407,
      "grad_norm": 0.5035712122917175,
      "learning_rate": 1.0060411603284634e-05,
      "loss": 0.1862,
      "step": 27585
    },
    {
      "epoch": 0.49816191074614147,
      "grad_norm": 0.5458855628967285,
      "learning_rate": 1.0057575409890272e-05,
      "loss": 0.2431,
      "step": 27590
    },
    {
      "epoch": 0.49825219017904215,
      "grad_norm": 0.3478755056858063,
      "learning_rate": 1.0054739211864387e-05,
      "loss": 0.3048,
      "step": 27595
    },
    {
      "epoch": 0.4983424696119429,
      "grad_norm": 0.4352189600467682,
      "learning_rate": 1.0051903009435133e-05,
      "loss": 0.2288,
      "step": 27600
    },
    {
      "epoch": 0.4984327490448436,
      "grad_norm": 0.4167369306087494,
      "learning_rate": 1.0049066802830655e-05,
      "loss": 0.1723,
      "step": 27605
    },
    {
      "epoch": 0.4985230284777443,
      "grad_norm": 1.0207589864730835,
      "learning_rate": 1.0046230592279112e-05,
      "loss": 0.2857,
      "step": 27610
    },
    {
      "epoch": 0.498613307910645,
      "grad_norm": 0.699300229549408,
      "learning_rate": 1.0043394378008656e-05,
      "loss": 0.2225,
      "step": 27615
    },
    {
      "epoch": 0.49870358734354575,
      "grad_norm": 0.4939732849597931,
      "learning_rate": 1.0040558160247436e-05,
      "loss": 0.162,
      "step": 27620
    },
    {
      "epoch": 0.49879386677644644,
      "grad_norm": 0.34428349137306213,
      "learning_rate": 1.0037721939223603e-05,
      "loss": 0.2868,
      "step": 27625
    },
    {
      "epoch": 0.4988841462093472,
      "grad_norm": 0.4207374155521393,
      "learning_rate": 1.0034885715165318e-05,
      "loss": 0.1792,
      "step": 27630
    },
    {
      "epoch": 0.49897442564224787,
      "grad_norm": 0.41916459798812866,
      "learning_rate": 1.003204948830073e-05,
      "loss": 0.309,
      "step": 27635
    },
    {
      "epoch": 0.4990647050751486,
      "grad_norm": 0.4053615927696228,
      "learning_rate": 1.0029213258857993e-05,
      "loss": 0.2872,
      "step": 27640
    },
    {
      "epoch": 0.4991549845080493,
      "grad_norm": 0.37571027874946594,
      "learning_rate": 1.0026377027065264e-05,
      "loss": 0.195,
      "step": 27645
    },
    {
      "epoch": 0.49924526394095003,
      "grad_norm": 0.46607398986816406,
      "learning_rate": 1.0023540793150694e-05,
      "loss": 0.1856,
      "step": 27650
    },
    {
      "epoch": 0.4993355433738507,
      "grad_norm": 0.6066296696662903,
      "learning_rate": 1.0020704557342438e-05,
      "loss": 0.2583,
      "step": 27655
    },
    {
      "epoch": 0.49942582280675146,
      "grad_norm": 0.3336362838745117,
      "learning_rate": 1.001786831986865e-05,
      "loss": 0.1857,
      "step": 27660
    },
    {
      "epoch": 0.49951610223965215,
      "grad_norm": 0.5273229479789734,
      "learning_rate": 1.0015032080957487e-05,
      "loss": 0.2838,
      "step": 27665
    },
    {
      "epoch": 0.4996063816725529,
      "grad_norm": 0.4261868894100189,
      "learning_rate": 1.0012195840837103e-05,
      "loss": 0.2801,
      "step": 27670
    },
    {
      "epoch": 0.4996966611054536,
      "grad_norm": 0.6468798518180847,
      "learning_rate": 1.0009359599735653e-05,
      "loss": 0.3093,
      "step": 27675
    },
    {
      "epoch": 0.4997869405383543,
      "grad_norm": 0.6814336776733398,
      "learning_rate": 1.0006523357881288e-05,
      "loss": 0.3179,
      "step": 27680
    },
    {
      "epoch": 0.499877219971255,
      "grad_norm": 0.4324711263179779,
      "learning_rate": 1.000368711550217e-05,
      "loss": 0.2867,
      "step": 27685
    },
    {
      "epoch": 0.49996749940415575,
      "grad_norm": 0.31334424018859863,
      "learning_rate": 1.0000850872826447e-05,
      "loss": 0.2088,
      "step": 27690
    },
    {
      "epoch": 0.5000577788370565,
      "grad_norm": 0.3834947347640991,
      "learning_rate": 9.99801463008228e-06,
      "loss": 0.3434,
      "step": 27695
    },
    {
      "epoch": 0.5001480582699571,
      "grad_norm": 0.2797144651412964,
      "learning_rate": 9.995178387497817e-06,
      "loss": 0.2136,
      "step": 27700
    },
    {
      "epoch": 0.5002383377028579,
      "grad_norm": 0.48420146107673645,
      "learning_rate": 9.992342145301222e-06,
      "loss": 0.3488,
      "step": 27705
    },
    {
      "epoch": 0.5003286171357586,
      "grad_norm": 0.3981853425502777,
      "learning_rate": 9.989505903720643e-06,
      "loss": 0.2418,
      "step": 27710
    },
    {
      "epoch": 0.5004188965686593,
      "grad_norm": 0.43650510907173157,
      "learning_rate": 9.986669662984242e-06,
      "loss": 0.2245,
      "step": 27715
    },
    {
      "epoch": 0.50050917600156,
      "grad_norm": 0.42851582169532776,
      "learning_rate": 9.983833423320166e-06,
      "loss": 0.2556,
      "step": 27720
    },
    {
      "epoch": 0.5005994554344607,
      "grad_norm": 0.9285907745361328,
      "learning_rate": 9.980997184956574e-06,
      "loss": 0.2418,
      "step": 27725
    },
    {
      "epoch": 0.5006897348673615,
      "grad_norm": 0.2296239137649536,
      "learning_rate": 9.978160948121623e-06,
      "loss": 0.2344,
      "step": 27730
    },
    {
      "epoch": 0.5007800143002622,
      "grad_norm": 0.5208185315132141,
      "learning_rate": 9.975324713043461e-06,
      "loss": 0.2269,
      "step": 27735
    },
    {
      "epoch": 0.5008702937331629,
      "grad_norm": 0.4143800139427185,
      "learning_rate": 9.972488479950249e-06,
      "loss": 0.2573,
      "step": 27740
    },
    {
      "epoch": 0.5009605731660636,
      "grad_norm": 0.3906264901161194,
      "learning_rate": 9.969652249070137e-06,
      "loss": 0.1876,
      "step": 27745
    },
    {
      "epoch": 0.5010508525989643,
      "grad_norm": 0.6202120184898376,
      "learning_rate": 9.966816020631284e-06,
      "loss": 0.2853,
      "step": 27750
    },
    {
      "epoch": 0.501141132031865,
      "grad_norm": 0.5595226883888245,
      "learning_rate": 9.963979794861836e-06,
      "loss": 0.2391,
      "step": 27755
    },
    {
      "epoch": 0.5012314114647658,
      "grad_norm": 0.3227062523365021,
      "learning_rate": 9.961143571989958e-06,
      "loss": 0.229,
      "step": 27760
    },
    {
      "epoch": 0.5013216908976664,
      "grad_norm": 0.565716028213501,
      "learning_rate": 9.958307352243791e-06,
      "loss": 0.2199,
      "step": 27765
    },
    {
      "epoch": 0.5014119703305672,
      "grad_norm": 0.5395566821098328,
      "learning_rate": 9.955471135851498e-06,
      "loss": 0.2633,
      "step": 27770
    },
    {
      "epoch": 0.5015022497634679,
      "grad_norm": 0.39155447483062744,
      "learning_rate": 9.952634923041225e-06,
      "loss": 0.2835,
      "step": 27775
    },
    {
      "epoch": 0.5015925291963687,
      "grad_norm": 0.5883301496505737,
      "learning_rate": 9.949798714041133e-06,
      "loss": 0.2501,
      "step": 27780
    },
    {
      "epoch": 0.5016828086292693,
      "grad_norm": 0.7940782308578491,
      "learning_rate": 9.946962509079365e-06,
      "loss": 0.2023,
      "step": 27785
    },
    {
      "epoch": 0.50177308806217,
      "grad_norm": 0.39916032552719116,
      "learning_rate": 9.94412630838408e-06,
      "loss": 0.2586,
      "step": 27790
    },
    {
      "epoch": 0.5018633674950708,
      "grad_norm": 0.5648788213729858,
      "learning_rate": 9.941290112183426e-06,
      "loss": 0.2484,
      "step": 27795
    },
    {
      "epoch": 0.5019536469279715,
      "grad_norm": 0.41418156027793884,
      "learning_rate": 9.938453920705558e-06,
      "loss": 0.242,
      "step": 27800
    },
    {
      "epoch": 0.5020439263608721,
      "grad_norm": 0.6468842625617981,
      "learning_rate": 9.935617734178623e-06,
      "loss": 0.2861,
      "step": 27805
    },
    {
      "epoch": 0.5021342057937729,
      "grad_norm": 0.4214017689228058,
      "learning_rate": 9.932781552830773e-06,
      "loss": 0.2731,
      "step": 27810
    },
    {
      "epoch": 0.5022244852266736,
      "grad_norm": 0.4069287180900574,
      "learning_rate": 9.92994537689016e-06,
      "loss": 0.2039,
      "step": 27815
    },
    {
      "epoch": 0.5023147646595744,
      "grad_norm": 0.20495177805423737,
      "learning_rate": 9.927109206584927e-06,
      "loss": 0.3151,
      "step": 27820
    },
    {
      "epoch": 0.502405044092475,
      "grad_norm": 0.5869178771972656,
      "learning_rate": 9.924273042143234e-06,
      "loss": 0.195,
      "step": 27825
    },
    {
      "epoch": 0.5024953235253757,
      "grad_norm": 0.4586544632911682,
      "learning_rate": 9.921436883793221e-06,
      "loss": 0.246,
      "step": 27830
    },
    {
      "epoch": 0.5025856029582765,
      "grad_norm": 0.3754526674747467,
      "learning_rate": 9.91860073176304e-06,
      "loss": 0.2455,
      "step": 27835
    },
    {
      "epoch": 0.5026758823911772,
      "grad_norm": 0.3354601562023163,
      "learning_rate": 9.915764586280838e-06,
      "loss": 0.2015,
      "step": 27840
    },
    {
      "epoch": 0.5027661618240779,
      "grad_norm": 0.49308615922927856,
      "learning_rate": 9.912928447574765e-06,
      "loss": 0.1974,
      "step": 27845
    },
    {
      "epoch": 0.5028564412569786,
      "grad_norm": 0.44969847798347473,
      "learning_rate": 9.910092315872964e-06,
      "loss": 0.2235,
      "step": 27850
    },
    {
      "epoch": 0.5029467206898793,
      "grad_norm": 0.46338215470314026,
      "learning_rate": 9.907256191403582e-06,
      "loss": 0.3144,
      "step": 27855
    },
    {
      "epoch": 0.5030370001227801,
      "grad_norm": 0.4935147166252136,
      "learning_rate": 9.904420074394762e-06,
      "loss": 0.2248,
      "step": 27860
    },
    {
      "epoch": 0.5031272795556807,
      "grad_norm": 0.5603335499763489,
      "learning_rate": 9.90158396507466e-06,
      "loss": 0.2033,
      "step": 27865
    },
    {
      "epoch": 0.5032175589885814,
      "grad_norm": 0.5353633165359497,
      "learning_rate": 9.898747863671404e-06,
      "loss": 0.296,
      "step": 27870
    },
    {
      "epoch": 0.5033078384214822,
      "grad_norm": 0.3191150724887848,
      "learning_rate": 9.895911770413152e-06,
      "loss": 0.2318,
      "step": 27875
    },
    {
      "epoch": 0.5033981178543829,
      "grad_norm": 0.6160486936569214,
      "learning_rate": 9.893075685528037e-06,
      "loss": 0.2417,
      "step": 27880
    },
    {
      "epoch": 0.5034883972872836,
      "grad_norm": 0.6783169507980347,
      "learning_rate": 9.89023960924421e-06,
      "loss": 0.2349,
      "step": 27885
    },
    {
      "epoch": 0.5035786767201843,
      "grad_norm": 0.6717947721481323,
      "learning_rate": 9.887403541789804e-06,
      "loss": 0.2927,
      "step": 27890
    },
    {
      "epoch": 0.503668956153085,
      "grad_norm": 0.36918288469314575,
      "learning_rate": 9.88456748339297e-06,
      "loss": 0.3937,
      "step": 27895
    },
    {
      "epoch": 0.5037592355859858,
      "grad_norm": 0.4017901122570038,
      "learning_rate": 9.881731434281838e-06,
      "loss": 0.1621,
      "step": 27900
    },
    {
      "epoch": 0.5038495150188864,
      "grad_norm": 0.5388122797012329,
      "learning_rate": 9.878895394684554e-06,
      "loss": 0.3263,
      "step": 27905
    },
    {
      "epoch": 0.5039397944517872,
      "grad_norm": 0.5903985500335693,
      "learning_rate": 9.876059364829257e-06,
      "loss": 0.2077,
      "step": 27910
    },
    {
      "epoch": 0.5040300738846879,
      "grad_norm": 0.36499443650245667,
      "learning_rate": 9.873223344944078e-06,
      "loss": 0.1709,
      "step": 27915
    },
    {
      "epoch": 0.5041203533175886,
      "grad_norm": 0.5463023781776428,
      "learning_rate": 9.870387335257162e-06,
      "loss": 0.2483,
      "step": 27920
    },
    {
      "epoch": 0.5042106327504893,
      "grad_norm": 0.4573571979999542,
      "learning_rate": 9.867551335996641e-06,
      "loss": 0.2362,
      "step": 27925
    },
    {
      "epoch": 0.50430091218339,
      "grad_norm": 0.3257332444190979,
      "learning_rate": 9.864715347390656e-06,
      "loss": 0.2524,
      "step": 27930
    },
    {
      "epoch": 0.5043911916162908,
      "grad_norm": 0.45857176184654236,
      "learning_rate": 9.861879369667334e-06,
      "loss": 0.2533,
      "step": 27935
    },
    {
      "epoch": 0.5044814710491915,
      "grad_norm": 0.4118990898132324,
      "learning_rate": 9.859043403054813e-06,
      "loss": 0.2963,
      "step": 27940
    },
    {
      "epoch": 0.5045717504820921,
      "grad_norm": 0.3844326436519623,
      "learning_rate": 9.856207447781225e-06,
      "loss": 0.2782,
      "step": 27945
    },
    {
      "epoch": 0.5046620299149929,
      "grad_norm": 0.6469088792800903,
      "learning_rate": 9.853371504074703e-06,
      "loss": 0.2838,
      "step": 27950
    },
    {
      "epoch": 0.5047523093478936,
      "grad_norm": 0.3941665291786194,
      "learning_rate": 9.850535572163377e-06,
      "loss": 0.2362,
      "step": 27955
    },
    {
      "epoch": 0.5048425887807944,
      "grad_norm": 0.4124378263950348,
      "learning_rate": 9.84769965227538e-06,
      "loss": 0.3276,
      "step": 27960
    },
    {
      "epoch": 0.504932868213695,
      "grad_norm": 0.6127313375473022,
      "learning_rate": 9.844863744638836e-06,
      "loss": 0.2372,
      "step": 27965
    },
    {
      "epoch": 0.5050231476465957,
      "grad_norm": 0.522842288017273,
      "learning_rate": 9.842027849481876e-06,
      "loss": 0.2822,
      "step": 27970
    },
    {
      "epoch": 0.5051134270794965,
      "grad_norm": 0.34865602850914,
      "learning_rate": 9.839191967032626e-06,
      "loss": 0.2357,
      "step": 27975
    },
    {
      "epoch": 0.5052037065123972,
      "grad_norm": 0.3709487020969391,
      "learning_rate": 9.836356097519217e-06,
      "loss": 0.1587,
      "step": 27980
    },
    {
      "epoch": 0.5052939859452978,
      "grad_norm": 0.26086875796318054,
      "learning_rate": 9.833520241169767e-06,
      "loss": 0.2099,
      "step": 27985
    },
    {
      "epoch": 0.5053842653781986,
      "grad_norm": 0.40823107957839966,
      "learning_rate": 9.830684398212403e-06,
      "loss": 0.2308,
      "step": 27990
    },
    {
      "epoch": 0.5054745448110993,
      "grad_norm": 0.47882601618766785,
      "learning_rate": 9.827848568875248e-06,
      "loss": 0.2709,
      "step": 27995
    },
    {
      "epoch": 0.5055648242440001,
      "grad_norm": 0.45627081394195557,
      "learning_rate": 9.825012753386427e-06,
      "loss": 0.2869,
      "step": 28000
    },
    {
      "epoch": 0.5056551036769007,
      "grad_norm": 0.8625367879867554,
      "learning_rate": 9.822176951974058e-06,
      "loss": 0.2206,
      "step": 28005
    },
    {
      "epoch": 0.5057453831098014,
      "grad_norm": 0.7245298027992249,
      "learning_rate": 9.819341164866255e-06,
      "loss": 0.2835,
      "step": 28010
    },
    {
      "epoch": 0.5058356625427022,
      "grad_norm": 0.3014034032821655,
      "learning_rate": 9.816505392291146e-06,
      "loss": 0.263,
      "step": 28015
    },
    {
      "epoch": 0.5059259419756029,
      "grad_norm": 0.40651869773864746,
      "learning_rate": 9.813669634476842e-06,
      "loss": 0.2666,
      "step": 28020
    },
    {
      "epoch": 0.5060162214085036,
      "grad_norm": 0.6607047319412231,
      "learning_rate": 9.810833891651463e-06,
      "loss": 0.277,
      "step": 28025
    },
    {
      "epoch": 0.5061065008414043,
      "grad_norm": 0.5369471311569214,
      "learning_rate": 9.80799816404312e-06,
      "loss": 0.1842,
      "step": 28030
    },
    {
      "epoch": 0.506196780274305,
      "grad_norm": 0.5690027475357056,
      "learning_rate": 9.805162451879931e-06,
      "loss": 0.2852,
      "step": 28035
    },
    {
      "epoch": 0.5062870597072058,
      "grad_norm": 0.5109096765518188,
      "learning_rate": 9.802326755390002e-06,
      "loss": 0.2249,
      "step": 28040
    },
    {
      "epoch": 0.5063773391401064,
      "grad_norm": 0.2942725419998169,
      "learning_rate": 9.799491074801454e-06,
      "loss": 0.2417,
      "step": 28045
    },
    {
      "epoch": 0.5064676185730072,
      "grad_norm": 0.667260468006134,
      "learning_rate": 9.796655410342386e-06,
      "loss": 0.2583,
      "step": 28050
    },
    {
      "epoch": 0.5065578980059079,
      "grad_norm": 0.37191522121429443,
      "learning_rate": 9.793819762240915e-06,
      "loss": 0.2346,
      "step": 28055
    },
    {
      "epoch": 0.5066481774388086,
      "grad_norm": 0.5072374939918518,
      "learning_rate": 9.790984130725144e-06,
      "loss": 0.2721,
      "step": 28060
    },
    {
      "epoch": 0.5067384568717093,
      "grad_norm": 0.4050954282283783,
      "learning_rate": 9.788148516023181e-06,
      "loss": 0.1833,
      "step": 28065
    },
    {
      "epoch": 0.50682873630461,
      "grad_norm": 0.5642181038856506,
      "learning_rate": 9.785312918363129e-06,
      "loss": 0.1655,
      "step": 28070
    },
    {
      "epoch": 0.5069190157375107,
      "grad_norm": 0.7335848808288574,
      "learning_rate": 9.78247733797309e-06,
      "loss": 0.2526,
      "step": 28075
    },
    {
      "epoch": 0.5070092951704115,
      "grad_norm": 0.5068244934082031,
      "learning_rate": 9.779641775081169e-06,
      "loss": 0.1532,
      "step": 28080
    },
    {
      "epoch": 0.5070995746033121,
      "grad_norm": 0.5320443511009216,
      "learning_rate": 9.776806229915466e-06,
      "loss": 0.1299,
      "step": 28085
    },
    {
      "epoch": 0.5071898540362129,
      "grad_norm": 0.7844002842903137,
      "learning_rate": 9.773970702704079e-06,
      "loss": 0.2342,
      "step": 28090
    },
    {
      "epoch": 0.5072801334691136,
      "grad_norm": 0.41133514046669006,
      "learning_rate": 9.771135193675104e-06,
      "loss": 0.2925,
      "step": 28095
    },
    {
      "epoch": 0.5073704129020143,
      "grad_norm": 0.359630286693573,
      "learning_rate": 9.76829970305664e-06,
      "loss": 0.1802,
      "step": 28100
    },
    {
      "epoch": 0.507460692334915,
      "grad_norm": 0.6197757720947266,
      "learning_rate": 9.765464231076778e-06,
      "loss": 0.3045,
      "step": 28105
    },
    {
      "epoch": 0.5075509717678157,
      "grad_norm": 0.5487251877784729,
      "learning_rate": 9.762628777963618e-06,
      "loss": 0.3134,
      "step": 28110
    },
    {
      "epoch": 0.5076412512007165,
      "grad_norm": 0.40097513794898987,
      "learning_rate": 9.759793343945242e-06,
      "loss": 0.2812,
      "step": 28115
    },
    {
      "epoch": 0.5077315306336172,
      "grad_norm": 0.41617217659950256,
      "learning_rate": 9.756957929249747e-06,
      "loss": 0.389,
      "step": 28120
    },
    {
      "epoch": 0.5078218100665178,
      "grad_norm": 0.3712722361087799,
      "learning_rate": 9.75412253410522e-06,
      "loss": 0.2858,
      "step": 28125
    },
    {
      "epoch": 0.5079120894994186,
      "grad_norm": 0.520553469657898,
      "learning_rate": 9.751287158739748e-06,
      "loss": 0.2226,
      "step": 28130
    },
    {
      "epoch": 0.5080023689323193,
      "grad_norm": 0.2953408658504486,
      "learning_rate": 9.748451803381414e-06,
      "loss": 0.2733,
      "step": 28135
    },
    {
      "epoch": 0.5080926483652201,
      "grad_norm": 0.4719192683696747,
      "learning_rate": 9.745616468258305e-06,
      "loss": 0.1934,
      "step": 28140
    },
    {
      "epoch": 0.5081829277981208,
      "grad_norm": 0.5132400989532471,
      "learning_rate": 9.742781153598499e-06,
      "loss": 0.2645,
      "step": 28145
    },
    {
      "epoch": 0.5082732072310214,
      "grad_norm": 0.4969528019428253,
      "learning_rate": 9.739945859630082e-06,
      "loss": 0.2432,
      "step": 28150
    },
    {
      "epoch": 0.5083634866639222,
      "grad_norm": 0.42903774976730347,
      "learning_rate": 9.737110586581125e-06,
      "loss": 0.1691,
      "step": 28155
    },
    {
      "epoch": 0.5084537660968229,
      "grad_norm": 0.4738997519016266,
      "learning_rate": 9.734275334679717e-06,
      "loss": 0.2364,
      "step": 28160
    },
    {
      "epoch": 0.5085440455297237,
      "grad_norm": 0.4509764611721039,
      "learning_rate": 9.731440104153921e-06,
      "loss": 0.252,
      "step": 28165
    },
    {
      "epoch": 0.5086343249626243,
      "grad_norm": 0.7997400164604187,
      "learning_rate": 9.728604895231819e-06,
      "loss": 0.2634,
      "step": 28170
    },
    {
      "epoch": 0.508724604395525,
      "grad_norm": 0.5595467686653137,
      "learning_rate": 9.725769708141476e-06,
      "loss": 0.3151,
      "step": 28175
    },
    {
      "epoch": 0.5088148838284258,
      "grad_norm": 0.6390929222106934,
      "learning_rate": 9.722934543110972e-06,
      "loss": 0.2556,
      "step": 28180
    },
    {
      "epoch": 0.5089051632613265,
      "grad_norm": 0.5860655903816223,
      "learning_rate": 9.720099400368369e-06,
      "loss": 0.2759,
      "step": 28185
    },
    {
      "epoch": 0.5089954426942271,
      "grad_norm": 0.4354051649570465,
      "learning_rate": 9.71726428014173e-06,
      "loss": 0.3003,
      "step": 28190
    },
    {
      "epoch": 0.5090857221271279,
      "grad_norm": 0.41822466254234314,
      "learning_rate": 9.714429182659127e-06,
      "loss": 0.2826,
      "step": 28195
    },
    {
      "epoch": 0.5091760015600286,
      "grad_norm": 0.4065977931022644,
      "learning_rate": 9.711594108148618e-06,
      "loss": 0.346,
      "step": 28200
    },
    {
      "epoch": 0.5092662809929294,
      "grad_norm": 0.9190388321876526,
      "learning_rate": 9.708759056838266e-06,
      "loss": 0.2788,
      "step": 28205
    },
    {
      "epoch": 0.50935656042583,
      "grad_norm": 0.689323902130127,
      "learning_rate": 9.70592402895613e-06,
      "loss": 0.2443,
      "step": 28210
    },
    {
      "epoch": 0.5094468398587307,
      "grad_norm": 0.24661704897880554,
      "learning_rate": 9.703089024730269e-06,
      "loss": 0.2348,
      "step": 28215
    },
    {
      "epoch": 0.5095371192916315,
      "grad_norm": 0.7008240818977356,
      "learning_rate": 9.700254044388735e-06,
      "loss": 0.2048,
      "step": 28220
    },
    {
      "epoch": 0.5096273987245322,
      "grad_norm": 0.3344728350639343,
      "learning_rate": 9.697419088159589e-06,
      "loss": 0.2509,
      "step": 28225
    },
    {
      "epoch": 0.5097176781574329,
      "grad_norm": 0.4515267610549927,
      "learning_rate": 9.694584156270872e-06,
      "loss": 0.2501,
      "step": 28230
    },
    {
      "epoch": 0.5098079575903336,
      "grad_norm": 0.46323707699775696,
      "learning_rate": 9.691749248950642e-06,
      "loss": 0.2469,
      "step": 28235
    },
    {
      "epoch": 0.5098982370232343,
      "grad_norm": 0.5240350365638733,
      "learning_rate": 9.688914366426942e-06,
      "loss": 0.3135,
      "step": 28240
    },
    {
      "epoch": 0.5099885164561351,
      "grad_norm": 0.47504279017448425,
      "learning_rate": 9.686079508927823e-06,
      "loss": 0.224,
      "step": 28245
    },
    {
      "epoch": 0.5100787958890357,
      "grad_norm": 0.2030981481075287,
      "learning_rate": 9.683244676681321e-06,
      "loss": 0.2984,
      "step": 28250
    },
    {
      "epoch": 0.5101690753219364,
      "grad_norm": 0.41279366612434387,
      "learning_rate": 9.680409869915482e-06,
      "loss": 0.2637,
      "step": 28255
    },
    {
      "epoch": 0.5102593547548372,
      "grad_norm": 0.7960222959518433,
      "learning_rate": 9.677575088858346e-06,
      "loss": 0.3638,
      "step": 28260
    },
    {
      "epoch": 0.5103496341877379,
      "grad_norm": 0.5688741207122803,
      "learning_rate": 9.674740333737952e-06,
      "loss": 0.2385,
      "step": 28265
    },
    {
      "epoch": 0.5104399136206386,
      "grad_norm": 0.2537517547607422,
      "learning_rate": 9.671905604782327e-06,
      "loss": 0.2378,
      "step": 28270
    },
    {
      "epoch": 0.5105301930535393,
      "grad_norm": 0.549437403678894,
      "learning_rate": 9.669070902219519e-06,
      "loss": 0.2271,
      "step": 28275
    },
    {
      "epoch": 0.51062047248644,
      "grad_norm": 0.6095982193946838,
      "learning_rate": 9.666236226277548e-06,
      "loss": 0.3151,
      "step": 28280
    },
    {
      "epoch": 0.5107107519193408,
      "grad_norm": 0.3918159008026123,
      "learning_rate": 9.663401577184443e-06,
      "loss": 0.1602,
      "step": 28285
    },
    {
      "epoch": 0.5108010313522414,
      "grad_norm": 0.6957646012306213,
      "learning_rate": 9.660566955168239e-06,
      "loss": 0.398,
      "step": 28290
    },
    {
      "epoch": 0.5108913107851422,
      "grad_norm": 0.4322679936885834,
      "learning_rate": 9.657732360456951e-06,
      "loss": 0.2227,
      "step": 28295
    },
    {
      "epoch": 0.5109815902180429,
      "grad_norm": 0.5070502758026123,
      "learning_rate": 9.654897793278607e-06,
      "loss": 0.1661,
      "step": 28300
    },
    {
      "epoch": 0.5110718696509436,
      "grad_norm": 0.2714361846446991,
      "learning_rate": 9.652063253861227e-06,
      "loss": 0.27,
      "step": 28305
    },
    {
      "epoch": 0.5111621490838443,
      "grad_norm": 0.4892715811729431,
      "learning_rate": 9.649228742432829e-06,
      "loss": 0.2854,
      "step": 28310
    },
    {
      "epoch": 0.511252428516745,
      "grad_norm": 0.44520044326782227,
      "learning_rate": 9.646394259221426e-06,
      "loss": 0.2486,
      "step": 28315
    },
    {
      "epoch": 0.5113427079496458,
      "grad_norm": 0.2912124693393707,
      "learning_rate": 9.643559804455037e-06,
      "loss": 0.2223,
      "step": 28320
    },
    {
      "epoch": 0.5114329873825465,
      "grad_norm": 0.2975776493549347,
      "learning_rate": 9.640725378361666e-06,
      "loss": 0.1979,
      "step": 28325
    },
    {
      "epoch": 0.5115232668154471,
      "grad_norm": 0.2630120813846588,
      "learning_rate": 9.637890981169331e-06,
      "loss": 0.3285,
      "step": 28330
    },
    {
      "epoch": 0.5116135462483479,
      "grad_norm": 0.3051246106624603,
      "learning_rate": 9.635056613106029e-06,
      "loss": 0.2442,
      "step": 28335
    },
    {
      "epoch": 0.5117038256812486,
      "grad_norm": 0.4141266942024231,
      "learning_rate": 9.632222274399773e-06,
      "loss": 0.2679,
      "step": 28340
    },
    {
      "epoch": 0.5117941051141494,
      "grad_norm": 0.44740721583366394,
      "learning_rate": 9.629387965278558e-06,
      "loss": 0.2184,
      "step": 28345
    },
    {
      "epoch": 0.51188438454705,
      "grad_norm": 0.31492388248443604,
      "learning_rate": 9.62655368597039e-06,
      "loss": 0.2088,
      "step": 28350
    },
    {
      "epoch": 0.5119746639799507,
      "grad_norm": 0.38207197189331055,
      "learning_rate": 9.623719436703258e-06,
      "loss": 0.2585,
      "step": 28355
    },
    {
      "epoch": 0.5120649434128515,
      "grad_norm": 0.4363836944103241,
      "learning_rate": 9.620885217705167e-06,
      "loss": 0.2704,
      "step": 28360
    },
    {
      "epoch": 0.5121552228457522,
      "grad_norm": 0.4479374289512634,
      "learning_rate": 9.618051029204101e-06,
      "loss": 0.2479,
      "step": 28365
    },
    {
      "epoch": 0.5122455022786528,
      "grad_norm": 0.2569049596786499,
      "learning_rate": 9.615216871428053e-06,
      "loss": 0.187,
      "step": 28370
    },
    {
      "epoch": 0.5123357817115536,
      "grad_norm": 0.3158555030822754,
      "learning_rate": 9.612382744605012e-06,
      "loss": 0.2949,
      "step": 28375
    },
    {
      "epoch": 0.5124260611444543,
      "grad_norm": 0.37251681089401245,
      "learning_rate": 9.609548648962957e-06,
      "loss": 0.2353,
      "step": 28380
    },
    {
      "epoch": 0.5125163405773551,
      "grad_norm": 0.4977036416530609,
      "learning_rate": 9.606714584729877e-06,
      "loss": 0.2804,
      "step": 28385
    },
    {
      "epoch": 0.5126066200102557,
      "grad_norm": 0.6994753479957581,
      "learning_rate": 9.603880552133746e-06,
      "loss": 0.2686,
      "step": 28390
    },
    {
      "epoch": 0.5126968994431564,
      "grad_norm": 0.5312997698783875,
      "learning_rate": 9.60104655140255e-06,
      "loss": 0.2027,
      "step": 28395
    },
    {
      "epoch": 0.5127871788760572,
      "grad_norm": 0.33299341797828674,
      "learning_rate": 9.598212582764253e-06,
      "loss": 0.2144,
      "step": 28400
    },
    {
      "epoch": 0.5128774583089579,
      "grad_norm": 0.3861413300037384,
      "learning_rate": 9.595378646446837e-06,
      "loss": 0.2526,
      "step": 28405
    },
    {
      "epoch": 0.5129677377418586,
      "grad_norm": 0.4368645250797272,
      "learning_rate": 9.592544742678263e-06,
      "loss": 0.3002,
      "step": 28410
    },
    {
      "epoch": 0.5130580171747593,
      "grad_norm": 0.49991172552108765,
      "learning_rate": 9.589710871686503e-06,
      "loss": 0.2944,
      "step": 28415
    },
    {
      "epoch": 0.51314829660766,
      "grad_norm": 0.6610318422317505,
      "learning_rate": 9.58687703369952e-06,
      "loss": 0.2078,
      "step": 28420
    },
    {
      "epoch": 0.5132385760405608,
      "grad_norm": 0.35077694058418274,
      "learning_rate": 9.584043228945278e-06,
      "loss": 0.2142,
      "step": 28425
    },
    {
      "epoch": 0.5133288554734614,
      "grad_norm": 0.5012067556381226,
      "learning_rate": 9.581209457651732e-06,
      "loss": 0.2189,
      "step": 28430
    },
    {
      "epoch": 0.5134191349063622,
      "grad_norm": 0.435333788394928,
      "learning_rate": 9.578375720046841e-06,
      "loss": 0.3415,
      "step": 28435
    },
    {
      "epoch": 0.5135094143392629,
      "grad_norm": 0.35054147243499756,
      "learning_rate": 9.575542016358556e-06,
      "loss": 0.1883,
      "step": 28440
    },
    {
      "epoch": 0.5135996937721636,
      "grad_norm": 0.3902779519557953,
      "learning_rate": 9.572708346814836e-06,
      "loss": 0.2516,
      "step": 28445
    },
    {
      "epoch": 0.5136899732050643,
      "grad_norm": 0.4505479037761688,
      "learning_rate": 9.569874711643616e-06,
      "loss": 0.2091,
      "step": 28450
    },
    {
      "epoch": 0.513780252637965,
      "grad_norm": 0.4965774714946747,
      "learning_rate": 9.56704111107285e-06,
      "loss": 0.2457,
      "step": 28455
    },
    {
      "epoch": 0.5138705320708657,
      "grad_norm": 0.661870539188385,
      "learning_rate": 9.564207545330482e-06,
      "loss": 0.1949,
      "step": 28460
    },
    {
      "epoch": 0.5139608115037665,
      "grad_norm": 0.6075689792633057,
      "learning_rate": 9.561374014644443e-06,
      "loss": 0.2481,
      "step": 28465
    },
    {
      "epoch": 0.5140510909366671,
      "grad_norm": 0.36879128217697144,
      "learning_rate": 9.55854051924268e-06,
      "loss": 0.2393,
      "step": 28470
    },
    {
      "epoch": 0.5141413703695679,
      "grad_norm": 0.6054572463035583,
      "learning_rate": 9.555707059353118e-06,
      "loss": 0.1553,
      "step": 28475
    },
    {
      "epoch": 0.5142316498024686,
      "grad_norm": 0.5598069429397583,
      "learning_rate": 9.552873635203696e-06,
      "loss": 0.2656,
      "step": 28480
    },
    {
      "epoch": 0.5143219292353693,
      "grad_norm": 0.4982486367225647,
      "learning_rate": 9.550040247022337e-06,
      "loss": 0.1776,
      "step": 28485
    },
    {
      "epoch": 0.51441220866827,
      "grad_norm": 0.6931257843971252,
      "learning_rate": 9.547206895036973e-06,
      "loss": 0.2757,
      "step": 28490
    },
    {
      "epoch": 0.5145024881011707,
      "grad_norm": 0.31999334692955017,
      "learning_rate": 9.544373579475518e-06,
      "loss": 0.2901,
      "step": 28495
    },
    {
      "epoch": 0.5145927675340715,
      "grad_norm": 0.4983128309249878,
      "learning_rate": 9.541540300565897e-06,
      "loss": 0.0993,
      "step": 28500
    },
    {
      "epoch": 0.5146830469669722,
      "grad_norm": 0.5988265872001648,
      "learning_rate": 9.538707058536024e-06,
      "loss": 0.232,
      "step": 28505
    },
    {
      "epoch": 0.5147733263998728,
      "grad_norm": 0.5505892634391785,
      "learning_rate": 9.535873853613818e-06,
      "loss": 0.3384,
      "step": 28510
    },
    {
      "epoch": 0.5148636058327736,
      "grad_norm": 0.342400461435318,
      "learning_rate": 9.533040686027183e-06,
      "loss": 0.1917,
      "step": 28515
    },
    {
      "epoch": 0.5149538852656743,
      "grad_norm": 0.3296906054019928,
      "learning_rate": 9.53020755600403e-06,
      "loss": 0.2454,
      "step": 28520
    },
    {
      "epoch": 0.5150441646985751,
      "grad_norm": 0.256753146648407,
      "learning_rate": 9.527374463772262e-06,
      "loss": 0.2047,
      "step": 28525
    },
    {
      "epoch": 0.5151344441314757,
      "grad_norm": 0.350436806678772,
      "learning_rate": 9.524541409559787e-06,
      "loss": 0.3151,
      "step": 28530
    },
    {
      "epoch": 0.5152247235643764,
      "grad_norm": 0.39108237624168396,
      "learning_rate": 9.52170839359449e-06,
      "loss": 0.2864,
      "step": 28535
    },
    {
      "epoch": 0.5153150029972772,
      "grad_norm": 0.3559560477733612,
      "learning_rate": 9.518875416104284e-06,
      "loss": 0.1883,
      "step": 28540
    },
    {
      "epoch": 0.5154052824301779,
      "grad_norm": 0.40671393275260925,
      "learning_rate": 9.516042477317049e-06,
      "loss": 0.2448,
      "step": 28545
    },
    {
      "epoch": 0.5154955618630787,
      "grad_norm": 0.5957432389259338,
      "learning_rate": 9.513209577460679e-06,
      "loss": 0.2479,
      "step": 28550
    },
    {
      "epoch": 0.5155858412959793,
      "grad_norm": 0.4093860685825348,
      "learning_rate": 9.510376716763062e-06,
      "loss": 0.256,
      "step": 28555
    },
    {
      "epoch": 0.51567612072888,
      "grad_norm": 0.5931309461593628,
      "learning_rate": 9.507543895452072e-06,
      "loss": 0.2012,
      "step": 28560
    },
    {
      "epoch": 0.5157664001617808,
      "grad_norm": 0.5393112897872925,
      "learning_rate": 9.504711113755599e-06,
      "loss": 0.3817,
      "step": 28565
    },
    {
      "epoch": 0.5158566795946815,
      "grad_norm": 0.3842506408691406,
      "learning_rate": 9.501878371901512e-06,
      "loss": 0.2598,
      "step": 28570
    },
    {
      "epoch": 0.5159469590275821,
      "grad_norm": 0.44890961050987244,
      "learning_rate": 9.499045670117693e-06,
      "loss": 0.2563,
      "step": 28575
    },
    {
      "epoch": 0.5160372384604829,
      "grad_norm": 0.36904144287109375,
      "learning_rate": 9.496213008632006e-06,
      "loss": 0.242,
      "step": 28580
    },
    {
      "epoch": 0.5161275178933836,
      "grad_norm": 0.4298558235168457,
      "learning_rate": 9.493380387672318e-06,
      "loss": 0.3049,
      "step": 28585
    },
    {
      "epoch": 0.5162177973262844,
      "grad_norm": 0.36553990840911865,
      "learning_rate": 9.490547807466494e-06,
      "loss": 0.2364,
      "step": 28590
    },
    {
      "epoch": 0.516308076759185,
      "grad_norm": 0.6329402327537537,
      "learning_rate": 9.487715268242399e-06,
      "loss": 0.2018,
      "step": 28595
    },
    {
      "epoch": 0.5163983561920857,
      "grad_norm": 0.34179818630218506,
      "learning_rate": 9.48488277022788e-06,
      "loss": 0.3315,
      "step": 28600
    },
    {
      "epoch": 0.5164886356249865,
      "grad_norm": 0.46953728795051575,
      "learning_rate": 9.482050313650803e-06,
      "loss": 0.2657,
      "step": 28605
    },
    {
      "epoch": 0.5165789150578872,
      "grad_norm": 0.31157082319259644,
      "learning_rate": 9.479217898739007e-06,
      "loss": 0.2288,
      "step": 28610
    },
    {
      "epoch": 0.5166691944907879,
      "grad_norm": 0.4610215425491333,
      "learning_rate": 9.476385525720347e-06,
      "loss": 0.2543,
      "step": 28615
    },
    {
      "epoch": 0.5167594739236886,
      "grad_norm": 0.2740088701248169,
      "learning_rate": 9.473553194822663e-06,
      "loss": 0.2235,
      "step": 28620
    },
    {
      "epoch": 0.5168497533565893,
      "grad_norm": 0.5691370964050293,
      "learning_rate": 9.470720906273802e-06,
      "loss": 0.2223,
      "step": 28625
    },
    {
      "epoch": 0.5169400327894901,
      "grad_norm": 2.938239097595215,
      "learning_rate": 9.46788866030159e-06,
      "loss": 0.2719,
      "step": 28630
    },
    {
      "epoch": 0.5170303122223907,
      "grad_norm": 0.3056987524032593,
      "learning_rate": 9.46505645713387e-06,
      "loss": 0.2005,
      "step": 28635
    },
    {
      "epoch": 0.5171205916552915,
      "grad_norm": 0.746619462966919,
      "learning_rate": 9.462224296998465e-06,
      "loss": 0.2983,
      "step": 28640
    },
    {
      "epoch": 0.5172108710881922,
      "grad_norm": 0.5389447808265686,
      "learning_rate": 9.45939218012321e-06,
      "loss": 0.3343,
      "step": 28645
    },
    {
      "epoch": 0.5173011505210929,
      "grad_norm": 0.3819253444671631,
      "learning_rate": 9.456560106735922e-06,
      "loss": 0.2068,
      "step": 28650
    },
    {
      "epoch": 0.5173914299539936,
      "grad_norm": 0.3484255373477936,
      "learning_rate": 9.453728077064418e-06,
      "loss": 0.2132,
      "step": 28655
    },
    {
      "epoch": 0.5174817093868943,
      "grad_norm": 0.3534581661224365,
      "learning_rate": 9.450896091336525e-06,
      "loss": 0.1971,
      "step": 28660
    },
    {
      "epoch": 0.517571988819795,
      "grad_norm": 0.7409288883209229,
      "learning_rate": 9.448064149780043e-06,
      "loss": 0.1966,
      "step": 28665
    },
    {
      "epoch": 0.5176622682526958,
      "grad_norm": 0.36281609535217285,
      "learning_rate": 9.445232252622792e-06,
      "loss": 0.253,
      "step": 28670
    },
    {
      "epoch": 0.5177525476855964,
      "grad_norm": 0.4301600754261017,
      "learning_rate": 9.44240040009257e-06,
      "loss": 0.2761,
      "step": 28675
    },
    {
      "epoch": 0.5178428271184972,
      "grad_norm": 0.47277727723121643,
      "learning_rate": 9.439568592417182e-06,
      "loss": 0.2723,
      "step": 28680
    },
    {
      "epoch": 0.5179331065513979,
      "grad_norm": 0.884819507598877,
      "learning_rate": 9.436736829824425e-06,
      "loss": 0.2927,
      "step": 28685
    },
    {
      "epoch": 0.5180233859842986,
      "grad_norm": 0.37751275300979614,
      "learning_rate": 9.433905112542098e-06,
      "loss": 0.2587,
      "step": 28690
    },
    {
      "epoch": 0.5181136654171993,
      "grad_norm": 0.5852225422859192,
      "learning_rate": 9.431073440797987e-06,
      "loss": 0.2184,
      "step": 28695
    },
    {
      "epoch": 0.5182039448501,
      "grad_norm": 0.4749210476875305,
      "learning_rate": 9.42824181481988e-06,
      "loss": 0.3444,
      "step": 28700
    },
    {
      "epoch": 0.5182942242830008,
      "grad_norm": 0.2995322644710541,
      "learning_rate": 9.425410234835562e-06,
      "loss": 0.1988,
      "step": 28705
    },
    {
      "epoch": 0.5183845037159015,
      "grad_norm": 0.5372695922851562,
      "learning_rate": 9.422578701072816e-06,
      "loss": 0.212,
      "step": 28710
    },
    {
      "epoch": 0.5184747831488021,
      "grad_norm": 0.37080609798431396,
      "learning_rate": 9.419747213759412e-06,
      "loss": 0.2895,
      "step": 28715
    },
    {
      "epoch": 0.5185650625817029,
      "grad_norm": 0.16475968062877655,
      "learning_rate": 9.416915773123128e-06,
      "loss": 0.226,
      "step": 28720
    },
    {
      "epoch": 0.5186553420146036,
      "grad_norm": 0.5332977771759033,
      "learning_rate": 9.414084379391729e-06,
      "loss": 0.2959,
      "step": 28725
    },
    {
      "epoch": 0.5187456214475044,
      "grad_norm": 0.5752127766609192,
      "learning_rate": 9.411253032792981e-06,
      "loss": 0.2302,
      "step": 28730
    },
    {
      "epoch": 0.518835900880405,
      "grad_norm": 0.9700411558151245,
      "learning_rate": 9.40842173355465e-06,
      "loss": 0.3014,
      "step": 28735
    },
    {
      "epoch": 0.5189261803133057,
      "grad_norm": 0.66222083568573,
      "learning_rate": 9.405590481904485e-06,
      "loss": 0.2954,
      "step": 28740
    },
    {
      "epoch": 0.5190164597462065,
      "grad_norm": 0.3779366612434387,
      "learning_rate": 9.402759278070246e-06,
      "loss": 0.2753,
      "step": 28745
    },
    {
      "epoch": 0.5191067391791072,
      "grad_norm": 0.5842838883399963,
      "learning_rate": 9.399928122279678e-06,
      "loss": 0.2887,
      "step": 28750
    },
    {
      "epoch": 0.5191970186120078,
      "grad_norm": 0.7366604208946228,
      "learning_rate": 9.397097014760535e-06,
      "loss": 0.2706,
      "step": 28755
    },
    {
      "epoch": 0.5192872980449086,
      "grad_norm": 0.4361642301082611,
      "learning_rate": 9.39426595574055e-06,
      "loss": 0.1711,
      "step": 28760
    },
    {
      "epoch": 0.5193775774778093,
      "grad_norm": 0.6456303596496582,
      "learning_rate": 9.391434945447463e-06,
      "loss": 0.2822,
      "step": 28765
    },
    {
      "epoch": 0.5194678569107101,
      "grad_norm": 0.5982512831687927,
      "learning_rate": 9.388603984109011e-06,
      "loss": 0.2058,
      "step": 28770
    },
    {
      "epoch": 0.5195581363436107,
      "grad_norm": 0.45432358980178833,
      "learning_rate": 9.385773071952928e-06,
      "loss": 0.2887,
      "step": 28775
    },
    {
      "epoch": 0.5196484157765114,
      "grad_norm": 0.3200981318950653,
      "learning_rate": 9.38294220920693e-06,
      "loss": 0.2377,
      "step": 28780
    },
    {
      "epoch": 0.5197386952094122,
      "grad_norm": 0.5555515289306641,
      "learning_rate": 9.380111396098748e-06,
      "loss": 0.2141,
      "step": 28785
    },
    {
      "epoch": 0.5198289746423129,
      "grad_norm": 0.390817254781723,
      "learning_rate": 9.377280632856094e-06,
      "loss": 0.2542,
      "step": 28790
    },
    {
      "epoch": 0.5199192540752136,
      "grad_norm": 0.46619880199432373,
      "learning_rate": 9.374449919706687e-06,
      "loss": 0.2805,
      "step": 28795
    },
    {
      "epoch": 0.5200095335081143,
      "grad_norm": 0.2127184420824051,
      "learning_rate": 9.371619256878236e-06,
      "loss": 0.153,
      "step": 28800
    },
    {
      "epoch": 0.520099812941015,
      "grad_norm": 0.7713163495063782,
      "learning_rate": 9.368788644598449e-06,
      "loss": 0.1596,
      "step": 28805
    },
    {
      "epoch": 0.5201900923739158,
      "grad_norm": 0.3343033194541931,
      "learning_rate": 9.365958083095023e-06,
      "loss": 0.2513,
      "step": 28810
    },
    {
      "epoch": 0.5202803718068164,
      "grad_norm": 0.5713295936584473,
      "learning_rate": 9.36312757259566e-06,
      "loss": 0.2409,
      "step": 28815
    },
    {
      "epoch": 0.5203706512397172,
      "grad_norm": 0.7380322813987732,
      "learning_rate": 9.360297113328052e-06,
      "loss": 0.3282,
      "step": 28820
    },
    {
      "epoch": 0.5204609306726179,
      "grad_norm": 0.6009113788604736,
      "learning_rate": 9.357466705519895e-06,
      "loss": 0.2825,
      "step": 28825
    },
    {
      "epoch": 0.5205512101055186,
      "grad_norm": 0.42540380358695984,
      "learning_rate": 9.354636349398868e-06,
      "loss": 0.205,
      "step": 28830
    },
    {
      "epoch": 0.5206414895384193,
      "grad_norm": 0.7028429508209229,
      "learning_rate": 9.35180604519265e-06,
      "loss": 0.2577,
      "step": 28835
    },
    {
      "epoch": 0.52073176897132,
      "grad_norm": 0.40496763586997986,
      "learning_rate": 9.34897579312893e-06,
      "loss": 0.2601,
      "step": 28840
    },
    {
      "epoch": 0.5208220484042207,
      "grad_norm": 0.4968245327472687,
      "learning_rate": 9.346145593435369e-06,
      "loss": 0.3397,
      "step": 28845
    },
    {
      "epoch": 0.5209123278371215,
      "grad_norm": 0.33956199884414673,
      "learning_rate": 9.343315446339643e-06,
      "loss": 0.2776,
      "step": 28850
    },
    {
      "epoch": 0.5210026072700221,
      "grad_norm": 0.5444244742393494,
      "learning_rate": 9.340485352069411e-06,
      "loss": 0.1788,
      "step": 28855
    },
    {
      "epoch": 0.5210928867029229,
      "grad_norm": 0.6133355498313904,
      "learning_rate": 9.33765531085234e-06,
      "loss": 0.1939,
      "step": 28860
    },
    {
      "epoch": 0.5211831661358236,
      "grad_norm": 0.6628184914588928,
      "learning_rate": 9.334825322916084e-06,
      "loss": 0.2833,
      "step": 28865
    },
    {
      "epoch": 0.5212734455687243,
      "grad_norm": 0.39197325706481934,
      "learning_rate": 9.331995388488296e-06,
      "loss": 0.1877,
      "step": 28870
    },
    {
      "epoch": 0.521363725001625,
      "grad_norm": 0.42335212230682373,
      "learning_rate": 9.329165507796619e-06,
      "loss": 0.3009,
      "step": 28875
    },
    {
      "epoch": 0.5214540044345257,
      "grad_norm": 0.61272794008255,
      "learning_rate": 9.326335681068703e-06,
      "loss": 0.3649,
      "step": 28880
    },
    {
      "epoch": 0.5215442838674265,
      "grad_norm": 1.6266295909881592,
      "learning_rate": 9.32350590853218e-06,
      "loss": 0.2248,
      "step": 28885
    },
    {
      "epoch": 0.5216345633003272,
      "grad_norm": 0.4243999719619751,
      "learning_rate": 9.32067619041469e-06,
      "loss": 0.2775,
      "step": 28890
    },
    {
      "epoch": 0.5217248427332278,
      "grad_norm": 0.4876992106437683,
      "learning_rate": 9.31784652694386e-06,
      "loss": 0.2488,
      "step": 28895
    },
    {
      "epoch": 0.5218151221661286,
      "grad_norm": 0.3923934996128082,
      "learning_rate": 9.315016918347318e-06,
      "loss": 0.2923,
      "step": 28900
    },
    {
      "epoch": 0.5219054015990293,
      "grad_norm": 0.452089786529541,
      "learning_rate": 9.312187364852683e-06,
      "loss": 0.2258,
      "step": 28905
    },
    {
      "epoch": 0.5219956810319301,
      "grad_norm": 0.619631826877594,
      "learning_rate": 9.309357866687578e-06,
      "loss": 0.3225,
      "step": 28910
    },
    {
      "epoch": 0.5220859604648307,
      "grad_norm": 0.3504837155342102,
      "learning_rate": 9.306528424079606e-06,
      "loss": 0.238,
      "step": 28915
    },
    {
      "epoch": 0.5221762398977314,
      "grad_norm": 0.3095729351043701,
      "learning_rate": 9.303699037256384e-06,
      "loss": 0.23,
      "step": 28920
    },
    {
      "epoch": 0.5222665193306322,
      "grad_norm": 0.48953840136528015,
      "learning_rate": 9.30086970644551e-06,
      "loss": 0.2503,
      "step": 28925
    },
    {
      "epoch": 0.5223567987635329,
      "grad_norm": 0.4810202717781067,
      "learning_rate": 9.298040431874584e-06,
      "loss": 0.3123,
      "step": 28930
    },
    {
      "epoch": 0.5224470781964337,
      "grad_norm": 0.3621257245540619,
      "learning_rate": 9.295211213771206e-06,
      "loss": 0.1933,
      "step": 28935
    },
    {
      "epoch": 0.5225373576293343,
      "grad_norm": 0.3242107331752777,
      "learning_rate": 9.292382052362955e-06,
      "loss": 0.2264,
      "step": 28940
    },
    {
      "epoch": 0.522627637062235,
      "grad_norm": 0.7714591026306152,
      "learning_rate": 9.289552947877428e-06,
      "loss": 0.2266,
      "step": 28945
    },
    {
      "epoch": 0.5227179164951358,
      "grad_norm": 0.3129047751426697,
      "learning_rate": 9.286723900542199e-06,
      "loss": 0.2461,
      "step": 28950
    },
    {
      "epoch": 0.5228081959280365,
      "grad_norm": 0.5419982075691223,
      "learning_rate": 9.283894910584848e-06,
      "loss": 0.1536,
      "step": 28955
    },
    {
      "epoch": 0.5228984753609371,
      "grad_norm": 0.39662978053092957,
      "learning_rate": 9.28106597823294e-06,
      "loss": 0.2931,
      "step": 28960
    },
    {
      "epoch": 0.5229887547938379,
      "grad_norm": 0.45678624510765076,
      "learning_rate": 9.278237103714053e-06,
      "loss": 0.2421,
      "step": 28965
    },
    {
      "epoch": 0.5230790342267386,
      "grad_norm": 0.5336384177207947,
      "learning_rate": 9.275408287255738e-06,
      "loss": 0.228,
      "step": 28970
    },
    {
      "epoch": 0.5231693136596394,
      "grad_norm": 0.5243151187896729,
      "learning_rate": 9.272579529085564e-06,
      "loss": 0.2832,
      "step": 28975
    },
    {
      "epoch": 0.52325959309254,
      "grad_norm": 0.4475562870502472,
      "learning_rate": 9.269750829431073e-06,
      "loss": 0.2219,
      "step": 28980
    },
    {
      "epoch": 0.5233498725254407,
      "grad_norm": 0.4101310670375824,
      "learning_rate": 9.26692218851982e-06,
      "loss": 0.2665,
      "step": 28985
    },
    {
      "epoch": 0.5234401519583415,
      "grad_norm": 0.5185317993164062,
      "learning_rate": 9.264093606579346e-06,
      "loss": 0.2303,
      "step": 28990
    },
    {
      "epoch": 0.5235304313912422,
      "grad_norm": 0.5011386871337891,
      "learning_rate": 9.261265083837192e-06,
      "loss": 0.1616,
      "step": 28995
    },
    {
      "epoch": 0.5236207108241429,
      "grad_norm": 0.2844066917896271,
      "learning_rate": 9.258436620520888e-06,
      "loss": 0.2353,
      "step": 29000
    },
    {
      "epoch": 0.5237109902570436,
      "grad_norm": 0.4489717483520508,
      "learning_rate": 9.255608216857972e-06,
      "loss": 0.3063,
      "step": 29005
    },
    {
      "epoch": 0.5238012696899443,
      "grad_norm": 0.433130145072937,
      "learning_rate": 9.25277987307596e-06,
      "loss": 0.2162,
      "step": 29010
    },
    {
      "epoch": 0.5238915491228451,
      "grad_norm": 0.4645278751850128,
      "learning_rate": 9.249951589402371e-06,
      "loss": 0.262,
      "step": 29015
    },
    {
      "epoch": 0.5239818285557457,
      "grad_norm": 0.46296849846839905,
      "learning_rate": 9.24712336606473e-06,
      "loss": 0.2047,
      "step": 29020
    },
    {
      "epoch": 0.5240721079886465,
      "grad_norm": 0.5372308492660522,
      "learning_rate": 9.244295203290533e-06,
      "loss": 0.1853,
      "step": 29025
    },
    {
      "epoch": 0.5241623874215472,
      "grad_norm": 0.4358402490615845,
      "learning_rate": 9.241467101307298e-06,
      "loss": 0.1978,
      "step": 29030
    },
    {
      "epoch": 0.5242526668544479,
      "grad_norm": 0.730522871017456,
      "learning_rate": 9.238639060342515e-06,
      "loss": 0.2102,
      "step": 29035
    },
    {
      "epoch": 0.5243429462873486,
      "grad_norm": 0.5406123995780945,
      "learning_rate": 9.235811080623688e-06,
      "loss": 0.3067,
      "step": 29040
    },
    {
      "epoch": 0.5244332257202493,
      "grad_norm": 0.4470652639865875,
      "learning_rate": 9.232983162378301e-06,
      "loss": 0.2268,
      "step": 29045
    },
    {
      "epoch": 0.52452350515315,
      "grad_norm": 0.4234684705734253,
      "learning_rate": 9.230155305833843e-06,
      "loss": 0.2263,
      "step": 29050
    },
    {
      "epoch": 0.5246137845860508,
      "grad_norm": 1.0069785118103027,
      "learning_rate": 9.227327511217792e-06,
      "loss": 0.2249,
      "step": 29055
    },
    {
      "epoch": 0.5247040640189514,
      "grad_norm": 0.7058039307594299,
      "learning_rate": 9.224499778757625e-06,
      "loss": 0.2124,
      "step": 29060
    },
    {
      "epoch": 0.5247943434518522,
      "grad_norm": 0.5813343524932861,
      "learning_rate": 9.221672108680812e-06,
      "loss": 0.2509,
      "step": 29065
    },
    {
      "epoch": 0.5248846228847529,
      "grad_norm": 0.24793928861618042,
      "learning_rate": 9.218844501214824e-06,
      "loss": 0.2581,
      "step": 29070
    },
    {
      "epoch": 0.5249749023176536,
      "grad_norm": 0.3823011815547943,
      "learning_rate": 9.21601695658711e-06,
      "loss": 0.2139,
      "step": 29075
    },
    {
      "epoch": 0.5250651817505543,
      "grad_norm": 0.506294846534729,
      "learning_rate": 9.213189475025133e-06,
      "loss": 0.2607,
      "step": 29080
    },
    {
      "epoch": 0.525155461183455,
      "grad_norm": 0.7374309301376343,
      "learning_rate": 9.210362056756342e-06,
      "loss": 0.1908,
      "step": 29085
    },
    {
      "epoch": 0.5252457406163558,
      "grad_norm": 0.4976447820663452,
      "learning_rate": 9.207534702008187e-06,
      "loss": 0.3187,
      "step": 29090
    },
    {
      "epoch": 0.5253360200492565,
      "grad_norm": 1.3869751691818237,
      "learning_rate": 9.204707411008099e-06,
      "loss": 0.2927,
      "step": 29095
    },
    {
      "epoch": 0.5254262994821571,
      "grad_norm": 0.5298295617103577,
      "learning_rate": 9.201880183983518e-06,
      "loss": 0.2162,
      "step": 29100
    },
    {
      "epoch": 0.5255165789150579,
      "grad_norm": 0.2758086025714874,
      "learning_rate": 9.199053021161876e-06,
      "loss": 0.2581,
      "step": 29105
    },
    {
      "epoch": 0.5256068583479586,
      "grad_norm": 0.5744746923446655,
      "learning_rate": 9.196225922770592e-06,
      "loss": 0.27,
      "step": 29110
    },
    {
      "epoch": 0.5256971377808594,
      "grad_norm": 0.5285934805870056,
      "learning_rate": 9.193398889037088e-06,
      "loss": 0.3273,
      "step": 29115
    },
    {
      "epoch": 0.52578741721376,
      "grad_norm": 0.38951054215431213,
      "learning_rate": 9.19057192018878e-06,
      "loss": 0.2136,
      "step": 29120
    },
    {
      "epoch": 0.5258776966466607,
      "grad_norm": 0.5702796578407288,
      "learning_rate": 9.187745016453074e-06,
      "loss": 0.2056,
      "step": 29125
    },
    {
      "epoch": 0.5259679760795615,
      "grad_norm": 0.5056290626525879,
      "learning_rate": 9.184918178057378e-06,
      "loss": 0.2869,
      "step": 29130
    },
    {
      "epoch": 0.5260582555124622,
      "grad_norm": 0.648164689540863,
      "learning_rate": 9.18209140522909e-06,
      "loss": 0.2259,
      "step": 29135
    },
    {
      "epoch": 0.5261485349453628,
      "grad_norm": 0.3163846731185913,
      "learning_rate": 9.179264698195598e-06,
      "loss": 0.2527,
      "step": 29140
    },
    {
      "epoch": 0.5262388143782636,
      "grad_norm": 0.35842669010162354,
      "learning_rate": 9.176438057184295e-06,
      "loss": 0.2556,
      "step": 29145
    },
    {
      "epoch": 0.5263290938111643,
      "grad_norm": 0.9657479524612427,
      "learning_rate": 9.173611482422562e-06,
      "loss": 0.3573,
      "step": 29150
    },
    {
      "epoch": 0.5264193732440651,
      "grad_norm": 0.5299378037452698,
      "learning_rate": 9.17078497413778e-06,
      "loss": 0.236,
      "step": 29155
    },
    {
      "epoch": 0.5265096526769657,
      "grad_norm": 0.2969253957271576,
      "learning_rate": 9.167958532557313e-06,
      "loss": 0.1682,
      "step": 29160
    },
    {
      "epoch": 0.5265999321098664,
      "grad_norm": 0.5945093035697937,
      "learning_rate": 9.165132157908536e-06,
      "loss": 0.2323,
      "step": 29165
    },
    {
      "epoch": 0.5266902115427672,
      "grad_norm": 0.5590220093727112,
      "learning_rate": 9.162305850418805e-06,
      "loss": 0.2607,
      "step": 29170
    },
    {
      "epoch": 0.5267804909756679,
      "grad_norm": 0.3752111494541168,
      "learning_rate": 9.15947961031548e-06,
      "loss": 0.1746,
      "step": 29175
    },
    {
      "epoch": 0.5268707704085686,
      "grad_norm": 0.5085766315460205,
      "learning_rate": 9.156653437825904e-06,
      "loss": 0.2402,
      "step": 29180
    },
    {
      "epoch": 0.5269610498414693,
      "grad_norm": 0.2286229431629181,
      "learning_rate": 9.153827333177434e-06,
      "loss": 0.2185,
      "step": 29185
    },
    {
      "epoch": 0.52705132927437,
      "grad_norm": 0.44430407881736755,
      "learning_rate": 9.151001296597398e-06,
      "loss": 0.2004,
      "step": 29190
    },
    {
      "epoch": 0.5271416087072708,
      "grad_norm": 0.35909855365753174,
      "learning_rate": 9.148175328313136e-06,
      "loss": 0.2301,
      "step": 29195
    },
    {
      "epoch": 0.5272318881401714,
      "grad_norm": 0.3297822177410126,
      "learning_rate": 9.145349428551979e-06,
      "loss": 0.2516,
      "step": 29200
    },
    {
      "epoch": 0.5273221675730722,
      "grad_norm": 0.51425701379776,
      "learning_rate": 9.142523597541241e-06,
      "loss": 0.2572,
      "step": 29205
    },
    {
      "epoch": 0.5274124470059729,
      "grad_norm": 0.804265022277832,
      "learning_rate": 9.13969783550825e-06,
      "loss": 0.2921,
      "step": 29210
    },
    {
      "epoch": 0.5275027264388736,
      "grad_norm": 0.3839067816734314,
      "learning_rate": 9.136872142680309e-06,
      "loss": 0.2301,
      "step": 29215
    },
    {
      "epoch": 0.5275930058717743,
      "grad_norm": 0.7506440281867981,
      "learning_rate": 9.134046519284733e-06,
      "loss": 0.1969,
      "step": 29220
    },
    {
      "epoch": 0.527683285304675,
      "grad_norm": 0.5303971767425537,
      "learning_rate": 9.131220965548814e-06,
      "loss": 0.1988,
      "step": 29225
    },
    {
      "epoch": 0.5277735647375758,
      "grad_norm": 0.47842785716056824,
      "learning_rate": 9.128395481699855e-06,
      "loss": 0.2993,
      "step": 29230
    },
    {
      "epoch": 0.5278638441704765,
      "grad_norm": 0.35133376717567444,
      "learning_rate": 9.125570067965138e-06,
      "loss": 0.2038,
      "step": 29235
    },
    {
      "epoch": 0.5279541236033771,
      "grad_norm": 0.704802393913269,
      "learning_rate": 9.122744724571956e-06,
      "loss": 0.205,
      "step": 29240
    },
    {
      "epoch": 0.5280444030362779,
      "grad_norm": 0.29799526929855347,
      "learning_rate": 9.119919451747578e-06,
      "loss": 0.2352,
      "step": 29245
    },
    {
      "epoch": 0.5281346824691786,
      "grad_norm": 0.538298487663269,
      "learning_rate": 9.117094249719287e-06,
      "loss": 0.1288,
      "step": 29250
    },
    {
      "epoch": 0.5282249619020793,
      "grad_norm": 0.60027015209198,
      "learning_rate": 9.11426911871434e-06,
      "loss": 0.2194,
      "step": 29255
    },
    {
      "epoch": 0.52831524133498,
      "grad_norm": 0.5876868367195129,
      "learning_rate": 9.111444058960004e-06,
      "loss": 0.1387,
      "step": 29260
    },
    {
      "epoch": 0.5284055207678807,
      "grad_norm": 0.32721805572509766,
      "learning_rate": 9.10861907068353e-06,
      "loss": 0.2288,
      "step": 29265
    },
    {
      "epoch": 0.5284958002007815,
      "grad_norm": 0.3952575922012329,
      "learning_rate": 9.105794154112177e-06,
      "loss": 0.2973,
      "step": 29270
    },
    {
      "epoch": 0.5285860796336822,
      "grad_norm": 0.41881096363067627,
      "learning_rate": 9.102969309473178e-06,
      "loss": 0.2401,
      "step": 29275
    },
    {
      "epoch": 0.5286763590665828,
      "grad_norm": 0.6066296696662903,
      "learning_rate": 9.100144536993777e-06,
      "loss": 0.2784,
      "step": 29280
    },
    {
      "epoch": 0.5287666384994836,
      "grad_norm": 0.39961621165275574,
      "learning_rate": 9.097319836901209e-06,
      "loss": 0.2199,
      "step": 29285
    },
    {
      "epoch": 0.5288569179323843,
      "grad_norm": 0.4351910650730133,
      "learning_rate": 9.094495209422693e-06,
      "loss": 0.2384,
      "step": 29290
    },
    {
      "epoch": 0.5289471973652851,
      "grad_norm": 0.47434571385383606,
      "learning_rate": 9.091670654785455e-06,
      "loss": 0.207,
      "step": 29295
    },
    {
      "epoch": 0.5290374767981857,
      "grad_norm": 0.37549877166748047,
      "learning_rate": 9.088846173216707e-06,
      "loss": 0.2085,
      "step": 29300
    },
    {
      "epoch": 0.5291277562310864,
      "grad_norm": 0.46450865268707275,
      "learning_rate": 9.086021764943664e-06,
      "loss": 0.227,
      "step": 29305
    },
    {
      "epoch": 0.5292180356639872,
      "grad_norm": 0.6716872453689575,
      "learning_rate": 9.083197430193517e-06,
      "loss": 0.2121,
      "step": 29310
    },
    {
      "epoch": 0.5293083150968879,
      "grad_norm": 0.34130343794822693,
      "learning_rate": 9.080373169193477e-06,
      "loss": 0.2925,
      "step": 29315
    },
    {
      "epoch": 0.5293985945297885,
      "grad_norm": 0.4936409294605255,
      "learning_rate": 9.077548982170727e-06,
      "loss": 0.2717,
      "step": 29320
    },
    {
      "epoch": 0.5294888739626893,
      "grad_norm": 0.7294759154319763,
      "learning_rate": 9.074724869352457e-06,
      "loss": 0.2836,
      "step": 29325
    },
    {
      "epoch": 0.52957915339559,
      "grad_norm": 0.6342700123786926,
      "learning_rate": 9.071900830965839e-06,
      "loss": 0.2111,
      "step": 29330
    },
    {
      "epoch": 0.5296694328284908,
      "grad_norm": 0.58757084608078,
      "learning_rate": 9.069076867238058e-06,
      "loss": 0.3456,
      "step": 29335
    },
    {
      "epoch": 0.5297597122613915,
      "grad_norm": 0.4558350145816803,
      "learning_rate": 9.06625297839627e-06,
      "loss": 0.2422,
      "step": 29340
    },
    {
      "epoch": 0.5298499916942921,
      "grad_norm": 0.4043460786342621,
      "learning_rate": 9.063429164667641e-06,
      "loss": 0.2341,
      "step": 29345
    },
    {
      "epoch": 0.5299402711271929,
      "grad_norm": 0.34301161766052246,
      "learning_rate": 9.060605426279327e-06,
      "loss": 0.3049,
      "step": 29350
    },
    {
      "epoch": 0.5300305505600936,
      "grad_norm": 0.5473177433013916,
      "learning_rate": 9.05778176345848e-06,
      "loss": 0.3061,
      "step": 29355
    },
    {
      "epoch": 0.5301208299929944,
      "grad_norm": 0.4395565390586853,
      "learning_rate": 9.054958176432232e-06,
      "loss": 0.2773,
      "step": 29360
    },
    {
      "epoch": 0.530211109425895,
      "grad_norm": 0.5470331311225891,
      "learning_rate": 9.052134665427734e-06,
      "loss": 0.3942,
      "step": 29365
    },
    {
      "epoch": 0.5303013888587957,
      "grad_norm": 0.4692026376724243,
      "learning_rate": 9.049311230672105e-06,
      "loss": 0.2163,
      "step": 29370
    },
    {
      "epoch": 0.5303916682916965,
      "grad_norm": 0.401054710149765,
      "learning_rate": 9.04648787239248e-06,
      "loss": 0.3039,
      "step": 29375
    },
    {
      "epoch": 0.5304819477245972,
      "grad_norm": 0.5699111223220825,
      "learning_rate": 9.043664590815973e-06,
      "loss": 0.2137,
      "step": 29380
    },
    {
      "epoch": 0.5305722271574979,
      "grad_norm": 0.4167243242263794,
      "learning_rate": 9.040841386169694e-06,
      "loss": 0.1648,
      "step": 29385
    },
    {
      "epoch": 0.5306625065903986,
      "grad_norm": 0.6566867828369141,
      "learning_rate": 9.038018258680753e-06,
      "loss": 0.3274,
      "step": 29390
    },
    {
      "epoch": 0.5307527860232993,
      "grad_norm": 0.46134153008461,
      "learning_rate": 9.035195208576247e-06,
      "loss": 0.2062,
      "step": 29395
    },
    {
      "epoch": 0.5308430654562001,
      "grad_norm": 0.35193225741386414,
      "learning_rate": 9.032372236083275e-06,
      "loss": 0.2503,
      "step": 29400
    },
    {
      "epoch": 0.5309333448891007,
      "grad_norm": 0.36347219347953796,
      "learning_rate": 9.029549341428918e-06,
      "loss": 0.243,
      "step": 29405
    },
    {
      "epoch": 0.5310236243220015,
      "grad_norm": 0.8077743649482727,
      "learning_rate": 9.026726524840261e-06,
      "loss": 0.2307,
      "step": 29410
    },
    {
      "epoch": 0.5311139037549022,
      "grad_norm": 0.5789394974708557,
      "learning_rate": 9.02390378654438e-06,
      "loss": 0.2531,
      "step": 29415
    },
    {
      "epoch": 0.5312041831878029,
      "grad_norm": 0.4585419297218323,
      "learning_rate": 9.021081126768344e-06,
      "loss": 0.1938,
      "step": 29420
    },
    {
      "epoch": 0.5312944626207036,
      "grad_norm": 0.39548465609550476,
      "learning_rate": 9.01825854573921e-06,
      "loss": 0.2606,
      "step": 29425
    },
    {
      "epoch": 0.5313847420536043,
      "grad_norm": 0.42935436964035034,
      "learning_rate": 9.01543604368404e-06,
      "loss": 0.2124,
      "step": 29430
    },
    {
      "epoch": 0.531475021486505,
      "grad_norm": 0.2676616311073303,
      "learning_rate": 9.012613620829881e-06,
      "loss": 0.167,
      "step": 29435
    },
    {
      "epoch": 0.5315653009194058,
      "grad_norm": 0.3309231698513031,
      "learning_rate": 9.009791277403782e-06,
      "loss": 0.2867,
      "step": 29440
    },
    {
      "epoch": 0.5316555803523064,
      "grad_norm": 0.2971544563770294,
      "learning_rate": 9.006969013632765e-06,
      "loss": 0.2161,
      "step": 29445
    },
    {
      "epoch": 0.5317458597852072,
      "grad_norm": 0.42876797914505005,
      "learning_rate": 9.00414682974388e-06,
      "loss": 0.2147,
      "step": 29450
    },
    {
      "epoch": 0.5318361392181079,
      "grad_norm": 0.47307348251342773,
      "learning_rate": 9.001324725964138e-06,
      "loss": 0.2491,
      "step": 29455
    },
    {
      "epoch": 0.5319264186510086,
      "grad_norm": 0.48468372225761414,
      "learning_rate": 8.998502702520562e-06,
      "loss": 0.2941,
      "step": 29460
    },
    {
      "epoch": 0.5320166980839093,
      "grad_norm": 0.6423856019973755,
      "learning_rate": 8.995680759640162e-06,
      "loss": 0.2658,
      "step": 29465
    },
    {
      "epoch": 0.53210697751681,
      "grad_norm": 0.3446263074874878,
      "learning_rate": 8.992858897549945e-06,
      "loss": 0.2299,
      "step": 29470
    },
    {
      "epoch": 0.5321972569497108,
      "grad_norm": 0.2555557191371918,
      "learning_rate": 8.990037116476905e-06,
      "loss": 0.2368,
      "step": 29475
    },
    {
      "epoch": 0.5322875363826115,
      "grad_norm": 0.36306333541870117,
      "learning_rate": 8.987215416648033e-06,
      "loss": 0.2372,
      "step": 29480
    },
    {
      "epoch": 0.5323778158155121,
      "grad_norm": 0.6085293889045715,
      "learning_rate": 8.984393798290322e-06,
      "loss": 0.2426,
      "step": 29485
    },
    {
      "epoch": 0.5324680952484129,
      "grad_norm": 0.32165977358818054,
      "learning_rate": 8.981572261630743e-06,
      "loss": 0.2472,
      "step": 29490
    },
    {
      "epoch": 0.5325583746813136,
      "grad_norm": 0.3249320983886719,
      "learning_rate": 8.97875080689627e-06,
      "loss": 0.3114,
      "step": 29495
    },
    {
      "epoch": 0.5326486541142144,
      "grad_norm": 0.2133033275604248,
      "learning_rate": 8.97592943431387e-06,
      "loss": 0.1943,
      "step": 29500
    },
    {
      "epoch": 0.532738933547115,
      "grad_norm": 0.600243866443634,
      "learning_rate": 8.973108144110504e-06,
      "loss": 0.2675,
      "step": 29505
    },
    {
      "epoch": 0.5328292129800157,
      "grad_norm": 0.5490859746932983,
      "learning_rate": 8.970286936513114e-06,
      "loss": 0.2145,
      "step": 29510
    },
    {
      "epoch": 0.5329194924129165,
      "grad_norm": 0.24173220992088318,
      "learning_rate": 8.967465811748663e-06,
      "loss": 0.1804,
      "step": 29515
    },
    {
      "epoch": 0.5330097718458172,
      "grad_norm": 0.49922430515289307,
      "learning_rate": 8.964644770044074e-06,
      "loss": 0.2265,
      "step": 29520
    },
    {
      "epoch": 0.5331000512787178,
      "grad_norm": 0.7988704442977905,
      "learning_rate": 8.961823811626288e-06,
      "loss": 0.2428,
      "step": 29525
    },
    {
      "epoch": 0.5331903307116186,
      "grad_norm": 0.539200484752655,
      "learning_rate": 8.959002936722226e-06,
      "loss": 0.3329,
      "step": 29530
    },
    {
      "epoch": 0.5332806101445193,
      "grad_norm": 0.47633564472198486,
      "learning_rate": 8.956182145558812e-06,
      "loss": 0.2948,
      "step": 29535
    },
    {
      "epoch": 0.5333708895774201,
      "grad_norm": 0.5710396766662598,
      "learning_rate": 8.953361438362953e-06,
      "loss": 0.2617,
      "step": 29540
    },
    {
      "epoch": 0.5334611690103207,
      "grad_norm": 0.3195212483406067,
      "learning_rate": 8.950540815361556e-06,
      "loss": 0.2382,
      "step": 29545
    },
    {
      "epoch": 0.5335514484432214,
      "grad_norm": 0.4052313268184662,
      "learning_rate": 8.947720276781521e-06,
      "loss": 0.2092,
      "step": 29550
    },
    {
      "epoch": 0.5336417278761222,
      "grad_norm": 0.5307109951972961,
      "learning_rate": 8.944899822849742e-06,
      "loss": 0.255,
      "step": 29555
    },
    {
      "epoch": 0.5337320073090229,
      "grad_norm": 0.24876217544078827,
      "learning_rate": 8.942079453793095e-06,
      "loss": 0.2055,
      "step": 29560
    },
    {
      "epoch": 0.5338222867419236,
      "grad_norm": 0.6838825941085815,
      "learning_rate": 8.939259169838468e-06,
      "loss": 0.3401,
      "step": 29565
    },
    {
      "epoch": 0.5339125661748243,
      "grad_norm": 0.21149404346942902,
      "learning_rate": 8.936438971212731e-06,
      "loss": 0.2823,
      "step": 29570
    },
    {
      "epoch": 0.534002845607725,
      "grad_norm": 0.6761206388473511,
      "learning_rate": 8.93361885814274e-06,
      "loss": 0.3155,
      "step": 29575
    },
    {
      "epoch": 0.5340931250406258,
      "grad_norm": 0.37084195017814636,
      "learning_rate": 8.930798830855362e-06,
      "loss": 0.211,
      "step": 29580
    },
    {
      "epoch": 0.5341834044735264,
      "grad_norm": 0.25710147619247437,
      "learning_rate": 8.927978889577442e-06,
      "loss": 0.299,
      "step": 29585
    },
    {
      "epoch": 0.5342736839064272,
      "grad_norm": 0.9091326594352722,
      "learning_rate": 8.925159034535827e-06,
      "loss": 0.1839,
      "step": 29590
    },
    {
      "epoch": 0.5343639633393279,
      "grad_norm": 0.28009116649627686,
      "learning_rate": 8.922339265957352e-06,
      "loss": 0.1992,
      "step": 29595
    },
    {
      "epoch": 0.5344542427722286,
      "grad_norm": 0.28533223271369934,
      "learning_rate": 8.919519584068852e-06,
      "loss": 0.3137,
      "step": 29600
    },
    {
      "epoch": 0.5345445222051293,
      "grad_norm": 0.7236104607582092,
      "learning_rate": 8.91669998909714e-06,
      "loss": 0.28,
      "step": 29605
    },
    {
      "epoch": 0.53463480163803,
      "grad_norm": 0.9480448961257935,
      "learning_rate": 8.91388048126904e-06,
      "loss": 0.2162,
      "step": 29610
    },
    {
      "epoch": 0.5347250810709308,
      "grad_norm": 0.6658946871757507,
      "learning_rate": 8.911061060811357e-06,
      "loss": 0.2499,
      "step": 29615
    },
    {
      "epoch": 0.5348153605038315,
      "grad_norm": 0.46006113290786743,
      "learning_rate": 8.908241727950899e-06,
      "loss": 0.1842,
      "step": 29620
    },
    {
      "epoch": 0.5349056399367321,
      "grad_norm": 0.21971310675144196,
      "learning_rate": 8.90542248291445e-06,
      "loss": 0.1928,
      "step": 29625
    },
    {
      "epoch": 0.5349959193696329,
      "grad_norm": 0.33377861976623535,
      "learning_rate": 8.902603325928807e-06,
      "loss": 0.2457,
      "step": 29630
    },
    {
      "epoch": 0.5350861988025336,
      "grad_norm": 0.34162864089012146,
      "learning_rate": 8.899784257220745e-06,
      "loss": 0.1545,
      "step": 29635
    },
    {
      "epoch": 0.5351764782354343,
      "grad_norm": 0.5479598045349121,
      "learning_rate": 8.896965277017044e-06,
      "loss": 0.2917,
      "step": 29640
    },
    {
      "epoch": 0.535266757668335,
      "grad_norm": 0.46999359130859375,
      "learning_rate": 8.894146385544462e-06,
      "loss": 0.2553,
      "step": 29645
    },
    {
      "epoch": 0.5353570371012357,
      "grad_norm": 0.3225659430027008,
      "learning_rate": 8.89132758302977e-06,
      "loss": 0.2486,
      "step": 29650
    },
    {
      "epoch": 0.5354473165341365,
      "grad_norm": 0.4857708811759949,
      "learning_rate": 8.88850886969971e-06,
      "loss": 0.1502,
      "step": 29655
    },
    {
      "epoch": 0.5355375959670372,
      "grad_norm": 0.6933872103691101,
      "learning_rate": 8.885690245781027e-06,
      "loss": 0.2487,
      "step": 29660
    },
    {
      "epoch": 0.5356278753999378,
      "grad_norm": 0.5495598316192627,
      "learning_rate": 8.882871711500466e-06,
      "loss": 0.1984,
      "step": 29665
    },
    {
      "epoch": 0.5357181548328386,
      "grad_norm": 0.4869239330291748,
      "learning_rate": 8.88005326708475e-06,
      "loss": 0.3247,
      "step": 29670
    },
    {
      "epoch": 0.5358084342657393,
      "grad_norm": 0.8459439873695374,
      "learning_rate": 8.877234912760608e-06,
      "loss": 0.2102,
      "step": 29675
    },
    {
      "epoch": 0.5358987136986401,
      "grad_norm": 0.5308342576026917,
      "learning_rate": 8.874416648754752e-06,
      "loss": 0.1865,
      "step": 29680
    },
    {
      "epoch": 0.5359889931315407,
      "grad_norm": 0.32921162247657776,
      "learning_rate": 8.871598475293896e-06,
      "loss": 0.2981,
      "step": 29685
    },
    {
      "epoch": 0.5360792725644414,
      "grad_norm": 0.6725593209266663,
      "learning_rate": 8.868780392604735e-06,
      "loss": 0.2554,
      "step": 29690
    },
    {
      "epoch": 0.5361695519973422,
      "grad_norm": 0.48778754472732544,
      "learning_rate": 8.86596240091397e-06,
      "loss": 0.2012,
      "step": 29695
    },
    {
      "epoch": 0.5362598314302429,
      "grad_norm": 0.4959346055984497,
      "learning_rate": 8.86314450044828e-06,
      "loss": 0.3329,
      "step": 29700
    },
    {
      "epoch": 0.5363501108631435,
      "grad_norm": 0.3672713339328766,
      "learning_rate": 8.860326691434353e-06,
      "loss": 0.2391,
      "step": 29705
    },
    {
      "epoch": 0.5364403902960443,
      "grad_norm": 0.4734179675579071,
      "learning_rate": 8.857508974098855e-06,
      "loss": 0.2225,
      "step": 29710
    },
    {
      "epoch": 0.536530669728945,
      "grad_norm": 0.39586877822875977,
      "learning_rate": 8.854691348668459e-06,
      "loss": 0.2175,
      "step": 29715
    },
    {
      "epoch": 0.5366209491618458,
      "grad_norm": 0.43276047706604004,
      "learning_rate": 8.851873815369809e-06,
      "loss": 0.1962,
      "step": 29720
    },
    {
      "epoch": 0.5367112285947465,
      "grad_norm": 0.5799977779388428,
      "learning_rate": 8.849056374429567e-06,
      "loss": 0.2642,
      "step": 29725
    },
    {
      "epoch": 0.5368015080276471,
      "grad_norm": 0.3819090723991394,
      "learning_rate": 8.84623902607437e-06,
      "loss": 0.2946,
      "step": 29730
    },
    {
      "epoch": 0.5368917874605479,
      "grad_norm": 0.5902660489082336,
      "learning_rate": 8.843421770530859e-06,
      "loss": 0.2203,
      "step": 29735
    },
    {
      "epoch": 0.5369820668934486,
      "grad_norm": 0.4830811321735382,
      "learning_rate": 8.840604608025651e-06,
      "loss": 0.2205,
      "step": 29740
    },
    {
      "epoch": 0.5370723463263494,
      "grad_norm": 0.3830168843269348,
      "learning_rate": 8.837787538785377e-06,
      "loss": 0.2467,
      "step": 29745
    },
    {
      "epoch": 0.53716262575925,
      "grad_norm": 0.691828727722168,
      "learning_rate": 8.834970563036646e-06,
      "loss": 0.2306,
      "step": 29750
    },
    {
      "epoch": 0.5372529051921507,
      "grad_norm": 0.6907765865325928,
      "learning_rate": 8.83215368100606e-06,
      "loss": 0.2701,
      "step": 29755
    },
    {
      "epoch": 0.5373431846250515,
      "grad_norm": 0.5170467495918274,
      "learning_rate": 8.829336892920222e-06,
      "loss": 0.2605,
      "step": 29760
    },
    {
      "epoch": 0.5374334640579522,
      "grad_norm": 1.5071393251419067,
      "learning_rate": 8.826520199005714e-06,
      "loss": 0.2684,
      "step": 29765
    },
    {
      "epoch": 0.5375237434908529,
      "grad_norm": 0.46036243438720703,
      "learning_rate": 8.823703599489128e-06,
      "loss": 0.2093,
      "step": 29770
    },
    {
      "epoch": 0.5376140229237536,
      "grad_norm": 0.4511943757534027,
      "learning_rate": 8.820887094597034e-06,
      "loss": 0.2095,
      "step": 29775
    },
    {
      "epoch": 0.5377043023566543,
      "grad_norm": 0.5065392851829529,
      "learning_rate": 8.818070684556005e-06,
      "loss": 0.3288,
      "step": 29780
    },
    {
      "epoch": 0.5377945817895551,
      "grad_norm": 0.5359420776367188,
      "learning_rate": 8.815254369592589e-06,
      "loss": 0.2837,
      "step": 29785
    },
    {
      "epoch": 0.5378848612224557,
      "grad_norm": 0.6591514348983765,
      "learning_rate": 8.812438149933347e-06,
      "loss": 0.1904,
      "step": 29790
    },
    {
      "epoch": 0.5379751406553565,
      "grad_norm": 0.37006279826164246,
      "learning_rate": 8.809622025804821e-06,
      "loss": 0.2878,
      "step": 29795
    },
    {
      "epoch": 0.5380654200882572,
      "grad_norm": 0.5443669557571411,
      "learning_rate": 8.806805997433551e-06,
      "loss": 0.2524,
      "step": 29800
    },
    {
      "epoch": 0.5381556995211579,
      "grad_norm": 0.41262099146842957,
      "learning_rate": 8.80399006504606e-06,
      "loss": 0.2925,
      "step": 29805
    },
    {
      "epoch": 0.5382459789540586,
      "grad_norm": 0.40744054317474365,
      "learning_rate": 8.801174228868875e-06,
      "loss": 0.2424,
      "step": 29810
    },
    {
      "epoch": 0.5383362583869593,
      "grad_norm": 0.4741535484790802,
      "learning_rate": 8.798358489128505e-06,
      "loss": 0.3022,
      "step": 29815
    },
    {
      "epoch": 0.53842653781986,
      "grad_norm": 0.30072957277297974,
      "learning_rate": 8.79554284605146e-06,
      "loss": 0.2032,
      "step": 29820
    },
    {
      "epoch": 0.5385168172527608,
      "grad_norm": 0.5489832162857056,
      "learning_rate": 8.792727299864233e-06,
      "loss": 0.2193,
      "step": 29825
    },
    {
      "epoch": 0.5386070966856614,
      "grad_norm": 0.5997200012207031,
      "learning_rate": 8.789911850793318e-06,
      "loss": 0.2775,
      "step": 29830
    },
    {
      "epoch": 0.5386973761185622,
      "grad_norm": 0.4494878351688385,
      "learning_rate": 8.787096499065196e-06,
      "loss": 0.3104,
      "step": 29835
    },
    {
      "epoch": 0.5387876555514629,
      "grad_norm": 0.2640324532985687,
      "learning_rate": 8.784281244906344e-06,
      "loss": 0.2495,
      "step": 29840
    },
    {
      "epoch": 0.5388779349843636,
      "grad_norm": 0.6722503900527954,
      "learning_rate": 8.781466088543228e-06,
      "loss": 0.2428,
      "step": 29845
    },
    {
      "epoch": 0.5389682144172643,
      "grad_norm": 0.36204028129577637,
      "learning_rate": 8.7786510302023e-06,
      "loss": 0.2151,
      "step": 29850
    },
    {
      "epoch": 0.539058493850165,
      "grad_norm": 0.4201105237007141,
      "learning_rate": 8.775836070110022e-06,
      "loss": 0.1772,
      "step": 29855
    },
    {
      "epoch": 0.5391487732830658,
      "grad_norm": 0.487784743309021,
      "learning_rate": 8.773021208492826e-06,
      "loss": 0.2394,
      "step": 29860
    },
    {
      "epoch": 0.5392390527159665,
      "grad_norm": 0.3053692877292633,
      "learning_rate": 8.77020644557716e-06,
      "loss": 0.2204,
      "step": 29865
    },
    {
      "epoch": 0.5393293321488671,
      "grad_norm": 0.5000454783439636,
      "learning_rate": 8.767391781589439e-06,
      "loss": 0.2511,
      "step": 29870
    },
    {
      "epoch": 0.5394196115817679,
      "grad_norm": 0.5024951696395874,
      "learning_rate": 8.764577216756088e-06,
      "loss": 0.2015,
      "step": 29875
    },
    {
      "epoch": 0.5395098910146686,
      "grad_norm": 0.5001381039619446,
      "learning_rate": 8.761762751303517e-06,
      "loss": 0.1156,
      "step": 29880
    },
    {
      "epoch": 0.5396001704475694,
      "grad_norm": 0.2666400969028473,
      "learning_rate": 8.758948385458133e-06,
      "loss": 0.1681,
      "step": 29885
    },
    {
      "epoch": 0.53969044988047,
      "grad_norm": 0.47668972611427307,
      "learning_rate": 8.756134119446323e-06,
      "loss": 0.3536,
      "step": 29890
    },
    {
      "epoch": 0.5397807293133707,
      "grad_norm": 0.332539439201355,
      "learning_rate": 8.753319953494484e-06,
      "loss": 0.2453,
      "step": 29895
    },
    {
      "epoch": 0.5398710087462715,
      "grad_norm": 0.5720506310462952,
      "learning_rate": 8.750505887828985e-06,
      "loss": 0.2368,
      "step": 29900
    },
    {
      "epoch": 0.5399612881791722,
      "grad_norm": 0.30869919061660767,
      "learning_rate": 8.747691922676208e-06,
      "loss": 0.2336,
      "step": 29905
    },
    {
      "epoch": 0.5400515676120728,
      "grad_norm": 0.5556881427764893,
      "learning_rate": 8.744878058262507e-06,
      "loss": 0.2798,
      "step": 29910
    },
    {
      "epoch": 0.5401418470449736,
      "grad_norm": 4.185696601867676,
      "learning_rate": 8.742064294814246e-06,
      "loss": 0.2371,
      "step": 29915
    },
    {
      "epoch": 0.5402321264778743,
      "grad_norm": 0.3491530120372772,
      "learning_rate": 8.73925063255776e-06,
      "loss": 0.2588,
      "step": 29920
    },
    {
      "epoch": 0.5403224059107751,
      "grad_norm": 0.5603716373443604,
      "learning_rate": 8.736437071719397e-06,
      "loss": 0.2899,
      "step": 29925
    },
    {
      "epoch": 0.5404126853436757,
      "grad_norm": 0.6623853445053101,
      "learning_rate": 8.733623612525488e-06,
      "loss": 0.2577,
      "step": 29930
    },
    {
      "epoch": 0.5405029647765764,
      "grad_norm": 0.43849846720695496,
      "learning_rate": 8.730810255202347e-06,
      "loss": 0.1973,
      "step": 29935
    },
    {
      "epoch": 0.5405932442094772,
      "grad_norm": 0.5376783609390259,
      "learning_rate": 8.727996999976295e-06,
      "loss": 0.2428,
      "step": 29940
    },
    {
      "epoch": 0.5406835236423779,
      "grad_norm": 0.4216878116130829,
      "learning_rate": 8.725183847073633e-06,
      "loss": 0.3253,
      "step": 29945
    },
    {
      "epoch": 0.5407738030752786,
      "grad_norm": 0.4465843439102173,
      "learning_rate": 8.722370796720667e-06,
      "loss": 0.2747,
      "step": 29950
    },
    {
      "epoch": 0.5408640825081793,
      "grad_norm": 0.5713403224945068,
      "learning_rate": 8.719557849143676e-06,
      "loss": 0.1876,
      "step": 29955
    },
    {
      "epoch": 0.54095436194108,
      "grad_norm": 0.5041961073875427,
      "learning_rate": 8.716745004568949e-06,
      "loss": 0.2545,
      "step": 29960
    },
    {
      "epoch": 0.5410446413739808,
      "grad_norm": 0.3148564398288727,
      "learning_rate": 8.713932263222755e-06,
      "loss": 0.1554,
      "step": 29965
    },
    {
      "epoch": 0.5411349208068814,
      "grad_norm": 0.5915099382400513,
      "learning_rate": 8.71111962533136e-06,
      "loss": 0.2664,
      "step": 29970
    },
    {
      "epoch": 0.5412252002397822,
      "grad_norm": 0.4450978934764862,
      "learning_rate": 8.70830709112102e-06,
      "loss": 0.2685,
      "step": 29975
    },
    {
      "epoch": 0.5413154796726829,
      "grad_norm": 0.4449928402900696,
      "learning_rate": 8.705494660817987e-06,
      "loss": 0.2886,
      "step": 29980
    },
    {
      "epoch": 0.5414057591055836,
      "grad_norm": 0.5779251456260681,
      "learning_rate": 8.702682334648492e-06,
      "loss": 0.1615,
      "step": 29985
    },
    {
      "epoch": 0.5414960385384843,
      "grad_norm": 0.4866010844707489,
      "learning_rate": 8.699870112838772e-06,
      "loss": 0.2066,
      "step": 29990
    },
    {
      "epoch": 0.541586317971385,
      "grad_norm": 0.651261031627655,
      "learning_rate": 8.697057995615046e-06,
      "loss": 0.2259,
      "step": 29995
    },
    {
      "epoch": 0.5416765974042858,
      "grad_norm": 0.5383086204528809,
      "learning_rate": 8.694245983203538e-06,
      "loss": 0.2452,
      "step": 30000
    },
    {
      "epoch": 0.5417668768371865,
      "grad_norm": 0.5351186990737915,
      "learning_rate": 8.69143407583044e-06,
      "loss": 0.2553,
      "step": 30005
    },
    {
      "epoch": 0.5418571562700871,
      "grad_norm": 0.5017138123512268,
      "learning_rate": 8.68862227372196e-06,
      "loss": 0.2409,
      "step": 30010
    },
    {
      "epoch": 0.5419474357029879,
      "grad_norm": 0.5619741082191467,
      "learning_rate": 8.68581057710428e-06,
      "loss": 0.3087,
      "step": 30015
    },
    {
      "epoch": 0.5420377151358886,
      "grad_norm": 0.3829174041748047,
      "learning_rate": 8.682998986203589e-06,
      "loss": 0.2271,
      "step": 30020
    },
    {
      "epoch": 0.5421279945687894,
      "grad_norm": 0.3206920623779297,
      "learning_rate": 8.680187501246051e-06,
      "loss": 0.2102,
      "step": 30025
    },
    {
      "epoch": 0.54221827400169,
      "grad_norm": 0.4522305428981781,
      "learning_rate": 8.677376122457832e-06,
      "loss": 0.2373,
      "step": 30030
    },
    {
      "epoch": 0.5423085534345907,
      "grad_norm": 0.6030825972557068,
      "learning_rate": 8.67456485006509e-06,
      "loss": 0.2398,
      "step": 30035
    },
    {
      "epoch": 0.5423988328674915,
      "grad_norm": 0.4817292392253876,
      "learning_rate": 8.671753684293964e-06,
      "loss": 0.249,
      "step": 30040
    },
    {
      "epoch": 0.5424891123003922,
      "grad_norm": 0.5090333223342896,
      "learning_rate": 8.668942625370604e-06,
      "loss": 0.2429,
      "step": 30045
    },
    {
      "epoch": 0.5425793917332928,
      "grad_norm": 0.4519498944282532,
      "learning_rate": 8.666131673521126e-06,
      "loss": 0.2883,
      "step": 30050
    },
    {
      "epoch": 0.5426696711661936,
      "grad_norm": 0.3915843069553375,
      "learning_rate": 8.663320828971659e-06,
      "loss": 0.2167,
      "step": 30055
    },
    {
      "epoch": 0.5427599505990943,
      "grad_norm": 0.7703441977500916,
      "learning_rate": 8.660510091948312e-06,
      "loss": 0.1479,
      "step": 30060
    },
    {
      "epoch": 0.5428502300319951,
      "grad_norm": 0.649986207485199,
      "learning_rate": 8.657699462677193e-06,
      "loss": 0.1445,
      "step": 30065
    },
    {
      "epoch": 0.5429405094648957,
      "grad_norm": 0.5072412490844727,
      "learning_rate": 8.654888941384387e-06,
      "loss": 0.2073,
      "step": 30070
    },
    {
      "epoch": 0.5430307888977964,
      "grad_norm": 0.32711225748062134,
      "learning_rate": 8.652078528295989e-06,
      "loss": 0.2987,
      "step": 30075
    },
    {
      "epoch": 0.5431210683306972,
      "grad_norm": 0.3813733160495758,
      "learning_rate": 8.649268223638072e-06,
      "loss": 0.2443,
      "step": 30080
    },
    {
      "epoch": 0.5432113477635979,
      "grad_norm": 0.3937578499317169,
      "learning_rate": 8.646458027636707e-06,
      "loss": 0.2543,
      "step": 30085
    },
    {
      "epoch": 0.5433016271964985,
      "grad_norm": 0.19544045627117157,
      "learning_rate": 8.64364794051795e-06,
      "loss": 0.1893,
      "step": 30090
    },
    {
      "epoch": 0.5433919066293993,
      "grad_norm": 0.382412850856781,
      "learning_rate": 8.640837962507858e-06,
      "loss": 0.2122,
      "step": 30095
    },
    {
      "epoch": 0.5434821860623,
      "grad_norm": 0.6230947375297546,
      "learning_rate": 8.638028093832464e-06,
      "loss": 0.2841,
      "step": 30100
    },
    {
      "epoch": 0.5435724654952008,
      "grad_norm": 0.42004403471946716,
      "learning_rate": 8.635218334717813e-06,
      "loss": 0.2005,
      "step": 30105
    },
    {
      "epoch": 0.5436627449281014,
      "grad_norm": 0.46466878056526184,
      "learning_rate": 8.63240868538992e-06,
      "loss": 0.1808,
      "step": 30110
    },
    {
      "epoch": 0.5437530243610021,
      "grad_norm": 0.4530866742134094,
      "learning_rate": 8.629599146074811e-06,
      "loss": 0.2173,
      "step": 30115
    },
    {
      "epoch": 0.5438433037939029,
      "grad_norm": 0.635175347328186,
      "learning_rate": 8.626789716998482e-06,
      "loss": 0.2622,
      "step": 30120
    },
    {
      "epoch": 0.5439335832268036,
      "grad_norm": 0.3577985465526581,
      "learning_rate": 8.623980398386937e-06,
      "loss": 0.3134,
      "step": 30125
    },
    {
      "epoch": 0.5440238626597044,
      "grad_norm": 0.6977483034133911,
      "learning_rate": 8.621171190466167e-06,
      "loss": 0.1519,
      "step": 30130
    },
    {
      "epoch": 0.544114142092605,
      "grad_norm": 0.4724690318107605,
      "learning_rate": 8.618362093462145e-06,
      "loss": 0.2143,
      "step": 30135
    },
    {
      "epoch": 0.5442044215255057,
      "grad_norm": 0.5454923510551453,
      "learning_rate": 8.615553107600851e-06,
      "loss": 0.2889,
      "step": 30140
    },
    {
      "epoch": 0.5442947009584065,
      "grad_norm": 0.5041444301605225,
      "learning_rate": 8.61274423310824e-06,
      "loss": 0.2603,
      "step": 30145
    },
    {
      "epoch": 0.5443849803913072,
      "grad_norm": 0.5674653053283691,
      "learning_rate": 8.609935470210274e-06,
      "loss": 0.2271,
      "step": 30150
    },
    {
      "epoch": 0.5444752598242079,
      "grad_norm": 0.5120981335639954,
      "learning_rate": 8.607126819132885e-06,
      "loss": 0.2082,
      "step": 30155
    },
    {
      "epoch": 0.5445655392571086,
      "grad_norm": 1.0488464832305908,
      "learning_rate": 8.604318280102024e-06,
      "loss": 0.28,
      "step": 30160
    },
    {
      "epoch": 0.5446558186900093,
      "grad_norm": 0.3890886902809143,
      "learning_rate": 8.601509853343605e-06,
      "loss": 0.2601,
      "step": 30165
    },
    {
      "epoch": 0.5447460981229101,
      "grad_norm": 0.5292909741401672,
      "learning_rate": 8.598701539083552e-06,
      "loss": 0.2329,
      "step": 30170
    },
    {
      "epoch": 0.5448363775558107,
      "grad_norm": 0.4403056800365448,
      "learning_rate": 8.59589333754777e-06,
      "loss": 0.2538,
      "step": 30175
    },
    {
      "epoch": 0.5449266569887115,
      "grad_norm": 0.313803493976593,
      "learning_rate": 8.593085248962165e-06,
      "loss": 0.2003,
      "step": 30180
    },
    {
      "epoch": 0.5450169364216122,
      "grad_norm": 0.3524719774723053,
      "learning_rate": 8.590277273552618e-06,
      "loss": 0.2941,
      "step": 30185
    },
    {
      "epoch": 0.5451072158545129,
      "grad_norm": 0.4764222502708435,
      "learning_rate": 8.587469411545017e-06,
      "loss": 0.2931,
      "step": 30190
    },
    {
      "epoch": 0.5451974952874136,
      "grad_norm": 0.8016039729118347,
      "learning_rate": 8.584661663165228e-06,
      "loss": 0.2096,
      "step": 30195
    },
    {
      "epoch": 0.5452877747203143,
      "grad_norm": 0.29778075218200684,
      "learning_rate": 8.581854028639123e-06,
      "loss": 0.2752,
      "step": 30200
    },
    {
      "epoch": 0.545378054153215,
      "grad_norm": 0.4600449204444885,
      "learning_rate": 8.579046508192546e-06,
      "loss": 0.2206,
      "step": 30205
    },
    {
      "epoch": 0.5454683335861158,
      "grad_norm": 0.46086183190345764,
      "learning_rate": 8.576239102051349e-06,
      "loss": 0.2971,
      "step": 30210
    },
    {
      "epoch": 0.5455586130190164,
      "grad_norm": 0.7456693649291992,
      "learning_rate": 8.573431810441366e-06,
      "loss": 0.3141,
      "step": 30215
    },
    {
      "epoch": 0.5456488924519172,
      "grad_norm": 0.4374391436576843,
      "learning_rate": 8.570624633588416e-06,
      "loss": 0.2165,
      "step": 30220
    },
    {
      "epoch": 0.5457391718848179,
      "grad_norm": 0.3902812600135803,
      "learning_rate": 8.567817571718328e-06,
      "loss": 0.2213,
      "step": 30225
    },
    {
      "epoch": 0.5458294513177187,
      "grad_norm": 0.49425461888313293,
      "learning_rate": 8.565010625056897e-06,
      "loss": 0.2629,
      "step": 30230
    },
    {
      "epoch": 0.5459197307506193,
      "grad_norm": 0.27138713002204895,
      "learning_rate": 8.562203793829933e-06,
      "loss": 0.2566,
      "step": 30235
    },
    {
      "epoch": 0.54601001018352,
      "grad_norm": 0.322450190782547,
      "learning_rate": 8.559397078263217e-06,
      "loss": 0.3462,
      "step": 30240
    },
    {
      "epoch": 0.5461002896164208,
      "grad_norm": 0.5785377025604248,
      "learning_rate": 8.556590478582534e-06,
      "loss": 0.2137,
      "step": 30245
    },
    {
      "epoch": 0.5461905690493215,
      "grad_norm": 0.5261700749397278,
      "learning_rate": 8.55378399501365e-06,
      "loss": 0.2801,
      "step": 30250
    },
    {
      "epoch": 0.5462808484822221,
      "grad_norm": 0.8050107955932617,
      "learning_rate": 8.550977627782332e-06,
      "loss": 0.1884,
      "step": 30255
    },
    {
      "epoch": 0.5463711279151229,
      "grad_norm": 0.2777807414531708,
      "learning_rate": 8.548171377114324e-06,
      "loss": 0.2244,
      "step": 30260
    },
    {
      "epoch": 0.5464614073480236,
      "grad_norm": 0.3774738311767578,
      "learning_rate": 8.545365243235378e-06,
      "loss": 0.2375,
      "step": 30265
    },
    {
      "epoch": 0.5465516867809244,
      "grad_norm": 0.5623134970664978,
      "learning_rate": 8.542559226371217e-06,
      "loss": 0.2624,
      "step": 30270
    },
    {
      "epoch": 0.546641966213825,
      "grad_norm": 0.28870970010757446,
      "learning_rate": 8.539753326747573e-06,
      "loss": 0.2838,
      "step": 30275
    },
    {
      "epoch": 0.5467322456467257,
      "grad_norm": 0.5625777244567871,
      "learning_rate": 8.536947544590153e-06,
      "loss": 0.2504,
      "step": 30280
    },
    {
      "epoch": 0.5468225250796265,
      "grad_norm": 0.4875805675983429,
      "learning_rate": 8.534141880124672e-06,
      "loss": 0.2305,
      "step": 30285
    },
    {
      "epoch": 0.5469128045125272,
      "grad_norm": 0.4243050217628479,
      "learning_rate": 8.53133633357681e-06,
      "loss": 0.2689,
      "step": 30290
    },
    {
      "epoch": 0.5470030839454278,
      "grad_norm": 0.5069988369941711,
      "learning_rate": 8.52853090517227e-06,
      "loss": 0.2417,
      "step": 30295
    },
    {
      "epoch": 0.5470933633783286,
      "grad_norm": 0.40005379915237427,
      "learning_rate": 8.525725595136718e-06,
      "loss": 0.2321,
      "step": 30300
    },
    {
      "epoch": 0.5471836428112293,
      "grad_norm": 0.5960733890533447,
      "learning_rate": 8.52292040369582e-06,
      "loss": 0.1913,
      "step": 30305
    },
    {
      "epoch": 0.5472739222441301,
      "grad_norm": 0.4210493266582489,
      "learning_rate": 8.520115331075241e-06,
      "loss": 0.1685,
      "step": 30310
    },
    {
      "epoch": 0.5473642016770307,
      "grad_norm": 0.44696950912475586,
      "learning_rate": 8.517310377500619e-06,
      "loss": 0.2598,
      "step": 30315
    },
    {
      "epoch": 0.5474544811099314,
      "grad_norm": 0.3658466935157776,
      "learning_rate": 8.514505543197599e-06,
      "loss": 0.2373,
      "step": 30320
    },
    {
      "epoch": 0.5475447605428322,
      "grad_norm": 0.5077930688858032,
      "learning_rate": 8.511700828391807e-06,
      "loss": 0.2969,
      "step": 30325
    },
    {
      "epoch": 0.5476350399757329,
      "grad_norm": 0.2905920147895813,
      "learning_rate": 8.508896233308866e-06,
      "loss": 0.2063,
      "step": 30330
    },
    {
      "epoch": 0.5477253194086336,
      "grad_norm": 0.42059463262557983,
      "learning_rate": 8.506091758174375e-06,
      "loss": 0.2191,
      "step": 30335
    },
    {
      "epoch": 0.5478155988415343,
      "grad_norm": 0.4557342529296875,
      "learning_rate": 8.503287403213944e-06,
      "loss": 0.2594,
      "step": 30340
    },
    {
      "epoch": 0.547905878274435,
      "grad_norm": 0.5572250485420227,
      "learning_rate": 8.500483168653158e-06,
      "loss": 0.3236,
      "step": 30345
    },
    {
      "epoch": 0.5479961577073358,
      "grad_norm": 0.5110666155815125,
      "learning_rate": 8.497679054717603e-06,
      "loss": 0.2828,
      "step": 30350
    },
    {
      "epoch": 0.5480864371402364,
      "grad_norm": 0.5500931739807129,
      "learning_rate": 8.494875061632837e-06,
      "loss": 0.1782,
      "step": 30355
    },
    {
      "epoch": 0.5481767165731372,
      "grad_norm": 0.6230002641677856,
      "learning_rate": 8.492071189624436e-06,
      "loss": 0.3495,
      "step": 30360
    },
    {
      "epoch": 0.5482669960060379,
      "grad_norm": 0.5025086998939514,
      "learning_rate": 8.48926743891794e-06,
      "loss": 0.1751,
      "step": 30365
    },
    {
      "epoch": 0.5483572754389386,
      "grad_norm": 0.26423999667167664,
      "learning_rate": 8.486463809738896e-06,
      "loss": 0.2421,
      "step": 30370
    },
    {
      "epoch": 0.5484475548718393,
      "grad_norm": 0.3416552245616913,
      "learning_rate": 8.483660302312832e-06,
      "loss": 0.1558,
      "step": 30375
    },
    {
      "epoch": 0.54853783430474,
      "grad_norm": 0.35250940918922424,
      "learning_rate": 8.480856916865276e-06,
      "loss": 0.232,
      "step": 30380
    },
    {
      "epoch": 0.5486281137376408,
      "grad_norm": 0.19040101766586304,
      "learning_rate": 8.478053653621733e-06,
      "loss": 0.2796,
      "step": 30385
    },
    {
      "epoch": 0.5487183931705415,
      "grad_norm": 0.31666895747184753,
      "learning_rate": 8.475250512807709e-06,
      "loss": 0.2196,
      "step": 30390
    },
    {
      "epoch": 0.5488086726034421,
      "grad_norm": 0.36421525478363037,
      "learning_rate": 8.472447494648699e-06,
      "loss": 0.2334,
      "step": 30395
    },
    {
      "epoch": 0.5488989520363429,
      "grad_norm": 0.36844000220298767,
      "learning_rate": 8.469644599370174e-06,
      "loss": 0.2541,
      "step": 30400
    },
    {
      "epoch": 0.5489892314692436,
      "grad_norm": 0.9060449600219727,
      "learning_rate": 8.46684182719762e-06,
      "loss": 0.2211,
      "step": 30405
    },
    {
      "epoch": 0.5490795109021444,
      "grad_norm": 0.3441360294818878,
      "learning_rate": 8.46403917835649e-06,
      "loss": 0.2631,
      "step": 30410
    },
    {
      "epoch": 0.549169790335045,
      "grad_norm": 0.4760313034057617,
      "learning_rate": 8.461236653072245e-06,
      "loss": 0.1989,
      "step": 30415
    },
    {
      "epoch": 0.5492600697679457,
      "grad_norm": 0.20249541103839874,
      "learning_rate": 8.458434251570315e-06,
      "loss": 0.2381,
      "step": 30420
    },
    {
      "epoch": 0.5493503492008465,
      "grad_norm": 0.4496363699436188,
      "learning_rate": 8.45563197407615e-06,
      "loss": 0.2252,
      "step": 30425
    },
    {
      "epoch": 0.5494406286337472,
      "grad_norm": 0.5233907103538513,
      "learning_rate": 8.452829820815159e-06,
      "loss": 0.2399,
      "step": 30430
    },
    {
      "epoch": 0.5495309080666478,
      "grad_norm": 0.5437569618225098,
      "learning_rate": 8.450027792012763e-06,
      "loss": 0.2259,
      "step": 30435
    },
    {
      "epoch": 0.5496211874995486,
      "grad_norm": 0.4562883973121643,
      "learning_rate": 8.447225887894358e-06,
      "loss": 0.1948,
      "step": 30440
    },
    {
      "epoch": 0.5497114669324493,
      "grad_norm": 0.47309333086013794,
      "learning_rate": 8.444424108685346e-06,
      "loss": 0.255,
      "step": 30445
    },
    {
      "epoch": 0.5498017463653501,
      "grad_norm": 0.4495987296104431,
      "learning_rate": 8.4416224546111e-06,
      "loss": 0.2512,
      "step": 30450
    },
    {
      "epoch": 0.5498920257982507,
      "grad_norm": 0.6639607548713684,
      "learning_rate": 8.438820925897e-06,
      "loss": 0.2563,
      "step": 30455
    },
    {
      "epoch": 0.5499823052311514,
      "grad_norm": 0.5853849649429321,
      "learning_rate": 8.436019522768402e-06,
      "loss": 0.2935,
      "step": 30460
    },
    {
      "epoch": 0.5500725846640522,
      "grad_norm": 0.570285975933075,
      "learning_rate": 8.433218245450666e-06,
      "loss": 0.2587,
      "step": 30465
    },
    {
      "epoch": 0.5501628640969529,
      "grad_norm": 0.42344382405281067,
      "learning_rate": 8.430417094169128e-06,
      "loss": 0.1531,
      "step": 30470
    },
    {
      "epoch": 0.5502531435298536,
      "grad_norm": 0.576492428779602,
      "learning_rate": 8.427616069149123e-06,
      "loss": 0.1968,
      "step": 30475
    },
    {
      "epoch": 0.5503434229627543,
      "grad_norm": 0.40256741642951965,
      "learning_rate": 8.424815170615972e-06,
      "loss": 0.2192,
      "step": 30480
    },
    {
      "epoch": 0.550433702395655,
      "grad_norm": 0.30532220005989075,
      "learning_rate": 8.42201439879499e-06,
      "loss": 0.1906,
      "step": 30485
    },
    {
      "epoch": 0.5505239818285558,
      "grad_norm": 0.29087671637535095,
      "learning_rate": 8.419213753911478e-06,
      "loss": 0.2323,
      "step": 30490
    },
    {
      "epoch": 0.5506142612614564,
      "grad_norm": 0.38298240303993225,
      "learning_rate": 8.416413236190723e-06,
      "loss": 0.2341,
      "step": 30495
    },
    {
      "epoch": 0.5507045406943571,
      "grad_norm": 0.33192840218544006,
      "learning_rate": 8.41361284585801e-06,
      "loss": 0.1706,
      "step": 30500
    },
    {
      "epoch": 0.5507948201272579,
      "grad_norm": 0.38980022072792053,
      "learning_rate": 8.410812583138608e-06,
      "loss": 0.2188,
      "step": 30505
    },
    {
      "epoch": 0.5508850995601586,
      "grad_norm": 0.6038150191307068,
      "learning_rate": 8.408012448257784e-06,
      "loss": 0.307,
      "step": 30510
    },
    {
      "epoch": 0.5509753789930594,
      "grad_norm": 0.39100486040115356,
      "learning_rate": 8.405212441440778e-06,
      "loss": 0.3175,
      "step": 30515
    },
    {
      "epoch": 0.55106565842596,
      "grad_norm": 0.2844029366970062,
      "learning_rate": 8.402412562912839e-06,
      "loss": 0.1746,
      "step": 30520
    },
    {
      "epoch": 0.5511559378588607,
      "grad_norm": 0.5369822382926941,
      "learning_rate": 8.399612812899191e-06,
      "loss": 0.2315,
      "step": 30525
    },
    {
      "epoch": 0.5512462172917615,
      "grad_norm": 0.5778539180755615,
      "learning_rate": 8.396813191625061e-06,
      "loss": 0.2859,
      "step": 30530
    },
    {
      "epoch": 0.5513364967246622,
      "grad_norm": 0.42409980297088623,
      "learning_rate": 8.39401369931565e-06,
      "loss": 0.2003,
      "step": 30535
    },
    {
      "epoch": 0.5514267761575629,
      "grad_norm": 0.32716935873031616,
      "learning_rate": 8.391214336196162e-06,
      "loss": 0.2488,
      "step": 30540
    },
    {
      "epoch": 0.5515170555904636,
      "grad_norm": 0.4676063656806946,
      "learning_rate": 8.38841510249178e-06,
      "loss": 0.2995,
      "step": 30545
    },
    {
      "epoch": 0.5516073350233643,
      "grad_norm": 0.8899534940719604,
      "learning_rate": 8.38561599842769e-06,
      "loss": 0.2397,
      "step": 30550
    },
    {
      "epoch": 0.5516976144562651,
      "grad_norm": 0.37093839049339294,
      "learning_rate": 8.38281702422905e-06,
      "loss": 0.2477,
      "step": 30555
    },
    {
      "epoch": 0.5517878938891657,
      "grad_norm": 0.45466116070747375,
      "learning_rate": 8.38001818012103e-06,
      "loss": 0.2345,
      "step": 30560
    },
    {
      "epoch": 0.5518781733220665,
      "grad_norm": 0.3915674388408661,
      "learning_rate": 8.377219466328763e-06,
      "loss": 0.1595,
      "step": 30565
    },
    {
      "epoch": 0.5519684527549672,
      "grad_norm": 0.4925644099712372,
      "learning_rate": 8.374420883077395e-06,
      "loss": 0.286,
      "step": 30570
    },
    {
      "epoch": 0.5520587321878679,
      "grad_norm": 3.358842134475708,
      "learning_rate": 8.371622430592048e-06,
      "loss": 0.22,
      "step": 30575
    },
    {
      "epoch": 0.5521490116207686,
      "grad_norm": 0.38430550694465637,
      "learning_rate": 8.368824109097833e-06,
      "loss": 0.2727,
      "step": 30580
    },
    {
      "epoch": 0.5522392910536693,
      "grad_norm": 0.36558714509010315,
      "learning_rate": 8.36602591881986e-06,
      "loss": 0.1963,
      "step": 30585
    },
    {
      "epoch": 0.55232957048657,
      "grad_norm": 0.6529437899589539,
      "learning_rate": 8.36322785998322e-06,
      "loss": 0.2819,
      "step": 30590
    },
    {
      "epoch": 0.5524198499194708,
      "grad_norm": 0.334749698638916,
      "learning_rate": 8.360429932813005e-06,
      "loss": 0.2205,
      "step": 30595
    },
    {
      "epoch": 0.5525101293523714,
      "grad_norm": 0.6322199702262878,
      "learning_rate": 8.357632137534275e-06,
      "loss": 0.1497,
      "step": 30600
    },
    {
      "epoch": 0.5526004087852722,
      "grad_norm": 0.41495248675346375,
      "learning_rate": 8.3548344743721e-06,
      "loss": 0.2921,
      "step": 30605
    },
    {
      "epoch": 0.5526906882181729,
      "grad_norm": 0.7760505676269531,
      "learning_rate": 8.352036943551527e-06,
      "loss": 0.4004,
      "step": 30610
    },
    {
      "epoch": 0.5527809676510737,
      "grad_norm": 0.4055958688259125,
      "learning_rate": 8.349239545297605e-06,
      "loss": 0.2173,
      "step": 30615
    },
    {
      "epoch": 0.5528712470839743,
      "grad_norm": 0.4020351469516754,
      "learning_rate": 8.346442279835356e-06,
      "loss": 0.1487,
      "step": 30620
    },
    {
      "epoch": 0.552961526516875,
      "grad_norm": 0.45829981565475464,
      "learning_rate": 8.343645147389807e-06,
      "loss": 0.3023,
      "step": 30625
    },
    {
      "epoch": 0.5530518059497758,
      "grad_norm": 0.6017299890518188,
      "learning_rate": 8.340848148185959e-06,
      "loss": 0.2992,
      "step": 30630
    },
    {
      "epoch": 0.5531420853826765,
      "grad_norm": 0.4230525493621826,
      "learning_rate": 8.338051282448816e-06,
      "loss": 0.2808,
      "step": 30635
    },
    {
      "epoch": 0.5532323648155771,
      "grad_norm": 0.5015904307365417,
      "learning_rate": 8.335254550403361e-06,
      "loss": 0.2763,
      "step": 30640
    },
    {
      "epoch": 0.5533226442484779,
      "grad_norm": 0.5678372979164124,
      "learning_rate": 8.332457952274578e-06,
      "loss": 0.202,
      "step": 30645
    },
    {
      "epoch": 0.5534129236813786,
      "grad_norm": 0.3022325038909912,
      "learning_rate": 8.329661488287425e-06,
      "loss": 0.1977,
      "step": 30650
    },
    {
      "epoch": 0.5535032031142794,
      "grad_norm": 0.541505753993988,
      "learning_rate": 8.326865158666863e-06,
      "loss": 0.2346,
      "step": 30655
    },
    {
      "epoch": 0.55359348254718,
      "grad_norm": 0.3254064917564392,
      "learning_rate": 8.324068963637832e-06,
      "loss": 0.1988,
      "step": 30660
    },
    {
      "epoch": 0.5536837619800807,
      "grad_norm": 0.553368091583252,
      "learning_rate": 8.321272903425273e-06,
      "loss": 0.2904,
      "step": 30665
    },
    {
      "epoch": 0.5537740414129815,
      "grad_norm": 0.5332421660423279,
      "learning_rate": 8.3184769782541e-06,
      "loss": 0.2829,
      "step": 30670
    },
    {
      "epoch": 0.5538643208458822,
      "grad_norm": 0.7650421261787415,
      "learning_rate": 8.315681188349225e-06,
      "loss": 0.3384,
      "step": 30675
    },
    {
      "epoch": 0.5539546002787828,
      "grad_norm": 0.4014601409435272,
      "learning_rate": 8.312885533935555e-06,
      "loss": 0.2257,
      "step": 30680
    },
    {
      "epoch": 0.5540448797116836,
      "grad_norm": 0.764105498790741,
      "learning_rate": 8.310090015237977e-06,
      "loss": 0.211,
      "step": 30685
    },
    {
      "epoch": 0.5541351591445843,
      "grad_norm": 0.3718617558479309,
      "learning_rate": 8.307294632481373e-06,
      "loss": 0.2753,
      "step": 30690
    },
    {
      "epoch": 0.5542254385774851,
      "grad_norm": 0.3817625045776367,
      "learning_rate": 8.304499385890605e-06,
      "loss": 0.1838,
      "step": 30695
    },
    {
      "epoch": 0.5543157180103857,
      "grad_norm": 0.3545447587966919,
      "learning_rate": 8.301704275690535e-06,
      "loss": 0.226,
      "step": 30700
    },
    {
      "epoch": 0.5544059974432864,
      "grad_norm": 0.49707674980163574,
      "learning_rate": 8.298909302106007e-06,
      "loss": 0.2114,
      "step": 30705
    },
    {
      "epoch": 0.5544962768761872,
      "grad_norm": 0.4153108298778534,
      "learning_rate": 8.296114465361861e-06,
      "loss": 0.2627,
      "step": 30710
    },
    {
      "epoch": 0.5545865563090879,
      "grad_norm": 0.6245543956756592,
      "learning_rate": 8.293319765682915e-06,
      "loss": 0.2159,
      "step": 30715
    },
    {
      "epoch": 0.5546768357419886,
      "grad_norm": 0.3766345977783203,
      "learning_rate": 8.290525203293987e-06,
      "loss": 0.2383,
      "step": 30720
    },
    {
      "epoch": 0.5547671151748893,
      "grad_norm": 0.4611589312553406,
      "learning_rate": 8.287730778419874e-06,
      "loss": 0.2811,
      "step": 30725
    },
    {
      "epoch": 0.55485739460779,
      "grad_norm": 0.4506969451904297,
      "learning_rate": 8.284936491285375e-06,
      "loss": 0.2702,
      "step": 30730
    },
    {
      "epoch": 0.5549476740406908,
      "grad_norm": 0.510202169418335,
      "learning_rate": 8.282142342115262e-06,
      "loss": 0.1874,
      "step": 30735
    },
    {
      "epoch": 0.5550379534735914,
      "grad_norm": 0.2654961347579956,
      "learning_rate": 8.279348331134309e-06,
      "loss": 0.3008,
      "step": 30740
    },
    {
      "epoch": 0.5551282329064922,
      "grad_norm": 0.4245106875896454,
      "learning_rate": 8.27655445856727e-06,
      "loss": 0.1683,
      "step": 30745
    },
    {
      "epoch": 0.5552185123393929,
      "grad_norm": 0.4185675382614136,
      "learning_rate": 8.273760724638898e-06,
      "loss": 0.2506,
      "step": 30750
    },
    {
      "epoch": 0.5553087917722936,
      "grad_norm": 0.4537559151649475,
      "learning_rate": 8.270967129573922e-06,
      "loss": 0.1892,
      "step": 30755
    },
    {
      "epoch": 0.5553990712051943,
      "grad_norm": 0.8804592490196228,
      "learning_rate": 8.268173673597072e-06,
      "loss": 0.2101,
      "step": 30760
    },
    {
      "epoch": 0.555489350638095,
      "grad_norm": 0.32881176471710205,
      "learning_rate": 8.265380356933058e-06,
      "loss": 0.1157,
      "step": 30765
    },
    {
      "epoch": 0.5555796300709958,
      "grad_norm": 0.529468834400177,
      "learning_rate": 8.262587179806579e-06,
      "loss": 0.2144,
      "step": 30770
    },
    {
      "epoch": 0.5556699095038965,
      "grad_norm": 0.3170679211616516,
      "learning_rate": 8.259794142442334e-06,
      "loss": 0.199,
      "step": 30775
    },
    {
      "epoch": 0.5557601889367971,
      "grad_norm": 0.4670516848564148,
      "learning_rate": 8.257001245064992e-06,
      "loss": 0.3112,
      "step": 30780
    },
    {
      "epoch": 0.5558504683696979,
      "grad_norm": 0.46530792117118835,
      "learning_rate": 8.254208487899231e-06,
      "loss": 0.2668,
      "step": 30785
    },
    {
      "epoch": 0.5559407478025986,
      "grad_norm": 0.47129592299461365,
      "learning_rate": 8.251415871169701e-06,
      "loss": 0.1948,
      "step": 30790
    },
    {
      "epoch": 0.5560310272354994,
      "grad_norm": 0.5364168882369995,
      "learning_rate": 8.248623395101056e-06,
      "loss": 0.2282,
      "step": 30795
    },
    {
      "epoch": 0.5561213066684,
      "grad_norm": 0.483106791973114,
      "learning_rate": 8.245831059917921e-06,
      "loss": 0.2011,
      "step": 30800
    },
    {
      "epoch": 0.5562115861013007,
      "grad_norm": 0.42646482586860657,
      "learning_rate": 8.243038865844923e-06,
      "loss": 0.3091,
      "step": 30805
    },
    {
      "epoch": 0.5563018655342015,
      "grad_norm": 0.49516451358795166,
      "learning_rate": 8.240246813106674e-06,
      "loss": 0.2004,
      "step": 30810
    },
    {
      "epoch": 0.5563921449671022,
      "grad_norm": 0.41743355989456177,
      "learning_rate": 8.237454901927775e-06,
      "loss": 0.252,
      "step": 30815
    },
    {
      "epoch": 0.5564824244000028,
      "grad_norm": 0.46710988879203796,
      "learning_rate": 8.234663132532812e-06,
      "loss": 0.2055,
      "step": 30820
    },
    {
      "epoch": 0.5565727038329036,
      "grad_norm": 0.48925188183784485,
      "learning_rate": 8.231871505146371e-06,
      "loss": 0.2268,
      "step": 30825
    },
    {
      "epoch": 0.5566629832658043,
      "grad_norm": 0.48817548155784607,
      "learning_rate": 8.229080019993007e-06,
      "loss": 0.2631,
      "step": 30830
    },
    {
      "epoch": 0.5567532626987051,
      "grad_norm": 0.40382134914398193,
      "learning_rate": 8.22628867729728e-06,
      "loss": 0.1754,
      "step": 30835
    },
    {
      "epoch": 0.5568435421316057,
      "grad_norm": 0.6790839433670044,
      "learning_rate": 8.223497477283732e-06,
      "loss": 0.2099,
      "step": 30840
    },
    {
      "epoch": 0.5569338215645064,
      "grad_norm": 0.5202447772026062,
      "learning_rate": 8.2207064201769e-06,
      "loss": 0.1921,
      "step": 30845
    },
    {
      "epoch": 0.5570241009974072,
      "grad_norm": 0.364751398563385,
      "learning_rate": 8.217915506201295e-06,
      "loss": 0.2326,
      "step": 30850
    },
    {
      "epoch": 0.5571143804303079,
      "grad_norm": 0.48742225766181946,
      "learning_rate": 8.215124735581432e-06,
      "loss": 0.3181,
      "step": 30855
    },
    {
      "epoch": 0.5572046598632086,
      "grad_norm": 0.5118826031684875,
      "learning_rate": 8.21233410854181e-06,
      "loss": 0.2389,
      "step": 30860
    },
    {
      "epoch": 0.5572949392961093,
      "grad_norm": 0.7876133322715759,
      "learning_rate": 8.209543625306906e-06,
      "loss": 0.2736,
      "step": 30865
    },
    {
      "epoch": 0.55738521872901,
      "grad_norm": 0.533372700214386,
      "learning_rate": 8.206753286101203e-06,
      "loss": 0.1628,
      "step": 30870
    },
    {
      "epoch": 0.5574754981619108,
      "grad_norm": 0.39007723331451416,
      "learning_rate": 8.203963091149156e-06,
      "loss": 0.2262,
      "step": 30875
    },
    {
      "epoch": 0.5575657775948114,
      "grad_norm": 0.14329008758068085,
      "learning_rate": 8.201173040675223e-06,
      "loss": 0.2332,
      "step": 30880
    },
    {
      "epoch": 0.5576560570277121,
      "grad_norm": 0.5678766369819641,
      "learning_rate": 8.198383134903839e-06,
      "loss": 0.2012,
      "step": 30885
    },
    {
      "epoch": 0.5577463364606129,
      "grad_norm": 0.4618380069732666,
      "learning_rate": 8.195593374059435e-06,
      "loss": 0.3058,
      "step": 30890
    },
    {
      "epoch": 0.5578366158935136,
      "grad_norm": 0.48835116624832153,
      "learning_rate": 8.192803758366421e-06,
      "loss": 0.2008,
      "step": 30895
    },
    {
      "epoch": 0.5579268953264143,
      "grad_norm": 0.4542429745197296,
      "learning_rate": 8.190014288049208e-06,
      "loss": 0.2599,
      "step": 30900
    },
    {
      "epoch": 0.558017174759315,
      "grad_norm": 0.5513837933540344,
      "learning_rate": 8.187224963332184e-06,
      "loss": 0.2203,
      "step": 30905
    },
    {
      "epoch": 0.5581074541922157,
      "grad_norm": 0.3412122130393982,
      "learning_rate": 8.184435784439735e-06,
      "loss": 0.2659,
      "step": 30910
    },
    {
      "epoch": 0.5581977336251165,
      "grad_norm": 0.3571067750453949,
      "learning_rate": 8.181646751596223e-06,
      "loss": 0.2321,
      "step": 30915
    },
    {
      "epoch": 0.5582880130580172,
      "grad_norm": 0.7699437737464905,
      "learning_rate": 8.17885786502601e-06,
      "loss": 0.2148,
      "step": 30920
    },
    {
      "epoch": 0.5583782924909179,
      "grad_norm": 0.4913608729839325,
      "learning_rate": 8.17606912495344e-06,
      "loss": 0.2209,
      "step": 30925
    },
    {
      "epoch": 0.5584685719238186,
      "grad_norm": 0.5387104153633118,
      "learning_rate": 8.173280531602852e-06,
      "loss": 0.3064,
      "step": 30930
    },
    {
      "epoch": 0.5585588513567193,
      "grad_norm": 0.4020043909549713,
      "learning_rate": 8.170492085198555e-06,
      "loss": 0.2623,
      "step": 30935
    },
    {
      "epoch": 0.5586491307896201,
      "grad_norm": 0.4475884735584259,
      "learning_rate": 8.167703785964877e-06,
      "loss": 0.2165,
      "step": 30940
    },
    {
      "epoch": 0.5587394102225207,
      "grad_norm": 0.26421064138412476,
      "learning_rate": 8.164915634126103e-06,
      "loss": 0.187,
      "step": 30945
    },
    {
      "epoch": 0.5588296896554215,
      "grad_norm": 0.43203189969062805,
      "learning_rate": 8.162127629906524e-06,
      "loss": 0.2383,
      "step": 30950
    },
    {
      "epoch": 0.5589199690883222,
      "grad_norm": 0.37303704023361206,
      "learning_rate": 8.159339773530418e-06,
      "loss": 0.2566,
      "step": 30955
    },
    {
      "epoch": 0.5590102485212229,
      "grad_norm": 0.3759598135948181,
      "learning_rate": 8.15655206522204e-06,
      "loss": 0.2628,
      "step": 30960
    },
    {
      "epoch": 0.5591005279541236,
      "grad_norm": 0.2845647633075714,
      "learning_rate": 8.153764505205647e-06,
      "loss": 0.2133,
      "step": 30965
    },
    {
      "epoch": 0.5591908073870243,
      "grad_norm": 0.4346495270729065,
      "learning_rate": 8.150977093705474e-06,
      "loss": 0.1625,
      "step": 30970
    },
    {
      "epoch": 0.559281086819925,
      "grad_norm": 0.5033888816833496,
      "learning_rate": 8.148189830945755e-06,
      "loss": 0.2136,
      "step": 30975
    },
    {
      "epoch": 0.5593713662528258,
      "grad_norm": 0.34858712553977966,
      "learning_rate": 8.145402717150694e-06,
      "loss": 0.2012,
      "step": 30980
    },
    {
      "epoch": 0.5594616456857264,
      "grad_norm": 0.4071477949619293,
      "learning_rate": 8.142615752544504e-06,
      "loss": 0.2675,
      "step": 30985
    },
    {
      "epoch": 0.5595519251186272,
      "grad_norm": 0.15057164430618286,
      "learning_rate": 8.13982893735137e-06,
      "loss": 0.2205,
      "step": 30990
    },
    {
      "epoch": 0.5596422045515279,
      "grad_norm": 0.5766903758049011,
      "learning_rate": 8.137042271795478e-06,
      "loss": 0.2327,
      "step": 30995
    },
    {
      "epoch": 0.5597324839844287,
      "grad_norm": 0.3336317837238312,
      "learning_rate": 8.134255756100983e-06,
      "loss": 0.207,
      "step": 31000
    },
    {
      "epoch": 0.5598227634173293,
      "grad_norm": 0.42386963963508606,
      "learning_rate": 8.131469390492053e-06,
      "loss": 0.2551,
      "step": 31005
    },
    {
      "epoch": 0.55991304285023,
      "grad_norm": 0.5276187062263489,
      "learning_rate": 8.128683175192821e-06,
      "loss": 0.2054,
      "step": 31010
    },
    {
      "epoch": 0.5600033222831308,
      "grad_norm": 0.8299673795700073,
      "learning_rate": 8.125897110427426e-06,
      "loss": 0.2904,
      "step": 31015
    },
    {
      "epoch": 0.5600936017160315,
      "grad_norm": 0.49706336855888367,
      "learning_rate": 8.123111196419976e-06,
      "loss": 0.2738,
      "step": 31020
    },
    {
      "epoch": 0.5601838811489321,
      "grad_norm": 0.8560537695884705,
      "learning_rate": 8.120325433394593e-06,
      "loss": 0.2854,
      "step": 31025
    },
    {
      "epoch": 0.5602741605818329,
      "grad_norm": 0.3300696611404419,
      "learning_rate": 8.117539821575355e-06,
      "loss": 0.3097,
      "step": 31030
    },
    {
      "epoch": 0.5603644400147336,
      "grad_norm": 0.31206315755844116,
      "learning_rate": 8.114754361186355e-06,
      "loss": 0.2221,
      "step": 31035
    },
    {
      "epoch": 0.5604547194476344,
      "grad_norm": 0.5207038521766663,
      "learning_rate": 8.111969052451661e-06,
      "loss": 0.1785,
      "step": 31040
    },
    {
      "epoch": 0.560544998880535,
      "grad_norm": 0.3452952802181244,
      "learning_rate": 8.109183895595325e-06,
      "loss": 0.2017,
      "step": 31045
    },
    {
      "epoch": 0.5606352783134357,
      "grad_norm": 0.5093872547149658,
      "learning_rate": 8.1063988908414e-06,
      "loss": 0.2114,
      "step": 31050
    },
    {
      "epoch": 0.5607255577463365,
      "grad_norm": 0.4056823253631592,
      "learning_rate": 8.103614038413915e-06,
      "loss": 0.2386,
      "step": 31055
    },
    {
      "epoch": 0.5608158371792372,
      "grad_norm": 0.4714640974998474,
      "learning_rate": 8.100829338536895e-06,
      "loss": 0.1914,
      "step": 31060
    },
    {
      "epoch": 0.5609061166121379,
      "grad_norm": 0.435139000415802,
      "learning_rate": 8.098044791434341e-06,
      "loss": 0.1605,
      "step": 31065
    },
    {
      "epoch": 0.5609963960450386,
      "grad_norm": 0.9441424012184143,
      "learning_rate": 8.09526039733026e-06,
      "loss": 0.1847,
      "step": 31070
    },
    {
      "epoch": 0.5610866754779393,
      "grad_norm": 0.369119256734848,
      "learning_rate": 8.092476156448629e-06,
      "loss": 0.1681,
      "step": 31075
    },
    {
      "epoch": 0.5611769549108401,
      "grad_norm": 0.44150403141975403,
      "learning_rate": 8.089692069013424e-06,
      "loss": 0.2804,
      "step": 31080
    },
    {
      "epoch": 0.5612672343437407,
      "grad_norm": 0.5297713279724121,
      "learning_rate": 8.086908135248599e-06,
      "loss": 0.2043,
      "step": 31085
    },
    {
      "epoch": 0.5613575137766414,
      "grad_norm": 0.4562094509601593,
      "learning_rate": 8.08412435537811e-06,
      "loss": 0.237,
      "step": 31090
    },
    {
      "epoch": 0.5614477932095422,
      "grad_norm": 0.5448012351989746,
      "learning_rate": 8.081340729625883e-06,
      "loss": 0.2525,
      "step": 31095
    },
    {
      "epoch": 0.5615380726424429,
      "grad_norm": 0.528515636920929,
      "learning_rate": 8.078557258215847e-06,
      "loss": 0.3582,
      "step": 31100
    },
    {
      "epoch": 0.5616283520753436,
      "grad_norm": 0.4745613932609558,
      "learning_rate": 8.075773941371907e-06,
      "loss": 0.2392,
      "step": 31105
    },
    {
      "epoch": 0.5617186315082443,
      "grad_norm": 0.2186175137758255,
      "learning_rate": 8.072990779317965e-06,
      "loss": 0.1902,
      "step": 31110
    },
    {
      "epoch": 0.561808910941145,
      "grad_norm": 0.5137507319450378,
      "learning_rate": 8.070207772277902e-06,
      "loss": 0.1964,
      "step": 31115
    },
    {
      "epoch": 0.5618991903740458,
      "grad_norm": 0.5128169059753418,
      "learning_rate": 8.067424920475595e-06,
      "loss": 0.2262,
      "step": 31120
    },
    {
      "epoch": 0.5619894698069464,
      "grad_norm": 0.33977770805358887,
      "learning_rate": 8.064642224134898e-06,
      "loss": 0.278,
      "step": 31125
    },
    {
      "epoch": 0.5620797492398472,
      "grad_norm": 0.3779820501804352,
      "learning_rate": 8.061859683479667e-06,
      "loss": 0.1944,
      "step": 31130
    },
    {
      "epoch": 0.5621700286727479,
      "grad_norm": 0.5303663611412048,
      "learning_rate": 8.059077298733734e-06,
      "loss": 0.2224,
      "step": 31135
    },
    {
      "epoch": 0.5622603081056486,
      "grad_norm": 0.5823243260383606,
      "learning_rate": 8.056295070120917e-06,
      "loss": 0.2282,
      "step": 31140
    },
    {
      "epoch": 0.5623505875385493,
      "grad_norm": 0.6812359690666199,
      "learning_rate": 8.053512997865029e-06,
      "loss": 0.2798,
      "step": 31145
    },
    {
      "epoch": 0.56244086697145,
      "grad_norm": 0.3564703166484833,
      "learning_rate": 8.050731082189869e-06,
      "loss": 0.1607,
      "step": 31150
    },
    {
      "epoch": 0.5625311464043508,
      "grad_norm": 0.9905957579612732,
      "learning_rate": 8.047949323319224e-06,
      "loss": 0.2861,
      "step": 31155
    },
    {
      "epoch": 0.5626214258372515,
      "grad_norm": 0.4461768567562103,
      "learning_rate": 8.045167721476859e-06,
      "loss": 0.349,
      "step": 31160
    },
    {
      "epoch": 0.5627117052701521,
      "grad_norm": 0.29345831274986267,
      "learning_rate": 8.042386276886539e-06,
      "loss": 0.3028,
      "step": 31165
    },
    {
      "epoch": 0.5628019847030529,
      "grad_norm": 0.283610999584198,
      "learning_rate": 8.039604989772008e-06,
      "loss": 0.2405,
      "step": 31170
    },
    {
      "epoch": 0.5628922641359536,
      "grad_norm": 0.648016631603241,
      "learning_rate": 8.036823860357006e-06,
      "loss": 0.1439,
      "step": 31175
    },
    {
      "epoch": 0.5629825435688544,
      "grad_norm": 0.426203191280365,
      "learning_rate": 8.034042888865248e-06,
      "loss": 0.2108,
      "step": 31180
    },
    {
      "epoch": 0.563072823001755,
      "grad_norm": 0.3136540949344635,
      "learning_rate": 8.031262075520448e-06,
      "loss": 0.2479,
      "step": 31185
    },
    {
      "epoch": 0.5631631024346557,
      "grad_norm": 0.44544175267219543,
      "learning_rate": 8.028481420546296e-06,
      "loss": 0.3021,
      "step": 31190
    },
    {
      "epoch": 0.5632533818675565,
      "grad_norm": 0.4027963876724243,
      "learning_rate": 8.025700924166485e-06,
      "loss": 0.183,
      "step": 31195
    },
    {
      "epoch": 0.5633436613004572,
      "grad_norm": 0.5416713356971741,
      "learning_rate": 8.022920586604672e-06,
      "loss": 0.2933,
      "step": 31200
    },
    {
      "epoch": 0.5634339407333578,
      "grad_norm": 0.74769526720047,
      "learning_rate": 8.020140408084532e-06,
      "loss": 0.1967,
      "step": 31205
    },
    {
      "epoch": 0.5635242201662586,
      "grad_norm": 0.43441691994667053,
      "learning_rate": 8.017360388829695e-06,
      "loss": 0.1863,
      "step": 31210
    },
    {
      "epoch": 0.5636144995991593,
      "grad_norm": 0.38378357887268066,
      "learning_rate": 8.014580529063802e-06,
      "loss": 0.2372,
      "step": 31215
    },
    {
      "epoch": 0.5637047790320601,
      "grad_norm": 0.321071594953537,
      "learning_rate": 8.011800829010471e-06,
      "loss": 0.2101,
      "step": 31220
    },
    {
      "epoch": 0.5637950584649607,
      "grad_norm": 0.2700953483581543,
      "learning_rate": 8.009021288893306e-06,
      "loss": 0.2174,
      "step": 31225
    },
    {
      "epoch": 0.5638853378978614,
      "grad_norm": 0.5070574879646301,
      "learning_rate": 8.006241908935904e-06,
      "loss": 0.2694,
      "step": 31230
    },
    {
      "epoch": 0.5639756173307622,
      "grad_norm": 0.2650270462036133,
      "learning_rate": 8.003462689361842e-06,
      "loss": 0.2226,
      "step": 31235
    },
    {
      "epoch": 0.5640658967636629,
      "grad_norm": 0.4303354322910309,
      "learning_rate": 8.000683630394694e-06,
      "loss": 0.2199,
      "step": 31240
    },
    {
      "epoch": 0.5641561761965636,
      "grad_norm": 0.4189873933792114,
      "learning_rate": 7.997904732258007e-06,
      "loss": 0.1785,
      "step": 31245
    },
    {
      "epoch": 0.5642464556294643,
      "grad_norm": 0.3254397511482239,
      "learning_rate": 7.995125995175331e-06,
      "loss": 0.3304,
      "step": 31250
    },
    {
      "epoch": 0.564336735062365,
      "grad_norm": 0.48033246397972107,
      "learning_rate": 7.99234741937019e-06,
      "loss": 0.2742,
      "step": 31255
    },
    {
      "epoch": 0.5644270144952658,
      "grad_norm": 0.5633978843688965,
      "learning_rate": 7.989569005066105e-06,
      "loss": 0.1938,
      "step": 31260
    },
    {
      "epoch": 0.5645172939281664,
      "grad_norm": 0.4215300381183624,
      "learning_rate": 7.986790752486571e-06,
      "loss": 0.2811,
      "step": 31265
    },
    {
      "epoch": 0.5646075733610671,
      "grad_norm": 0.49446821212768555,
      "learning_rate": 7.984012661855088e-06,
      "loss": 0.2229,
      "step": 31270
    },
    {
      "epoch": 0.5646978527939679,
      "grad_norm": 0.5683329701423645,
      "learning_rate": 7.981234733395126e-06,
      "loss": 0.3294,
      "step": 31275
    },
    {
      "epoch": 0.5647881322268686,
      "grad_norm": 0.3849588930606842,
      "learning_rate": 7.978456967330153e-06,
      "loss": 0.2557,
      "step": 31280
    },
    {
      "epoch": 0.5648784116597693,
      "grad_norm": 0.6493882536888123,
      "learning_rate": 7.975679363883617e-06,
      "loss": 0.2778,
      "step": 31285
    },
    {
      "epoch": 0.56496869109267,
      "grad_norm": 0.372500479221344,
      "learning_rate": 7.972901923278964e-06,
      "loss": 0.1883,
      "step": 31290
    },
    {
      "epoch": 0.5650589705255707,
      "grad_norm": 0.6053361296653748,
      "learning_rate": 7.970124645739607e-06,
      "loss": 0.2206,
      "step": 31295
    },
    {
      "epoch": 0.5651492499584715,
      "grad_norm": 0.533922016620636,
      "learning_rate": 7.967347531488967e-06,
      "loss": 0.3048,
      "step": 31300
    },
    {
      "epoch": 0.5652395293913721,
      "grad_norm": 0.7627965211868286,
      "learning_rate": 7.964570580750435e-06,
      "loss": 0.2308,
      "step": 31305
    },
    {
      "epoch": 0.5653298088242729,
      "grad_norm": 0.40491706132888794,
      "learning_rate": 7.961793793747407e-06,
      "loss": 0.2021,
      "step": 31310
    },
    {
      "epoch": 0.5654200882571736,
      "grad_norm": 0.258077472448349,
      "learning_rate": 7.959017170703247e-06,
      "loss": 0.2252,
      "step": 31315
    },
    {
      "epoch": 0.5655103676900743,
      "grad_norm": 0.5910612940788269,
      "learning_rate": 7.956240711841311e-06,
      "loss": 0.3261,
      "step": 31320
    },
    {
      "epoch": 0.5656006471229751,
      "grad_norm": 0.22161012887954712,
      "learning_rate": 7.953464417384959e-06,
      "loss": 0.2188,
      "step": 31325
    },
    {
      "epoch": 0.5656909265558757,
      "grad_norm": 0.20220966637134552,
      "learning_rate": 7.950688287557506e-06,
      "loss": 0.2747,
      "step": 31330
    },
    {
      "epoch": 0.5657812059887765,
      "grad_norm": 0.40949514508247375,
      "learning_rate": 7.947912322582288e-06,
      "loss": 0.1938,
      "step": 31335
    },
    {
      "epoch": 0.5658714854216772,
      "grad_norm": 0.718718945980072,
      "learning_rate": 7.945136522682596e-06,
      "loss": 0.1716,
      "step": 31340
    },
    {
      "epoch": 0.5659617648545779,
      "grad_norm": 0.5266116857528687,
      "learning_rate": 7.942360888081735e-06,
      "loss": 0.3613,
      "step": 31345
    },
    {
      "epoch": 0.5660520442874786,
      "grad_norm": 0.4842294454574585,
      "learning_rate": 7.939585419002979e-06,
      "loss": 0.2993,
      "step": 31350
    },
    {
      "epoch": 0.5661423237203793,
      "grad_norm": 0.571361780166626,
      "learning_rate": 7.936810115669597e-06,
      "loss": 0.2093,
      "step": 31355
    },
    {
      "epoch": 0.5662326031532801,
      "grad_norm": 0.5041170120239258,
      "learning_rate": 7.934034978304838e-06,
      "loss": 0.2402,
      "step": 31360
    },
    {
      "epoch": 0.5663228825861808,
      "grad_norm": 0.4833357036113739,
      "learning_rate": 7.931260007131947e-06,
      "loss": 0.2389,
      "step": 31365
    },
    {
      "epoch": 0.5664131620190814,
      "grad_norm": 0.34955894947052,
      "learning_rate": 7.928485202374143e-06,
      "loss": 0.2534,
      "step": 31370
    },
    {
      "epoch": 0.5665034414519822,
      "grad_norm": 0.4219762682914734,
      "learning_rate": 7.925710564254649e-06,
      "loss": 0.3005,
      "step": 31375
    },
    {
      "epoch": 0.5665937208848829,
      "grad_norm": 0.6094689965248108,
      "learning_rate": 7.922936092996652e-06,
      "loss": 0.2565,
      "step": 31380
    },
    {
      "epoch": 0.5666840003177837,
      "grad_norm": 0.30353668332099915,
      "learning_rate": 7.920161788823348e-06,
      "loss": 0.2227,
      "step": 31385
    },
    {
      "epoch": 0.5667742797506843,
      "grad_norm": 0.4556208550930023,
      "learning_rate": 7.917387651957904e-06,
      "loss": 0.1577,
      "step": 31390
    },
    {
      "epoch": 0.566864559183585,
      "grad_norm": 0.29068723320961,
      "learning_rate": 7.914613682623484e-06,
      "loss": 0.2559,
      "step": 31395
    },
    {
      "epoch": 0.5669548386164858,
      "grad_norm": 0.4912876784801483,
      "learning_rate": 7.911839881043229e-06,
      "loss": 0.1858,
      "step": 31400
    },
    {
      "epoch": 0.5670451180493865,
      "grad_norm": 2.5178728103637695,
      "learning_rate": 7.909066247440276e-06,
      "loss": 0.2582,
      "step": 31405
    },
    {
      "epoch": 0.5671353974822871,
      "grad_norm": 0.3810131549835205,
      "learning_rate": 7.906292782037742e-06,
      "loss": 0.1843,
      "step": 31410
    },
    {
      "epoch": 0.5672256769151879,
      "grad_norm": 0.30515459179878235,
      "learning_rate": 7.903519485058725e-06,
      "loss": 0.299,
      "step": 31415
    },
    {
      "epoch": 0.5673159563480886,
      "grad_norm": 0.5835896730422974,
      "learning_rate": 7.90074635672633e-06,
      "loss": 0.2202,
      "step": 31420
    },
    {
      "epoch": 0.5674062357809894,
      "grad_norm": 0.6488636136054993,
      "learning_rate": 7.897973397263622e-06,
      "loss": 0.307,
      "step": 31425
    },
    {
      "epoch": 0.56749651521389,
      "grad_norm": 0.6058459877967834,
      "learning_rate": 7.89520060689367e-06,
      "loss": 0.2541,
      "step": 31430
    },
    {
      "epoch": 0.5675867946467907,
      "grad_norm": 0.4164639711380005,
      "learning_rate": 7.892427985839528e-06,
      "loss": 0.228,
      "step": 31435
    },
    {
      "epoch": 0.5676770740796915,
      "grad_norm": 0.4838501214981079,
      "learning_rate": 7.889655534324233e-06,
      "loss": 0.203,
      "step": 31440
    },
    {
      "epoch": 0.5677673535125922,
      "grad_norm": 0.6691818237304688,
      "learning_rate": 7.886883252570802e-06,
      "loss": 0.253,
      "step": 31445
    },
    {
      "epoch": 0.5678576329454929,
      "grad_norm": 0.18681277334690094,
      "learning_rate": 7.884111140802253e-06,
      "loss": 0.1815,
      "step": 31450
    },
    {
      "epoch": 0.5679479123783936,
      "grad_norm": 0.2960595488548279,
      "learning_rate": 7.881339199241577e-06,
      "loss": 0.1957,
      "step": 31455
    },
    {
      "epoch": 0.5680381918112943,
      "grad_norm": 0.18710044026374817,
      "learning_rate": 7.87856742811176e-06,
      "loss": 0.1687,
      "step": 31460
    },
    {
      "epoch": 0.5681284712441951,
      "grad_norm": 0.5640111565589905,
      "learning_rate": 7.875795827635765e-06,
      "loss": 0.2155,
      "step": 31465
    },
    {
      "epoch": 0.5682187506770957,
      "grad_norm": 0.31908056139945984,
      "learning_rate": 7.873024398036558e-06,
      "loss": 0.2998,
      "step": 31470
    },
    {
      "epoch": 0.5683090301099964,
      "grad_norm": 0.5567411780357361,
      "learning_rate": 7.870253139537068e-06,
      "loss": 0.2355,
      "step": 31475
    },
    {
      "epoch": 0.5683993095428972,
      "grad_norm": 0.582668662071228,
      "learning_rate": 7.86748205236023e-06,
      "loss": 0.2492,
      "step": 31480
    },
    {
      "epoch": 0.5684895889757979,
      "grad_norm": 0.3465553820133209,
      "learning_rate": 7.864711136728956e-06,
      "loss": 0.2323,
      "step": 31485
    },
    {
      "epoch": 0.5685798684086986,
      "grad_norm": 0.5908859372138977,
      "learning_rate": 7.861940392866149e-06,
      "loss": 0.2537,
      "step": 31490
    },
    {
      "epoch": 0.5686701478415993,
      "grad_norm": 0.2922111749649048,
      "learning_rate": 7.859169820994688e-06,
      "loss": 0.2066,
      "step": 31495
    },
    {
      "epoch": 0.5687604272745,
      "grad_norm": 0.44060826301574707,
      "learning_rate": 7.856399421337451e-06,
      "loss": 0.3348,
      "step": 31500
    },
    {
      "epoch": 0.5688507067074008,
      "grad_norm": 0.2694040536880493,
      "learning_rate": 7.853629194117297e-06,
      "loss": 0.167,
      "step": 31505
    },
    {
      "epoch": 0.5689409861403014,
      "grad_norm": 0.3127096891403198,
      "learning_rate": 7.850859139557066e-06,
      "loss": 0.1711,
      "step": 31510
    },
    {
      "epoch": 0.5690312655732022,
      "grad_norm": 0.3559912443161011,
      "learning_rate": 7.848089257879593e-06,
      "loss": 0.1844,
      "step": 31515
    },
    {
      "epoch": 0.5691215450061029,
      "grad_norm": 0.34979164600372314,
      "learning_rate": 7.84531954930769e-06,
      "loss": 0.2305,
      "step": 31520
    },
    {
      "epoch": 0.5692118244390036,
      "grad_norm": 0.8576284050941467,
      "learning_rate": 7.842550014064166e-06,
      "loss": 0.2346,
      "step": 31525
    },
    {
      "epoch": 0.5693021038719043,
      "grad_norm": 0.34732529520988464,
      "learning_rate": 7.839780652371804e-06,
      "loss": 0.1926,
      "step": 31530
    },
    {
      "epoch": 0.569392383304805,
      "grad_norm": 0.3402683734893799,
      "learning_rate": 7.837011464453388e-06,
      "loss": 0.1806,
      "step": 31535
    },
    {
      "epoch": 0.5694826627377058,
      "grad_norm": 0.4833114445209503,
      "learning_rate": 7.83424245053167e-06,
      "loss": 0.329,
      "step": 31540
    },
    {
      "epoch": 0.5695729421706065,
      "grad_norm": 0.35603630542755127,
      "learning_rate": 7.8314736108294e-06,
      "loss": 0.1452,
      "step": 31545
    },
    {
      "epoch": 0.5696632216035071,
      "grad_norm": 0.5631104707717896,
      "learning_rate": 7.82870494556931e-06,
      "loss": 0.198,
      "step": 31550
    },
    {
      "epoch": 0.5697535010364079,
      "grad_norm": 0.36028844118118286,
      "learning_rate": 7.825936454974126e-06,
      "loss": 0.3105,
      "step": 31555
    },
    {
      "epoch": 0.5698437804693086,
      "grad_norm": 0.4279186725616455,
      "learning_rate": 7.823168139266542e-06,
      "loss": 0.1424,
      "step": 31560
    },
    {
      "epoch": 0.5699340599022094,
      "grad_norm": 0.3231222331523895,
      "learning_rate": 7.820399998669254e-06,
      "loss": 0.3305,
      "step": 31565
    },
    {
      "epoch": 0.57002433933511,
      "grad_norm": 0.4611665904521942,
      "learning_rate": 7.81763203340494e-06,
      "loss": 0.1877,
      "step": 31570
    },
    {
      "epoch": 0.5701146187680107,
      "grad_norm": 0.2889973223209381,
      "learning_rate": 7.814864243696265e-06,
      "loss": 0.2433,
      "step": 31575
    },
    {
      "epoch": 0.5702048982009115,
      "grad_norm": 0.38997986912727356,
      "learning_rate": 7.812096629765869e-06,
      "loss": 0.2144,
      "step": 31580
    },
    {
      "epoch": 0.5702951776338122,
      "grad_norm": 0.4267489016056061,
      "learning_rate": 7.809329191836394e-06,
      "loss": 0.2195,
      "step": 31585
    },
    {
      "epoch": 0.5703854570667128,
      "grad_norm": 0.123191699385643,
      "learning_rate": 7.806561930130458e-06,
      "loss": 0.2302,
      "step": 31590
    },
    {
      "epoch": 0.5704757364996136,
      "grad_norm": 0.5885843634605408,
      "learning_rate": 7.803794844870665e-06,
      "loss": 0.2811,
      "step": 31595
    },
    {
      "epoch": 0.5705660159325143,
      "grad_norm": 0.4821052849292755,
      "learning_rate": 7.80102793627961e-06,
      "loss": 0.3085,
      "step": 31600
    },
    {
      "epoch": 0.5706562953654151,
      "grad_norm": 0.4820989668369293,
      "learning_rate": 7.79826120457987e-06,
      "loss": 0.2794,
      "step": 31605
    },
    {
      "epoch": 0.5707465747983157,
      "grad_norm": 0.38792115449905396,
      "learning_rate": 7.795494649994007e-06,
      "loss": 0.1767,
      "step": 31610
    },
    {
      "epoch": 0.5708368542312164,
      "grad_norm": 0.3137624263763428,
      "learning_rate": 7.79272827274457e-06,
      "loss": 0.2751,
      "step": 31615
    },
    {
      "epoch": 0.5709271336641172,
      "grad_norm": 0.3731110692024231,
      "learning_rate": 7.7899620730541e-06,
      "loss": 0.2167,
      "step": 31620
    },
    {
      "epoch": 0.5710174130970179,
      "grad_norm": 0.33237844705581665,
      "learning_rate": 7.787196051145108e-06,
      "loss": 0.1972,
      "step": 31625
    },
    {
      "epoch": 0.5711076925299186,
      "grad_norm": 0.49587783217430115,
      "learning_rate": 7.784430207240105e-06,
      "loss": 0.293,
      "step": 31630
    },
    {
      "epoch": 0.5711979719628193,
      "grad_norm": 0.5755000710487366,
      "learning_rate": 7.781664541561584e-06,
      "loss": 0.2391,
      "step": 31635
    },
    {
      "epoch": 0.57128825139572,
      "grad_norm": 0.3513915538787842,
      "learning_rate": 7.778899054332024e-06,
      "loss": 0.2511,
      "step": 31640
    },
    {
      "epoch": 0.5713785308286208,
      "grad_norm": 0.39008522033691406,
      "learning_rate": 7.776133745773882e-06,
      "loss": 0.2076,
      "step": 31645
    },
    {
      "epoch": 0.5714688102615214,
      "grad_norm": 0.45953187346458435,
      "learning_rate": 7.773368616109616e-06,
      "loss": 0.2583,
      "step": 31650
    },
    {
      "epoch": 0.5715590896944222,
      "grad_norm": 0.24481430649757385,
      "learning_rate": 7.77060366556165e-06,
      "loss": 0.2592,
      "step": 31655
    },
    {
      "epoch": 0.5716493691273229,
      "grad_norm": 0.13567975163459778,
      "learning_rate": 7.767838894352413e-06,
      "loss": 0.2453,
      "step": 31660
    },
    {
      "epoch": 0.5717396485602236,
      "grad_norm": 0.5409976840019226,
      "learning_rate": 7.765074302704307e-06,
      "loss": 0.2279,
      "step": 31665
    },
    {
      "epoch": 0.5718299279931243,
      "grad_norm": 0.4621885120868683,
      "learning_rate": 7.762309890839726e-06,
      "loss": 0.2845,
      "step": 31670
    },
    {
      "epoch": 0.571920207426025,
      "grad_norm": 0.3942146897315979,
      "learning_rate": 7.75954565898104e-06,
      "loss": 0.2943,
      "step": 31675
    },
    {
      "epoch": 0.5720104868589257,
      "grad_norm": 0.6451921463012695,
      "learning_rate": 7.75678160735062e-06,
      "loss": 0.2337,
      "step": 31680
    },
    {
      "epoch": 0.5721007662918265,
      "grad_norm": 0.422530859708786,
      "learning_rate": 7.754017736170812e-06,
      "loss": 0.2573,
      "step": 31685
    },
    {
      "epoch": 0.5721910457247271,
      "grad_norm": 0.36159053444862366,
      "learning_rate": 7.751254045663943e-06,
      "loss": 0.2493,
      "step": 31690
    },
    {
      "epoch": 0.5722813251576279,
      "grad_norm": 0.4430783987045288,
      "learning_rate": 7.748490536052337e-06,
      "loss": 0.3274,
      "step": 31695
    },
    {
      "epoch": 0.5723716045905286,
      "grad_norm": 0.8427174091339111,
      "learning_rate": 7.745727207558297e-06,
      "loss": 0.2285,
      "step": 31700
    },
    {
      "epoch": 0.5724618840234293,
      "grad_norm": 0.6411793231964111,
      "learning_rate": 7.742964060404117e-06,
      "loss": 0.2674,
      "step": 31705
    },
    {
      "epoch": 0.5725521634563301,
      "grad_norm": 0.5644553303718567,
      "learning_rate": 7.740201094812062e-06,
      "loss": 0.2603,
      "step": 31710
    },
    {
      "epoch": 0.5726424428892307,
      "grad_norm": 0.3624364137649536,
      "learning_rate": 7.737438311004402e-06,
      "loss": 0.2745,
      "step": 31715
    },
    {
      "epoch": 0.5727327223221315,
      "grad_norm": 0.46105074882507324,
      "learning_rate": 7.734675709203377e-06,
      "loss": 0.3357,
      "step": 31720
    },
    {
      "epoch": 0.5728230017550322,
      "grad_norm": 0.43723955750465393,
      "learning_rate": 7.731913289631224e-06,
      "loss": 0.2978,
      "step": 31725
    },
    {
      "epoch": 0.5729132811879329,
      "grad_norm": 0.5800166726112366,
      "learning_rate": 7.729151052510152e-06,
      "loss": 0.1662,
      "step": 31730
    },
    {
      "epoch": 0.5730035606208336,
      "grad_norm": 0.2920645773410797,
      "learning_rate": 7.726388998062372e-06,
      "loss": 0.2246,
      "step": 31735
    },
    {
      "epoch": 0.5730938400537343,
      "grad_norm": 0.6474556922912598,
      "learning_rate": 7.723627126510063e-06,
      "loss": 0.2873,
      "step": 31740
    },
    {
      "epoch": 0.5731841194866351,
      "grad_norm": 0.3297922909259796,
      "learning_rate": 7.720865438075401e-06,
      "loss": 0.1754,
      "step": 31745
    },
    {
      "epoch": 0.5732743989195358,
      "grad_norm": 0.5210285782814026,
      "learning_rate": 7.718103932980545e-06,
      "loss": 0.2546,
      "step": 31750
    },
    {
      "epoch": 0.5733646783524364,
      "grad_norm": 0.47570544481277466,
      "learning_rate": 7.715342611447638e-06,
      "loss": 0.2637,
      "step": 31755
    },
    {
      "epoch": 0.5734549577853372,
      "grad_norm": 0.5185160040855408,
      "learning_rate": 7.712581473698804e-06,
      "loss": 0.2799,
      "step": 31760
    },
    {
      "epoch": 0.5735452372182379,
      "grad_norm": 0.340856671333313,
      "learning_rate": 7.709820519956162e-06,
      "loss": 0.2269,
      "step": 31765
    },
    {
      "epoch": 0.5736355166511387,
      "grad_norm": 0.3276570737361908,
      "learning_rate": 7.707059750441803e-06,
      "loss": 0.2887,
      "step": 31770
    },
    {
      "epoch": 0.5737257960840393,
      "grad_norm": 0.26869645714759827,
      "learning_rate": 7.704299165377822e-06,
      "loss": 0.264,
      "step": 31775
    },
    {
      "epoch": 0.57381607551694,
      "grad_norm": 0.4268071949481964,
      "learning_rate": 7.701538764986281e-06,
      "loss": 0.242,
      "step": 31780
    },
    {
      "epoch": 0.5739063549498408,
      "grad_norm": 0.722257137298584,
      "learning_rate": 7.69877854948923e-06,
      "loss": 0.2949,
      "step": 31785
    },
    {
      "epoch": 0.5739966343827415,
      "grad_norm": 0.4261210858821869,
      "learning_rate": 7.696018519108717e-06,
      "loss": 0.2736,
      "step": 31790
    },
    {
      "epoch": 0.5740869138156421,
      "grad_norm": 0.4993356168270111,
      "learning_rate": 7.69325867406676e-06,
      "loss": 0.3129,
      "step": 31795
    },
    {
      "epoch": 0.5741771932485429,
      "grad_norm": 0.5304279923439026,
      "learning_rate": 7.690499014585375e-06,
      "loss": 0.2423,
      "step": 31800
    },
    {
      "epoch": 0.5742674726814436,
      "grad_norm": 0.48582887649536133,
      "learning_rate": 7.687739540886549e-06,
      "loss": 0.2367,
      "step": 31805
    },
    {
      "epoch": 0.5743577521143444,
      "grad_norm": 0.4940268397331238,
      "learning_rate": 7.684980253192267e-06,
      "loss": 0.3923,
      "step": 31810
    },
    {
      "epoch": 0.574448031547245,
      "grad_norm": 0.32182779908180237,
      "learning_rate": 7.682221151724489e-06,
      "loss": 0.1695,
      "step": 31815
    },
    {
      "epoch": 0.5745383109801457,
      "grad_norm": 0.3178272247314453,
      "learning_rate": 7.679462236705172e-06,
      "loss": 0.3104,
      "step": 31820
    },
    {
      "epoch": 0.5746285904130465,
      "grad_norm": 0.24268971383571625,
      "learning_rate": 7.676703508356243e-06,
      "loss": 0.2052,
      "step": 31825
    },
    {
      "epoch": 0.5747188698459472,
      "grad_norm": 0.4058937430381775,
      "learning_rate": 7.673944966899625e-06,
      "loss": 0.1739,
      "step": 31830
    },
    {
      "epoch": 0.5748091492788479,
      "grad_norm": 0.5511011481285095,
      "learning_rate": 7.671186612557223e-06,
      "loss": 0.2352,
      "step": 31835
    },
    {
      "epoch": 0.5748994287117486,
      "grad_norm": 0.37227874994277954,
      "learning_rate": 7.668428445550929e-06,
      "loss": 0.2458,
      "step": 31840
    },
    {
      "epoch": 0.5749897081446493,
      "grad_norm": 0.36259961128234863,
      "learning_rate": 7.665670466102607e-06,
      "loss": 0.2716,
      "step": 31845
    },
    {
      "epoch": 0.5750799875775501,
      "grad_norm": 0.4528560936450958,
      "learning_rate": 7.66291267443413e-06,
      "loss": 0.2813,
      "step": 31850
    },
    {
      "epoch": 0.5751702670104507,
      "grad_norm": 0.5381774306297302,
      "learning_rate": 7.660155070767333e-06,
      "loss": 0.1767,
      "step": 31855
    },
    {
      "epoch": 0.5752605464433515,
      "grad_norm": 0.382550448179245,
      "learning_rate": 7.657397655324049e-06,
      "loss": 0.2337,
      "step": 31860
    },
    {
      "epoch": 0.5753508258762522,
      "grad_norm": 0.32369717955589294,
      "learning_rate": 7.654640428326095e-06,
      "loss": 0.2262,
      "step": 31865
    },
    {
      "epoch": 0.5754411053091529,
      "grad_norm": 0.3207179307937622,
      "learning_rate": 7.65188338999526e-06,
      "loss": 0.2847,
      "step": 31870
    },
    {
      "epoch": 0.5755313847420536,
      "grad_norm": 0.43713873624801636,
      "learning_rate": 7.649126540553337e-06,
      "loss": 0.2279,
      "step": 31875
    },
    {
      "epoch": 0.5756216641749543,
      "grad_norm": 0.9669202566146851,
      "learning_rate": 7.64636988022209e-06,
      "loss": 0.2079,
      "step": 31880
    },
    {
      "epoch": 0.575711943607855,
      "grad_norm": 0.38378670811653137,
      "learning_rate": 7.643613409223277e-06,
      "loss": 0.2033,
      "step": 31885
    },
    {
      "epoch": 0.5758022230407558,
      "grad_norm": 0.5948004126548767,
      "learning_rate": 7.640857127778627e-06,
      "loss": 0.2914,
      "step": 31890
    },
    {
      "epoch": 0.5758925024736564,
      "grad_norm": 0.4891936779022217,
      "learning_rate": 7.638101036109873e-06,
      "loss": 0.2132,
      "step": 31895
    },
    {
      "epoch": 0.5759827819065572,
      "grad_norm": 0.2774111330509186,
      "learning_rate": 7.635345134438716e-06,
      "loss": 0.2747,
      "step": 31900
    },
    {
      "epoch": 0.5760730613394579,
      "grad_norm": 0.4318869411945343,
      "learning_rate": 7.632589422986853e-06,
      "loss": 0.226,
      "step": 31905
    },
    {
      "epoch": 0.5761633407723586,
      "grad_norm": 0.3877994120121002,
      "learning_rate": 7.62983390197595e-06,
      "loss": 0.2415,
      "step": 31910
    },
    {
      "epoch": 0.5762536202052593,
      "grad_norm": 0.6382290124893188,
      "learning_rate": 7.6270785716276875e-06,
      "loss": 0.2094,
      "step": 31915
    },
    {
      "epoch": 0.57634389963816,
      "grad_norm": 0.6943665146827698,
      "learning_rate": 7.624323432163694e-06,
      "loss": 0.2478,
      "step": 31920
    },
    {
      "epoch": 0.5764341790710608,
      "grad_norm": 0.3250424265861511,
      "learning_rate": 7.621568483805611e-06,
      "loss": 0.2835,
      "step": 31925
    },
    {
      "epoch": 0.5765244585039615,
      "grad_norm": 0.36512619256973267,
      "learning_rate": 7.618813726775048e-06,
      "loss": 0.2425,
      "step": 31930
    },
    {
      "epoch": 0.5766147379368621,
      "grad_norm": 0.454352468252182,
      "learning_rate": 7.6160591612936115e-06,
      "loss": 0.2637,
      "step": 31935
    },
    {
      "epoch": 0.5767050173697629,
      "grad_norm": 0.30730774998664856,
      "learning_rate": 7.613304787582879e-06,
      "loss": 0.3078,
      "step": 31940
    },
    {
      "epoch": 0.5767952968026636,
      "grad_norm": 0.3757621943950653,
      "learning_rate": 7.610550605864425e-06,
      "loss": 0.2881,
      "step": 31945
    },
    {
      "epoch": 0.5768855762355644,
      "grad_norm": 0.29341045022010803,
      "learning_rate": 7.607796616359799e-06,
      "loss": 0.2679,
      "step": 31950
    },
    {
      "epoch": 0.576975855668465,
      "grad_norm": 0.3899611830711365,
      "learning_rate": 7.605042819290549e-06,
      "loss": 0.2254,
      "step": 31955
    },
    {
      "epoch": 0.5770661351013657,
      "grad_norm": 0.6977159976959229,
      "learning_rate": 7.602289214878187e-06,
      "loss": 0.3099,
      "step": 31960
    },
    {
      "epoch": 0.5771564145342665,
      "grad_norm": 0.43100249767303467,
      "learning_rate": 7.5995358033442225e-06,
      "loss": 0.2899,
      "step": 31965
    },
    {
      "epoch": 0.5772466939671672,
      "grad_norm": 0.35019803047180176,
      "learning_rate": 7.596782584910154e-06,
      "loss": 0.1998,
      "step": 31970
    },
    {
      "epoch": 0.5773369734000678,
      "grad_norm": 0.3060927987098694,
      "learning_rate": 7.5940295597974474e-06,
      "loss": 0.2194,
      "step": 31975
    },
    {
      "epoch": 0.5774272528329686,
      "grad_norm": 0.3989834487438202,
      "learning_rate": 7.591276728227576e-06,
      "loss": 0.2037,
      "step": 31980
    },
    {
      "epoch": 0.5775175322658693,
      "grad_norm": 0.3755948543548584,
      "learning_rate": 7.588524090421973e-06,
      "loss": 0.2055,
      "step": 31985
    },
    {
      "epoch": 0.5776078116987701,
      "grad_norm": 0.34243911504745483,
      "learning_rate": 7.585771646602078e-06,
      "loss": 0.2308,
      "step": 31990
    },
    {
      "epoch": 0.5776980911316707,
      "grad_norm": 0.3287220895290375,
      "learning_rate": 7.583019396989299e-06,
      "loss": 0.2884,
      "step": 31995
    },
    {
      "epoch": 0.5777883705645714,
      "grad_norm": 0.7058871388435364,
      "learning_rate": 7.580267341805039e-06,
      "loss": 0.2626,
      "step": 32000
    },
    {
      "epoch": 0.5778786499974722,
      "grad_norm": 0.41973498463630676,
      "learning_rate": 7.577515481270675e-06,
      "loss": 0.2699,
      "step": 32005
    },
    {
      "epoch": 0.5779689294303729,
      "grad_norm": 0.3011660873889923,
      "learning_rate": 7.574763815607579e-06,
      "loss": 0.1988,
      "step": 32010
    },
    {
      "epoch": 0.5780592088632736,
      "grad_norm": 0.3949172794818878,
      "learning_rate": 7.572012345037098e-06,
      "loss": 0.2731,
      "step": 32015
    },
    {
      "epoch": 0.5781494882961743,
      "grad_norm": 0.3967495858669281,
      "learning_rate": 7.569261069780577e-06,
      "loss": 0.23,
      "step": 32020
    },
    {
      "epoch": 0.578239767729075,
      "grad_norm": 0.390185683965683,
      "learning_rate": 7.566509990059323e-06,
      "loss": 0.197,
      "step": 32025
    },
    {
      "epoch": 0.5783300471619758,
      "grad_norm": 0.5422402620315552,
      "learning_rate": 7.563759106094652e-06,
      "loss": 0.1938,
      "step": 32030
    },
    {
      "epoch": 0.5784203265948764,
      "grad_norm": 0.44239312410354614,
      "learning_rate": 7.561008418107843e-06,
      "loss": 0.2468,
      "step": 32035
    },
    {
      "epoch": 0.5785106060277772,
      "grad_norm": 0.6479213833808899,
      "learning_rate": 7.558257926320177e-06,
      "loss": 0.1806,
      "step": 32040
    },
    {
      "epoch": 0.5786008854606779,
      "grad_norm": 0.3620591163635254,
      "learning_rate": 7.555507630952905e-06,
      "loss": 0.2389,
      "step": 32045
    },
    {
      "epoch": 0.5786911648935786,
      "grad_norm": 0.3467635214328766,
      "learning_rate": 7.552757532227276e-06,
      "loss": 0.2576,
      "step": 32050
    },
    {
      "epoch": 0.5787814443264793,
      "grad_norm": 0.7262997031211853,
      "learning_rate": 7.5500076303645076e-06,
      "loss": 0.3137,
      "step": 32055
    },
    {
      "epoch": 0.57887172375938,
      "grad_norm": 0.4510249197483063,
      "learning_rate": 7.54725792558581e-06,
      "loss": 0.1948,
      "step": 32060
    },
    {
      "epoch": 0.5789620031922807,
      "grad_norm": 0.4553157687187195,
      "learning_rate": 7.544508418112384e-06,
      "loss": 0.2515,
      "step": 32065
    },
    {
      "epoch": 0.5790522826251815,
      "grad_norm": 0.38665473461151123,
      "learning_rate": 7.541759108165397e-06,
      "loss": 0.2272,
      "step": 32070
    },
    {
      "epoch": 0.5791425620580821,
      "grad_norm": 0.4248069226741791,
      "learning_rate": 7.53900999596602e-06,
      "loss": 0.2566,
      "step": 32075
    },
    {
      "epoch": 0.5792328414909829,
      "grad_norm": 0.4079587757587433,
      "learning_rate": 7.536261081735393e-06,
      "loss": 0.2314,
      "step": 32080
    },
    {
      "epoch": 0.5793231209238836,
      "grad_norm": 0.5409365892410278,
      "learning_rate": 7.533512365694654e-06,
      "loss": 0.2204,
      "step": 32085
    },
    {
      "epoch": 0.5794134003567843,
      "grad_norm": 0.3574941158294678,
      "learning_rate": 7.530763848064907e-06,
      "loss": 0.1702,
      "step": 32090
    },
    {
      "epoch": 0.579503679789685,
      "grad_norm": 0.32814323902130127,
      "learning_rate": 7.5280155290672575e-06,
      "loss": 0.183,
      "step": 32095
    },
    {
      "epoch": 0.5795939592225857,
      "grad_norm": 0.5189509391784668,
      "learning_rate": 7.525267408922784e-06,
      "loss": 0.1933,
      "step": 32100
    },
    {
      "epoch": 0.5796842386554865,
      "grad_norm": 0.5349361300468445,
      "learning_rate": 7.522519487852558e-06,
      "loss": 0.2453,
      "step": 32105
    },
    {
      "epoch": 0.5797745180883872,
      "grad_norm": 0.6736632585525513,
      "learning_rate": 7.519771766077619e-06,
      "loss": 0.1977,
      "step": 32110
    },
    {
      "epoch": 0.579864797521288,
      "grad_norm": 0.2429421991109848,
      "learning_rate": 7.517024243819017e-06,
      "loss": 0.1993,
      "step": 32115
    },
    {
      "epoch": 0.5799550769541886,
      "grad_norm": 0.3097635805606842,
      "learning_rate": 7.5142769212977565e-06,
      "loss": 0.1585,
      "step": 32120
    },
    {
      "epoch": 0.5800453563870893,
      "grad_norm": 0.30012813210487366,
      "learning_rate": 7.511529798734847e-06,
      "loss": 0.2413,
      "step": 32125
    },
    {
      "epoch": 0.5801356358199901,
      "grad_norm": 0.4978336691856384,
      "learning_rate": 7.508782876351268e-06,
      "loss": 0.2074,
      "step": 32130
    },
    {
      "epoch": 0.5802259152528908,
      "grad_norm": 0.287640243768692,
      "learning_rate": 7.5060361543680005e-06,
      "loss": 0.2102,
      "step": 32135
    },
    {
      "epoch": 0.5803161946857914,
      "grad_norm": 0.33799034357070923,
      "learning_rate": 7.503289633005988e-06,
      "loss": 0.2385,
      "step": 32140
    },
    {
      "epoch": 0.5804064741186922,
      "grad_norm": 0.4288344979286194,
      "learning_rate": 7.500543312486171e-06,
      "loss": 0.3285,
      "step": 32145
    },
    {
      "epoch": 0.5804967535515929,
      "grad_norm": 0.4685920178890228,
      "learning_rate": 7.497797193029475e-06,
      "loss": 0.2195,
      "step": 32150
    },
    {
      "epoch": 0.5805870329844937,
      "grad_norm": 0.3669572174549103,
      "learning_rate": 7.495051274856797e-06,
      "loss": 0.2371,
      "step": 32155
    },
    {
      "epoch": 0.5806773124173943,
      "grad_norm": 0.28180867433547974,
      "learning_rate": 7.492305558189034e-06,
      "loss": 0.1351,
      "step": 32160
    },
    {
      "epoch": 0.580767591850295,
      "grad_norm": 0.33606743812561035,
      "learning_rate": 7.489560043247055e-06,
      "loss": 0.2492,
      "step": 32165
    },
    {
      "epoch": 0.5808578712831958,
      "grad_norm": 0.2527913451194763,
      "learning_rate": 7.486814730251721e-06,
      "loss": 0.236,
      "step": 32170
    },
    {
      "epoch": 0.5809481507160965,
      "grad_norm": 0.6620507836341858,
      "learning_rate": 7.484069619423864e-06,
      "loss": 0.2081,
      "step": 32175
    },
    {
      "epoch": 0.5810384301489971,
      "grad_norm": 0.3179631233215332,
      "learning_rate": 7.481324710984319e-06,
      "loss": 0.1368,
      "step": 32180
    },
    {
      "epoch": 0.5811287095818979,
      "grad_norm": 0.49275219440460205,
      "learning_rate": 7.478580005153884e-06,
      "loss": 0.2072,
      "step": 32185
    },
    {
      "epoch": 0.5812189890147986,
      "grad_norm": 0.6345303654670715,
      "learning_rate": 7.475835502153358e-06,
      "loss": 0.2794,
      "step": 32190
    },
    {
      "epoch": 0.5813092684476994,
      "grad_norm": 0.32289382815361023,
      "learning_rate": 7.47309120220351e-06,
      "loss": 0.2749,
      "step": 32195
    },
    {
      "epoch": 0.5813995478806,
      "grad_norm": 0.47582992911338806,
      "learning_rate": 7.470347105525108e-06,
      "loss": 0.1932,
      "step": 32200
    },
    {
      "epoch": 0.5814898273135007,
      "grad_norm": 0.6040963530540466,
      "learning_rate": 7.467603212338884e-06,
      "loss": 0.2711,
      "step": 32205
    },
    {
      "epoch": 0.5815801067464015,
      "grad_norm": 0.3335484564304352,
      "learning_rate": 7.464859522865572e-06,
      "loss": 0.2152,
      "step": 32210
    },
    {
      "epoch": 0.5816703861793022,
      "grad_norm": 0.6475858688354492,
      "learning_rate": 7.462116037325877e-06,
      "loss": 0.2821,
      "step": 32215
    },
    {
      "epoch": 0.5817606656122029,
      "grad_norm": 0.325191855430603,
      "learning_rate": 7.4593727559404984e-06,
      "loss": 0.2612,
      "step": 32220
    },
    {
      "epoch": 0.5818509450451036,
      "grad_norm": 0.5029043555259705,
      "learning_rate": 7.456629678930105e-06,
      "loss": 0.2514,
      "step": 32225
    },
    {
      "epoch": 0.5819412244780043,
      "grad_norm": 0.5356692671775818,
      "learning_rate": 7.453886806515365e-06,
      "loss": 0.2775,
      "step": 32230
    },
    {
      "epoch": 0.5820315039109051,
      "grad_norm": 0.41439715027809143,
      "learning_rate": 7.45114413891692e-06,
      "loss": 0.2357,
      "step": 32235
    },
    {
      "epoch": 0.5821217833438057,
      "grad_norm": 0.6161235570907593,
      "learning_rate": 7.4484016763553924e-06,
      "loss": 0.2713,
      "step": 32240
    },
    {
      "epoch": 0.5822120627767065,
      "grad_norm": 0.5415605902671814,
      "learning_rate": 7.445659419051403e-06,
      "loss": 0.1144,
      "step": 32245
    },
    {
      "epoch": 0.5823023422096072,
      "grad_norm": 0.45095542073249817,
      "learning_rate": 7.442917367225537e-06,
      "loss": 0.2546,
      "step": 32250
    },
    {
      "epoch": 0.5823926216425079,
      "grad_norm": 0.4577171504497528,
      "learning_rate": 7.44017552109838e-06,
      "loss": 0.2644,
      "step": 32255
    },
    {
      "epoch": 0.5824829010754086,
      "grad_norm": 0.5528016686439514,
      "learning_rate": 7.437433880890488e-06,
      "loss": 0.2314,
      "step": 32260
    },
    {
      "epoch": 0.5825731805083093,
      "grad_norm": 0.3215571343898773,
      "learning_rate": 7.434692446822413e-06,
      "loss": 0.195,
      "step": 32265
    },
    {
      "epoch": 0.58266345994121,
      "grad_norm": 0.2885240614414215,
      "learning_rate": 7.431951219114676e-06,
      "loss": 0.2912,
      "step": 32270
    },
    {
      "epoch": 0.5827537393741108,
      "grad_norm": 0.41558414697647095,
      "learning_rate": 7.4292101979877924e-06,
      "loss": 0.2877,
      "step": 32275
    },
    {
      "epoch": 0.5828440188070114,
      "grad_norm": 0.6126312613487244,
      "learning_rate": 7.426469383662256e-06,
      "loss": 0.2178,
      "step": 32280
    },
    {
      "epoch": 0.5829342982399122,
      "grad_norm": 0.39613085985183716,
      "learning_rate": 7.42372877635855e-06,
      "loss": 0.2333,
      "step": 32285
    },
    {
      "epoch": 0.5830245776728129,
      "grad_norm": 0.2641087472438812,
      "learning_rate": 7.420988376297128e-06,
      "loss": 0.2064,
      "step": 32290
    },
    {
      "epoch": 0.5831148571057136,
      "grad_norm": 0.2797715365886688,
      "learning_rate": 7.418248183698442e-06,
      "loss": 0.3491,
      "step": 32295
    },
    {
      "epoch": 0.5832051365386143,
      "grad_norm": 0.3410707712173462,
      "learning_rate": 7.415508198782917e-06,
      "loss": 0.1643,
      "step": 32300
    },
    {
      "epoch": 0.583295415971515,
      "grad_norm": 0.6921510696411133,
      "learning_rate": 7.412768421770969e-06,
      "loss": 0.2448,
      "step": 32305
    },
    {
      "epoch": 0.5833856954044158,
      "grad_norm": 0.3319537341594696,
      "learning_rate": 7.410028852882986e-06,
      "loss": 0.2072,
      "step": 32310
    },
    {
      "epoch": 0.5834759748373165,
      "grad_norm": 0.46647992730140686,
      "learning_rate": 7.407289492339358e-06,
      "loss": 0.2774,
      "step": 32315
    },
    {
      "epoch": 0.5835662542702171,
      "grad_norm": 0.36552879214286804,
      "learning_rate": 7.404550340360434e-06,
      "loss": 0.3033,
      "step": 32320
    },
    {
      "epoch": 0.5836565337031179,
      "grad_norm": 0.6458025574684143,
      "learning_rate": 7.401811397166566e-06,
      "loss": 0.2489,
      "step": 32325
    },
    {
      "epoch": 0.5837468131360186,
      "grad_norm": 0.42477449774742126,
      "learning_rate": 7.399072662978083e-06,
      "loss": 0.2144,
      "step": 32330
    },
    {
      "epoch": 0.5838370925689194,
      "grad_norm": 0.3511560559272766,
      "learning_rate": 7.3963341380152896e-06,
      "loss": 0.1876,
      "step": 32335
    },
    {
      "epoch": 0.58392737200182,
      "grad_norm": 0.27430459856987,
      "learning_rate": 7.393595822498486e-06,
      "loss": 0.2967,
      "step": 32340
    },
    {
      "epoch": 0.5840176514347207,
      "grad_norm": 0.39398086071014404,
      "learning_rate": 7.3908577166479465e-06,
      "loss": 0.1741,
      "step": 32345
    },
    {
      "epoch": 0.5841079308676215,
      "grad_norm": 0.3841555714607239,
      "learning_rate": 7.3881198206839375e-06,
      "loss": 0.189,
      "step": 32350
    },
    {
      "epoch": 0.5841982103005222,
      "grad_norm": 2.1301867961883545,
      "learning_rate": 7.385382134826695e-06,
      "loss": 0.2152,
      "step": 32355
    },
    {
      "epoch": 0.5842884897334228,
      "grad_norm": 0.3354480564594269,
      "learning_rate": 7.382644659296451e-06,
      "loss": 0.3059,
      "step": 32360
    },
    {
      "epoch": 0.5843787691663236,
      "grad_norm": 0.5217795372009277,
      "learning_rate": 7.379907394313412e-06,
      "loss": 0.1943,
      "step": 32365
    },
    {
      "epoch": 0.5844690485992243,
      "grad_norm": 0.4222526252269745,
      "learning_rate": 7.3771703400977745e-06,
      "loss": 0.2983,
      "step": 32370
    },
    {
      "epoch": 0.5845593280321251,
      "grad_norm": 0.5341085195541382,
      "learning_rate": 7.374433496869713e-06,
      "loss": 0.2763,
      "step": 32375
    },
    {
      "epoch": 0.5846496074650257,
      "grad_norm": 2.0725042819976807,
      "learning_rate": 7.37169686484939e-06,
      "loss": 0.3016,
      "step": 32380
    },
    {
      "epoch": 0.5847398868979264,
      "grad_norm": 0.2955710291862488,
      "learning_rate": 7.368960444256941e-06,
      "loss": 0.2444,
      "step": 32385
    },
    {
      "epoch": 0.5848301663308272,
      "grad_norm": 0.47486481070518494,
      "learning_rate": 7.3662242353124965e-06,
      "loss": 0.2634,
      "step": 32390
    },
    {
      "epoch": 0.5849204457637279,
      "grad_norm": 0.4509022831916809,
      "learning_rate": 7.36348823823616e-06,
      "loss": 0.2499,
      "step": 32395
    },
    {
      "epoch": 0.5850107251966286,
      "grad_norm": 0.3450268805027008,
      "learning_rate": 7.3607524532480325e-06,
      "loss": 0.2775,
      "step": 32400
    },
    {
      "epoch": 0.5851010046295293,
      "grad_norm": 0.35095328092575073,
      "learning_rate": 7.358016880568174e-06,
      "loss": 0.2789,
      "step": 32405
    },
    {
      "epoch": 0.58519128406243,
      "grad_norm": 0.37022092938423157,
      "learning_rate": 7.355281520416652e-06,
      "loss": 0.2526,
      "step": 32410
    },
    {
      "epoch": 0.5852815634953308,
      "grad_norm": 0.6604461669921875,
      "learning_rate": 7.352546373013499e-06,
      "loss": 0.2138,
      "step": 32415
    },
    {
      "epoch": 0.5853718429282314,
      "grad_norm": 0.41028112173080444,
      "learning_rate": 7.349811438578747e-06,
      "loss": 0.1188,
      "step": 32420
    },
    {
      "epoch": 0.5854621223611322,
      "grad_norm": 2.1903560161590576,
      "learning_rate": 7.347076717332393e-06,
      "loss": 0.154,
      "step": 32425
    },
    {
      "epoch": 0.5855524017940329,
      "grad_norm": 0.5380918383598328,
      "learning_rate": 7.344342209494427e-06,
      "loss": 0.2395,
      "step": 32430
    },
    {
      "epoch": 0.5856426812269336,
      "grad_norm": 0.6822723150253296,
      "learning_rate": 7.341607915284824e-06,
      "loss": 0.2498,
      "step": 32435
    },
    {
      "epoch": 0.5857329606598343,
      "grad_norm": 0.6931403875350952,
      "learning_rate": 7.3388738349235335e-06,
      "loss": 0.2981,
      "step": 32440
    },
    {
      "epoch": 0.585823240092735,
      "grad_norm": 0.5006046891212463,
      "learning_rate": 7.336139968630499e-06,
      "loss": 0.2418,
      "step": 32445
    },
    {
      "epoch": 0.5859135195256358,
      "grad_norm": 0.38321810960769653,
      "learning_rate": 7.333406316625631e-06,
      "loss": 0.1812,
      "step": 32450
    },
    {
      "epoch": 0.5860037989585365,
      "grad_norm": 0.28258997201919556,
      "learning_rate": 7.330672879128839e-06,
      "loss": 0.2136,
      "step": 32455
    },
    {
      "epoch": 0.5860940783914371,
      "grad_norm": 0.4144040048122406,
      "learning_rate": 7.3279396563600056e-06,
      "loss": 0.2228,
      "step": 32460
    },
    {
      "epoch": 0.5861843578243379,
      "grad_norm": 0.45015108585357666,
      "learning_rate": 7.325206648539003e-06,
      "loss": 0.2909,
      "step": 32465
    },
    {
      "epoch": 0.5862746372572386,
      "grad_norm": 0.3811364769935608,
      "learning_rate": 7.322473855885672e-06,
      "loss": 0.247,
      "step": 32470
    },
    {
      "epoch": 0.5863649166901393,
      "grad_norm": 0.2765819728374481,
      "learning_rate": 7.319741278619854e-06,
      "loss": 0.2313,
      "step": 32475
    },
    {
      "epoch": 0.58645519612304,
      "grad_norm": 0.7468791007995605,
      "learning_rate": 7.317008916961361e-06,
      "loss": 0.2351,
      "step": 32480
    },
    {
      "epoch": 0.5865454755559407,
      "grad_norm": 0.6331852078437805,
      "learning_rate": 7.3142767711299985e-06,
      "loss": 0.1694,
      "step": 32485
    },
    {
      "epoch": 0.5866357549888415,
      "grad_norm": 0.18905432522296906,
      "learning_rate": 7.311544841345538e-06,
      "loss": 0.2926,
      "step": 32490
    },
    {
      "epoch": 0.5867260344217422,
      "grad_norm": 0.6289650797843933,
      "learning_rate": 7.3088131278277495e-06,
      "loss": 0.2271,
      "step": 32495
    },
    {
      "epoch": 0.586816313854643,
      "grad_norm": 0.5063753724098206,
      "learning_rate": 7.306081630796378e-06,
      "loss": 0.2422,
      "step": 32500
    },
    {
      "epoch": 0.5869065932875436,
      "grad_norm": 0.48694613575935364,
      "learning_rate": 7.303350350471153e-06,
      "loss": 0.171,
      "step": 32505
    },
    {
      "epoch": 0.5869968727204443,
      "grad_norm": 0.39464613795280457,
      "learning_rate": 7.300619287071787e-06,
      "loss": 0.2704,
      "step": 32510
    },
    {
      "epoch": 0.5870871521533451,
      "grad_norm": 0.7511200904846191,
      "learning_rate": 7.297888440817971e-06,
      "loss": 0.2126,
      "step": 32515
    },
    {
      "epoch": 0.5871774315862458,
      "grad_norm": 0.36095550656318665,
      "learning_rate": 7.2951578119293845e-06,
      "loss": 0.2151,
      "step": 32520
    },
    {
      "epoch": 0.5872677110191464,
      "grad_norm": 0.692754328250885,
      "learning_rate": 7.292427400625683e-06,
      "loss": 0.3606,
      "step": 32525
    },
    {
      "epoch": 0.5873579904520472,
      "grad_norm": 0.35889407992362976,
      "learning_rate": 7.2896972071265175e-06,
      "loss": 0.2457,
      "step": 32530
    },
    {
      "epoch": 0.5874482698849479,
      "grad_norm": 0.5592451691627502,
      "learning_rate": 7.2869672316515006e-06,
      "loss": 0.2534,
      "step": 32535
    },
    {
      "epoch": 0.5875385493178487,
      "grad_norm": 0.5456913113594055,
      "learning_rate": 7.284237474420246e-06,
      "loss": 0.2476,
      "step": 32540
    },
    {
      "epoch": 0.5876288287507493,
      "grad_norm": 0.44270724058151245,
      "learning_rate": 7.281507935652339e-06,
      "loss": 0.2677,
      "step": 32545
    },
    {
      "epoch": 0.58771910818365,
      "grad_norm": 0.3586839735507965,
      "learning_rate": 7.278778615567357e-06,
      "loss": 0.355,
      "step": 32550
    },
    {
      "epoch": 0.5878093876165508,
      "grad_norm": 0.9107780456542969,
      "learning_rate": 7.276049514384846e-06,
      "loss": 0.1568,
      "step": 32555
    },
    {
      "epoch": 0.5878996670494515,
      "grad_norm": 0.588352382183075,
      "learning_rate": 7.27332063232435e-06,
      "loss": 0.2128,
      "step": 32560
    },
    {
      "epoch": 0.5879899464823521,
      "grad_norm": 0.3175543248653412,
      "learning_rate": 7.2705919696053815e-06,
      "loss": 0.1631,
      "step": 32565
    },
    {
      "epoch": 0.5880802259152529,
      "grad_norm": 0.5141485333442688,
      "learning_rate": 7.267863526447447e-06,
      "loss": 0.3225,
      "step": 32570
    },
    {
      "epoch": 0.5881705053481536,
      "grad_norm": 0.38632819056510925,
      "learning_rate": 7.265135303070024e-06,
      "loss": 0.2094,
      "step": 32575
    },
    {
      "epoch": 0.5882607847810544,
      "grad_norm": 0.459495484828949,
      "learning_rate": 7.262407299692586e-06,
      "loss": 0.2506,
      "step": 32580
    },
    {
      "epoch": 0.588351064213955,
      "grad_norm": 0.9379053711891174,
      "learning_rate": 7.259679516534573e-06,
      "loss": 0.2083,
      "step": 32585
    },
    {
      "epoch": 0.5884413436468557,
      "grad_norm": 0.494588702917099,
      "learning_rate": 7.256951953815421e-06,
      "loss": 0.2465,
      "step": 32590
    },
    {
      "epoch": 0.5885316230797565,
      "grad_norm": 0.6023609638214111,
      "learning_rate": 7.2542246117545365e-06,
      "loss": 0.2024,
      "step": 32595
    },
    {
      "epoch": 0.5886219025126572,
      "grad_norm": 0.2940469980239868,
      "learning_rate": 7.2514974905713245e-06,
      "loss": 0.2355,
      "step": 32600
    },
    {
      "epoch": 0.5887121819455579,
      "grad_norm": 0.33728957176208496,
      "learning_rate": 7.248770590485153e-06,
      "loss": 0.2037,
      "step": 32605
    },
    {
      "epoch": 0.5888024613784586,
      "grad_norm": 0.3679303824901581,
      "learning_rate": 7.246043911715382e-06,
      "loss": 0.1134,
      "step": 32610
    },
    {
      "epoch": 0.5888927408113593,
      "grad_norm": 0.5711658596992493,
      "learning_rate": 7.24331745448136e-06,
      "loss": 0.3191,
      "step": 32615
    },
    {
      "epoch": 0.5889830202442601,
      "grad_norm": 0.6049374341964722,
      "learning_rate": 7.2405912190024e-06,
      "loss": 0.1782,
      "step": 32620
    },
    {
      "epoch": 0.5890732996771607,
      "grad_norm": 0.5340908765792847,
      "learning_rate": 7.237865205497817e-06,
      "loss": 0.3328,
      "step": 32625
    },
    {
      "epoch": 0.5891635791100615,
      "grad_norm": 0.3141300082206726,
      "learning_rate": 7.235139414186893e-06,
      "loss": 0.3038,
      "step": 32630
    },
    {
      "epoch": 0.5892538585429622,
      "grad_norm": 0.36305537819862366,
      "learning_rate": 7.232413845288904e-06,
      "loss": 0.1934,
      "step": 32635
    },
    {
      "epoch": 0.5893441379758629,
      "grad_norm": 0.4140510559082031,
      "learning_rate": 7.229688499023094e-06,
      "loss": 0.2361,
      "step": 32640
    },
    {
      "epoch": 0.5894344174087636,
      "grad_norm": 0.4857751429080963,
      "learning_rate": 7.226963375608709e-06,
      "loss": 0.2085,
      "step": 32645
    },
    {
      "epoch": 0.5895246968416643,
      "grad_norm": 0.5394529700279236,
      "learning_rate": 7.224238475264953e-06,
      "loss": 0.2648,
      "step": 32650
    },
    {
      "epoch": 0.589614976274565,
      "grad_norm": 0.7949308753013611,
      "learning_rate": 7.221513798211032e-06,
      "loss": 0.2172,
      "step": 32655
    },
    {
      "epoch": 0.5897052557074658,
      "grad_norm": 0.6003508567810059,
      "learning_rate": 7.218789344666122e-06,
      "loss": 0.3284,
      "step": 32660
    },
    {
      "epoch": 0.5897955351403664,
      "grad_norm": 0.377872109413147,
      "learning_rate": 7.216065114849393e-06,
      "loss": 0.2713,
      "step": 32665
    },
    {
      "epoch": 0.5898858145732672,
      "grad_norm": 0.4554804265499115,
      "learning_rate": 7.2133411089799795e-06,
      "loss": 0.2544,
      "step": 32670
    },
    {
      "epoch": 0.5899760940061679,
      "grad_norm": 0.4866463541984558,
      "learning_rate": 7.210617327277015e-06,
      "loss": 0.2311,
      "step": 32675
    },
    {
      "epoch": 0.5900663734390686,
      "grad_norm": 0.522548496723175,
      "learning_rate": 7.207893769959604e-06,
      "loss": 0.2194,
      "step": 32680
    },
    {
      "epoch": 0.5901566528719693,
      "grad_norm": 0.4356546103954315,
      "learning_rate": 7.205170437246843e-06,
      "loss": 0.2707,
      "step": 32685
    },
    {
      "epoch": 0.59024693230487,
      "grad_norm": 0.24004743993282318,
      "learning_rate": 7.202447329357793e-06,
      "loss": 0.2574,
      "step": 32690
    },
    {
      "epoch": 0.5903372117377708,
      "grad_norm": 0.28580644726753235,
      "learning_rate": 7.199724446511522e-06,
      "loss": 0.1791,
      "step": 32695
    },
    {
      "epoch": 0.5904274911706715,
      "grad_norm": 0.5693247318267822,
      "learning_rate": 7.197001788927056e-06,
      "loss": 0.2817,
      "step": 32700
    },
    {
      "epoch": 0.5905177706035721,
      "grad_norm": 0.46006160974502563,
      "learning_rate": 7.194279356823416e-06,
      "loss": 0.3109,
      "step": 32705
    },
    {
      "epoch": 0.5906080500364729,
      "grad_norm": 0.38426798582077026,
      "learning_rate": 7.191557150419605e-06,
      "loss": 0.2108,
      "step": 32710
    },
    {
      "epoch": 0.5906983294693736,
      "grad_norm": 1.0084168910980225,
      "learning_rate": 7.188835169934598e-06,
      "loss": 0.2506,
      "step": 32715
    },
    {
      "epoch": 0.5907886089022744,
      "grad_norm": 0.4460688829421997,
      "learning_rate": 7.1861134155873654e-06,
      "loss": 0.265,
      "step": 32720
    },
    {
      "epoch": 0.590878888335175,
      "grad_norm": 1.0033804178237915,
      "learning_rate": 7.183391887596847e-06,
      "loss": 0.1897,
      "step": 32725
    },
    {
      "epoch": 0.5909691677680757,
      "grad_norm": 0.34205472469329834,
      "learning_rate": 7.180670586181977e-06,
      "loss": 0.2416,
      "step": 32730
    },
    {
      "epoch": 0.5910594472009765,
      "grad_norm": 0.7768039107322693,
      "learning_rate": 7.177949511561655e-06,
      "loss": 0.2599,
      "step": 32735
    },
    {
      "epoch": 0.5911497266338772,
      "grad_norm": 0.49444419145584106,
      "learning_rate": 7.175228663954779e-06,
      "loss": 0.2712,
      "step": 32740
    },
    {
      "epoch": 0.5912400060667778,
      "grad_norm": 0.1657266914844513,
      "learning_rate": 7.172508043580217e-06,
      "loss": 0.2728,
      "step": 32745
    },
    {
      "epoch": 0.5913302854996786,
      "grad_norm": 0.5309212803840637,
      "learning_rate": 7.169787650656828e-06,
      "loss": 0.3384,
      "step": 32750
    },
    {
      "epoch": 0.5914205649325793,
      "grad_norm": 0.7677445411682129,
      "learning_rate": 7.16706748540344e-06,
      "loss": 0.1973,
      "step": 32755
    },
    {
      "epoch": 0.5915108443654801,
      "grad_norm": 0.4121074080467224,
      "learning_rate": 7.164347548038882e-06,
      "loss": 0.2144,
      "step": 32760
    },
    {
      "epoch": 0.5916011237983807,
      "grad_norm": 0.5791410207748413,
      "learning_rate": 7.161627838781941e-06,
      "loss": 0.2754,
      "step": 32765
    },
    {
      "epoch": 0.5916914032312814,
      "grad_norm": 0.5827525854110718,
      "learning_rate": 7.158908357851406e-06,
      "loss": 0.251,
      "step": 32770
    },
    {
      "epoch": 0.5917816826641822,
      "grad_norm": 0.7230418920516968,
      "learning_rate": 7.156189105466034e-06,
      "loss": 0.2533,
      "step": 32775
    },
    {
      "epoch": 0.5918719620970829,
      "grad_norm": 0.40231138467788696,
      "learning_rate": 7.153470081844578e-06,
      "loss": 0.1542,
      "step": 32780
    },
    {
      "epoch": 0.5919622415299836,
      "grad_norm": 0.771411657333374,
      "learning_rate": 7.150751287205754e-06,
      "loss": 0.2126,
      "step": 32785
    },
    {
      "epoch": 0.5920525209628843,
      "grad_norm": 0.5192276835441589,
      "learning_rate": 7.148032721768271e-06,
      "loss": 0.2392,
      "step": 32790
    },
    {
      "epoch": 0.592142800395785,
      "grad_norm": 0.5683626532554626,
      "learning_rate": 7.145314385750824e-06,
      "loss": 0.3019,
      "step": 32795
    },
    {
      "epoch": 0.5922330798286858,
      "grad_norm": 0.3925086557865143,
      "learning_rate": 7.142596279372076e-06,
      "loss": 0.278,
      "step": 32800
    },
    {
      "epoch": 0.5923233592615864,
      "grad_norm": 0.2804316580295563,
      "learning_rate": 7.139878402850683e-06,
      "loss": 0.156,
      "step": 32805
    },
    {
      "epoch": 0.5924136386944872,
      "grad_norm": 0.4466037452220917,
      "learning_rate": 7.137160756405274e-06,
      "loss": 0.2184,
      "step": 32810
    },
    {
      "epoch": 0.5925039181273879,
      "grad_norm": 0.38808631896972656,
      "learning_rate": 7.134443340254474e-06,
      "loss": 0.1733,
      "step": 32815
    },
    {
      "epoch": 0.5925941975602886,
      "grad_norm": 0.344827264547348,
      "learning_rate": 7.131726154616866e-06,
      "loss": 0.2711,
      "step": 32820
    },
    {
      "epoch": 0.5926844769931893,
      "grad_norm": 0.4249587953090668,
      "learning_rate": 7.12900919971104e-06,
      "loss": 0.2162,
      "step": 32825
    },
    {
      "epoch": 0.59277475642609,
      "grad_norm": 0.4515315592288971,
      "learning_rate": 7.1262924757555455e-06,
      "loss": 0.2211,
      "step": 32830
    },
    {
      "epoch": 0.5928650358589908,
      "grad_norm": 0.5152904391288757,
      "learning_rate": 7.12357598296893e-06,
      "loss": 0.2673,
      "step": 32835
    },
    {
      "epoch": 0.5929553152918915,
      "grad_norm": 0.4642157554626465,
      "learning_rate": 7.120859721569711e-06,
      "loss": 0.2428,
      "step": 32840
    },
    {
      "epoch": 0.5930455947247921,
      "grad_norm": 0.24216291308403015,
      "learning_rate": 7.118143691776398e-06,
      "loss": 0.2074,
      "step": 32845
    },
    {
      "epoch": 0.5931358741576929,
      "grad_norm": 0.5426397323608398,
      "learning_rate": 7.115427893807467e-06,
      "loss": 0.2045,
      "step": 32850
    },
    {
      "epoch": 0.5932261535905936,
      "grad_norm": 0.6953004598617554,
      "learning_rate": 7.112712327881393e-06,
      "loss": 0.2555,
      "step": 32855
    },
    {
      "epoch": 0.5933164330234943,
      "grad_norm": 0.5227711200714111,
      "learning_rate": 7.1099969942166156e-06,
      "loss": 0.2336,
      "step": 32860
    },
    {
      "epoch": 0.593406712456395,
      "grad_norm": 0.38676440715789795,
      "learning_rate": 7.107281893031572e-06,
      "loss": 0.1827,
      "step": 32865
    },
    {
      "epoch": 0.5934969918892957,
      "grad_norm": 0.4496510922908783,
      "learning_rate": 7.104567024544665e-06,
      "loss": 0.1892,
      "step": 32870
    },
    {
      "epoch": 0.5935872713221965,
      "grad_norm": 0.4076612889766693,
      "learning_rate": 7.1018523889742906e-06,
      "loss": 0.1925,
      "step": 32875
    },
    {
      "epoch": 0.5936775507550972,
      "grad_norm": 0.44218653440475464,
      "learning_rate": 7.099137986538822e-06,
      "loss": 0.2463,
      "step": 32880
    },
    {
      "epoch": 0.5937678301879978,
      "grad_norm": 0.32306867837905884,
      "learning_rate": 7.096423817456605e-06,
      "loss": 0.191,
      "step": 32885
    },
    {
      "epoch": 0.5938581096208986,
      "grad_norm": 0.3924742043018341,
      "learning_rate": 7.093709881945987e-06,
      "loss": 0.2466,
      "step": 32890
    },
    {
      "epoch": 0.5939483890537993,
      "grad_norm": 0.2994252145290375,
      "learning_rate": 7.090996180225274e-06,
      "loss": 0.1696,
      "step": 32895
    },
    {
      "epoch": 0.5940386684867001,
      "grad_norm": 0.3526621162891388,
      "learning_rate": 7.0882827125127695e-06,
      "loss": 0.2948,
      "step": 32900
    },
    {
      "epoch": 0.5941289479196008,
      "grad_norm": 0.3149809241294861,
      "learning_rate": 7.085569479026748e-06,
      "loss": 0.2851,
      "step": 32905
    },
    {
      "epoch": 0.5942192273525014,
      "grad_norm": 0.5721583366394043,
      "learning_rate": 7.0828564799854786e-06,
      "loss": 0.2253,
      "step": 32910
    },
    {
      "epoch": 0.5943095067854022,
      "grad_norm": 0.45385658740997314,
      "learning_rate": 7.0801437156071884e-06,
      "loss": 0.1538,
      "step": 32915
    },
    {
      "epoch": 0.5943997862183029,
      "grad_norm": 0.22235175967216492,
      "learning_rate": 7.0774311861101105e-06,
      "loss": 0.3371,
      "step": 32920
    },
    {
      "epoch": 0.5944900656512037,
      "grad_norm": 0.44551515579223633,
      "learning_rate": 7.074718891712442e-06,
      "loss": 0.3237,
      "step": 32925
    },
    {
      "epoch": 0.5945803450841043,
      "grad_norm": 0.42134904861450195,
      "learning_rate": 7.072006832632376e-06,
      "loss": 0.3219,
      "step": 32930
    },
    {
      "epoch": 0.594670624517005,
      "grad_norm": 0.327347993850708,
      "learning_rate": 7.069295009088065e-06,
      "loss": 0.2123,
      "step": 32935
    },
    {
      "epoch": 0.5947609039499058,
      "grad_norm": 0.300333708524704,
      "learning_rate": 7.066583421297666e-06,
      "loss": 0.2104,
      "step": 32940
    },
    {
      "epoch": 0.5948511833828065,
      "grad_norm": 0.6798262000083923,
      "learning_rate": 7.063872069479299e-06,
      "loss": 0.2733,
      "step": 32945
    },
    {
      "epoch": 0.5949414628157071,
      "grad_norm": 0.3732709586620331,
      "learning_rate": 7.0611609538510805e-06,
      "loss": 0.2049,
      "step": 32950
    },
    {
      "epoch": 0.5950317422486079,
      "grad_norm": 0.3809985816478729,
      "learning_rate": 7.0584500746310914e-06,
      "loss": 0.2661,
      "step": 32955
    },
    {
      "epoch": 0.5951220216815086,
      "grad_norm": 0.35851752758026123,
      "learning_rate": 7.055739432037412e-06,
      "loss": 0.2184,
      "step": 32960
    },
    {
      "epoch": 0.5952123011144094,
      "grad_norm": 0.5948862433433533,
      "learning_rate": 7.053029026288085e-06,
      "loss": 0.1534,
      "step": 32965
    },
    {
      "epoch": 0.59530258054731,
      "grad_norm": 0.4455398917198181,
      "learning_rate": 7.050318857601147e-06,
      "loss": 0.1618,
      "step": 32970
    },
    {
      "epoch": 0.5953928599802107,
      "grad_norm": 0.3599335551261902,
      "learning_rate": 7.047608926194613e-06,
      "loss": 0.26,
      "step": 32975
    },
    {
      "epoch": 0.5954831394131115,
      "grad_norm": 0.37924250960350037,
      "learning_rate": 7.0448992322864715e-06,
      "loss": 0.2788,
      "step": 32980
    },
    {
      "epoch": 0.5955734188460122,
      "grad_norm": 0.3209358751773834,
      "learning_rate": 7.042189776094704e-06,
      "loss": 0.1891,
      "step": 32985
    },
    {
      "epoch": 0.5956636982789129,
      "grad_norm": 0.5446608662605286,
      "learning_rate": 7.03948055783726e-06,
      "loss": 0.1934,
      "step": 32990
    },
    {
      "epoch": 0.5957539777118136,
      "grad_norm": 0.28908249735832214,
      "learning_rate": 7.0367715777320845e-06,
      "loss": 0.2733,
      "step": 32995
    },
    {
      "epoch": 0.5958442571447143,
      "grad_norm": 0.38413670659065247,
      "learning_rate": 7.034062835997087e-06,
      "loss": 0.1384,
      "step": 33000
    },
    {
      "epoch": 0.5959345365776151,
      "grad_norm": 0.5856424570083618,
      "learning_rate": 7.031354332850171e-06,
      "loss": 0.1553,
      "step": 33005
    },
    {
      "epoch": 0.5960248160105157,
      "grad_norm": 0.3369595408439636,
      "learning_rate": 7.028646068509214e-06,
      "loss": 0.2167,
      "step": 33010
    },
    {
      "epoch": 0.5961150954434165,
      "grad_norm": 0.7805834412574768,
      "learning_rate": 7.02593804319208e-06,
      "loss": 0.221,
      "step": 33015
    },
    {
      "epoch": 0.5962053748763172,
      "grad_norm": 0.6036295294761658,
      "learning_rate": 7.0232302571166025e-06,
      "loss": 0.1743,
      "step": 33020
    },
    {
      "epoch": 0.5962956543092179,
      "grad_norm": 0.26456376910209656,
      "learning_rate": 7.020522710500612e-06,
      "loss": 0.1735,
      "step": 33025
    },
    {
      "epoch": 0.5963859337421186,
      "grad_norm": 0.3909691572189331,
      "learning_rate": 7.017815403561903e-06,
      "loss": 0.176,
      "step": 33030
    },
    {
      "epoch": 0.5964762131750193,
      "grad_norm": 0.450097918510437,
      "learning_rate": 7.015108336518264e-06,
      "loss": 0.2082,
      "step": 33035
    },
    {
      "epoch": 0.59656649260792,
      "grad_norm": 0.3938777446746826,
      "learning_rate": 7.012401509587456e-06,
      "loss": 0.2654,
      "step": 33040
    },
    {
      "epoch": 0.5966567720408208,
      "grad_norm": 0.6687896847724915,
      "learning_rate": 7.009694922987227e-06,
      "loss": 0.2151,
      "step": 33045
    },
    {
      "epoch": 0.5967470514737214,
      "grad_norm": 0.6720011234283447,
      "learning_rate": 7.006988576935296e-06,
      "loss": 0.2315,
      "step": 33050
    },
    {
      "epoch": 0.5968373309066222,
      "grad_norm": 0.45662808418273926,
      "learning_rate": 7.0042824716493774e-06,
      "loss": 0.2633,
      "step": 33055
    },
    {
      "epoch": 0.5969276103395229,
      "grad_norm": 0.39432138204574585,
      "learning_rate": 7.001576607347148e-06,
      "loss": 0.2786,
      "step": 33060
    },
    {
      "epoch": 0.5970178897724236,
      "grad_norm": 0.2883530557155609,
      "learning_rate": 6.998870984246286e-06,
      "loss": 0.1384,
      "step": 33065
    },
    {
      "epoch": 0.5971081692053243,
      "grad_norm": 0.6928462982177734,
      "learning_rate": 6.9961656025644284e-06,
      "loss": 0.1864,
      "step": 33070
    },
    {
      "epoch": 0.597198448638225,
      "grad_norm": 0.4874042272567749,
      "learning_rate": 6.993460462519208e-06,
      "loss": 0.2796,
      "step": 33075
    },
    {
      "epoch": 0.5972887280711258,
      "grad_norm": 0.4482225477695465,
      "learning_rate": 6.990755564328237e-06,
      "loss": 0.2195,
      "step": 33080
    },
    {
      "epoch": 0.5973790075040265,
      "grad_norm": 0.43891894817352295,
      "learning_rate": 6.988050908209096e-06,
      "loss": 0.1704,
      "step": 33085
    },
    {
      "epoch": 0.5974692869369271,
      "grad_norm": 0.39821308851242065,
      "learning_rate": 6.985346494379367e-06,
      "loss": 0.1956,
      "step": 33090
    },
    {
      "epoch": 0.5975595663698279,
      "grad_norm": 0.8562854528427124,
      "learning_rate": 6.98264232305659e-06,
      "loss": 0.2367,
      "step": 33095
    },
    {
      "epoch": 0.5976498458027286,
      "grad_norm": 0.511570394039154,
      "learning_rate": 6.9799383944583e-06,
      "loss": 0.2898,
      "step": 33100
    },
    {
      "epoch": 0.5977401252356294,
      "grad_norm": 0.21128128468990326,
      "learning_rate": 6.977234708802007e-06,
      "loss": 0.2213,
      "step": 33105
    },
    {
      "epoch": 0.59783040466853,
      "grad_norm": 0.5206707715988159,
      "learning_rate": 6.974531266305209e-06,
      "loss": 0.2196,
      "step": 33110
    },
    {
      "epoch": 0.5979206841014307,
      "grad_norm": 0.6969338655471802,
      "learning_rate": 6.971828067185369e-06,
      "loss": 0.2228,
      "step": 33115
    },
    {
      "epoch": 0.5980109635343315,
      "grad_norm": 5.248167037963867,
      "learning_rate": 6.969125111659945e-06,
      "loss": 0.2213,
      "step": 33120
    },
    {
      "epoch": 0.5981012429672322,
      "grad_norm": 0.26893338561058044,
      "learning_rate": 6.966422399946367e-06,
      "loss": 0.2328,
      "step": 33125
    },
    {
      "epoch": 0.5981915224001328,
      "grad_norm": 0.29222434759140015,
      "learning_rate": 6.9637199322620565e-06,
      "loss": 0.2759,
      "step": 33130
    },
    {
      "epoch": 0.5982818018330336,
      "grad_norm": 0.2278512418270111,
      "learning_rate": 6.961017708824395e-06,
      "loss": 0.2972,
      "step": 33135
    },
    {
      "epoch": 0.5983720812659343,
      "grad_norm": 0.3829987049102783,
      "learning_rate": 6.958315729850769e-06,
      "loss": 0.2152,
      "step": 33140
    },
    {
      "epoch": 0.5984623606988351,
      "grad_norm": 0.4814859628677368,
      "learning_rate": 6.9556139955585225e-06,
      "loss": 0.2729,
      "step": 33145
    },
    {
      "epoch": 0.5985526401317357,
      "grad_norm": 0.5911121368408203,
      "learning_rate": 6.952912506164997e-06,
      "loss": 0.1972,
      "step": 33150
    },
    {
      "epoch": 0.5986429195646364,
      "grad_norm": 0.5242965817451477,
      "learning_rate": 6.950211261887509e-06,
      "loss": 0.2519,
      "step": 33155
    },
    {
      "epoch": 0.5987331989975372,
      "grad_norm": 0.3182903528213501,
      "learning_rate": 6.947510262943347e-06,
      "loss": 0.1171,
      "step": 33160
    },
    {
      "epoch": 0.5988234784304379,
      "grad_norm": 0.6158583164215088,
      "learning_rate": 6.944809509549793e-06,
      "loss": 0.2047,
      "step": 33165
    },
    {
      "epoch": 0.5989137578633386,
      "grad_norm": 0.6978740692138672,
      "learning_rate": 6.942109001924098e-06,
      "loss": 0.1853,
      "step": 33170
    },
    {
      "epoch": 0.5990040372962393,
      "grad_norm": 0.46398329734802246,
      "learning_rate": 6.939408740283504e-06,
      "loss": 0.1128,
      "step": 33175
    },
    {
      "epoch": 0.59909431672914,
      "grad_norm": 0.4759972393512726,
      "learning_rate": 6.936708724845221e-06,
      "loss": 0.2341,
      "step": 33180
    },
    {
      "epoch": 0.5991845961620408,
      "grad_norm": 0.7083093523979187,
      "learning_rate": 6.9340089558264505e-06,
      "loss": 0.2312,
      "step": 33185
    },
    {
      "epoch": 0.5992748755949414,
      "grad_norm": 0.5469648838043213,
      "learning_rate": 6.931309433444367e-06,
      "loss": 0.288,
      "step": 33190
    },
    {
      "epoch": 0.5993651550278422,
      "grad_norm": 0.8999037146568298,
      "learning_rate": 6.92861015791613e-06,
      "loss": 0.2401,
      "step": 33195
    },
    {
      "epoch": 0.5994554344607429,
      "grad_norm": 1.071241021156311,
      "learning_rate": 6.925911129458871e-06,
      "loss": 0.2853,
      "step": 33200
    },
    {
      "epoch": 0.5995457138936436,
      "grad_norm": 0.4732885956764221,
      "learning_rate": 6.923212348289714e-06,
      "loss": 0.2811,
      "step": 33205
    },
    {
      "epoch": 0.5996359933265443,
      "grad_norm": 0.31907421350479126,
      "learning_rate": 6.92051381462575e-06,
      "loss": 0.1849,
      "step": 33210
    },
    {
      "epoch": 0.599726272759445,
      "grad_norm": 0.7735522389411926,
      "learning_rate": 6.917815528684062e-06,
      "loss": 0.2664,
      "step": 33215
    },
    {
      "epoch": 0.5998165521923458,
      "grad_norm": 0.41051408648490906,
      "learning_rate": 6.915117490681703e-06,
      "loss": 0.2254,
      "step": 33220
    },
    {
      "epoch": 0.5999068316252465,
      "grad_norm": 0.8531498312950134,
      "learning_rate": 6.912419700835717e-06,
      "loss": 0.3404,
      "step": 33225
    },
    {
      "epoch": 0.5999971110581471,
      "grad_norm": 0.40225955843925476,
      "learning_rate": 6.9097221593631126e-06,
      "loss": 0.2003,
      "step": 33230
    },
    {
      "epoch": 0.6000873904910479,
      "grad_norm": 0.3249049782752991,
      "learning_rate": 6.907024866480895e-06,
      "loss": 0.2132,
      "step": 33235
    },
    {
      "epoch": 0.6001776699239486,
      "grad_norm": 0.6028874516487122,
      "learning_rate": 6.904327822406036e-06,
      "loss": 0.2482,
      "step": 33240
    },
    {
      "epoch": 0.6002679493568494,
      "grad_norm": 0.3450670540332794,
      "learning_rate": 6.9016310273555e-06,
      "loss": 0.1889,
      "step": 33245
    },
    {
      "epoch": 0.60035822878975,
      "grad_norm": 0.5652427077293396,
      "learning_rate": 6.898934481546219e-06,
      "loss": 0.2245,
      "step": 33250
    },
    {
      "epoch": 0.6004485082226507,
      "grad_norm": 0.7112207412719727,
      "learning_rate": 6.89623818519511e-06,
      "loss": 0.2746,
      "step": 33255
    },
    {
      "epoch": 0.6005387876555515,
      "grad_norm": 0.5124552249908447,
      "learning_rate": 6.893542138519076e-06,
      "loss": 0.2693,
      "step": 33260
    },
    {
      "epoch": 0.6006290670884522,
      "grad_norm": 0.4290370047092438,
      "learning_rate": 6.890846341734987e-06,
      "loss": 0.1819,
      "step": 33265
    },
    {
      "epoch": 0.6007193465213528,
      "grad_norm": 0.45447850227355957,
      "learning_rate": 6.888150795059705e-06,
      "loss": 0.3399,
      "step": 33270
    },
    {
      "epoch": 0.6008096259542536,
      "grad_norm": 0.28496402502059937,
      "learning_rate": 6.885455498710065e-06,
      "loss": 0.2014,
      "step": 33275
    },
    {
      "epoch": 0.6008999053871543,
      "grad_norm": 0.3123587369918823,
      "learning_rate": 6.882760452902888e-06,
      "loss": 0.2066,
      "step": 33280
    },
    {
      "epoch": 0.6009901848200551,
      "grad_norm": 0.5603709816932678,
      "learning_rate": 6.880065657854965e-06,
      "loss": 0.2138,
      "step": 33285
    },
    {
      "epoch": 0.6010804642529558,
      "grad_norm": 0.3382869362831116,
      "learning_rate": 6.87737111378308e-06,
      "loss": 0.2526,
      "step": 33290
    },
    {
      "epoch": 0.6011707436858564,
      "grad_norm": 0.500745415687561,
      "learning_rate": 6.874676820903983e-06,
      "loss": 0.3137,
      "step": 33295
    },
    {
      "epoch": 0.6012610231187572,
      "grad_norm": 0.3618326187133789,
      "learning_rate": 6.871982779434412e-06,
      "loss": 0.2506,
      "step": 33300
    },
    {
      "epoch": 0.6013513025516579,
      "grad_norm": 0.481738418340683,
      "learning_rate": 6.869288989591082e-06,
      "loss": 0.2275,
      "step": 33305
    },
    {
      "epoch": 0.6014415819845587,
      "grad_norm": 0.4610210657119751,
      "learning_rate": 6.8665954515906955e-06,
      "loss": 0.2111,
      "step": 33310
    },
    {
      "epoch": 0.6015318614174593,
      "grad_norm": 0.583780825138092,
      "learning_rate": 6.863902165649917e-06,
      "loss": 0.2995,
      "step": 33315
    },
    {
      "epoch": 0.60162214085036,
      "grad_norm": 0.4351842701435089,
      "learning_rate": 6.861209131985412e-06,
      "loss": 0.206,
      "step": 33320
    },
    {
      "epoch": 0.6017124202832608,
      "grad_norm": 0.6149066090583801,
      "learning_rate": 6.858516350813806e-06,
      "loss": 0.3505,
      "step": 33325
    },
    {
      "epoch": 0.6018026997161615,
      "grad_norm": 0.3864111304283142,
      "learning_rate": 6.855823822351726e-06,
      "loss": 0.3704,
      "step": 33330
    },
    {
      "epoch": 0.6018929791490621,
      "grad_norm": 0.1760164052248001,
      "learning_rate": 6.853131546815751e-06,
      "loss": 0.1318,
      "step": 33335
    },
    {
      "epoch": 0.6019832585819629,
      "grad_norm": 0.18229033052921295,
      "learning_rate": 6.85043952442247e-06,
      "loss": 0.2728,
      "step": 33340
    },
    {
      "epoch": 0.6020735380148636,
      "grad_norm": 0.3698432445526123,
      "learning_rate": 6.8477477553884275e-06,
      "loss": 0.2804,
      "step": 33345
    },
    {
      "epoch": 0.6021638174477644,
      "grad_norm": 0.6074759364128113,
      "learning_rate": 6.845056239930157e-06,
      "loss": 0.231,
      "step": 33350
    },
    {
      "epoch": 0.602254096880665,
      "grad_norm": 0.4252072274684906,
      "learning_rate": 6.8423649782641774e-06,
      "loss": 0.2622,
      "step": 33355
    },
    {
      "epoch": 0.6023443763135657,
      "grad_norm": 0.35315191745758057,
      "learning_rate": 6.839673970606973e-06,
      "loss": 0.256,
      "step": 33360
    },
    {
      "epoch": 0.6024346557464665,
      "grad_norm": 0.49989739060401917,
      "learning_rate": 6.836983217175022e-06,
      "loss": 0.3722,
      "step": 33365
    },
    {
      "epoch": 0.6025249351793672,
      "grad_norm": 0.5511475205421448,
      "learning_rate": 6.834292718184772e-06,
      "loss": 0.2877,
      "step": 33370
    },
    {
      "epoch": 0.6026152146122679,
      "grad_norm": 0.341680645942688,
      "learning_rate": 6.831602473852659e-06,
      "loss": 0.2114,
      "step": 33375
    },
    {
      "epoch": 0.6027054940451686,
      "grad_norm": 0.4739898145198822,
      "learning_rate": 6.8289124843950874e-06,
      "loss": 0.2597,
      "step": 33380
    },
    {
      "epoch": 0.6027957734780693,
      "grad_norm": 0.3317372798919678,
      "learning_rate": 6.826222750028453e-06,
      "loss": 0.195,
      "step": 33385
    },
    {
      "epoch": 0.6028860529109701,
      "grad_norm": 0.5364915728569031,
      "learning_rate": 6.823533270969119e-06,
      "loss": 0.2198,
      "step": 33390
    },
    {
      "epoch": 0.6029763323438707,
      "grad_norm": 0.280256986618042,
      "learning_rate": 6.820844047433444e-06,
      "loss": 0.2108,
      "step": 33395
    },
    {
      "epoch": 0.6030666117767715,
      "grad_norm": 0.4123166501522064,
      "learning_rate": 6.818155079637745e-06,
      "loss": 0.2774,
      "step": 33400
    },
    {
      "epoch": 0.6031568912096722,
      "grad_norm": 0.2371968924999237,
      "learning_rate": 6.815466367798338e-06,
      "loss": 0.2645,
      "step": 33405
    },
    {
      "epoch": 0.6032471706425729,
      "grad_norm": 0.3424372375011444,
      "learning_rate": 6.8127779121315075e-06,
      "loss": 0.2064,
      "step": 33410
    },
    {
      "epoch": 0.6033374500754736,
      "grad_norm": 0.47352439165115356,
      "learning_rate": 6.8100897128535206e-06,
      "loss": 0.2816,
      "step": 33415
    },
    {
      "epoch": 0.6034277295083743,
      "grad_norm": 0.5806667804718018,
      "learning_rate": 6.807401770180622e-06,
      "loss": 0.2438,
      "step": 33420
    },
    {
      "epoch": 0.603518008941275,
      "grad_norm": 0.4509647488594055,
      "learning_rate": 6.804714084329043e-06,
      "loss": 0.2502,
      "step": 33425
    },
    {
      "epoch": 0.6036082883741758,
      "grad_norm": 0.6704498529434204,
      "learning_rate": 6.802026655514982e-06,
      "loss": 0.2453,
      "step": 33430
    },
    {
      "epoch": 0.6036985678070764,
      "grad_norm": 0.517916202545166,
      "learning_rate": 6.7993394839546235e-06,
      "loss": 0.3107,
      "step": 33435
    },
    {
      "epoch": 0.6037888472399772,
      "grad_norm": 0.42207249999046326,
      "learning_rate": 6.796652569864135e-06,
      "loss": 0.2529,
      "step": 33440
    },
    {
      "epoch": 0.6038791266728779,
      "grad_norm": 0.4871932864189148,
      "learning_rate": 6.7939659134596534e-06,
      "loss": 0.1876,
      "step": 33445
    },
    {
      "epoch": 0.6039694061057787,
      "grad_norm": 0.42853614687919617,
      "learning_rate": 6.791279514957306e-06,
      "loss": 0.1931,
      "step": 33450
    },
    {
      "epoch": 0.6040596855386793,
      "grad_norm": 0.36403781175613403,
      "learning_rate": 6.78859337457319e-06,
      "loss": 0.2714,
      "step": 33455
    },
    {
      "epoch": 0.60414996497158,
      "grad_norm": 0.643302857875824,
      "learning_rate": 6.785907492523391e-06,
      "loss": 0.2594,
      "step": 33460
    },
    {
      "epoch": 0.6042402444044808,
      "grad_norm": 0.5370124578475952,
      "learning_rate": 6.7832218690239636e-06,
      "loss": 0.3383,
      "step": 33465
    },
    {
      "epoch": 0.6043305238373815,
      "grad_norm": 0.5137472152709961,
      "learning_rate": 6.780536504290949e-06,
      "loss": 0.2418,
      "step": 33470
    },
    {
      "epoch": 0.6044208032702821,
      "grad_norm": 0.48361706733703613,
      "learning_rate": 6.777851398540363e-06,
      "loss": 0.2246,
      "step": 33475
    },
    {
      "epoch": 0.6045110827031829,
      "grad_norm": 0.2475363165140152,
      "learning_rate": 6.775166551988207e-06,
      "loss": 0.1396,
      "step": 33480
    },
    {
      "epoch": 0.6046013621360836,
      "grad_norm": 0.5070680975914001,
      "learning_rate": 6.772481964850453e-06,
      "loss": 0.3403,
      "step": 33485
    },
    {
      "epoch": 0.6046916415689844,
      "grad_norm": 0.5311540365219116,
      "learning_rate": 6.769797637343063e-06,
      "loss": 0.2733,
      "step": 33490
    },
    {
      "epoch": 0.604781921001885,
      "grad_norm": 0.44159597158432007,
      "learning_rate": 6.767113569681963e-06,
      "loss": 0.1231,
      "step": 33495
    },
    {
      "epoch": 0.6048722004347857,
      "grad_norm": 0.3537275791168213,
      "learning_rate": 6.764429762083071e-06,
      "loss": 0.2573,
      "step": 33500
    },
    {
      "epoch": 0.6049624798676865,
      "grad_norm": 0.6137586236000061,
      "learning_rate": 6.761746214762281e-06,
      "loss": 0.2545,
      "step": 33505
    },
    {
      "epoch": 0.6050527593005872,
      "grad_norm": 0.5679592490196228,
      "learning_rate": 6.759062927935466e-06,
      "loss": 0.2441,
      "step": 33510
    },
    {
      "epoch": 0.6051430387334878,
      "grad_norm": 0.43349897861480713,
      "learning_rate": 6.75637990181847e-06,
      "loss": 0.2459,
      "step": 33515
    },
    {
      "epoch": 0.6052333181663886,
      "grad_norm": 0.5576727390289307,
      "learning_rate": 6.7536971366271305e-06,
      "loss": 0.2069,
      "step": 33520
    },
    {
      "epoch": 0.6053235975992893,
      "grad_norm": 0.3747663199901581,
      "learning_rate": 6.7510146325772555e-06,
      "loss": 0.2893,
      "step": 33525
    },
    {
      "epoch": 0.6054138770321901,
      "grad_norm": 0.7237586379051208,
      "learning_rate": 6.748332389884627e-06,
      "loss": 0.2664,
      "step": 33530
    },
    {
      "epoch": 0.6055041564650907,
      "grad_norm": 0.44492822885513306,
      "learning_rate": 6.745650408765017e-06,
      "loss": 0.2212,
      "step": 33535
    },
    {
      "epoch": 0.6055944358979914,
      "grad_norm": 0.31927186250686646,
      "learning_rate": 6.742968689434169e-06,
      "loss": 0.1558,
      "step": 33540
    },
    {
      "epoch": 0.6056847153308922,
      "grad_norm": 0.6788272857666016,
      "learning_rate": 6.7402872321078094e-06,
      "loss": 0.2176,
      "step": 33545
    },
    {
      "epoch": 0.6057749947637929,
      "grad_norm": 0.5264540910720825,
      "learning_rate": 6.737606037001642e-06,
      "loss": 0.1845,
      "step": 33550
    },
    {
      "epoch": 0.6058652741966936,
      "grad_norm": 0.5646959543228149,
      "learning_rate": 6.734925104331351e-06,
      "loss": 0.2625,
      "step": 33555
    },
    {
      "epoch": 0.6059555536295943,
      "grad_norm": 0.4939166009426117,
      "learning_rate": 6.732244434312594e-06,
      "loss": 0.319,
      "step": 33560
    },
    {
      "epoch": 0.606045833062495,
      "grad_norm": 0.25671443343162537,
      "learning_rate": 6.729564027161013e-06,
      "loss": 0.2953,
      "step": 33565
    },
    {
      "epoch": 0.6061361124953958,
      "grad_norm": 0.5742185711860657,
      "learning_rate": 6.7268838830922275e-06,
      "loss": 0.1592,
      "step": 33570
    },
    {
      "epoch": 0.6062263919282964,
      "grad_norm": 0.24970318377017975,
      "learning_rate": 6.724204002321839e-06,
      "loss": 0.231,
      "step": 33575
    },
    {
      "epoch": 0.6063166713611972,
      "grad_norm": 0.27477550506591797,
      "learning_rate": 6.721524385065417e-06,
      "loss": 0.186,
      "step": 33580
    },
    {
      "epoch": 0.6064069507940979,
      "grad_norm": 0.4840497374534607,
      "learning_rate": 6.7188450315385236e-06,
      "loss": 0.1251,
      "step": 33585
    },
    {
      "epoch": 0.6064972302269986,
      "grad_norm": 0.38732531666755676,
      "learning_rate": 6.716165941956689e-06,
      "loss": 0.1924,
      "step": 33590
    },
    {
      "epoch": 0.6065875096598993,
      "grad_norm": 0.2982494831085205,
      "learning_rate": 6.713487116535433e-06,
      "loss": 0.219,
      "step": 33595
    },
    {
      "epoch": 0.6066777890928,
      "grad_norm": 0.3711354434490204,
      "learning_rate": 6.710808555490237e-06,
      "loss": 0.2206,
      "step": 33600
    },
    {
      "epoch": 0.6067680685257008,
      "grad_norm": 0.3120940923690796,
      "learning_rate": 6.708130259036585e-06,
      "loss": 0.2555,
      "step": 33605
    },
    {
      "epoch": 0.6068583479586015,
      "grad_norm": 0.3166637420654297,
      "learning_rate": 6.705452227389914e-06,
      "loss": 0.2232,
      "step": 33610
    },
    {
      "epoch": 0.6069486273915021,
      "grad_norm": 0.787291944026947,
      "learning_rate": 6.70277446076566e-06,
      "loss": 0.2896,
      "step": 33615
    },
    {
      "epoch": 0.6070389068244029,
      "grad_norm": 0.4427154064178467,
      "learning_rate": 6.7000969593792295e-06,
      "loss": 0.3057,
      "step": 33620
    },
    {
      "epoch": 0.6071291862573036,
      "grad_norm": 0.3680169880390167,
      "learning_rate": 6.697419723446001e-06,
      "loss": 0.2339,
      "step": 33625
    },
    {
      "epoch": 0.6072194656902044,
      "grad_norm": 0.5025822520256042,
      "learning_rate": 6.694742753181347e-06,
      "loss": 0.2132,
      "step": 33630
    },
    {
      "epoch": 0.607309745123105,
      "grad_norm": 0.5601755976676941,
      "learning_rate": 6.692066048800605e-06,
      "loss": 0.1726,
      "step": 33635
    },
    {
      "epoch": 0.6074000245560057,
      "grad_norm": 0.3275012969970703,
      "learning_rate": 6.689389610519102e-06,
      "loss": 0.1813,
      "step": 33640
    },
    {
      "epoch": 0.6074903039889065,
      "grad_norm": 0.3714092969894409,
      "learning_rate": 6.6867134385521305e-06,
      "loss": 0.1963,
      "step": 33645
    },
    {
      "epoch": 0.6075805834218072,
      "grad_norm": 0.7068204283714294,
      "learning_rate": 6.684037533114976e-06,
      "loss": 0.1556,
      "step": 33650
    },
    {
      "epoch": 0.6076708628547078,
      "grad_norm": 0.4423868656158447,
      "learning_rate": 6.681361894422889e-06,
      "loss": 0.2443,
      "step": 33655
    },
    {
      "epoch": 0.6077611422876086,
      "grad_norm": 1.2866357564926147,
      "learning_rate": 6.678686522691112e-06,
      "loss": 0.2191,
      "step": 33660
    },
    {
      "epoch": 0.6078514217205093,
      "grad_norm": 0.2865354120731354,
      "learning_rate": 6.676011418134851e-06,
      "loss": 0.2938,
      "step": 33665
    },
    {
      "epoch": 0.6079417011534101,
      "grad_norm": 0.48758894205093384,
      "learning_rate": 6.673336580969311e-06,
      "loss": 0.16,
      "step": 33670
    },
    {
      "epoch": 0.6080319805863107,
      "grad_norm": 0.7908018827438354,
      "learning_rate": 6.6706620114096496e-06,
      "loss": 0.2236,
      "step": 33675
    },
    {
      "epoch": 0.6081222600192114,
      "grad_norm": 0.6689500212669373,
      "learning_rate": 6.667987709671027e-06,
      "loss": 0.1365,
      "step": 33680
    },
    {
      "epoch": 0.6082125394521122,
      "grad_norm": 0.6612555384635925,
      "learning_rate": 6.665313675968564e-06,
      "loss": 0.3081,
      "step": 33685
    },
    {
      "epoch": 0.6083028188850129,
      "grad_norm": 0.601243793964386,
      "learning_rate": 6.662639910517374e-06,
      "loss": 0.2605,
      "step": 33690
    },
    {
      "epoch": 0.6083930983179137,
      "grad_norm": 0.5894544124603271,
      "learning_rate": 6.6599664135325345e-06,
      "loss": 0.2103,
      "step": 33695
    },
    {
      "epoch": 0.6084833777508143,
      "grad_norm": 0.5290404558181763,
      "learning_rate": 6.657293185229115e-06,
      "loss": 0.2908,
      "step": 33700
    },
    {
      "epoch": 0.608573657183715,
      "grad_norm": 0.4308032691478729,
      "learning_rate": 6.654620225822156e-06,
      "loss": 0.2222,
      "step": 33705
    },
    {
      "epoch": 0.6086639366166158,
      "grad_norm": 0.30124881863594055,
      "learning_rate": 6.651947535526672e-06,
      "loss": 0.202,
      "step": 33710
    },
    {
      "epoch": 0.6087542160495165,
      "grad_norm": 0.3160736560821533,
      "learning_rate": 6.64927511455767e-06,
      "loss": 0.2642,
      "step": 33715
    },
    {
      "epoch": 0.6088444954824171,
      "grad_norm": 0.3296411335468292,
      "learning_rate": 6.64660296313012e-06,
      "loss": 0.2427,
      "step": 33720
    },
    {
      "epoch": 0.6089347749153179,
      "grad_norm": 0.6400465369224548,
      "learning_rate": 6.6439310814589856e-06,
      "loss": 0.1672,
      "step": 33725
    },
    {
      "epoch": 0.6090250543482186,
      "grad_norm": 0.7744160890579224,
      "learning_rate": 6.641259469759186e-06,
      "loss": 0.179,
      "step": 33730
    },
    {
      "epoch": 0.6091153337811194,
      "grad_norm": 0.37067508697509766,
      "learning_rate": 6.6385881282456484e-06,
      "loss": 0.2073,
      "step": 33735
    },
    {
      "epoch": 0.60920561321402,
      "grad_norm": 0.4492238163948059,
      "learning_rate": 6.635917057133253e-06,
      "loss": 0.1919,
      "step": 33740
    },
    {
      "epoch": 0.6092958926469207,
      "grad_norm": 0.4266092777252197,
      "learning_rate": 6.633246256636873e-06,
      "loss": 0.2887,
      "step": 33745
    },
    {
      "epoch": 0.6093861720798215,
      "grad_norm": 0.5106934905052185,
      "learning_rate": 6.630575726971351e-06,
      "loss": 0.304,
      "step": 33750
    },
    {
      "epoch": 0.6094764515127222,
      "grad_norm": 0.26896047592163086,
      "learning_rate": 6.627905468351517e-06,
      "loss": 0.2235,
      "step": 33755
    },
    {
      "epoch": 0.6095667309456229,
      "grad_norm": 0.3424336016178131,
      "learning_rate": 6.6252354809921675e-06,
      "loss": 0.1777,
      "step": 33760
    },
    {
      "epoch": 0.6096570103785236,
      "grad_norm": 0.37666502594947815,
      "learning_rate": 6.622565765108087e-06,
      "loss": 0.2396,
      "step": 33765
    },
    {
      "epoch": 0.6097472898114243,
      "grad_norm": 0.6772646307945251,
      "learning_rate": 6.619896320914034e-06,
      "loss": 0.2772,
      "step": 33770
    },
    {
      "epoch": 0.6098375692443251,
      "grad_norm": 0.4182206988334656,
      "learning_rate": 6.61722714862475e-06,
      "loss": 0.2773,
      "step": 33775
    },
    {
      "epoch": 0.6099278486772257,
      "grad_norm": 0.5708869099617004,
      "learning_rate": 6.614558248454942e-06,
      "loss": 0.1938,
      "step": 33780
    },
    {
      "epoch": 0.6100181281101265,
      "grad_norm": 0.39538654685020447,
      "learning_rate": 6.611889620619311e-06,
      "loss": 0.2765,
      "step": 33785
    },
    {
      "epoch": 0.6101084075430272,
      "grad_norm": 0.6491618156433105,
      "learning_rate": 6.609221265332524e-06,
      "loss": 0.2431,
      "step": 33790
    },
    {
      "epoch": 0.6101986869759279,
      "grad_norm": 1.528210163116455,
      "learning_rate": 6.606553182809234e-06,
      "loss": 0.2736,
      "step": 33795
    },
    {
      "epoch": 0.6102889664088286,
      "grad_norm": 0.26595795154571533,
      "learning_rate": 6.603885373264072e-06,
      "loss": 0.2506,
      "step": 33800
    },
    {
      "epoch": 0.6103792458417293,
      "grad_norm": 0.43251991271972656,
      "learning_rate": 6.601217836911633e-06,
      "loss": 0.2007,
      "step": 33805
    },
    {
      "epoch": 0.61046952527463,
      "grad_norm": 0.33078595995903015,
      "learning_rate": 6.598550573966511e-06,
      "loss": 0.2649,
      "step": 33810
    },
    {
      "epoch": 0.6105598047075308,
      "grad_norm": 0.6174448728561401,
      "learning_rate": 6.595883584643261e-06,
      "loss": 0.2894,
      "step": 33815
    },
    {
      "epoch": 0.6106500841404314,
      "grad_norm": 0.5917649269104004,
      "learning_rate": 6.593216869156432e-06,
      "loss": 0.2367,
      "step": 33820
    },
    {
      "epoch": 0.6107403635733322,
      "grad_norm": 0.40432947874069214,
      "learning_rate": 6.590550427720531e-06,
      "loss": 0.2344,
      "step": 33825
    },
    {
      "epoch": 0.6108306430062329,
      "grad_norm": 0.26861852407455444,
      "learning_rate": 6.587884260550062e-06,
      "loss": 0.1964,
      "step": 33830
    },
    {
      "epoch": 0.6109209224391337,
      "grad_norm": 0.36993351578712463,
      "learning_rate": 6.585218367859493e-06,
      "loss": 0.2807,
      "step": 33835
    },
    {
      "epoch": 0.6110112018720343,
      "grad_norm": 0.30036458373069763,
      "learning_rate": 6.582552749863282e-06,
      "loss": 0.2971,
      "step": 33840
    },
    {
      "epoch": 0.611101481304935,
      "grad_norm": 0.3073712885379791,
      "learning_rate": 6.579887406775851e-06,
      "loss": 0.2099,
      "step": 33845
    },
    {
      "epoch": 0.6111917607378358,
      "grad_norm": 0.42312800884246826,
      "learning_rate": 6.577222338811615e-06,
      "loss": 0.2325,
      "step": 33850
    },
    {
      "epoch": 0.6112820401707365,
      "grad_norm": 0.3106360137462616,
      "learning_rate": 6.5745575461849535e-06,
      "loss": 0.3377,
      "step": 33855
    },
    {
      "epoch": 0.6113723196036371,
      "grad_norm": 0.40115922689437866,
      "learning_rate": 6.571893029110234e-06,
      "loss": 0.2981,
      "step": 33860
    },
    {
      "epoch": 0.6114625990365379,
      "grad_norm": 0.3473933935165405,
      "learning_rate": 6.569228787801791e-06,
      "loss": 0.217,
      "step": 33865
    },
    {
      "epoch": 0.6115528784694386,
      "grad_norm": 0.3707481026649475,
      "learning_rate": 6.566564822473955e-06,
      "loss": 0.2357,
      "step": 33870
    },
    {
      "epoch": 0.6116431579023394,
      "grad_norm": 0.618390679359436,
      "learning_rate": 6.563901133341011e-06,
      "loss": 0.1936,
      "step": 33875
    },
    {
      "epoch": 0.61173343733524,
      "grad_norm": 0.4362730383872986,
      "learning_rate": 6.561237720617239e-06,
      "loss": 0.2656,
      "step": 33880
    },
    {
      "epoch": 0.6118237167681407,
      "grad_norm": 0.4240482747554779,
      "learning_rate": 6.558574584516888e-06,
      "loss": 0.264,
      "step": 33885
    },
    {
      "epoch": 0.6119139962010415,
      "grad_norm": 0.3510347008705139,
      "learning_rate": 6.555911725254196e-06,
      "loss": 0.2546,
      "step": 33890
    },
    {
      "epoch": 0.6120042756339422,
      "grad_norm": 0.2657509744167328,
      "learning_rate": 6.553249143043361e-06,
      "loss": 0.205,
      "step": 33895
    },
    {
      "epoch": 0.6120945550668428,
      "grad_norm": 0.28988775610923767,
      "learning_rate": 6.55058683809857e-06,
      "loss": 0.2574,
      "step": 33900
    },
    {
      "epoch": 0.6121848344997436,
      "grad_norm": 0.28201043605804443,
      "learning_rate": 6.5479248106339944e-06,
      "loss": 0.1787,
      "step": 33905
    },
    {
      "epoch": 0.6122751139326443,
      "grad_norm": 0.5050878524780273,
      "learning_rate": 6.545263060863763e-06,
      "loss": 0.2254,
      "step": 33910
    },
    {
      "epoch": 0.6123653933655451,
      "grad_norm": 0.5506510734558105,
      "learning_rate": 6.542601589002002e-06,
      "loss": 0.2477,
      "step": 33915
    },
    {
      "epoch": 0.6124556727984457,
      "grad_norm": 0.4561326503753662,
      "learning_rate": 6.539940395262803e-06,
      "loss": 0.2256,
      "step": 33920
    },
    {
      "epoch": 0.6125459522313464,
      "grad_norm": 0.6151211261749268,
      "learning_rate": 6.537279479860248e-06,
      "loss": 0.1705,
      "step": 33925
    },
    {
      "epoch": 0.6126362316642472,
      "grad_norm": 0.33129143714904785,
      "learning_rate": 6.534618843008374e-06,
      "loss": 0.1842,
      "step": 33930
    },
    {
      "epoch": 0.6127265110971479,
      "grad_norm": 0.715975821018219,
      "learning_rate": 6.5319584849212255e-06,
      "loss": 0.2435,
      "step": 33935
    },
    {
      "epoch": 0.6128167905300486,
      "grad_norm": 0.452910840511322,
      "learning_rate": 6.529298405812799e-06,
      "loss": 0.2526,
      "step": 33940
    },
    {
      "epoch": 0.6129070699629493,
      "grad_norm": 0.4404039680957794,
      "learning_rate": 6.526638605897082e-06,
      "loss": 0.1775,
      "step": 33945
    },
    {
      "epoch": 0.61299734939585,
      "grad_norm": 0.2817777097225189,
      "learning_rate": 6.5239790853880325e-06,
      "loss": 0.2172,
      "step": 33950
    },
    {
      "epoch": 0.6130876288287508,
      "grad_norm": 0.4639744758605957,
      "learning_rate": 6.521319844499599e-06,
      "loss": 0.215,
      "step": 33955
    },
    {
      "epoch": 0.6131779082616514,
      "grad_norm": 0.7217711806297302,
      "learning_rate": 6.518660883445685e-06,
      "loss": 0.3002,
      "step": 33960
    },
    {
      "epoch": 0.6132681876945522,
      "grad_norm": 0.5533402562141418,
      "learning_rate": 6.516002202440195e-06,
      "loss": 0.2314,
      "step": 33965
    },
    {
      "epoch": 0.6133584671274529,
      "grad_norm": 0.3430977761745453,
      "learning_rate": 6.513343801696993e-06,
      "loss": 0.2209,
      "step": 33970
    },
    {
      "epoch": 0.6134487465603536,
      "grad_norm": 0.4223589301109314,
      "learning_rate": 6.510685681429937e-06,
      "loss": 0.2129,
      "step": 33975
    },
    {
      "epoch": 0.6135390259932543,
      "grad_norm": 0.3885827660560608,
      "learning_rate": 6.508027841852844e-06,
      "loss": 0.2364,
      "step": 33980
    },
    {
      "epoch": 0.613629305426155,
      "grad_norm": 0.26624947786331177,
      "learning_rate": 6.505370283179524e-06,
      "loss": 0.1584,
      "step": 33985
    },
    {
      "epoch": 0.6137195848590558,
      "grad_norm": 0.5280428528785706,
      "learning_rate": 6.502713005623758e-06,
      "loss": 0.2098,
      "step": 33990
    },
    {
      "epoch": 0.6138098642919565,
      "grad_norm": 0.42223653197288513,
      "learning_rate": 6.500056009399296e-06,
      "loss": 0.291,
      "step": 33995
    },
    {
      "epoch": 0.6139001437248571,
      "grad_norm": 0.4914424419403076,
      "learning_rate": 6.497399294719888e-06,
      "loss": 0.2807,
      "step": 34000
    },
    {
      "epoch": 0.6139904231577579,
      "grad_norm": 0.36946821212768555,
      "learning_rate": 6.494742861799237e-06,
      "loss": 0.28,
      "step": 34005
    },
    {
      "epoch": 0.6140807025906586,
      "grad_norm": 0.6688858866691589,
      "learning_rate": 6.492086710851039e-06,
      "loss": 0.327,
      "step": 34010
    },
    {
      "epoch": 0.6141709820235594,
      "grad_norm": 0.35073772072792053,
      "learning_rate": 6.489430842088957e-06,
      "loss": 0.3021,
      "step": 34015
    },
    {
      "epoch": 0.61426126145646,
      "grad_norm": 0.5166581273078918,
      "learning_rate": 6.486775255726645e-06,
      "loss": 0.1812,
      "step": 34020
    },
    {
      "epoch": 0.6143515408893607,
      "grad_norm": 0.37278056144714355,
      "learning_rate": 6.484119951977714e-06,
      "loss": 0.267,
      "step": 34025
    },
    {
      "epoch": 0.6144418203222615,
      "grad_norm": 0.44880130887031555,
      "learning_rate": 6.481464931055774e-06,
      "loss": 0.2423,
      "step": 34030
    },
    {
      "epoch": 0.6145320997551622,
      "grad_norm": 0.35509809851646423,
      "learning_rate": 6.478810193174395e-06,
      "loss": 0.2572,
      "step": 34035
    },
    {
      "epoch": 0.6146223791880628,
      "grad_norm": 0.5073139667510986,
      "learning_rate": 6.4761557385471385e-06,
      "loss": 0.2865,
      "step": 34040
    },
    {
      "epoch": 0.6147126586209636,
      "grad_norm": 0.4112986624240875,
      "learning_rate": 6.473501567387528e-06,
      "loss": 0.1997,
      "step": 34045
    },
    {
      "epoch": 0.6148029380538643,
      "grad_norm": 0.4614567160606384,
      "learning_rate": 6.470847679909078e-06,
      "loss": 0.3304,
      "step": 34050
    },
    {
      "epoch": 0.6148932174867651,
      "grad_norm": 0.45724689960479736,
      "learning_rate": 6.468194076325271e-06,
      "loss": 0.2825,
      "step": 34055
    },
    {
      "epoch": 0.6149834969196657,
      "grad_norm": 0.29275423288345337,
      "learning_rate": 6.465540756849575e-06,
      "loss": 0.2266,
      "step": 34060
    },
    {
      "epoch": 0.6150737763525664,
      "grad_norm": 0.4016673266887665,
      "learning_rate": 6.462887721695424e-06,
      "loss": 0.203,
      "step": 34065
    },
    {
      "epoch": 0.6151640557854672,
      "grad_norm": 0.3553887903690338,
      "learning_rate": 6.460234971076243e-06,
      "loss": 0.2053,
      "step": 34070
    },
    {
      "epoch": 0.6152543352183679,
      "grad_norm": 0.6590433716773987,
      "learning_rate": 6.45758250520542e-06,
      "loss": 0.2523,
      "step": 34075
    },
    {
      "epoch": 0.6153446146512687,
      "grad_norm": 0.38849779963493347,
      "learning_rate": 6.454930324296326e-06,
      "loss": 0.1843,
      "step": 34080
    },
    {
      "epoch": 0.6154348940841693,
      "grad_norm": 0.812056303024292,
      "learning_rate": 6.4522784285623165e-06,
      "loss": 0.1499,
      "step": 34085
    },
    {
      "epoch": 0.61552517351707,
      "grad_norm": 0.6710259318351746,
      "learning_rate": 6.449626818216709e-06,
      "loss": 0.2728,
      "step": 34090
    },
    {
      "epoch": 0.6156154529499708,
      "grad_norm": 0.46642282605171204,
      "learning_rate": 6.4469754934728115e-06,
      "loss": 0.1461,
      "step": 34095
    },
    {
      "epoch": 0.6157057323828715,
      "grad_norm": 0.583949089050293,
      "learning_rate": 6.444324454543902e-06,
      "loss": 0.1903,
      "step": 34100
    },
    {
      "epoch": 0.6157960118157721,
      "grad_norm": 0.3867447078227997,
      "learning_rate": 6.441673701643241e-06,
      "loss": 0.2485,
      "step": 34105
    },
    {
      "epoch": 0.6158862912486729,
      "grad_norm": 0.44788867235183716,
      "learning_rate": 6.439023234984054e-06,
      "loss": 0.1767,
      "step": 34110
    },
    {
      "epoch": 0.6159765706815736,
      "grad_norm": 0.35031911730766296,
      "learning_rate": 6.4363730547795615e-06,
      "loss": 0.2218,
      "step": 34115
    },
    {
      "epoch": 0.6160668501144744,
      "grad_norm": 0.524443507194519,
      "learning_rate": 6.433723161242943e-06,
      "loss": 0.2522,
      "step": 34120
    },
    {
      "epoch": 0.616157129547375,
      "grad_norm": 1.5180704593658447,
      "learning_rate": 6.43107355458737e-06,
      "loss": 0.296,
      "step": 34125
    },
    {
      "epoch": 0.6162474089802757,
      "grad_norm": 0.5914297699928284,
      "learning_rate": 6.428424235025979e-06,
      "loss": 0.2384,
      "step": 34130
    },
    {
      "epoch": 0.6163376884131765,
      "grad_norm": 0.3615359961986542,
      "learning_rate": 6.425775202771893e-06,
      "loss": 0.3157,
      "step": 34135
    },
    {
      "epoch": 0.6164279678460772,
      "grad_norm": 0.49133241176605225,
      "learning_rate": 6.4231264580382e-06,
      "loss": 0.2748,
      "step": 34140
    },
    {
      "epoch": 0.6165182472789779,
      "grad_norm": 0.3857567310333252,
      "learning_rate": 6.420478001037982e-06,
      "loss": 0.2871,
      "step": 34145
    },
    {
      "epoch": 0.6166085267118786,
      "grad_norm": 0.14246675372123718,
      "learning_rate": 6.4178298319842795e-06,
      "loss": 0.2136,
      "step": 34150
    },
    {
      "epoch": 0.6166988061447793,
      "grad_norm": 0.35812297463417053,
      "learning_rate": 6.415181951090126e-06,
      "loss": 0.2314,
      "step": 34155
    },
    {
      "epoch": 0.6167890855776801,
      "grad_norm": 0.4039628505706787,
      "learning_rate": 6.412534358568516e-06,
      "loss": 0.3472,
      "step": 34160
    },
    {
      "epoch": 0.6168793650105807,
      "grad_norm": 0.3499111533164978,
      "learning_rate": 6.409887054632436e-06,
      "loss": 0.2001,
      "step": 34165
    },
    {
      "epoch": 0.6169696444434815,
      "grad_norm": 0.4685569405555725,
      "learning_rate": 6.407240039494843e-06,
      "loss": 0.2612,
      "step": 34170
    },
    {
      "epoch": 0.6170599238763822,
      "grad_norm": 0.4339657425880432,
      "learning_rate": 6.404593313368661e-06,
      "loss": 0.2849,
      "step": 34175
    },
    {
      "epoch": 0.6171502033092829,
      "grad_norm": 0.20176739990711212,
      "learning_rate": 6.401946876466809e-06,
      "loss": 0.2056,
      "step": 34180
    },
    {
      "epoch": 0.6172404827421836,
      "grad_norm": 0.6576533317565918,
      "learning_rate": 6.3993007290021695e-06,
      "loss": 0.3022,
      "step": 34185
    },
    {
      "epoch": 0.6173307621750843,
      "grad_norm": 0.43607407808303833,
      "learning_rate": 6.396654871187606e-06,
      "loss": 0.2136,
      "step": 34190
    },
    {
      "epoch": 0.617421041607985,
      "grad_norm": 0.7124909162521362,
      "learning_rate": 6.394009303235959e-06,
      "loss": 0.227,
      "step": 34195
    },
    {
      "epoch": 0.6175113210408858,
      "grad_norm": 0.4305925965309143,
      "learning_rate": 6.39136402536005e-06,
      "loss": 0.2486,
      "step": 34200
    },
    {
      "epoch": 0.6176016004737864,
      "grad_norm": 0.7978116869926453,
      "learning_rate": 6.388719037772662e-06,
      "loss": 0.3101,
      "step": 34205
    },
    {
      "epoch": 0.6176918799066872,
      "grad_norm": 0.5606377720832825,
      "learning_rate": 6.386074340686575e-06,
      "loss": 0.2684,
      "step": 34210
    },
    {
      "epoch": 0.6177821593395879,
      "grad_norm": 0.578524112701416,
      "learning_rate": 6.383429934314526e-06,
      "loss": 0.1665,
      "step": 34215
    },
    {
      "epoch": 0.6178724387724887,
      "grad_norm": 0.3661254346370697,
      "learning_rate": 6.380785818869251e-06,
      "loss": 0.262,
      "step": 34220
    },
    {
      "epoch": 0.6179627182053893,
      "grad_norm": 0.30994969606399536,
      "learning_rate": 6.378141994563438e-06,
      "loss": 0.2091,
      "step": 34225
    },
    {
      "epoch": 0.61805299763829,
      "grad_norm": 0.5623824000358582,
      "learning_rate": 6.37549846160977e-06,
      "loss": 0.2085,
      "step": 34230
    },
    {
      "epoch": 0.6181432770711908,
      "grad_norm": 0.6264132261276245,
      "learning_rate": 6.3728552202208945e-06,
      "loss": 0.3464,
      "step": 34235
    },
    {
      "epoch": 0.6182335565040915,
      "grad_norm": 0.47519147396087646,
      "learning_rate": 6.3702122706094515e-06,
      "loss": 0.2653,
      "step": 34240
    },
    {
      "epoch": 0.6183238359369921,
      "grad_norm": 0.36290884017944336,
      "learning_rate": 6.367569612988033e-06,
      "loss": 0.2852,
      "step": 34245
    },
    {
      "epoch": 0.6184141153698929,
      "grad_norm": 0.5039083957672119,
      "learning_rate": 6.364927247569235e-06,
      "loss": 0.2122,
      "step": 34250
    },
    {
      "epoch": 0.6185043948027936,
      "grad_norm": 0.456661581993103,
      "learning_rate": 6.362285174565605e-06,
      "loss": 0.1835,
      "step": 34255
    },
    {
      "epoch": 0.6185946742356944,
      "grad_norm": 0.27523332834243774,
      "learning_rate": 6.359643394189688e-06,
      "loss": 0.2324,
      "step": 34260
    },
    {
      "epoch": 0.618684953668595,
      "grad_norm": 0.6493564248085022,
      "learning_rate": 6.357001906653993e-06,
      "loss": 0.1441,
      "step": 34265
    },
    {
      "epoch": 0.6187752331014957,
      "grad_norm": 0.4140636622905731,
      "learning_rate": 6.354360712171003e-06,
      "loss": 0.2299,
      "step": 34270
    },
    {
      "epoch": 0.6188655125343965,
      "grad_norm": 0.3528628647327423,
      "learning_rate": 6.35171981095319e-06,
      "loss": 0.2098,
      "step": 34275
    },
    {
      "epoch": 0.6189557919672972,
      "grad_norm": 0.40171143412590027,
      "learning_rate": 6.349079203212989e-06,
      "loss": 0.2456,
      "step": 34280
    },
    {
      "epoch": 0.6190460714001979,
      "grad_norm": 0.45002445578575134,
      "learning_rate": 6.346438889162827e-06,
      "loss": 0.1707,
      "step": 34285
    },
    {
      "epoch": 0.6191363508330986,
      "grad_norm": 0.6447219848632812,
      "learning_rate": 6.343798869015088e-06,
      "loss": 0.2391,
      "step": 34290
    },
    {
      "epoch": 0.6192266302659993,
      "grad_norm": 0.49332907795906067,
      "learning_rate": 6.34115914298215e-06,
      "loss": 0.1497,
      "step": 34295
    },
    {
      "epoch": 0.6193169096989001,
      "grad_norm": 0.32314741611480713,
      "learning_rate": 6.338519711276353e-06,
      "loss": 0.2177,
      "step": 34300
    },
    {
      "epoch": 0.6194071891318007,
      "grad_norm": 0.3138463497161865,
      "learning_rate": 6.3358805741100285e-06,
      "loss": 0.2568,
      "step": 34305
    },
    {
      "epoch": 0.6194974685647014,
      "grad_norm": 0.3489329516887665,
      "learning_rate": 6.333241731695466e-06,
      "loss": 0.2336,
      "step": 34310
    },
    {
      "epoch": 0.6195877479976022,
      "grad_norm": 0.290552020072937,
      "learning_rate": 6.33060318424495e-06,
      "loss": 0.1592,
      "step": 34315
    },
    {
      "epoch": 0.6196780274305029,
      "grad_norm": 0.37173569202423096,
      "learning_rate": 6.3279649319707245e-06,
      "loss": 0.1762,
      "step": 34320
    },
    {
      "epoch": 0.6197683068634036,
      "grad_norm": 0.49615028500556946,
      "learning_rate": 6.325326975085026e-06,
      "loss": 0.2648,
      "step": 34325
    },
    {
      "epoch": 0.6198585862963043,
      "grad_norm": 0.5045347213745117,
      "learning_rate": 6.322689313800052e-06,
      "loss": 0.3263,
      "step": 34330
    },
    {
      "epoch": 0.619948865729205,
      "grad_norm": 0.41068151593208313,
      "learning_rate": 6.320051948327991e-06,
      "loss": 0.159,
      "step": 34335
    },
    {
      "epoch": 0.6200391451621058,
      "grad_norm": 0.8643396496772766,
      "learning_rate": 6.3174148788809895e-06,
      "loss": 0.2559,
      "step": 34340
    },
    {
      "epoch": 0.6201294245950064,
      "grad_norm": 0.32451334595680237,
      "learning_rate": 6.314778105671189e-06,
      "loss": 0.344,
      "step": 34345
    },
    {
      "epoch": 0.6202197040279072,
      "grad_norm": 0.6597735285758972,
      "learning_rate": 6.3121416289106975e-06,
      "loss": 0.2211,
      "step": 34350
    },
    {
      "epoch": 0.6203099834608079,
      "grad_norm": 0.42996546626091003,
      "learning_rate": 6.3095054488115945e-06,
      "loss": 0.2387,
      "step": 34355
    },
    {
      "epoch": 0.6204002628937086,
      "grad_norm": 0.4882093369960785,
      "learning_rate": 6.306869565585949e-06,
      "loss": 0.242,
      "step": 34360
    },
    {
      "epoch": 0.6204905423266093,
      "grad_norm": 0.3846128582954407,
      "learning_rate": 6.304233979445791e-06,
      "loss": 0.1743,
      "step": 34365
    },
    {
      "epoch": 0.62058082175951,
      "grad_norm": 0.7323339581489563,
      "learning_rate": 6.301598690603143e-06,
      "loss": 0.2092,
      "step": 34370
    },
    {
      "epoch": 0.6206711011924108,
      "grad_norm": 0.21192646026611328,
      "learning_rate": 6.298963699269989e-06,
      "loss": 0.2194,
      "step": 34375
    },
    {
      "epoch": 0.6207613806253115,
      "grad_norm": 0.49501052498817444,
      "learning_rate": 6.296329005658295e-06,
      "loss": 0.1443,
      "step": 34380
    },
    {
      "epoch": 0.6208516600582121,
      "grad_norm": 0.3888206481933594,
      "learning_rate": 6.293694609980003e-06,
      "loss": 0.143,
      "step": 34385
    },
    {
      "epoch": 0.6209419394911129,
      "grad_norm": 0.5048769116401672,
      "learning_rate": 6.291060512447036e-06,
      "loss": 0.2263,
      "step": 34390
    },
    {
      "epoch": 0.6210322189240136,
      "grad_norm": 0.4002924859523773,
      "learning_rate": 6.2884267132712805e-06,
      "loss": 0.2666,
      "step": 34395
    },
    {
      "epoch": 0.6211224983569144,
      "grad_norm": 0.415455162525177,
      "learning_rate": 6.285793212664615e-06,
      "loss": 0.2045,
      "step": 34400
    },
    {
      "epoch": 0.621212777789815,
      "grad_norm": 0.28391233086586,
      "learning_rate": 6.283160010838876e-06,
      "loss": 0.1603,
      "step": 34405
    },
    {
      "epoch": 0.6213030572227157,
      "grad_norm": 0.6596155762672424,
      "learning_rate": 6.280527108005892e-06,
      "loss": 0.2308,
      "step": 34410
    },
    {
      "epoch": 0.6213933366556165,
      "grad_norm": 0.4737531244754791,
      "learning_rate": 6.2778945043774566e-06,
      "loss": 0.2599,
      "step": 34415
    },
    {
      "epoch": 0.6214836160885172,
      "grad_norm": 0.44785115122795105,
      "learning_rate": 6.275262200165351e-06,
      "loss": 0.3414,
      "step": 34420
    },
    {
      "epoch": 0.6215738955214178,
      "grad_norm": 0.23669376969337463,
      "learning_rate": 6.272630195581315e-06,
      "loss": 0.2331,
      "step": 34425
    },
    {
      "epoch": 0.6216641749543186,
      "grad_norm": 0.2137349247932434,
      "learning_rate": 6.2699984908370815e-06,
      "loss": 0.2572,
      "step": 34430
    },
    {
      "epoch": 0.6217544543872193,
      "grad_norm": 0.5028064846992493,
      "learning_rate": 6.267367086144348e-06,
      "loss": 0.2618,
      "step": 34435
    },
    {
      "epoch": 0.6218447338201201,
      "grad_norm": 0.5181789398193359,
      "learning_rate": 6.264735981714798e-06,
      "loss": 0.1782,
      "step": 34440
    },
    {
      "epoch": 0.6219350132530207,
      "grad_norm": 0.8552795052528381,
      "learning_rate": 6.262105177760078e-06,
      "loss": 0.2167,
      "step": 34445
    },
    {
      "epoch": 0.6220252926859214,
      "grad_norm": 0.4309433102607727,
      "learning_rate": 6.259474674491816e-06,
      "loss": 0.189,
      "step": 34450
    },
    {
      "epoch": 0.6221155721188222,
      "grad_norm": 0.4273078143596649,
      "learning_rate": 6.256844472121624e-06,
      "loss": 0.2871,
      "step": 34455
    },
    {
      "epoch": 0.6222058515517229,
      "grad_norm": 0.40720057487487793,
      "learning_rate": 6.254214570861075e-06,
      "loss": 0.2457,
      "step": 34460
    },
    {
      "epoch": 0.6222961309846236,
      "grad_norm": 0.2771539092063904,
      "learning_rate": 6.251584970921736e-06,
      "loss": 0.1588,
      "step": 34465
    },
    {
      "epoch": 0.6223864104175243,
      "grad_norm": 0.5335532426834106,
      "learning_rate": 6.2489556725151256e-06,
      "loss": 0.2535,
      "step": 34470
    },
    {
      "epoch": 0.622476689850425,
      "grad_norm": 0.37770241498947144,
      "learning_rate": 6.246326675852763e-06,
      "loss": 0.1764,
      "step": 34475
    },
    {
      "epoch": 0.6225669692833258,
      "grad_norm": 0.39951586723327637,
      "learning_rate": 6.2436979811461244e-06,
      "loss": 0.2941,
      "step": 34480
    },
    {
      "epoch": 0.6226572487162265,
      "grad_norm": 0.4991743862628937,
      "learning_rate": 6.241069588606675e-06,
      "loss": 0.1905,
      "step": 34485
    },
    {
      "epoch": 0.6227475281491271,
      "grad_norm": 0.44111600518226624,
      "learning_rate": 6.238441498445844e-06,
      "loss": 0.2377,
      "step": 34490
    },
    {
      "epoch": 0.6228378075820279,
      "grad_norm": 0.6911139488220215,
      "learning_rate": 6.235813710875048e-06,
      "loss": 0.2707,
      "step": 34495
    },
    {
      "epoch": 0.6229280870149286,
      "grad_norm": 0.5214601159095764,
      "learning_rate": 6.233186226105669e-06,
      "loss": 0.2626,
      "step": 34500
    },
    {
      "epoch": 0.6230183664478294,
      "grad_norm": 0.2713903784751892,
      "learning_rate": 6.2305590443490735e-06,
      "loss": 0.2807,
      "step": 34505
    },
    {
      "epoch": 0.62310864588073,
      "grad_norm": 0.38960278034210205,
      "learning_rate": 6.22793216581659e-06,
      "loss": 0.2792,
      "step": 34510
    },
    {
      "epoch": 0.6231989253136307,
      "grad_norm": 0.5309028625488281,
      "learning_rate": 6.225305590719546e-06,
      "loss": 0.3215,
      "step": 34515
    },
    {
      "epoch": 0.6232892047465315,
      "grad_norm": 0.6727641224861145,
      "learning_rate": 6.222679319269218e-06,
      "loss": 0.2728,
      "step": 34520
    },
    {
      "epoch": 0.6233794841794322,
      "grad_norm": 0.6221792697906494,
      "learning_rate": 6.220053351676876e-06,
      "loss": 0.266,
      "step": 34525
    },
    {
      "epoch": 0.6234697636123329,
      "grad_norm": 0.3391434848308563,
      "learning_rate": 6.217427688153757e-06,
      "loss": 0.1645,
      "step": 34530
    },
    {
      "epoch": 0.6235600430452336,
      "grad_norm": 0.24358294904232025,
      "learning_rate": 6.214802328911085e-06,
      "loss": 0.1537,
      "step": 34535
    },
    {
      "epoch": 0.6236503224781343,
      "grad_norm": 0.6168825626373291,
      "learning_rate": 6.212177274160041e-06,
      "loss": 0.2049,
      "step": 34540
    },
    {
      "epoch": 0.6237406019110351,
      "grad_norm": 0.23135057091712952,
      "learning_rate": 6.209552524111791e-06,
      "loss": 0.1655,
      "step": 34545
    },
    {
      "epoch": 0.6238308813439357,
      "grad_norm": 0.3386957049369812,
      "learning_rate": 6.206928078977488e-06,
      "loss": 0.2538,
      "step": 34550
    },
    {
      "epoch": 0.6239211607768365,
      "grad_norm": 0.35345205664634705,
      "learning_rate": 6.204303938968237e-06,
      "loss": 0.2356,
      "step": 34555
    },
    {
      "epoch": 0.6240114402097372,
      "grad_norm": 0.3981238007545471,
      "learning_rate": 6.20168010429514e-06,
      "loss": 0.3016,
      "step": 34560
    },
    {
      "epoch": 0.6241017196426379,
      "grad_norm": 0.4604952931404114,
      "learning_rate": 6.19905657516926e-06,
      "loss": 0.2168,
      "step": 34565
    },
    {
      "epoch": 0.6241919990755386,
      "grad_norm": 0.7224426865577698,
      "learning_rate": 6.196433351801647e-06,
      "loss": 0.215,
      "step": 34570
    },
    {
      "epoch": 0.6242822785084393,
      "grad_norm": 0.6132074594497681,
      "learning_rate": 6.193810434403309e-06,
      "loss": 0.2833,
      "step": 34575
    },
    {
      "epoch": 0.6243725579413401,
      "grad_norm": 0.41222119331359863,
      "learning_rate": 6.191187823185256e-06,
      "loss": 0.2253,
      "step": 34580
    },
    {
      "epoch": 0.6244628373742408,
      "grad_norm": 0.7935979962348938,
      "learning_rate": 6.188565518358445e-06,
      "loss": 0.194,
      "step": 34585
    },
    {
      "epoch": 0.6245531168071414,
      "grad_norm": 0.7769051790237427,
      "learning_rate": 6.185943520133828e-06,
      "loss": 0.1693,
      "step": 34590
    },
    {
      "epoch": 0.6246433962400422,
      "grad_norm": 0.3631523549556732,
      "learning_rate": 6.1833218287223215e-06,
      "loss": 0.2517,
      "step": 34595
    },
    {
      "epoch": 0.6247336756729429,
      "grad_norm": 0.35102444887161255,
      "learning_rate": 6.180700444334828e-06,
      "loss": 0.1909,
      "step": 34600
    },
    {
      "epoch": 0.6248239551058437,
      "grad_norm": 0.8616628050804138,
      "learning_rate": 6.178079367182212e-06,
      "loss": 0.2941,
      "step": 34605
    },
    {
      "epoch": 0.6249142345387443,
      "grad_norm": 0.4506857693195343,
      "learning_rate": 6.175458597475324e-06,
      "loss": 0.2509,
      "step": 34610
    },
    {
      "epoch": 0.625004513971645,
      "grad_norm": 0.40034666657447815,
      "learning_rate": 6.172838135424982e-06,
      "loss": 0.2744,
      "step": 34615
    },
    {
      "epoch": 0.6250947934045458,
      "grad_norm": 0.41605767607688904,
      "learning_rate": 6.17021798124199e-06,
      "loss": 0.2696,
      "step": 34620
    },
    {
      "epoch": 0.6251850728374465,
      "grad_norm": 0.5841496586799622,
      "learning_rate": 6.167598135137114e-06,
      "loss": 0.2184,
      "step": 34625
    },
    {
      "epoch": 0.6252753522703471,
      "grad_norm": 0.3614466190338135,
      "learning_rate": 6.164978597321104e-06,
      "loss": 0.2185,
      "step": 34630
    },
    {
      "epoch": 0.6253656317032479,
      "grad_norm": 0.471843421459198,
      "learning_rate": 6.162359368004685e-06,
      "loss": 0.2819,
      "step": 34635
    },
    {
      "epoch": 0.6254559111361486,
      "grad_norm": 0.696699321269989,
      "learning_rate": 6.159740447398546e-06,
      "loss": 0.2147,
      "step": 34640
    },
    {
      "epoch": 0.6255461905690494,
      "grad_norm": 0.5258233547210693,
      "learning_rate": 6.157121835713373e-06,
      "loss": 0.3034,
      "step": 34645
    },
    {
      "epoch": 0.62563647000195,
      "grad_norm": 0.2687463164329529,
      "learning_rate": 6.154503533159806e-06,
      "loss": 0.2262,
      "step": 34650
    },
    {
      "epoch": 0.6257267494348507,
      "grad_norm": 0.4536837935447693,
      "learning_rate": 6.15188553994847e-06,
      "loss": 0.3113,
      "step": 34655
    },
    {
      "epoch": 0.6258170288677515,
      "grad_norm": 0.2425563782453537,
      "learning_rate": 6.149267856289962e-06,
      "loss": 0.2513,
      "step": 34660
    },
    {
      "epoch": 0.6259073083006522,
      "grad_norm": 0.4896921217441559,
      "learning_rate": 6.146650482394861e-06,
      "loss": 0.2983,
      "step": 34665
    },
    {
      "epoch": 0.6259975877335529,
      "grad_norm": 0.4157010018825531,
      "learning_rate": 6.14403341847371e-06,
      "loss": 0.2094,
      "step": 34670
    },
    {
      "epoch": 0.6260878671664536,
      "grad_norm": 0.6023677587509155,
      "learning_rate": 6.141416664737034e-06,
      "loss": 0.2272,
      "step": 34675
    },
    {
      "epoch": 0.6261781465993543,
      "grad_norm": 0.8046973347663879,
      "learning_rate": 6.138800221395333e-06,
      "loss": 0.292,
      "step": 34680
    },
    {
      "epoch": 0.6262684260322551,
      "grad_norm": 0.3939763009548187,
      "learning_rate": 6.136184088659084e-06,
      "loss": 0.3303,
      "step": 34685
    },
    {
      "epoch": 0.6263587054651557,
      "grad_norm": 0.28620174527168274,
      "learning_rate": 6.133568266738729e-06,
      "loss": 0.2183,
      "step": 34690
    },
    {
      "epoch": 0.6264489848980564,
      "grad_norm": 0.5315285921096802,
      "learning_rate": 6.130952755844697e-06,
      "loss": 0.1653,
      "step": 34695
    },
    {
      "epoch": 0.6265392643309572,
      "grad_norm": 0.244839608669281,
      "learning_rate": 6.128337556187383e-06,
      "loss": 0.2486,
      "step": 34700
    },
    {
      "epoch": 0.6266295437638579,
      "grad_norm": 0.4421430230140686,
      "learning_rate": 6.125722667977163e-06,
      "loss": 0.2309,
      "step": 34705
    },
    {
      "epoch": 0.6267198231967586,
      "grad_norm": 0.4121013879776001,
      "learning_rate": 6.123108091424386e-06,
      "loss": 0.1869,
      "step": 34710
    },
    {
      "epoch": 0.6268101026296593,
      "grad_norm": 0.3209613561630249,
      "learning_rate": 6.120493826739378e-06,
      "loss": 0.2081,
      "step": 34715
    },
    {
      "epoch": 0.62690038206256,
      "grad_norm": 0.30417075753211975,
      "learning_rate": 6.117879874132434e-06,
      "loss": 0.192,
      "step": 34720
    },
    {
      "epoch": 0.6269906614954608,
      "grad_norm": 0.34549078345298767,
      "learning_rate": 6.115266233813825e-06,
      "loss": 0.2372,
      "step": 34725
    },
    {
      "epoch": 0.6270809409283614,
      "grad_norm": 0.3664049804210663,
      "learning_rate": 6.112652905993808e-06,
      "loss": 0.1923,
      "step": 34730
    },
    {
      "epoch": 0.6271712203612622,
      "grad_norm": 0.5543047189712524,
      "learning_rate": 6.110039890882598e-06,
      "loss": 0.199,
      "step": 34735
    },
    {
      "epoch": 0.6272614997941629,
      "grad_norm": 0.4151439964771271,
      "learning_rate": 6.1074271886903955e-06,
      "loss": 0.1908,
      "step": 34740
    },
    {
      "epoch": 0.6273517792270636,
      "grad_norm": 0.684883713722229,
      "learning_rate": 6.104814799627373e-06,
      "loss": 0.208,
      "step": 34745
    },
    {
      "epoch": 0.6274420586599643,
      "grad_norm": 0.5946511626243591,
      "learning_rate": 6.102202723903683e-06,
      "loss": 0.246,
      "step": 34750
    },
    {
      "epoch": 0.627532338092865,
      "grad_norm": 0.48237690329551697,
      "learning_rate": 6.09959096172944e-06,
      "loss": 0.2516,
      "step": 34755
    },
    {
      "epoch": 0.6276226175257658,
      "grad_norm": 0.6267902255058289,
      "learning_rate": 6.096979513314749e-06,
      "loss": 0.2279,
      "step": 34760
    },
    {
      "epoch": 0.6277128969586665,
      "grad_norm": 0.45958006381988525,
      "learning_rate": 6.094368378869675e-06,
      "loss": 0.2476,
      "step": 34765
    },
    {
      "epoch": 0.6278031763915671,
      "grad_norm": 0.35172709822654724,
      "learning_rate": 6.091757558604272e-06,
      "loss": 0.2897,
      "step": 34770
    },
    {
      "epoch": 0.6278934558244679,
      "grad_norm": 0.9599870443344116,
      "learning_rate": 6.089147052728552e-06,
      "loss": 0.1377,
      "step": 34775
    },
    {
      "epoch": 0.6279837352573686,
      "grad_norm": 0.48359227180480957,
      "learning_rate": 6.086536861452523e-06,
      "loss": 0.2726,
      "step": 34780
    },
    {
      "epoch": 0.6280740146902694,
      "grad_norm": 0.8267931342124939,
      "learning_rate": 6.083926984986146e-06,
      "loss": 0.362,
      "step": 34785
    },
    {
      "epoch": 0.62816429412317,
      "grad_norm": 0.3220060467720032,
      "learning_rate": 6.081317423539373e-06,
      "loss": 0.2479,
      "step": 34790
    },
    {
      "epoch": 0.6282545735560707,
      "grad_norm": 0.25943776965141296,
      "learning_rate": 6.078708177322119e-06,
      "loss": 0.1505,
      "step": 34795
    },
    {
      "epoch": 0.6283448529889715,
      "grad_norm": 0.4073909819126129,
      "learning_rate": 6.076099246544286e-06,
      "loss": 0.1963,
      "step": 34800
    },
    {
      "epoch": 0.6284351324218722,
      "grad_norm": 0.362216591835022,
      "learning_rate": 6.073490631415735e-06,
      "loss": 0.1303,
      "step": 34805
    },
    {
      "epoch": 0.6285254118547728,
      "grad_norm": 1.4766465425491333,
      "learning_rate": 6.070882332146315e-06,
      "loss": 0.2623,
      "step": 34810
    },
    {
      "epoch": 0.6286156912876736,
      "grad_norm": 0.3296297788619995,
      "learning_rate": 6.0682743489458475e-06,
      "loss": 0.2353,
      "step": 34815
    },
    {
      "epoch": 0.6287059707205743,
      "grad_norm": 0.3773016631603241,
      "learning_rate": 6.065666682024119e-06,
      "loss": 0.2789,
      "step": 34820
    },
    {
      "epoch": 0.6287962501534751,
      "grad_norm": 0.39737921953201294,
      "learning_rate": 6.063059331590899e-06,
      "loss": 0.23,
      "step": 34825
    },
    {
      "epoch": 0.6288865295863757,
      "grad_norm": 0.46278733015060425,
      "learning_rate": 6.060452297855932e-06,
      "loss": 0.2963,
      "step": 34830
    },
    {
      "epoch": 0.6289768090192764,
      "grad_norm": 0.2169886976480484,
      "learning_rate": 6.057845581028938e-06,
      "loss": 0.2966,
      "step": 34835
    },
    {
      "epoch": 0.6290670884521772,
      "grad_norm": 0.5035675764083862,
      "learning_rate": 6.0552391813196e-06,
      "loss": 0.2011,
      "step": 34840
    },
    {
      "epoch": 0.6291573678850779,
      "grad_norm": 0.46862685680389404,
      "learning_rate": 6.0526330989375925e-06,
      "loss": 0.2631,
      "step": 34845
    },
    {
      "epoch": 0.6292476473179786,
      "grad_norm": 0.3086468577384949,
      "learning_rate": 6.050027334092549e-06,
      "loss": 0.1986,
      "step": 34850
    },
    {
      "epoch": 0.6293379267508793,
      "grad_norm": 0.5292624831199646,
      "learning_rate": 6.0474218869940905e-06,
      "loss": 0.2383,
      "step": 34855
    },
    {
      "epoch": 0.62942820618378,
      "grad_norm": 0.3100101947784424,
      "learning_rate": 6.044816757851799e-06,
      "loss": 0.2109,
      "step": 34860
    },
    {
      "epoch": 0.6295184856166808,
      "grad_norm": 0.2324603945016861,
      "learning_rate": 6.042211946875248e-06,
      "loss": 0.2094,
      "step": 34865
    },
    {
      "epoch": 0.6296087650495814,
      "grad_norm": 0.5259659886360168,
      "learning_rate": 6.039607454273966e-06,
      "loss": 0.2353,
      "step": 34870
    },
    {
      "epoch": 0.6296990444824822,
      "grad_norm": 0.48725876212120056,
      "learning_rate": 6.037003280257471e-06,
      "loss": 0.1715,
      "step": 34875
    },
    {
      "epoch": 0.6297893239153829,
      "grad_norm": 0.34312328696250916,
      "learning_rate": 6.034399425035247e-06,
      "loss": 0.1883,
      "step": 34880
    },
    {
      "epoch": 0.6298796033482836,
      "grad_norm": 0.6141727566719055,
      "learning_rate": 6.031795888816758e-06,
      "loss": 0.1919,
      "step": 34885
    },
    {
      "epoch": 0.6299698827811844,
      "grad_norm": 0.3491286337375641,
      "learning_rate": 6.029192671811437e-06,
      "loss": 0.1581,
      "step": 34890
    },
    {
      "epoch": 0.630060162214085,
      "grad_norm": 0.33346256613731384,
      "learning_rate": 6.026589774228696e-06,
      "loss": 0.2257,
      "step": 34895
    },
    {
      "epoch": 0.6301504416469857,
      "grad_norm": 0.2053835690021515,
      "learning_rate": 6.0239871962779166e-06,
      "loss": 0.1951,
      "step": 34900
    },
    {
      "epoch": 0.6302407210798865,
      "grad_norm": 0.17909343540668488,
      "learning_rate": 6.0213849381684595e-06,
      "loss": 0.1976,
      "step": 34905
    },
    {
      "epoch": 0.6303310005127872,
      "grad_norm": 0.5308035016059875,
      "learning_rate": 6.01878300010966e-06,
      "loss": 0.2853,
      "step": 34910
    },
    {
      "epoch": 0.6304212799456879,
      "grad_norm": 0.5771810412406921,
      "learning_rate": 6.016181382310818e-06,
      "loss": 0.2938,
      "step": 34915
    },
    {
      "epoch": 0.6305115593785886,
      "grad_norm": 0.31199419498443604,
      "learning_rate": 6.0135800849812196e-06,
      "loss": 0.1956,
      "step": 34920
    },
    {
      "epoch": 0.6306018388114893,
      "grad_norm": 0.8073880076408386,
      "learning_rate": 6.010979108330118e-06,
      "loss": 0.2518,
      "step": 34925
    },
    {
      "epoch": 0.6306921182443901,
      "grad_norm": 0.8024955987930298,
      "learning_rate": 6.008378452566747e-06,
      "loss": 0.2603,
      "step": 34930
    },
    {
      "epoch": 0.6307823976772907,
      "grad_norm": 0.5539519190788269,
      "learning_rate": 6.005778117900306e-06,
      "loss": 0.2635,
      "step": 34935
    },
    {
      "epoch": 0.6308726771101915,
      "grad_norm": 0.6516796350479126,
      "learning_rate": 6.003178104539975e-06,
      "loss": 0.2426,
      "step": 34940
    },
    {
      "epoch": 0.6309629565430922,
      "grad_norm": 0.5461468696594238,
      "learning_rate": 6.000578412694904e-06,
      "loss": 0.1898,
      "step": 34945
    },
    {
      "epoch": 0.6310532359759929,
      "grad_norm": 0.231127068400383,
      "learning_rate": 5.997979042574225e-06,
      "loss": 0.2876,
      "step": 34950
    },
    {
      "epoch": 0.6311435154088936,
      "grad_norm": 0.5740653276443481,
      "learning_rate": 5.995379994387032e-06,
      "loss": 0.2504,
      "step": 34955
    },
    {
      "epoch": 0.6312337948417943,
      "grad_norm": 0.3960036635398865,
      "learning_rate": 5.992781268342403e-06,
      "loss": 0.2665,
      "step": 34960
    },
    {
      "epoch": 0.6313240742746951,
      "grad_norm": 0.2456202656030655,
      "learning_rate": 5.990182864649382e-06,
      "loss": 0.2269,
      "step": 34965
    },
    {
      "epoch": 0.6314143537075958,
      "grad_norm": 0.3660065233707428,
      "learning_rate": 5.987584783516999e-06,
      "loss": 0.1913,
      "step": 34970
    },
    {
      "epoch": 0.6315046331404964,
      "grad_norm": 0.7571065425872803,
      "learning_rate": 5.984987025154245e-06,
      "loss": 0.2416,
      "step": 34975
    },
    {
      "epoch": 0.6315949125733972,
      "grad_norm": 0.5113848447799683,
      "learning_rate": 5.982389589770097e-06,
      "loss": 0.2439,
      "step": 34980
    },
    {
      "epoch": 0.6316851920062979,
      "grad_norm": 0.5383720397949219,
      "learning_rate": 5.979792477573491e-06,
      "loss": 0.188,
      "step": 34985
    },
    {
      "epoch": 0.6317754714391987,
      "grad_norm": 0.3642173111438751,
      "learning_rate": 5.977195688773352e-06,
      "loss": 0.1801,
      "step": 34990
    },
    {
      "epoch": 0.6318657508720993,
      "grad_norm": 0.46961209177970886,
      "learning_rate": 5.974599223578574e-06,
      "loss": 0.1936,
      "step": 34995
    },
    {
      "epoch": 0.631956030305,
      "grad_norm": 0.27093908190727234,
      "learning_rate": 5.9720030821980165e-06,
      "loss": 0.2676,
      "step": 35000
    },
    {
      "epoch": 0.6320463097379008,
      "grad_norm": 0.48034536838531494,
      "learning_rate": 5.969407264840528e-06,
      "loss": 0.3011,
      "step": 35005
    },
    {
      "epoch": 0.6321365891708015,
      "grad_norm": 0.4448093771934509,
      "learning_rate": 5.966811771714918e-06,
      "loss": 0.277,
      "step": 35010
    },
    {
      "epoch": 0.6322268686037021,
      "grad_norm": 0.5743049383163452,
      "learning_rate": 5.96421660302998e-06,
      "loss": 0.2075,
      "step": 35015
    },
    {
      "epoch": 0.6323171480366029,
      "grad_norm": 0.6041221618652344,
      "learning_rate": 5.961621758994468e-06,
      "loss": 0.2375,
      "step": 35020
    },
    {
      "epoch": 0.6324074274695036,
      "grad_norm": 0.5226449966430664,
      "learning_rate": 5.959027239817127e-06,
      "loss": 0.3159,
      "step": 35025
    },
    {
      "epoch": 0.6324977069024044,
      "grad_norm": 0.4296029210090637,
      "learning_rate": 5.956433045706662e-06,
      "loss": 0.2542,
      "step": 35030
    },
    {
      "epoch": 0.632587986335305,
      "grad_norm": 0.3556138873100281,
      "learning_rate": 5.95383917687176e-06,
      "loss": 0.2856,
      "step": 35035
    },
    {
      "epoch": 0.6326782657682057,
      "grad_norm": 0.5710278749465942,
      "learning_rate": 5.951245633521077e-06,
      "loss": 0.2282,
      "step": 35040
    },
    {
      "epoch": 0.6327685452011065,
      "grad_norm": 0.3186838626861572,
      "learning_rate": 5.948652415863248e-06,
      "loss": 0.165,
      "step": 35045
    },
    {
      "epoch": 0.6328588246340072,
      "grad_norm": 0.25505760312080383,
      "learning_rate": 5.946059524106873e-06,
      "loss": 0.2014,
      "step": 35050
    },
    {
      "epoch": 0.6329491040669079,
      "grad_norm": 0.33168452978134155,
      "learning_rate": 5.943466958460536e-06,
      "loss": 0.2486,
      "step": 35055
    },
    {
      "epoch": 0.6330393834998086,
      "grad_norm": 0.6071575880050659,
      "learning_rate": 5.940874719132786e-06,
      "loss": 0.2014,
      "step": 35060
    },
    {
      "epoch": 0.6331296629327093,
      "grad_norm": 0.7817810773849487,
      "learning_rate": 5.938282806332156e-06,
      "loss": 0.3459,
      "step": 35065
    },
    {
      "epoch": 0.6332199423656101,
      "grad_norm": 0.4133113920688629,
      "learning_rate": 5.93569122026714e-06,
      "loss": 0.3335,
      "step": 35070
    },
    {
      "epoch": 0.6333102217985107,
      "grad_norm": 0.5362765192985535,
      "learning_rate": 5.933099961146216e-06,
      "loss": 0.2565,
      "step": 35075
    },
    {
      "epoch": 0.6334005012314115,
      "grad_norm": 0.37904590368270874,
      "learning_rate": 5.9305090291778285e-06,
      "loss": 0.2446,
      "step": 35080
    },
    {
      "epoch": 0.6334907806643122,
      "grad_norm": 0.6799839735031128,
      "learning_rate": 5.927918424570407e-06,
      "loss": 0.2036,
      "step": 35085
    },
    {
      "epoch": 0.6335810600972129,
      "grad_norm": 0.33658117055892944,
      "learning_rate": 5.925328147532339e-06,
      "loss": 0.2155,
      "step": 35090
    },
    {
      "epoch": 0.6336713395301136,
      "grad_norm": 0.25472602248191833,
      "learning_rate": 5.922738198271993e-06,
      "loss": 0.3382,
      "step": 35095
    },
    {
      "epoch": 0.6337616189630143,
      "grad_norm": 0.5699844360351562,
      "learning_rate": 5.920148576997716e-06,
      "loss": 0.3147,
      "step": 35100
    },
    {
      "epoch": 0.633851898395915,
      "grad_norm": 0.49318039417266846,
      "learning_rate": 5.917559283917822e-06,
      "loss": 0.2488,
      "step": 35105
    },
    {
      "epoch": 0.6339421778288158,
      "grad_norm": 0.5778481364250183,
      "learning_rate": 5.914970319240605e-06,
      "loss": 0.2135,
      "step": 35110
    },
    {
      "epoch": 0.6340324572617164,
      "grad_norm": 0.3343014717102051,
      "learning_rate": 5.9123816831743195e-06,
      "loss": 0.1618,
      "step": 35115
    },
    {
      "epoch": 0.6341227366946172,
      "grad_norm": 0.5119194388389587,
      "learning_rate": 5.9097933759272105e-06,
      "loss": 0.2919,
      "step": 35120
    },
    {
      "epoch": 0.6342130161275179,
      "grad_norm": 0.7334364652633667,
      "learning_rate": 5.907205397707485e-06,
      "loss": 0.1894,
      "step": 35125
    },
    {
      "epoch": 0.6343032955604186,
      "grad_norm": 0.5871245265007019,
      "learning_rate": 5.90461774872333e-06,
      "loss": 0.2589,
      "step": 35130
    },
    {
      "epoch": 0.6343935749933193,
      "grad_norm": 0.3996553421020508,
      "learning_rate": 5.902030429182898e-06,
      "loss": 0.1758,
      "step": 35135
    },
    {
      "epoch": 0.63448385442622,
      "grad_norm": 0.40675926208496094,
      "learning_rate": 5.899443439294323e-06,
      "loss": 0.2498,
      "step": 35140
    },
    {
      "epoch": 0.6345741338591208,
      "grad_norm": 0.3976016342639923,
      "learning_rate": 5.896856779265709e-06,
      "loss": 0.1763,
      "step": 35145
    },
    {
      "epoch": 0.6346644132920215,
      "grad_norm": 0.29940494894981384,
      "learning_rate": 5.89427044930514e-06,
      "loss": 0.276,
      "step": 35150
    },
    {
      "epoch": 0.6347546927249221,
      "grad_norm": 0.5581749081611633,
      "learning_rate": 5.891684449620655e-06,
      "loss": 0.2797,
      "step": 35155
    },
    {
      "epoch": 0.6348449721578229,
      "grad_norm": 0.27761101722717285,
      "learning_rate": 5.889098780420288e-06,
      "loss": 0.1851,
      "step": 35160
    },
    {
      "epoch": 0.6349352515907236,
      "grad_norm": 0.3999343514442444,
      "learning_rate": 5.886513441912034e-06,
      "loss": 0.2203,
      "step": 35165
    },
    {
      "epoch": 0.6350255310236244,
      "grad_norm": 0.5790286064147949,
      "learning_rate": 5.883928434303867e-06,
      "loss": 0.2046,
      "step": 35170
    },
    {
      "epoch": 0.635115810456525,
      "grad_norm": 0.4353305995464325,
      "learning_rate": 5.881343757803729e-06,
      "loss": 0.1458,
      "step": 35175
    },
    {
      "epoch": 0.6352060898894257,
      "grad_norm": 0.4601106643676758,
      "learning_rate": 5.878759412619544e-06,
      "loss": 0.162,
      "step": 35180
    },
    {
      "epoch": 0.6352963693223265,
      "grad_norm": 0.46851181983947754,
      "learning_rate": 5.876175398959198e-06,
      "loss": 0.2171,
      "step": 35185
    },
    {
      "epoch": 0.6353866487552272,
      "grad_norm": 1.0405105352401733,
      "learning_rate": 5.873591717030557e-06,
      "loss": 0.1599,
      "step": 35190
    },
    {
      "epoch": 0.6354769281881278,
      "grad_norm": 0.7821049690246582,
      "learning_rate": 5.871008367041463e-06,
      "loss": 0.2537,
      "step": 35195
    },
    {
      "epoch": 0.6355672076210286,
      "grad_norm": 0.5013983249664307,
      "learning_rate": 5.868425349199723e-06,
      "loss": 0.2079,
      "step": 35200
    },
    {
      "epoch": 0.6356574870539293,
      "grad_norm": 0.41896483302116394,
      "learning_rate": 5.865842663713124e-06,
      "loss": 0.2,
      "step": 35205
    },
    {
      "epoch": 0.6357477664868301,
      "grad_norm": 0.5238785743713379,
      "learning_rate": 5.863260310789423e-06,
      "loss": 0.2898,
      "step": 35210
    },
    {
      "epoch": 0.6358380459197307,
      "grad_norm": 0.2647916078567505,
      "learning_rate": 5.860678290636358e-06,
      "loss": 0.2858,
      "step": 35215
    },
    {
      "epoch": 0.6359283253526314,
      "grad_norm": 0.6003150343894958,
      "learning_rate": 5.858096603461625e-06,
      "loss": 0.308,
      "step": 35220
    },
    {
      "epoch": 0.6360186047855322,
      "grad_norm": 0.3803805708885193,
      "learning_rate": 5.855515249472906e-06,
      "loss": 0.2548,
      "step": 35225
    },
    {
      "epoch": 0.6361088842184329,
      "grad_norm": 0.696205735206604,
      "learning_rate": 5.852934228877851e-06,
      "loss": 0.2959,
      "step": 35230
    },
    {
      "epoch": 0.6361991636513336,
      "grad_norm": 0.7358081936836243,
      "learning_rate": 5.850353541884086e-06,
      "loss": 0.1871,
      "step": 35235
    },
    {
      "epoch": 0.6362894430842343,
      "grad_norm": 0.40561583638191223,
      "learning_rate": 5.847773188699207e-06,
      "loss": 0.2418,
      "step": 35240
    },
    {
      "epoch": 0.636379722517135,
      "grad_norm": 0.368505597114563,
      "learning_rate": 5.8451931695307894e-06,
      "loss": 0.3055,
      "step": 35245
    },
    {
      "epoch": 0.6364700019500358,
      "grad_norm": 0.36378854513168335,
      "learning_rate": 5.842613484586368e-06,
      "loss": 0.2551,
      "step": 35250
    },
    {
      "epoch": 0.6365602813829364,
      "grad_norm": 0.6695896983146667,
      "learning_rate": 5.840034134073468e-06,
      "loss": 0.1631,
      "step": 35255
    },
    {
      "epoch": 0.6366505608158372,
      "grad_norm": 0.5897035598754883,
      "learning_rate": 5.837455118199573e-06,
      "loss": 0.2288,
      "step": 35260
    },
    {
      "epoch": 0.6367408402487379,
      "grad_norm": 0.573806643486023,
      "learning_rate": 5.834876437172154e-06,
      "loss": 0.2627,
      "step": 35265
    },
    {
      "epoch": 0.6368311196816386,
      "grad_norm": 0.38577497005462646,
      "learning_rate": 5.832298091198637e-06,
      "loss": 0.1708,
      "step": 35270
    },
    {
      "epoch": 0.6369213991145394,
      "grad_norm": 0.7552957534790039,
      "learning_rate": 5.82972008048644e-06,
      "loss": 0.2758,
      "step": 35275
    },
    {
      "epoch": 0.63701167854744,
      "grad_norm": 0.4966082274913788,
      "learning_rate": 5.82714240524294e-06,
      "loss": 0.2313,
      "step": 35280
    },
    {
      "epoch": 0.6371019579803407,
      "grad_norm": 0.39154505729675293,
      "learning_rate": 5.824565065675493e-06,
      "loss": 0.1762,
      "step": 35285
    },
    {
      "epoch": 0.6371922374132415,
      "grad_norm": 0.5730046629905701,
      "learning_rate": 5.821988061991432e-06,
      "loss": 0.3759,
      "step": 35290
    },
    {
      "epoch": 0.6372825168461422,
      "grad_norm": 0.6496819853782654,
      "learning_rate": 5.819411394398049e-06,
      "loss": 0.2163,
      "step": 35295
    },
    {
      "epoch": 0.6373727962790429,
      "grad_norm": 0.45367497205734253,
      "learning_rate": 5.81683506310263e-06,
      "loss": 0.1538,
      "step": 35300
    },
    {
      "epoch": 0.6374630757119436,
      "grad_norm": 0.6378582715988159,
      "learning_rate": 5.814259068312409e-06,
      "loss": 0.1712,
      "step": 35305
    },
    {
      "epoch": 0.6375533551448443,
      "grad_norm": 0.36133307218551636,
      "learning_rate": 5.811683410234614e-06,
      "loss": 0.1385,
      "step": 35310
    },
    {
      "epoch": 0.6376436345777451,
      "grad_norm": 0.29456639289855957,
      "learning_rate": 5.809108089076436e-06,
      "loss": 0.2176,
      "step": 35315
    },
    {
      "epoch": 0.6377339140106457,
      "grad_norm": 0.5693657994270325,
      "learning_rate": 5.806533105045047e-06,
      "loss": 0.2002,
      "step": 35320
    },
    {
      "epoch": 0.6378241934435465,
      "grad_norm": 0.6943349242210388,
      "learning_rate": 5.803958458347576e-06,
      "loss": 0.169,
      "step": 35325
    },
    {
      "epoch": 0.6379144728764472,
      "grad_norm": 0.31143563985824585,
      "learning_rate": 5.8013841491911415e-06,
      "loss": 0.2483,
      "step": 35330
    },
    {
      "epoch": 0.6380047523093479,
      "grad_norm": 0.5593167543411255,
      "learning_rate": 5.798810177782822e-06,
      "loss": 0.2721,
      "step": 35335
    },
    {
      "epoch": 0.6380950317422486,
      "grad_norm": 0.2441977560520172,
      "learning_rate": 5.796236544329682e-06,
      "loss": 0.3029,
      "step": 35340
    },
    {
      "epoch": 0.6381853111751493,
      "grad_norm": 0.4800540804862976,
      "learning_rate": 5.793663249038743e-06,
      "loss": 0.1755,
      "step": 35345
    },
    {
      "epoch": 0.6382755906080501,
      "grad_norm": 0.4255560636520386,
      "learning_rate": 5.7910902921170145e-06,
      "loss": 0.2864,
      "step": 35350
    },
    {
      "epoch": 0.6383658700409508,
      "grad_norm": 0.7475911974906921,
      "learning_rate": 5.788517673771469e-06,
      "loss": 0.2881,
      "step": 35355
    },
    {
      "epoch": 0.6384561494738514,
      "grad_norm": 0.3153039216995239,
      "learning_rate": 5.7859453942090605e-06,
      "loss": 0.3081,
      "step": 35360
    },
    {
      "epoch": 0.6385464289067522,
      "grad_norm": 0.5710461735725403,
      "learning_rate": 5.783373453636705e-06,
      "loss": 0.1815,
      "step": 35365
    },
    {
      "epoch": 0.6386367083396529,
      "grad_norm": 0.4108995199203491,
      "learning_rate": 5.780801852261295e-06,
      "loss": 0.3069,
      "step": 35370
    },
    {
      "epoch": 0.6387269877725537,
      "grad_norm": 0.7596222162246704,
      "learning_rate": 5.778230590289697e-06,
      "loss": 0.2432,
      "step": 35375
    },
    {
      "epoch": 0.6388172672054543,
      "grad_norm": 0.41312968730926514,
      "learning_rate": 5.775659667928755e-06,
      "loss": 0.1997,
      "step": 35380
    },
    {
      "epoch": 0.638907546638355,
      "grad_norm": 0.4853145480155945,
      "learning_rate": 5.773089085385281e-06,
      "loss": 0.2101,
      "step": 35385
    },
    {
      "epoch": 0.6389978260712558,
      "grad_norm": 0.7995417714118958,
      "learning_rate": 5.770518842866055e-06,
      "loss": 0.2198,
      "step": 35390
    },
    {
      "epoch": 0.6390881055041565,
      "grad_norm": 0.8485028743743896,
      "learning_rate": 5.767948940577841e-06,
      "loss": 0.1597,
      "step": 35395
    },
    {
      "epoch": 0.6391783849370571,
      "grad_norm": 0.4845674932003021,
      "learning_rate": 5.76537937872736e-06,
      "loss": 0.2332,
      "step": 35400
    },
    {
      "epoch": 0.6392686643699579,
      "grad_norm": 0.41197094321250916,
      "learning_rate": 5.762810157521324e-06,
      "loss": 0.253,
      "step": 35405
    },
    {
      "epoch": 0.6393589438028586,
      "grad_norm": 0.4559381306171417,
      "learning_rate": 5.760241277166399e-06,
      "loss": 0.2397,
      "step": 35410
    },
    {
      "epoch": 0.6394492232357594,
      "grad_norm": 0.22541190683841705,
      "learning_rate": 5.757672737869237e-06,
      "loss": 0.2344,
      "step": 35415
    },
    {
      "epoch": 0.63953950266866,
      "grad_norm": 0.3318548798561096,
      "learning_rate": 5.755104539836458e-06,
      "loss": 0.2098,
      "step": 35420
    },
    {
      "epoch": 0.6396297821015607,
      "grad_norm": 0.6875835657119751,
      "learning_rate": 5.7525366832746595e-06,
      "loss": 0.2797,
      "step": 35425
    },
    {
      "epoch": 0.6397200615344615,
      "grad_norm": 0.4598506689071655,
      "learning_rate": 5.749969168390399e-06,
      "loss": 0.2842,
      "step": 35430
    },
    {
      "epoch": 0.6398103409673622,
      "grad_norm": 0.5909374952316284,
      "learning_rate": 5.747401995390223e-06,
      "loss": 0.2285,
      "step": 35435
    },
    {
      "epoch": 0.6399006204002629,
      "grad_norm": 0.507408857345581,
      "learning_rate": 5.744835164480633e-06,
      "loss": 0.2556,
      "step": 35440
    },
    {
      "epoch": 0.6399908998331636,
      "grad_norm": 0.3172169029712677,
      "learning_rate": 5.742268675868116e-06,
      "loss": 0.2692,
      "step": 35445
    },
    {
      "epoch": 0.6400811792660643,
      "grad_norm": 0.5158324837684631,
      "learning_rate": 5.739702529759126e-06,
      "loss": 0.2005,
      "step": 35450
    },
    {
      "epoch": 0.6401714586989651,
      "grad_norm": 0.4424983263015747,
      "learning_rate": 5.7371367263600975e-06,
      "loss": 0.2113,
      "step": 35455
    },
    {
      "epoch": 0.6402617381318657,
      "grad_norm": 0.4412218928337097,
      "learning_rate": 5.734571265877424e-06,
      "loss": 0.2653,
      "step": 35460
    },
    {
      "epoch": 0.6403520175647665,
      "grad_norm": 0.3951146602630615,
      "learning_rate": 5.732006148517477e-06,
      "loss": 0.2267,
      "step": 35465
    },
    {
      "epoch": 0.6404422969976672,
      "grad_norm": 0.5053501725196838,
      "learning_rate": 5.729441374486609e-06,
      "loss": 0.3119,
      "step": 35470
    },
    {
      "epoch": 0.6405325764305679,
      "grad_norm": 0.397300124168396,
      "learning_rate": 5.726876943991125e-06,
      "loss": 0.2443,
      "step": 35475
    },
    {
      "epoch": 0.6406228558634686,
      "grad_norm": 0.59889155626297,
      "learning_rate": 5.724312857237324e-06,
      "loss": 0.262,
      "step": 35480
    },
    {
      "epoch": 0.6407131352963693,
      "grad_norm": 0.43211787939071655,
      "learning_rate": 5.7217491144314674e-06,
      "loss": 0.1797,
      "step": 35485
    },
    {
      "epoch": 0.64080341472927,
      "grad_norm": 0.5028895735740662,
      "learning_rate": 5.7191857157797914e-06,
      "loss": 0.372,
      "step": 35490
    },
    {
      "epoch": 0.6408936941621708,
      "grad_norm": 0.4395959675312042,
      "learning_rate": 5.716622661488497e-06,
      "loss": 0.266,
      "step": 35495
    },
    {
      "epoch": 0.6409839735950714,
      "grad_norm": 0.29293063282966614,
      "learning_rate": 5.714059951763768e-06,
      "loss": 0.231,
      "step": 35500
    },
    {
      "epoch": 0.6410742530279722,
      "grad_norm": 0.9143898487091064,
      "learning_rate": 5.711497586811754e-06,
      "loss": 0.1584,
      "step": 35505
    },
    {
      "epoch": 0.6411645324608729,
      "grad_norm": 0.4413289427757263,
      "learning_rate": 5.708935566838576e-06,
      "loss": 0.1866,
      "step": 35510
    },
    {
      "epoch": 0.6412548118937736,
      "grad_norm": 0.4569292962551117,
      "learning_rate": 5.706373892050332e-06,
      "loss": 0.2071,
      "step": 35515
    },
    {
      "epoch": 0.6413450913266743,
      "grad_norm": 0.42200034856796265,
      "learning_rate": 5.703812562653097e-06,
      "loss": 0.1951,
      "step": 35520
    },
    {
      "epoch": 0.641435370759575,
      "grad_norm": 0.44226059317588806,
      "learning_rate": 5.7012515788529e-06,
      "loss": 0.2704,
      "step": 35525
    },
    {
      "epoch": 0.6415256501924758,
      "grad_norm": 0.565065860748291,
      "learning_rate": 5.698690940855762e-06,
      "loss": 0.2008,
      "step": 35530
    },
    {
      "epoch": 0.6416159296253765,
      "grad_norm": 0.20949938893318176,
      "learning_rate": 5.6961306488676605e-06,
      "loss": 0.2107,
      "step": 35535
    },
    {
      "epoch": 0.6417062090582771,
      "grad_norm": 0.994368314743042,
      "learning_rate": 5.6935707030945595e-06,
      "loss": 0.2555,
      "step": 35540
    },
    {
      "epoch": 0.6417964884911779,
      "grad_norm": 0.2960260808467865,
      "learning_rate": 5.691011103742379e-06,
      "loss": 0.2338,
      "step": 35545
    },
    {
      "epoch": 0.6418867679240786,
      "grad_norm": 0.31084543466567993,
      "learning_rate": 5.688451851017032e-06,
      "loss": 0.2752,
      "step": 35550
    },
    {
      "epoch": 0.6419770473569794,
      "grad_norm": 0.46179312467575073,
      "learning_rate": 5.685892945124385e-06,
      "loss": 0.2412,
      "step": 35555
    },
    {
      "epoch": 0.64206732678988,
      "grad_norm": 0.6409241557121277,
      "learning_rate": 5.683334386270282e-06,
      "loss": 0.2783,
      "step": 35560
    },
    {
      "epoch": 0.6421576062227807,
      "grad_norm": 0.3262818455696106,
      "learning_rate": 5.680776174660545e-06,
      "loss": 0.2545,
      "step": 35565
    },
    {
      "epoch": 0.6422478856556815,
      "grad_norm": 0.27425482869148254,
      "learning_rate": 5.678218310500957e-06,
      "loss": 0.2485,
      "step": 35570
    },
    {
      "epoch": 0.6423381650885822,
      "grad_norm": 0.5403268337249756,
      "learning_rate": 5.675660793997284e-06,
      "loss": 0.263,
      "step": 35575
    },
    {
      "epoch": 0.6424284445214828,
      "grad_norm": 0.38726648688316345,
      "learning_rate": 5.673103625355258e-06,
      "loss": 0.2678,
      "step": 35580
    },
    {
      "epoch": 0.6425187239543836,
      "grad_norm": 0.4004386365413666,
      "learning_rate": 5.670546804780593e-06,
      "loss": 0.3211,
      "step": 35585
    },
    {
      "epoch": 0.6426090033872843,
      "grad_norm": 0.6058290004730225,
      "learning_rate": 5.667990332478952e-06,
      "loss": 0.1706,
      "step": 35590
    },
    {
      "epoch": 0.6426992828201851,
      "grad_norm": 0.6734848022460938,
      "learning_rate": 5.665434208655997e-06,
      "loss": 0.314,
      "step": 35595
    },
    {
      "epoch": 0.6427895622530857,
      "grad_norm": 0.4511653482913971,
      "learning_rate": 5.66287843351734e-06,
      "loss": 0.1464,
      "step": 35600
    },
    {
      "epoch": 0.6428798416859864,
      "grad_norm": 0.32747411727905273,
      "learning_rate": 5.660323007268585e-06,
      "loss": 0.195,
      "step": 35605
    },
    {
      "epoch": 0.6429701211188872,
      "grad_norm": 0.6119348406791687,
      "learning_rate": 5.657767930115285e-06,
      "loss": 0.2925,
      "step": 35610
    },
    {
      "epoch": 0.6430604005517879,
      "grad_norm": 0.4540742039680481,
      "learning_rate": 5.655213202262986e-06,
      "loss": 0.2207,
      "step": 35615
    },
    {
      "epoch": 0.6431506799846886,
      "grad_norm": 0.2162233144044876,
      "learning_rate": 5.652658823917194e-06,
      "loss": 0.1626,
      "step": 35620
    },
    {
      "epoch": 0.6432409594175893,
      "grad_norm": 0.5270794630050659,
      "learning_rate": 5.650104795283396e-06,
      "loss": 0.2382,
      "step": 35625
    },
    {
      "epoch": 0.64333123885049,
      "grad_norm": 0.35142067074775696,
      "learning_rate": 5.647551116567036e-06,
      "loss": 0.2793,
      "step": 35630
    },
    {
      "epoch": 0.6434215182833908,
      "grad_norm": 0.3540780246257782,
      "learning_rate": 5.644997787973548e-06,
      "loss": 0.2156,
      "step": 35635
    },
    {
      "epoch": 0.6435117977162914,
      "grad_norm": 0.4894404709339142,
      "learning_rate": 5.642444809708321e-06,
      "loss": 0.1685,
      "step": 35640
    },
    {
      "epoch": 0.6436020771491922,
      "grad_norm": 0.5173498392105103,
      "learning_rate": 5.639892181976725e-06,
      "loss": 0.3287,
      "step": 35645
    },
    {
      "epoch": 0.6436923565820929,
      "grad_norm": 0.39868810772895813,
      "learning_rate": 5.637339904984105e-06,
      "loss": 0.2852,
      "step": 35650
    },
    {
      "epoch": 0.6437826360149936,
      "grad_norm": 0.41706356406211853,
      "learning_rate": 5.634787978935767e-06,
      "loss": 0.1923,
      "step": 35655
    },
    {
      "epoch": 0.6438729154478943,
      "grad_norm": 0.5662165880203247,
      "learning_rate": 5.632236404037002e-06,
      "loss": 0.2293,
      "step": 35660
    },
    {
      "epoch": 0.643963194880795,
      "grad_norm": 0.6363428235054016,
      "learning_rate": 5.629685180493058e-06,
      "loss": 0.2729,
      "step": 35665
    },
    {
      "epoch": 0.6440534743136958,
      "grad_norm": 0.7945840358734131,
      "learning_rate": 5.62713430850917e-06,
      "loss": 0.3457,
      "step": 35670
    },
    {
      "epoch": 0.6441437537465965,
      "grad_norm": 0.8168322443962097,
      "learning_rate": 5.624583788290529e-06,
      "loss": 0.3418,
      "step": 35675
    },
    {
      "epoch": 0.6442340331794972,
      "grad_norm": 0.5742879509925842,
      "learning_rate": 5.6220336200423086e-06,
      "loss": 0.1267,
      "step": 35680
    },
    {
      "epoch": 0.6443243126123979,
      "grad_norm": 0.7505717277526855,
      "learning_rate": 5.6194838039696545e-06,
      "loss": 0.2254,
      "step": 35685
    },
    {
      "epoch": 0.6444145920452986,
      "grad_norm": 0.7047513723373413,
      "learning_rate": 5.616934340277682e-06,
      "loss": 0.2286,
      "step": 35690
    },
    {
      "epoch": 0.6445048714781993,
      "grad_norm": 0.304597407579422,
      "learning_rate": 5.6143852291714705e-06,
      "loss": 0.2411,
      "step": 35695
    },
    {
      "epoch": 0.6445951509111001,
      "grad_norm": 0.34877634048461914,
      "learning_rate": 5.611836470856083e-06,
      "loss": 0.2565,
      "step": 35700
    },
    {
      "epoch": 0.6446854303440007,
      "grad_norm": 0.276822566986084,
      "learning_rate": 5.609288065536545e-06,
      "loss": 0.1882,
      "step": 35705
    },
    {
      "epoch": 0.6447757097769015,
      "grad_norm": 0.40141189098358154,
      "learning_rate": 5.606740013417858e-06,
      "loss": 0.2008,
      "step": 35710
    },
    {
      "epoch": 0.6448659892098022,
      "grad_norm": 0.4635002613067627,
      "learning_rate": 5.604192314704993e-06,
      "loss": 0.2469,
      "step": 35715
    },
    {
      "epoch": 0.644956268642703,
      "grad_norm": 0.3973727524280548,
      "learning_rate": 5.601644969602899e-06,
      "loss": 0.2499,
      "step": 35720
    },
    {
      "epoch": 0.6450465480756036,
      "grad_norm": 0.29262813925743103,
      "learning_rate": 5.599097978316487e-06,
      "loss": 0.1604,
      "step": 35725
    },
    {
      "epoch": 0.6451368275085043,
      "grad_norm": 0.32524144649505615,
      "learning_rate": 5.596551341050648e-06,
      "loss": 0.207,
      "step": 35730
    },
    {
      "epoch": 0.6452271069414051,
      "grad_norm": 0.5345089435577393,
      "learning_rate": 5.594005058010237e-06,
      "loss": 0.2494,
      "step": 35735
    },
    {
      "epoch": 0.6453173863743058,
      "grad_norm": 0.4536992013454437,
      "learning_rate": 5.591459129400081e-06,
      "loss": 0.207,
      "step": 35740
    },
    {
      "epoch": 0.6454076658072064,
      "grad_norm": 0.3381335735321045,
      "learning_rate": 5.588913555424984e-06,
      "loss": 0.1777,
      "step": 35745
    },
    {
      "epoch": 0.6454979452401072,
      "grad_norm": 0.40819457173347473,
      "learning_rate": 5.58636833628972e-06,
      "loss": 0.2212,
      "step": 35750
    },
    {
      "epoch": 0.6455882246730079,
      "grad_norm": 0.2871498167514801,
      "learning_rate": 5.583823472199037e-06,
      "loss": 0.1725,
      "step": 35755
    },
    {
      "epoch": 0.6456785041059087,
      "grad_norm": 0.32077890634536743,
      "learning_rate": 5.581278963357644e-06,
      "loss": 0.1816,
      "step": 35760
    },
    {
      "epoch": 0.6457687835388093,
      "grad_norm": 0.40499499440193176,
      "learning_rate": 5.578734809970234e-06,
      "loss": 0.3163,
      "step": 35765
    },
    {
      "epoch": 0.64585906297171,
      "grad_norm": 0.7485264539718628,
      "learning_rate": 5.5761910122414595e-06,
      "loss": 0.2407,
      "step": 35770
    },
    {
      "epoch": 0.6459493424046108,
      "grad_norm": 0.5958467125892639,
      "learning_rate": 5.573647570375954e-06,
      "loss": 0.2845,
      "step": 35775
    },
    {
      "epoch": 0.6460396218375115,
      "grad_norm": 0.40125805139541626,
      "learning_rate": 5.571104484578318e-06,
      "loss": 0.1859,
      "step": 35780
    },
    {
      "epoch": 0.6461299012704121,
      "grad_norm": 0.4314284026622772,
      "learning_rate": 5.56856175505313e-06,
      "loss": 0.3109,
      "step": 35785
    },
    {
      "epoch": 0.6462201807033129,
      "grad_norm": 0.4031558334827423,
      "learning_rate": 5.566019382004925e-06,
      "loss": 0.2476,
      "step": 35790
    },
    {
      "epoch": 0.6463104601362136,
      "grad_norm": 0.6834647059440613,
      "learning_rate": 5.563477365638226e-06,
      "loss": 0.2054,
      "step": 35795
    },
    {
      "epoch": 0.6464007395691144,
      "grad_norm": 0.2620033621788025,
      "learning_rate": 5.560935706157512e-06,
      "loss": 0.2788,
      "step": 35800
    },
    {
      "epoch": 0.646491019002015,
      "grad_norm": 0.7414353489875793,
      "learning_rate": 5.55839440376725e-06,
      "loss": 0.3292,
      "step": 35805
    },
    {
      "epoch": 0.6465812984349157,
      "grad_norm": 0.2887881398200989,
      "learning_rate": 5.555853458671856e-06,
      "loss": 0.2007,
      "step": 35810
    },
    {
      "epoch": 0.6466715778678165,
      "grad_norm": 0.6315146684646606,
      "learning_rate": 5.553312871075748e-06,
      "loss": 0.318,
      "step": 35815
    },
    {
      "epoch": 0.6467618573007172,
      "grad_norm": 0.4119872450828552,
      "learning_rate": 5.550772641183284e-06,
      "loss": 0.2603,
      "step": 35820
    },
    {
      "epoch": 0.6468521367336179,
      "grad_norm": 0.4186643064022064,
      "learning_rate": 5.548232769198818e-06,
      "loss": 0.209,
      "step": 35825
    },
    {
      "epoch": 0.6469424161665186,
      "grad_norm": 0.44736814498901367,
      "learning_rate": 5.5456932553266565e-06,
      "loss": 0.3051,
      "step": 35830
    },
    {
      "epoch": 0.6470326955994193,
      "grad_norm": 0.4290599822998047,
      "learning_rate": 5.543154099771082e-06,
      "loss": 0.1853,
      "step": 35835
    },
    {
      "epoch": 0.6471229750323201,
      "grad_norm": 0.40984511375427246,
      "learning_rate": 5.540615302736358e-06,
      "loss": 0.2482,
      "step": 35840
    },
    {
      "epoch": 0.6472132544652207,
      "grad_norm": 0.4099291265010834,
      "learning_rate": 5.538076864426708e-06,
      "loss": 0.2854,
      "step": 35845
    },
    {
      "epoch": 0.6473035338981215,
      "grad_norm": 0.34876224398612976,
      "learning_rate": 5.535538785046338e-06,
      "loss": 0.2069,
      "step": 35850
    },
    {
      "epoch": 0.6473938133310222,
      "grad_norm": 0.20505110919475555,
      "learning_rate": 5.533001064799408e-06,
      "loss": 0.2722,
      "step": 35855
    },
    {
      "epoch": 0.6474840927639229,
      "grad_norm": 0.4657360911369324,
      "learning_rate": 5.53046370389007e-06,
      "loss": 0.215,
      "step": 35860
    },
    {
      "epoch": 0.6475743721968236,
      "grad_norm": 0.4829479157924652,
      "learning_rate": 5.527926702522425e-06,
      "loss": 0.2516,
      "step": 35865
    },
    {
      "epoch": 0.6476646516297243,
      "grad_norm": 0.5626475214958191,
      "learning_rate": 5.525390060900566e-06,
      "loss": 0.2351,
      "step": 35870
    },
    {
      "epoch": 0.647754931062625,
      "grad_norm": 0.4846144914627075,
      "learning_rate": 5.522853779228534e-06,
      "loss": 0.3207,
      "step": 35875
    },
    {
      "epoch": 0.6478452104955258,
      "grad_norm": 0.6386345624923706,
      "learning_rate": 5.520317857710374e-06,
      "loss": 0.2153,
      "step": 35880
    },
    {
      "epoch": 0.6479354899284264,
      "grad_norm": 0.36993545293807983,
      "learning_rate": 5.517782296550065e-06,
      "loss": 0.1596,
      "step": 35885
    },
    {
      "epoch": 0.6480257693613272,
      "grad_norm": 0.34195250272750854,
      "learning_rate": 5.515247095951587e-06,
      "loss": 0.2695,
      "step": 35890
    },
    {
      "epoch": 0.6481160487942279,
      "grad_norm": 0.3296082317829132,
      "learning_rate": 5.5127122561188685e-06,
      "loss": 0.1724,
      "step": 35895
    },
    {
      "epoch": 0.6482063282271286,
      "grad_norm": 0.5959557890892029,
      "learning_rate": 5.510177777255827e-06,
      "loss": 0.2495,
      "step": 35900
    },
    {
      "epoch": 0.6482966076600293,
      "grad_norm": 0.47018685936927795,
      "learning_rate": 5.507643659566335e-06,
      "loss": 0.1836,
      "step": 35905
    },
    {
      "epoch": 0.64838688709293,
      "grad_norm": 0.61186283826828,
      "learning_rate": 5.505109903254249e-06,
      "loss": 0.1373,
      "step": 35910
    },
    {
      "epoch": 0.6484771665258308,
      "grad_norm": 0.5220193266868591,
      "learning_rate": 5.50257650852339e-06,
      "loss": 0.2653,
      "step": 35915
    },
    {
      "epoch": 0.6485674459587315,
      "grad_norm": 0.5862143635749817,
      "learning_rate": 5.500043475577555e-06,
      "loss": 0.2105,
      "step": 35920
    },
    {
      "epoch": 0.6486577253916321,
      "grad_norm": 0.48991987109184265,
      "learning_rate": 5.497510804620505e-06,
      "loss": 0.1884,
      "step": 35925
    },
    {
      "epoch": 0.6487480048245329,
      "grad_norm": 1.1300134658813477,
      "learning_rate": 5.4949784958559695e-06,
      "loss": 0.3095,
      "step": 35930
    },
    {
      "epoch": 0.6488382842574336,
      "grad_norm": 0.5286988615989685,
      "learning_rate": 5.492446549487664e-06,
      "loss": 0.2464,
      "step": 35935
    },
    {
      "epoch": 0.6489285636903344,
      "grad_norm": 0.49857473373413086,
      "learning_rate": 5.489914965719253e-06,
      "loss": 0.3035,
      "step": 35940
    },
    {
      "epoch": 0.649018843123235,
      "grad_norm": 0.523838996887207,
      "learning_rate": 5.4873837447543985e-06,
      "loss": 0.253,
      "step": 35945
    },
    {
      "epoch": 0.6491091225561357,
      "grad_norm": 0.28994113206863403,
      "learning_rate": 5.484852886796709e-06,
      "loss": 0.2581,
      "step": 35950
    },
    {
      "epoch": 0.6491994019890365,
      "grad_norm": 0.33836638927459717,
      "learning_rate": 5.482322392049779e-06,
      "loss": 0.2047,
      "step": 35955
    },
    {
      "epoch": 0.6492896814219372,
      "grad_norm": 0.3340393006801605,
      "learning_rate": 5.479792260717162e-06,
      "loss": 0.1438,
      "step": 35960
    },
    {
      "epoch": 0.6493799608548378,
      "grad_norm": 0.3953821361064911,
      "learning_rate": 5.477262493002399e-06,
      "loss": 0.3415,
      "step": 35965
    },
    {
      "epoch": 0.6494702402877386,
      "grad_norm": 0.4580709934234619,
      "learning_rate": 5.474733089108979e-06,
      "loss": 0.2731,
      "step": 35970
    },
    {
      "epoch": 0.6495605197206393,
      "grad_norm": 0.3954589366912842,
      "learning_rate": 5.472204049240381e-06,
      "loss": 0.4242,
      "step": 35975
    },
    {
      "epoch": 0.6496507991535401,
      "grad_norm": 0.43547070026397705,
      "learning_rate": 5.469675373600046e-06,
      "loss": 0.2182,
      "step": 35980
    },
    {
      "epoch": 0.6497410785864407,
      "grad_norm": 0.6323509812355042,
      "learning_rate": 5.467147062391392e-06,
      "loss": 0.2219,
      "step": 35985
    },
    {
      "epoch": 0.6498313580193414,
      "grad_norm": 0.5227145552635193,
      "learning_rate": 5.464619115817798e-06,
      "loss": 0.1728,
      "step": 35990
    },
    {
      "epoch": 0.6499216374522422,
      "grad_norm": 0.45743557810783386,
      "learning_rate": 5.462091534082623e-06,
      "loss": 0.2532,
      "step": 35995
    },
    {
      "epoch": 0.6500119168851429,
      "grad_norm": 0.28376418352127075,
      "learning_rate": 5.4595643173891865e-06,
      "loss": 0.1404,
      "step": 36000
    },
    {
      "epoch": 0.6501021963180436,
      "grad_norm": 0.39995232224464417,
      "learning_rate": 5.457037465940789e-06,
      "loss": 0.2931,
      "step": 36005
    },
    {
      "epoch": 0.6501924757509443,
      "grad_norm": 0.36790263652801514,
      "learning_rate": 5.4545109799407006e-06,
      "loss": 0.1568,
      "step": 36010
    },
    {
      "epoch": 0.650282755183845,
      "grad_norm": 1.251679539680481,
      "learning_rate": 5.45198485959215e-06,
      "loss": 0.2481,
      "step": 36015
    },
    {
      "epoch": 0.6503730346167458,
      "grad_norm": 0.5150558352470398,
      "learning_rate": 5.449459105098356e-06,
      "loss": 0.2474,
      "step": 36020
    },
    {
      "epoch": 0.6504633140496464,
      "grad_norm": 0.3952091932296753,
      "learning_rate": 5.446933716662486e-06,
      "loss": 0.2591,
      "step": 36025
    },
    {
      "epoch": 0.6505535934825472,
      "grad_norm": 1.0100435018539429,
      "learning_rate": 5.4444086944877e-06,
      "loss": 0.2566,
      "step": 36030
    },
    {
      "epoch": 0.6506438729154479,
      "grad_norm": 0.7629332542419434,
      "learning_rate": 5.4418840387771056e-06,
      "loss": 0.2562,
      "step": 36035
    },
    {
      "epoch": 0.6507341523483486,
      "grad_norm": 0.6848556399345398,
      "learning_rate": 5.439359749733801e-06,
      "loss": 0.2854,
      "step": 36040
    },
    {
      "epoch": 0.6508244317812493,
      "grad_norm": 0.2180611938238144,
      "learning_rate": 5.436835827560847e-06,
      "loss": 0.2099,
      "step": 36045
    },
    {
      "epoch": 0.65091471121415,
      "grad_norm": 0.2651490271091461,
      "learning_rate": 5.434312272461275e-06,
      "loss": 0.2515,
      "step": 36050
    },
    {
      "epoch": 0.6510049906470508,
      "grad_norm": 0.565922737121582,
      "learning_rate": 5.431789084638082e-06,
      "loss": 0.265,
      "step": 36055
    },
    {
      "epoch": 0.6510952700799515,
      "grad_norm": 0.4796663224697113,
      "learning_rate": 5.429266264294246e-06,
      "loss": 0.1974,
      "step": 36060
    },
    {
      "epoch": 0.6511855495128522,
      "grad_norm": 0.6576713919639587,
      "learning_rate": 5.426743811632705e-06,
      "loss": 0.2221,
      "step": 36065
    },
    {
      "epoch": 0.6512758289457529,
      "grad_norm": 0.4697735011577606,
      "learning_rate": 5.424221726856371e-06,
      "loss": 0.2401,
      "step": 36070
    },
    {
      "epoch": 0.6513661083786536,
      "grad_norm": 0.441311776638031,
      "learning_rate": 5.421700010168132e-06,
      "loss": 0.1525,
      "step": 36075
    },
    {
      "epoch": 0.6514563878115543,
      "grad_norm": 0.5874056816101074,
      "learning_rate": 5.419178661770841e-06,
      "loss": 0.2494,
      "step": 36080
    },
    {
      "epoch": 0.6515466672444551,
      "grad_norm": 0.3328922390937805,
      "learning_rate": 5.416657681867319e-06,
      "loss": 0.1765,
      "step": 36085
    },
    {
      "epoch": 0.6516369466773557,
      "grad_norm": 0.33248043060302734,
      "learning_rate": 5.414137070660366e-06,
      "loss": 0.2048,
      "step": 36090
    },
    {
      "epoch": 0.6517272261102565,
      "grad_norm": 0.48412153124809265,
      "learning_rate": 5.411616828352739e-06,
      "loss": 0.327,
      "step": 36095
    },
    {
      "epoch": 0.6518175055431572,
      "grad_norm": 0.5995043516159058,
      "learning_rate": 5.409096955147182e-06,
      "loss": 0.1911,
      "step": 36100
    },
    {
      "epoch": 0.651907784976058,
      "grad_norm": 0.5780115723609924,
      "learning_rate": 5.4065774512463905e-06,
      "loss": 0.2279,
      "step": 36105
    },
    {
      "epoch": 0.6519980644089586,
      "grad_norm": 0.19956594705581665,
      "learning_rate": 5.404058316853047e-06,
      "loss": 0.2053,
      "step": 36110
    },
    {
      "epoch": 0.6520883438418593,
      "grad_norm": 0.8025631904602051,
      "learning_rate": 5.4015395521698e-06,
      "loss": 0.3269,
      "step": 36115
    },
    {
      "epoch": 0.6521786232747601,
      "grad_norm": 0.6508951783180237,
      "learning_rate": 5.399021157399258e-06,
      "loss": 0.2592,
      "step": 36120
    },
    {
      "epoch": 0.6522689027076608,
      "grad_norm": 0.4481887221336365,
      "learning_rate": 5.396503132744016e-06,
      "loss": 0.194,
      "step": 36125
    },
    {
      "epoch": 0.6523591821405614,
      "grad_norm": 0.5886951684951782,
      "learning_rate": 5.39398547840662e-06,
      "loss": 0.2577,
      "step": 36130
    },
    {
      "epoch": 0.6524494615734622,
      "grad_norm": 0.43413764238357544,
      "learning_rate": 5.391468194589608e-06,
      "loss": 0.2112,
      "step": 36135
    },
    {
      "epoch": 0.6525397410063629,
      "grad_norm": 0.5776399970054626,
      "learning_rate": 5.388951281495467e-06,
      "loss": 0.2334,
      "step": 36140
    },
    {
      "epoch": 0.6526300204392637,
      "grad_norm": 0.3230409026145935,
      "learning_rate": 5.386434739326676e-06,
      "loss": 0.188,
      "step": 36145
    },
    {
      "epoch": 0.6527202998721643,
      "grad_norm": 0.3495572507381439,
      "learning_rate": 5.383918568285662e-06,
      "loss": 0.2011,
      "step": 36150
    },
    {
      "epoch": 0.652810579305065,
      "grad_norm": 0.3622836768627167,
      "learning_rate": 5.381402768574841e-06,
      "loss": 0.1843,
      "step": 36155
    },
    {
      "epoch": 0.6529008587379658,
      "grad_norm": 0.4878521263599396,
      "learning_rate": 5.378887340396584e-06,
      "loss": 0.1661,
      "step": 36160
    },
    {
      "epoch": 0.6529911381708665,
      "grad_norm": 0.45289501547813416,
      "learning_rate": 5.376372283953245e-06,
      "loss": 0.2874,
      "step": 36165
    },
    {
      "epoch": 0.6530814176037671,
      "grad_norm": 0.29209086298942566,
      "learning_rate": 5.373857599447135e-06,
      "loss": 0.2054,
      "step": 36170
    },
    {
      "epoch": 0.6531716970366679,
      "grad_norm": 0.6697922945022583,
      "learning_rate": 5.371343287080546e-06,
      "loss": 0.217,
      "step": 36175
    },
    {
      "epoch": 0.6532619764695686,
      "grad_norm": 0.6784439086914062,
      "learning_rate": 5.368829347055735e-06,
      "loss": 0.1929,
      "step": 36180
    },
    {
      "epoch": 0.6533522559024694,
      "grad_norm": 0.7056629657745361,
      "learning_rate": 5.366315779574936e-06,
      "loss": 0.2297,
      "step": 36185
    },
    {
      "epoch": 0.65344253533537,
      "grad_norm": 0.34243327379226685,
      "learning_rate": 5.363802584840338e-06,
      "loss": 0.1655,
      "step": 36190
    },
    {
      "epoch": 0.6535328147682707,
      "grad_norm": 0.4730566740036011,
      "learning_rate": 5.361289763054119e-06,
      "loss": 0.2395,
      "step": 36195
    },
    {
      "epoch": 0.6536230942011715,
      "grad_norm": 0.34965676069259644,
      "learning_rate": 5.35877731441841e-06,
      "loss": 0.2957,
      "step": 36200
    },
    {
      "epoch": 0.6537133736340722,
      "grad_norm": 0.5952594876289368,
      "learning_rate": 5.356265239135313e-06,
      "loss": 0.2028,
      "step": 36205
    },
    {
      "epoch": 0.6538036530669729,
      "grad_norm": 0.38088926672935486,
      "learning_rate": 5.353753537406923e-06,
      "loss": 0.3096,
      "step": 36210
    },
    {
      "epoch": 0.6538939324998736,
      "grad_norm": 0.32168248295783997,
      "learning_rate": 5.3512422094352765e-06,
      "loss": 0.2515,
      "step": 36215
    },
    {
      "epoch": 0.6539842119327743,
      "grad_norm": 0.6534793376922607,
      "learning_rate": 5.348731255422397e-06,
      "loss": 0.2103,
      "step": 36220
    },
    {
      "epoch": 0.6540744913656751,
      "grad_norm": 0.2537820041179657,
      "learning_rate": 5.346220675570266e-06,
      "loss": 0.1419,
      "step": 36225
    },
    {
      "epoch": 0.6541647707985757,
      "grad_norm": 0.38548409938812256,
      "learning_rate": 5.34371047008085e-06,
      "loss": 0.2454,
      "step": 36230
    },
    {
      "epoch": 0.6542550502314765,
      "grad_norm": 0.42382970452308655,
      "learning_rate": 5.341200639156068e-06,
      "loss": 0.2954,
      "step": 36235
    },
    {
      "epoch": 0.6543453296643772,
      "grad_norm": 0.5312998294830322,
      "learning_rate": 5.338691182997822e-06,
      "loss": 0.1951,
      "step": 36240
    },
    {
      "epoch": 0.6544356090972779,
      "grad_norm": 0.6543774008750916,
      "learning_rate": 5.3361821018079805e-06,
      "loss": 0.3208,
      "step": 36245
    },
    {
      "epoch": 0.6545258885301786,
      "grad_norm": 0.6860877275466919,
      "learning_rate": 5.3336733957883814e-06,
      "loss": 0.2424,
      "step": 36250
    },
    {
      "epoch": 0.6546161679630793,
      "grad_norm": 0.4910118579864502,
      "learning_rate": 5.331165065140827e-06,
      "loss": 0.2598,
      "step": 36255
    },
    {
      "epoch": 0.65470644739598,
      "grad_norm": 0.5055651664733887,
      "learning_rate": 5.328657110067103e-06,
      "loss": 0.2381,
      "step": 36260
    },
    {
      "epoch": 0.6547967268288808,
      "grad_norm": 0.4690207540988922,
      "learning_rate": 5.3261495307689446e-06,
      "loss": 0.2088,
      "step": 36265
    },
    {
      "epoch": 0.6548870062617814,
      "grad_norm": 0.570319652557373,
      "learning_rate": 5.323642327448075e-06,
      "loss": 0.2584,
      "step": 36270
    },
    {
      "epoch": 0.6549772856946822,
      "grad_norm": 0.6898211240768433,
      "learning_rate": 5.321135500306179e-06,
      "loss": 0.168,
      "step": 36275
    },
    {
      "epoch": 0.6550675651275829,
      "grad_norm": 0.43217116594314575,
      "learning_rate": 5.318629049544919e-06,
      "loss": 0.2906,
      "step": 36280
    },
    {
      "epoch": 0.6551578445604836,
      "grad_norm": 0.37369659543037415,
      "learning_rate": 5.316122975365914e-06,
      "loss": 0.2235,
      "step": 36285
    },
    {
      "epoch": 0.6552481239933843,
      "grad_norm": 0.16432589292526245,
      "learning_rate": 5.313617277970756e-06,
      "loss": 0.1311,
      "step": 36290
    },
    {
      "epoch": 0.655338403426285,
      "grad_norm": 0.7773046493530273,
      "learning_rate": 5.3111119575610195e-06,
      "loss": 0.1963,
      "step": 36295
    },
    {
      "epoch": 0.6554286828591858,
      "grad_norm": 0.4540032744407654,
      "learning_rate": 5.308607014338229e-06,
      "loss": 0.2944,
      "step": 36300
    },
    {
      "epoch": 0.6555189622920865,
      "grad_norm": 0.4507315456867218,
      "learning_rate": 5.306102448503897e-06,
      "loss": 0.3008,
      "step": 36305
    },
    {
      "epoch": 0.6556092417249871,
      "grad_norm": 0.35809871554374695,
      "learning_rate": 5.303598260259494e-06,
      "loss": 0.2297,
      "step": 36310
    },
    {
      "epoch": 0.6556995211578879,
      "grad_norm": 0.48642030358314514,
      "learning_rate": 5.301094449806469e-06,
      "loss": 0.1915,
      "step": 36315
    },
    {
      "epoch": 0.6557898005907886,
      "grad_norm": 0.7581939101219177,
      "learning_rate": 5.298591017346226e-06,
      "loss": 0.3413,
      "step": 36320
    },
    {
      "epoch": 0.6558800800236894,
      "grad_norm": 0.5003683567047119,
      "learning_rate": 5.296087963080159e-06,
      "loss": 0.2669,
      "step": 36325
    },
    {
      "epoch": 0.65597035945659,
      "grad_norm": 0.3249092698097229,
      "learning_rate": 5.293585287209609e-06,
      "loss": 0.2155,
      "step": 36330
    },
    {
      "epoch": 0.6560606388894907,
      "grad_norm": 0.5210465788841248,
      "learning_rate": 5.291082989935904e-06,
      "loss": 0.1858,
      "step": 36335
    },
    {
      "epoch": 0.6561509183223915,
      "grad_norm": 0.3324693739414215,
      "learning_rate": 5.288581071460335e-06,
      "loss": 0.2807,
      "step": 36340
    },
    {
      "epoch": 0.6562411977552922,
      "grad_norm": 0.44445836544036865,
      "learning_rate": 5.28607953198417e-06,
      "loss": 0.1725,
      "step": 36345
    },
    {
      "epoch": 0.6563314771881928,
      "grad_norm": 0.5412079691886902,
      "learning_rate": 5.283578371708626e-06,
      "loss": 0.2055,
      "step": 36350
    },
    {
      "epoch": 0.6564217566210936,
      "grad_norm": 0.48812752962112427,
      "learning_rate": 5.281077590834916e-06,
      "loss": 0.2257,
      "step": 36355
    },
    {
      "epoch": 0.6565120360539943,
      "grad_norm": 0.31539368629455566,
      "learning_rate": 5.278577189564199e-06,
      "loss": 0.2113,
      "step": 36360
    },
    {
      "epoch": 0.6566023154868951,
      "grad_norm": 0.7528833150863647,
      "learning_rate": 5.276077168097624e-06,
      "loss": 0.29,
      "step": 36365
    },
    {
      "epoch": 0.6566925949197957,
      "grad_norm": 0.3404152989387512,
      "learning_rate": 5.27357752663629e-06,
      "loss": 0.2261,
      "step": 36370
    },
    {
      "epoch": 0.6567828743526964,
      "grad_norm": 0.40433064103126526,
      "learning_rate": 5.27107826538128e-06,
      "loss": 0.2948,
      "step": 36375
    },
    {
      "epoch": 0.6568731537855972,
      "grad_norm": 0.38003936409950256,
      "learning_rate": 5.268579384533644e-06,
      "loss": 0.242,
      "step": 36380
    },
    {
      "epoch": 0.6569634332184979,
      "grad_norm": 0.35246750712394714,
      "learning_rate": 5.266080884294392e-06,
      "loss": 0.2454,
      "step": 36385
    },
    {
      "epoch": 0.6570537126513986,
      "grad_norm": 0.5807691216468811,
      "learning_rate": 5.263582764864519e-06,
      "loss": 0.3307,
      "step": 36390
    },
    {
      "epoch": 0.6571439920842993,
      "grad_norm": 0.402802437543869,
      "learning_rate": 5.26108502644497e-06,
      "loss": 0.2739,
      "step": 36395
    },
    {
      "epoch": 0.6572342715172,
      "grad_norm": 0.5744020342826843,
      "learning_rate": 5.258587669236676e-06,
      "loss": 0.2833,
      "step": 36400
    },
    {
      "epoch": 0.6573245509501008,
      "grad_norm": 0.3158230185508728,
      "learning_rate": 5.256090693440532e-06,
      "loss": 0.1906,
      "step": 36405
    },
    {
      "epoch": 0.6574148303830014,
      "grad_norm": 0.48997870087623596,
      "learning_rate": 5.253594099257402e-06,
      "loss": 0.2397,
      "step": 36410
    },
    {
      "epoch": 0.6575051098159022,
      "grad_norm": 0.3089967966079712,
      "learning_rate": 5.251097886888115e-06,
      "loss": 0.2195,
      "step": 36415
    },
    {
      "epoch": 0.6575953892488029,
      "grad_norm": 0.4702431559562683,
      "learning_rate": 5.248602056533478e-06,
      "loss": 0.2356,
      "step": 36420
    },
    {
      "epoch": 0.6576856686817036,
      "grad_norm": 0.48032915592193604,
      "learning_rate": 5.246106608394255e-06,
      "loss": 0.138,
      "step": 36425
    },
    {
      "epoch": 0.6577759481146043,
      "grad_norm": 0.3023422360420227,
      "learning_rate": 5.243611542671196e-06,
      "loss": 0.225,
      "step": 36430
    },
    {
      "epoch": 0.657866227547505,
      "grad_norm": 0.5493566989898682,
      "learning_rate": 5.241116859565002e-06,
      "loss": 0.1258,
      "step": 36435
    },
    {
      "epoch": 0.6579565069804058,
      "grad_norm": 0.4261058568954468,
      "learning_rate": 5.238622559276356e-06,
      "loss": 0.2396,
      "step": 36440
    },
    {
      "epoch": 0.6580467864133065,
      "grad_norm": 0.473242849111557,
      "learning_rate": 5.2361286420059064e-06,
      "loss": 0.1612,
      "step": 36445
    },
    {
      "epoch": 0.6581370658462071,
      "grad_norm": 0.3436768651008606,
      "learning_rate": 5.2336351079542755e-06,
      "loss": 0.221,
      "step": 36450
    },
    {
      "epoch": 0.6582273452791079,
      "grad_norm": 0.7097442150115967,
      "learning_rate": 5.231141957322041e-06,
      "loss": 0.1859,
      "step": 36455
    },
    {
      "epoch": 0.6583176247120086,
      "grad_norm": 0.5997614860534668,
      "learning_rate": 5.228649190309767e-06,
      "loss": 0.2897,
      "step": 36460
    },
    {
      "epoch": 0.6584079041449094,
      "grad_norm": 0.5260922908782959,
      "learning_rate": 5.226156807117971e-06,
      "loss": 0.2347,
      "step": 36465
    },
    {
      "epoch": 0.6584981835778101,
      "grad_norm": 0.23415659368038177,
      "learning_rate": 5.223664807947151e-06,
      "loss": 0.1616,
      "step": 36470
    },
    {
      "epoch": 0.6585884630107107,
      "grad_norm": 0.567236065864563,
      "learning_rate": 5.221173192997773e-06,
      "loss": 0.2294,
      "step": 36475
    },
    {
      "epoch": 0.6586787424436115,
      "grad_norm": 0.6114963293075562,
      "learning_rate": 5.218681962470263e-06,
      "loss": 0.1269,
      "step": 36480
    },
    {
      "epoch": 0.6587690218765122,
      "grad_norm": 0.7257540822029114,
      "learning_rate": 5.216191116565028e-06,
      "loss": 0.2045,
      "step": 36485
    },
    {
      "epoch": 0.658859301309413,
      "grad_norm": 0.5209638476371765,
      "learning_rate": 5.2137006554824345e-06,
      "loss": 0.26,
      "step": 36490
    },
    {
      "epoch": 0.6589495807423136,
      "grad_norm": 0.8320066332817078,
      "learning_rate": 5.211210579422825e-06,
      "loss": 0.2026,
      "step": 36495
    },
    {
      "epoch": 0.6590398601752143,
      "grad_norm": 0.4539569318294525,
      "learning_rate": 5.2087208885865036e-06,
      "loss": 0.3083,
      "step": 36500
    },
    {
      "epoch": 0.6591301396081151,
      "grad_norm": 0.22275879979133606,
      "learning_rate": 5.206231583173751e-06,
      "loss": 0.1345,
      "step": 36505
    },
    {
      "epoch": 0.6592204190410158,
      "grad_norm": 0.5147002339363098,
      "learning_rate": 5.203742663384813e-06,
      "loss": 0.1485,
      "step": 36510
    },
    {
      "epoch": 0.6593106984739164,
      "grad_norm": 0.507576048374176,
      "learning_rate": 5.20125412941991e-06,
      "loss": 0.1761,
      "step": 36515
    },
    {
      "epoch": 0.6594009779068172,
      "grad_norm": 0.40669941902160645,
      "learning_rate": 5.198765981479217e-06,
      "loss": 0.2278,
      "step": 36520
    },
    {
      "epoch": 0.6594912573397179,
      "grad_norm": 0.33739709854125977,
      "learning_rate": 5.1962782197628984e-06,
      "loss": 0.1623,
      "step": 36525
    },
    {
      "epoch": 0.6595815367726187,
      "grad_norm": 0.33824312686920166,
      "learning_rate": 5.193790844471065e-06,
      "loss": 0.3332,
      "step": 36530
    },
    {
      "epoch": 0.6596718162055193,
      "grad_norm": 0.5075321197509766,
      "learning_rate": 5.191303855803815e-06,
      "loss": 0.1877,
      "step": 36535
    },
    {
      "epoch": 0.65976209563842,
      "grad_norm": 0.3887553811073303,
      "learning_rate": 5.1888172539612045e-06,
      "loss": 0.2291,
      "step": 36540
    },
    {
      "epoch": 0.6598523750713208,
      "grad_norm": 0.35546180605888367,
      "learning_rate": 5.186331039143272e-06,
      "loss": 0.1708,
      "step": 36545
    },
    {
      "epoch": 0.6599426545042215,
      "grad_norm": 0.3968889117240906,
      "learning_rate": 5.183845211550004e-06,
      "loss": 0.3045,
      "step": 36550
    },
    {
      "epoch": 0.6600329339371221,
      "grad_norm": 0.2924818694591522,
      "learning_rate": 5.181359771381376e-06,
      "loss": 0.1616,
      "step": 36555
    },
    {
      "epoch": 0.6601232133700229,
      "grad_norm": 0.45263779163360596,
      "learning_rate": 5.178874718837318e-06,
      "loss": 0.2787,
      "step": 36560
    },
    {
      "epoch": 0.6602134928029236,
      "grad_norm": 0.4006568491458893,
      "learning_rate": 5.176390054117732e-06,
      "loss": 0.214,
      "step": 36565
    },
    {
      "epoch": 0.6603037722358244,
      "grad_norm": 0.30012068152427673,
      "learning_rate": 5.1739057774224965e-06,
      "loss": 0.1781,
      "step": 36570
    },
    {
      "epoch": 0.660394051668725,
      "grad_norm": 0.5558052659034729,
      "learning_rate": 5.1714218889514526e-06,
      "loss": 0.156,
      "step": 36575
    },
    {
      "epoch": 0.6604843311016257,
      "grad_norm": 0.4274640381336212,
      "learning_rate": 5.168938388904413e-06,
      "loss": 0.1808,
      "step": 36580
    },
    {
      "epoch": 0.6605746105345265,
      "grad_norm": 0.5439524054527283,
      "learning_rate": 5.166455277481153e-06,
      "loss": 0.3033,
      "step": 36585
    },
    {
      "epoch": 0.6606648899674272,
      "grad_norm": 0.4050881862640381,
      "learning_rate": 5.163972554881424e-06,
      "loss": 0.1802,
      "step": 36590
    },
    {
      "epoch": 0.6607551694003279,
      "grad_norm": 0.4661766588687897,
      "learning_rate": 5.161490221304939e-06,
      "loss": 0.1279,
      "step": 36595
    },
    {
      "epoch": 0.6608454488332286,
      "grad_norm": 0.5030481815338135,
      "learning_rate": 5.159008276951386e-06,
      "loss": 0.1636,
      "step": 36600
    },
    {
      "epoch": 0.6609357282661293,
      "grad_norm": 0.22832004725933075,
      "learning_rate": 5.156526722020419e-06,
      "loss": 0.2206,
      "step": 36605
    },
    {
      "epoch": 0.6610260076990301,
      "grad_norm": 0.3879418969154358,
      "learning_rate": 5.154045556711667e-06,
      "loss": 0.2062,
      "step": 36610
    },
    {
      "epoch": 0.6611162871319307,
      "grad_norm": 0.6533599495887756,
      "learning_rate": 5.151564781224712e-06,
      "loss": 0.2051,
      "step": 36615
    },
    {
      "epoch": 0.6612065665648315,
      "grad_norm": 0.6125730872154236,
      "learning_rate": 5.149084395759123e-06,
      "loss": 0.1912,
      "step": 36620
    },
    {
      "epoch": 0.6612968459977322,
      "grad_norm": 0.308635413646698,
      "learning_rate": 5.146604400514421e-06,
      "loss": 0.2386,
      "step": 36625
    },
    {
      "epoch": 0.6613871254306329,
      "grad_norm": 0.36153319478034973,
      "learning_rate": 5.144124795690111e-06,
      "loss": 0.2323,
      "step": 36630
    },
    {
      "epoch": 0.6614774048635336,
      "grad_norm": 0.3307132124900818,
      "learning_rate": 5.1416455814856515e-06,
      "loss": 0.1951,
      "step": 36635
    },
    {
      "epoch": 0.6615676842964343,
      "grad_norm": 0.4885282814502716,
      "learning_rate": 5.139166758100481e-06,
      "loss": 0.1321,
      "step": 36640
    },
    {
      "epoch": 0.661657963729335,
      "grad_norm": 0.5977446436882019,
      "learning_rate": 5.136688325734004e-06,
      "loss": 0.2655,
      "step": 36645
    },
    {
      "epoch": 0.6617482431622358,
      "grad_norm": 0.3410941958427429,
      "learning_rate": 5.134210284585595e-06,
      "loss": 0.1885,
      "step": 36650
    },
    {
      "epoch": 0.6618385225951364,
      "grad_norm": 0.4518417418003082,
      "learning_rate": 5.131732634854592e-06,
      "loss": 0.2936,
      "step": 36655
    },
    {
      "epoch": 0.6619288020280372,
      "grad_norm": 0.4261515736579895,
      "learning_rate": 5.129255376740297e-06,
      "loss": 0.1962,
      "step": 36660
    },
    {
      "epoch": 0.6620190814609379,
      "grad_norm": 0.5102550387382507,
      "learning_rate": 5.1267785104419945e-06,
      "loss": 0.1593,
      "step": 36665
    },
    {
      "epoch": 0.6621093608938386,
      "grad_norm": 0.46885019540786743,
      "learning_rate": 5.124302036158928e-06,
      "loss": 0.3083,
      "step": 36670
    },
    {
      "epoch": 0.6621996403267393,
      "grad_norm": 0.5711917877197266,
      "learning_rate": 5.121825954090317e-06,
      "loss": 0.2011,
      "step": 36675
    },
    {
      "epoch": 0.66228991975964,
      "grad_norm": 0.38981893658638,
      "learning_rate": 5.1193502644353365e-06,
      "loss": 0.187,
      "step": 36680
    },
    {
      "epoch": 0.6623801991925408,
      "grad_norm": 0.16882199048995972,
      "learning_rate": 5.1168749673931465e-06,
      "loss": 0.1494,
      "step": 36685
    },
    {
      "epoch": 0.6624704786254415,
      "grad_norm": 0.4587150514125824,
      "learning_rate": 5.114400063162857e-06,
      "loss": 0.2764,
      "step": 36690
    },
    {
      "epoch": 0.6625607580583421,
      "grad_norm": 0.39484915137290955,
      "learning_rate": 5.111925551943564e-06,
      "loss": 0.1908,
      "step": 36695
    },
    {
      "epoch": 0.6626510374912429,
      "grad_norm": 0.47309184074401855,
      "learning_rate": 5.1094514339343174e-06,
      "loss": 0.2775,
      "step": 36700
    },
    {
      "epoch": 0.6627413169241436,
      "grad_norm": 0.4850340187549591,
      "learning_rate": 5.106977709334144e-06,
      "loss": 0.1514,
      "step": 36705
    },
    {
      "epoch": 0.6628315963570444,
      "grad_norm": 0.3185379207134247,
      "learning_rate": 5.104504378342039e-06,
      "loss": 0.2157,
      "step": 36710
    },
    {
      "epoch": 0.662921875789945,
      "grad_norm": 0.4683305025100708,
      "learning_rate": 5.1020314411569675e-06,
      "loss": 0.2074,
      "step": 36715
    },
    {
      "epoch": 0.6630121552228457,
      "grad_norm": 0.4003462493419647,
      "learning_rate": 5.099558897977851e-06,
      "loss": 0.262,
      "step": 36720
    },
    {
      "epoch": 0.6631024346557465,
      "grad_norm": 0.49980586767196655,
      "learning_rate": 5.097086749003595e-06,
      "loss": 0.3328,
      "step": 36725
    },
    {
      "epoch": 0.6631927140886472,
      "grad_norm": 0.48928263783454895,
      "learning_rate": 5.094614994433058e-06,
      "loss": 0.1817,
      "step": 36730
    },
    {
      "epoch": 0.6632829935215478,
      "grad_norm": 0.25729429721832275,
      "learning_rate": 5.09214363446508e-06,
      "loss": 0.252,
      "step": 36735
    },
    {
      "epoch": 0.6633732729544486,
      "grad_norm": 0.49312248826026917,
      "learning_rate": 5.089672669298462e-06,
      "loss": 0.1915,
      "step": 36740
    },
    {
      "epoch": 0.6634635523873493,
      "grad_norm": 0.49311479926109314,
      "learning_rate": 5.0872020991319816e-06,
      "loss": 0.2368,
      "step": 36745
    },
    {
      "epoch": 0.6635538318202501,
      "grad_norm": 0.4821045398712158,
      "learning_rate": 5.084731924164373e-06,
      "loss": 0.2603,
      "step": 36750
    },
    {
      "epoch": 0.6636441112531507,
      "grad_norm": 0.47183868288993835,
      "learning_rate": 5.082262144594338e-06,
      "loss": 0.1998,
      "step": 36755
    },
    {
      "epoch": 0.6637343906860514,
      "grad_norm": 0.6003633737564087,
      "learning_rate": 5.079792760620562e-06,
      "loss": 0.2824,
      "step": 36760
    },
    {
      "epoch": 0.6638246701189522,
      "grad_norm": 0.5717042088508606,
      "learning_rate": 5.077323772441683e-06,
      "loss": 0.2828,
      "step": 36765
    },
    {
      "epoch": 0.6639149495518529,
      "grad_norm": 0.7113093733787537,
      "learning_rate": 5.074855180256313e-06,
      "loss": 0.246,
      "step": 36770
    },
    {
      "epoch": 0.6640052289847536,
      "grad_norm": 0.4163897931575775,
      "learning_rate": 5.072386984263036e-06,
      "loss": 0.1641,
      "step": 36775
    },
    {
      "epoch": 0.6640955084176543,
      "grad_norm": 0.5821138620376587,
      "learning_rate": 5.069919184660402e-06,
      "loss": 0.2837,
      "step": 36780
    },
    {
      "epoch": 0.664185787850555,
      "grad_norm": 0.2605414092540741,
      "learning_rate": 5.067451781646922e-06,
      "loss": 0.1363,
      "step": 36785
    },
    {
      "epoch": 0.6642760672834558,
      "grad_norm": 0.34480786323547363,
      "learning_rate": 5.064984775421085e-06,
      "loss": 0.2066,
      "step": 36790
    },
    {
      "epoch": 0.6643663467163564,
      "grad_norm": 0.5099565982818604,
      "learning_rate": 5.062518166181338e-06,
      "loss": 0.1796,
      "step": 36795
    },
    {
      "epoch": 0.6644566261492572,
      "grad_norm": 0.3806185722351074,
      "learning_rate": 5.060051954126106e-06,
      "loss": 0.2543,
      "step": 36800
    },
    {
      "epoch": 0.6645469055821579,
      "grad_norm": 0.47018367052078247,
      "learning_rate": 5.057586139453779e-06,
      "loss": 0.2296,
      "step": 36805
    },
    {
      "epoch": 0.6646371850150586,
      "grad_norm": 0.5153882503509521,
      "learning_rate": 5.0551207223627144e-06,
      "loss": 0.2,
      "step": 36810
    },
    {
      "epoch": 0.6647274644479593,
      "grad_norm": 0.8088284134864807,
      "learning_rate": 5.052655703051231e-06,
      "loss": 0.2412,
      "step": 36815
    },
    {
      "epoch": 0.66481774388086,
      "grad_norm": 0.5115001797676086,
      "learning_rate": 5.0501910817176305e-06,
      "loss": 0.2293,
      "step": 36820
    },
    {
      "epoch": 0.6649080233137608,
      "grad_norm": 0.5519483089447021,
      "learning_rate": 5.047726858560165e-06,
      "loss": 0.2607,
      "step": 36825
    },
    {
      "epoch": 0.6649983027466615,
      "grad_norm": 0.4139096736907959,
      "learning_rate": 5.04526303377707e-06,
      "loss": 0.223,
      "step": 36830
    },
    {
      "epoch": 0.6650885821795621,
      "grad_norm": 0.2258121818304062,
      "learning_rate": 5.0427996075665365e-06,
      "loss": 0.152,
      "step": 36835
    },
    {
      "epoch": 0.6651788616124629,
      "grad_norm": 0.19573138654232025,
      "learning_rate": 5.040336580126732e-06,
      "loss": 0.1931,
      "step": 36840
    },
    {
      "epoch": 0.6652691410453636,
      "grad_norm": 0.4620254337787628,
      "learning_rate": 5.037873951655794e-06,
      "loss": 0.1999,
      "step": 36845
    },
    {
      "epoch": 0.6653594204782644,
      "grad_norm": 0.7757815718650818,
      "learning_rate": 5.035411722351813e-06,
      "loss": 0.1894,
      "step": 36850
    },
    {
      "epoch": 0.6654496999111651,
      "grad_norm": 0.882802426815033,
      "learning_rate": 5.032949892412868e-06,
      "loss": 0.2849,
      "step": 36855
    },
    {
      "epoch": 0.6655399793440657,
      "grad_norm": 0.3408997356891632,
      "learning_rate": 5.030488462036987e-06,
      "loss": 0.2316,
      "step": 36860
    },
    {
      "epoch": 0.6656302587769665,
      "grad_norm": 0.5360017418861389,
      "learning_rate": 5.028027431422176e-06,
      "loss": 0.2421,
      "step": 36865
    },
    {
      "epoch": 0.6657205382098672,
      "grad_norm": 0.5048326253890991,
      "learning_rate": 5.025566800766409e-06,
      "loss": 0.3436,
      "step": 36870
    },
    {
      "epoch": 0.665810817642768,
      "grad_norm": 0.3691485822200775,
      "learning_rate": 5.023106570267627e-06,
      "loss": 0.1521,
      "step": 36875
    },
    {
      "epoch": 0.6659010970756686,
      "grad_norm": 0.24679838120937347,
      "learning_rate": 5.020646740123735e-06,
      "loss": 0.3498,
      "step": 36880
    },
    {
      "epoch": 0.6659913765085693,
      "grad_norm": 0.23643231391906738,
      "learning_rate": 5.018187310532611e-06,
      "loss": 0.2223,
      "step": 36885
    },
    {
      "epoch": 0.6660816559414701,
      "grad_norm": 0.399018794298172,
      "learning_rate": 5.015728281692094e-06,
      "loss": 0.2203,
      "step": 36890
    },
    {
      "epoch": 0.6661719353743708,
      "grad_norm": 0.4639451801776886,
      "learning_rate": 5.013269653800001e-06,
      "loss": 0.2392,
      "step": 36895
    },
    {
      "epoch": 0.6662622148072714,
      "grad_norm": 0.4985400140285492,
      "learning_rate": 5.0108114270541055e-06,
      "loss": 0.1535,
      "step": 36900
    },
    {
      "epoch": 0.6663524942401722,
      "grad_norm": 0.44081366062164307,
      "learning_rate": 5.008353601652154e-06,
      "loss": 0.2293,
      "step": 36905
    },
    {
      "epoch": 0.6664427736730729,
      "grad_norm": 0.3351982831954956,
      "learning_rate": 5.005896177791862e-06,
      "loss": 0.2259,
      "step": 36910
    },
    {
      "epoch": 0.6665330531059737,
      "grad_norm": 0.27997368574142456,
      "learning_rate": 5.003439155670916e-06,
      "loss": 0.1895,
      "step": 36915
    },
    {
      "epoch": 0.6666233325388743,
      "grad_norm": 0.607708752155304,
      "learning_rate": 5.000982535486958e-06,
      "loss": 0.296,
      "step": 36920
    },
    {
      "epoch": 0.666713611971775,
      "grad_norm": 0.30011940002441406,
      "learning_rate": 4.998526317437612e-06,
      "loss": 0.2256,
      "step": 36925
    },
    {
      "epoch": 0.6668038914046758,
      "grad_norm": 0.44001007080078125,
      "learning_rate": 4.996070501720456e-06,
      "loss": 0.2381,
      "step": 36930
    },
    {
      "epoch": 0.6668941708375765,
      "grad_norm": 0.43888503313064575,
      "learning_rate": 4.993615088533046e-06,
      "loss": 0.1804,
      "step": 36935
    },
    {
      "epoch": 0.6669844502704771,
      "grad_norm": 0.4655245542526245,
      "learning_rate": 4.991160078072907e-06,
      "loss": 0.2559,
      "step": 36940
    },
    {
      "epoch": 0.6670747297033779,
      "grad_norm": 1.812050223350525,
      "learning_rate": 4.988705470537516e-06,
      "loss": 0.2734,
      "step": 36945
    },
    {
      "epoch": 0.6671650091362786,
      "grad_norm": 0.44596067070961,
      "learning_rate": 4.9862512661243386e-06,
      "loss": 0.3315,
      "step": 36950
    },
    {
      "epoch": 0.6672552885691794,
      "grad_norm": 0.5072383284568787,
      "learning_rate": 4.9837974650307905e-06,
      "loss": 0.1918,
      "step": 36955
    },
    {
      "epoch": 0.66734556800208,
      "grad_norm": 0.30700528621673584,
      "learning_rate": 4.981344067454269e-06,
      "loss": 0.2694,
      "step": 36960
    },
    {
      "epoch": 0.6674358474349807,
      "grad_norm": 0.4336912930011749,
      "learning_rate": 4.9788910735921235e-06,
      "loss": 0.2399,
      "step": 36965
    },
    {
      "epoch": 0.6675261268678815,
      "grad_norm": 0.4809855818748474,
      "learning_rate": 4.9764384836416845e-06,
      "loss": 0.2984,
      "step": 36970
    },
    {
      "epoch": 0.6676164063007822,
      "grad_norm": 0.597041130065918,
      "learning_rate": 4.973986297800245e-06,
      "loss": 0.2233,
      "step": 36975
    },
    {
      "epoch": 0.6677066857336829,
      "grad_norm": 0.49554866552352905,
      "learning_rate": 4.971534516265069e-06,
      "loss": 0.2486,
      "step": 36980
    },
    {
      "epoch": 0.6677969651665836,
      "grad_norm": 0.4046391248703003,
      "learning_rate": 4.969083139233376e-06,
      "loss": 0.2427,
      "step": 36985
    },
    {
      "epoch": 0.6678872445994843,
      "grad_norm": 0.3640994429588318,
      "learning_rate": 4.966632166902371e-06,
      "loss": 0.2132,
      "step": 36990
    },
    {
      "epoch": 0.6679775240323851,
      "grad_norm": 0.5942362546920776,
      "learning_rate": 4.964181599469208e-06,
      "loss": 0.1826,
      "step": 36995
    },
    {
      "epoch": 0.6680678034652857,
      "grad_norm": 0.45490118861198425,
      "learning_rate": 4.96173143713102e-06,
      "loss": 0.2304,
      "step": 37000
    },
    {
      "epoch": 0.6681580828981865,
      "grad_norm": 0.8721827864646912,
      "learning_rate": 4.959281680084907e-06,
      "loss": 0.1934,
      "step": 37005
    },
    {
      "epoch": 0.6682483623310872,
      "grad_norm": 0.6914052367210388,
      "learning_rate": 4.956832328527939e-06,
      "loss": 0.2042,
      "step": 37010
    },
    {
      "epoch": 0.6683386417639879,
      "grad_norm": 0.6214689612388611,
      "learning_rate": 4.954383382657136e-06,
      "loss": 0.3038,
      "step": 37015
    },
    {
      "epoch": 0.6684289211968886,
      "grad_norm": 0.31522366404533386,
      "learning_rate": 4.95193484266951e-06,
      "loss": 0.2341,
      "step": 37020
    },
    {
      "epoch": 0.6685192006297893,
      "grad_norm": 0.590651273727417,
      "learning_rate": 4.9494867087620225e-06,
      "loss": 0.1615,
      "step": 37025
    },
    {
      "epoch": 0.66860948006269,
      "grad_norm": 0.20999670028686523,
      "learning_rate": 4.947038981131605e-06,
      "loss": 0.1288,
      "step": 37030
    },
    {
      "epoch": 0.6686997594955908,
      "grad_norm": 0.4950847923755646,
      "learning_rate": 4.944591659975163e-06,
      "loss": 0.2576,
      "step": 37035
    },
    {
      "epoch": 0.6687900389284914,
      "grad_norm": 0.5933647751808167,
      "learning_rate": 4.942144745489565e-06,
      "loss": 0.263,
      "step": 37040
    },
    {
      "epoch": 0.6688803183613922,
      "grad_norm": 1.0284976959228516,
      "learning_rate": 4.9396982378716526e-06,
      "loss": 0.2176,
      "step": 37045
    },
    {
      "epoch": 0.6689705977942929,
      "grad_norm": 0.3346775770187378,
      "learning_rate": 4.9372521373182214e-06,
      "loss": 0.2331,
      "step": 37050
    },
    {
      "epoch": 0.6690608772271937,
      "grad_norm": 0.4718320071697235,
      "learning_rate": 4.934806444026049e-06,
      "loss": 0.2758,
      "step": 37055
    },
    {
      "epoch": 0.6691511566600943,
      "grad_norm": 0.4274367690086365,
      "learning_rate": 4.932361158191868e-06,
      "loss": 0.1495,
      "step": 37060
    },
    {
      "epoch": 0.669241436092995,
      "grad_norm": 0.4431329369544983,
      "learning_rate": 4.929916280012384e-06,
      "loss": 0.2516,
      "step": 37065
    },
    {
      "epoch": 0.6693317155258958,
      "grad_norm": 0.46292510628700256,
      "learning_rate": 4.927471809684274e-06,
      "loss": 0.3105,
      "step": 37070
    },
    {
      "epoch": 0.6694219949587965,
      "grad_norm": 0.3133583068847656,
      "learning_rate": 4.925027747404181e-06,
      "loss": 0.1601,
      "step": 37075
    },
    {
      "epoch": 0.6695122743916971,
      "grad_norm": 0.4823862910270691,
      "learning_rate": 4.922584093368702e-06,
      "loss": 0.2668,
      "step": 37080
    },
    {
      "epoch": 0.6696025538245979,
      "grad_norm": 0.44316956400871277,
      "learning_rate": 4.9201408477744195e-06,
      "loss": 0.2384,
      "step": 37085
    },
    {
      "epoch": 0.6696928332574986,
      "grad_norm": 0.4984242022037506,
      "learning_rate": 4.9176980108178685e-06,
      "loss": 0.2973,
      "step": 37090
    },
    {
      "epoch": 0.6697831126903994,
      "grad_norm": 0.3840685784816742,
      "learning_rate": 4.915255582695565e-06,
      "loss": 0.2633,
      "step": 37095
    },
    {
      "epoch": 0.6698733921233,
      "grad_norm": 0.6775936484336853,
      "learning_rate": 4.912813563603975e-06,
      "loss": 0.2415,
      "step": 37100
    },
    {
      "epoch": 0.6699636715562007,
      "grad_norm": 0.2997501492500305,
      "learning_rate": 4.910371953739548e-06,
      "loss": 0.1747,
      "step": 37105
    },
    {
      "epoch": 0.6700539509891015,
      "grad_norm": 0.7513289451599121,
      "learning_rate": 4.907930753298691e-06,
      "loss": 0.2323,
      "step": 37110
    },
    {
      "epoch": 0.6701442304220022,
      "grad_norm": 0.5835806131362915,
      "learning_rate": 4.905489962477786e-06,
      "loss": 0.2219,
      "step": 37115
    },
    {
      "epoch": 0.6702345098549028,
      "grad_norm": 0.3198319375514984,
      "learning_rate": 4.903049581473173e-06,
      "loss": 0.1908,
      "step": 37120
    },
    {
      "epoch": 0.6703247892878036,
      "grad_norm": 0.33636313676834106,
      "learning_rate": 4.90060961048116e-06,
      "loss": 0.182,
      "step": 37125
    },
    {
      "epoch": 0.6704150687207043,
      "grad_norm": 0.4486347734928131,
      "learning_rate": 4.8981700496980255e-06,
      "loss": 0.2391,
      "step": 37130
    },
    {
      "epoch": 0.6705053481536051,
      "grad_norm": 0.2821575701236725,
      "learning_rate": 4.895730899320018e-06,
      "loss": 0.203,
      "step": 37135
    },
    {
      "epoch": 0.6705956275865057,
      "grad_norm": 0.2851160764694214,
      "learning_rate": 4.8932921595433516e-06,
      "loss": 0.1512,
      "step": 37140
    },
    {
      "epoch": 0.6706859070194064,
      "grad_norm": 0.6033151149749756,
      "learning_rate": 4.890853830564198e-06,
      "loss": 0.3092,
      "step": 37145
    },
    {
      "epoch": 0.6707761864523072,
      "grad_norm": 0.23315021395683289,
      "learning_rate": 4.8884159125787105e-06,
      "loss": 0.1523,
      "step": 37150
    },
    {
      "epoch": 0.6708664658852079,
      "grad_norm": 0.2469920516014099,
      "learning_rate": 4.885978405782994e-06,
      "loss": 0.2195,
      "step": 37155
    },
    {
      "epoch": 0.6709567453181086,
      "grad_norm": 0.9692341685295105,
      "learning_rate": 4.883541310373136e-06,
      "loss": 0.2526,
      "step": 37160
    },
    {
      "epoch": 0.6710470247510093,
      "grad_norm": 0.5049905776977539,
      "learning_rate": 4.881104626545175e-06,
      "loss": 0.153,
      "step": 37165
    },
    {
      "epoch": 0.67113730418391,
      "grad_norm": 0.428270548582077,
      "learning_rate": 4.8786683544951295e-06,
      "loss": 0.2615,
      "step": 37170
    },
    {
      "epoch": 0.6712275836168108,
      "grad_norm": 0.439714640378952,
      "learning_rate": 4.876232494418978e-06,
      "loss": 0.1528,
      "step": 37175
    },
    {
      "epoch": 0.6713178630497114,
      "grad_norm": 0.42459869384765625,
      "learning_rate": 4.873797046512675e-06,
      "loss": 0.2203,
      "step": 37180
    },
    {
      "epoch": 0.6714081424826122,
      "grad_norm": 0.5259662866592407,
      "learning_rate": 4.871362010972124e-06,
      "loss": 0.2435,
      "step": 37185
    },
    {
      "epoch": 0.6714984219155129,
      "grad_norm": 0.45902788639068604,
      "learning_rate": 4.8689273879932135e-06,
      "loss": 0.2823,
      "step": 37190
    },
    {
      "epoch": 0.6715887013484136,
      "grad_norm": 0.6105425357818604,
      "learning_rate": 4.866493177771784e-06,
      "loss": 0.289,
      "step": 37195
    },
    {
      "epoch": 0.6716789807813143,
      "grad_norm": 0.9166700839996338,
      "learning_rate": 4.864059380503655e-06,
      "loss": 0.2488,
      "step": 37200
    },
    {
      "epoch": 0.671769260214215,
      "grad_norm": 0.5581538677215576,
      "learning_rate": 4.861625996384611e-06,
      "loss": 0.3423,
      "step": 37205
    },
    {
      "epoch": 0.6718595396471158,
      "grad_norm": 0.16273179650306702,
      "learning_rate": 4.859193025610391e-06,
      "loss": 0.1914,
      "step": 37210
    },
    {
      "epoch": 0.6719498190800165,
      "grad_norm": 0.5495330691337585,
      "learning_rate": 4.856760468376719e-06,
      "loss": 0.2922,
      "step": 37215
    },
    {
      "epoch": 0.6720400985129171,
      "grad_norm": 0.4891622066497803,
      "learning_rate": 4.854328324879269e-06,
      "loss": 0.2126,
      "step": 37220
    },
    {
      "epoch": 0.6721303779458179,
      "grad_norm": 0.3623976409435272,
      "learning_rate": 4.851896595313697e-06,
      "loss": 0.1501,
      "step": 37225
    },
    {
      "epoch": 0.6722206573787186,
      "grad_norm": 0.6092562675476074,
      "learning_rate": 4.849465279875608e-06,
      "loss": 0.2624,
      "step": 37230
    },
    {
      "epoch": 0.6723109368116194,
      "grad_norm": 0.43685096502304077,
      "learning_rate": 4.84703437876059e-06,
      "loss": 0.2445,
      "step": 37235
    },
    {
      "epoch": 0.67240121624452,
      "grad_norm": 0.3236982524394989,
      "learning_rate": 4.844603892164192e-06,
      "loss": 0.287,
      "step": 37240
    },
    {
      "epoch": 0.6724914956774207,
      "grad_norm": 0.6211898326873779,
      "learning_rate": 4.842173820281929e-06,
      "loss": 0.1967,
      "step": 37245
    },
    {
      "epoch": 0.6725817751103215,
      "grad_norm": 0.35335037112236023,
      "learning_rate": 4.839744163309279e-06,
      "loss": 0.1927,
      "step": 37250
    },
    {
      "epoch": 0.6726720545432222,
      "grad_norm": 3.5796072483062744,
      "learning_rate": 4.837314921441695e-06,
      "loss": 0.3433,
      "step": 37255
    },
    {
      "epoch": 0.672762333976123,
      "grad_norm": 0.5667800903320312,
      "learning_rate": 4.834886094874587e-06,
      "loss": 0.1933,
      "step": 37260
    },
    {
      "epoch": 0.6728526134090236,
      "grad_norm": 0.238250270485878,
      "learning_rate": 4.832457683803337e-06,
      "loss": 0.2244,
      "step": 37265
    },
    {
      "epoch": 0.6729428928419243,
      "grad_norm": 0.9651192426681519,
      "learning_rate": 4.830029688423296e-06,
      "loss": 0.2784,
      "step": 37270
    },
    {
      "epoch": 0.6730331722748251,
      "grad_norm": 0.270931214094162,
      "learning_rate": 4.827602108929779e-06,
      "loss": 0.2955,
      "step": 37275
    },
    {
      "epoch": 0.6731234517077258,
      "grad_norm": 0.4074895977973938,
      "learning_rate": 4.825174945518064e-06,
      "loss": 0.306,
      "step": 37280
    },
    {
      "epoch": 0.6732137311406264,
      "grad_norm": 0.22945667803287506,
      "learning_rate": 4.822748198383402e-06,
      "loss": 0.2048,
      "step": 37285
    },
    {
      "epoch": 0.6733040105735272,
      "grad_norm": 0.7907798886299133,
      "learning_rate": 4.820321867721003e-06,
      "loss": 0.2558,
      "step": 37290
    },
    {
      "epoch": 0.6733942900064279,
      "grad_norm": 0.5409635901451111,
      "learning_rate": 4.817895953726053e-06,
      "loss": 0.271,
      "step": 37295
    },
    {
      "epoch": 0.6734845694393287,
      "grad_norm": 0.6511736512184143,
      "learning_rate": 4.815470456593692e-06,
      "loss": 0.3073,
      "step": 37300
    },
    {
      "epoch": 0.6735748488722293,
      "grad_norm": 0.560627281665802,
      "learning_rate": 4.813045376519038e-06,
      "loss": 0.2054,
      "step": 37305
    },
    {
      "epoch": 0.67366512830513,
      "grad_norm": 0.4113266170024872,
      "learning_rate": 4.810620713697173e-06,
      "loss": 0.1989,
      "step": 37310
    },
    {
      "epoch": 0.6737554077380308,
      "grad_norm": 0.5025131106376648,
      "learning_rate": 4.808196468323138e-06,
      "loss": 0.3788,
      "step": 37315
    },
    {
      "epoch": 0.6738456871709315,
      "grad_norm": 0.47875532507896423,
      "learning_rate": 4.805772640591954e-06,
      "loss": 0.2544,
      "step": 37320
    },
    {
      "epoch": 0.6739359666038321,
      "grad_norm": 0.6549875736236572,
      "learning_rate": 4.803349230698591e-06,
      "loss": 0.2717,
      "step": 37325
    },
    {
      "epoch": 0.6740262460367329,
      "grad_norm": 0.4416073262691498,
      "learning_rate": 4.800926238837998e-06,
      "loss": 0.2037,
      "step": 37330
    },
    {
      "epoch": 0.6741165254696336,
      "grad_norm": 0.17586322128772736,
      "learning_rate": 4.7985036652050895e-06,
      "loss": 0.2117,
      "step": 37335
    },
    {
      "epoch": 0.6742068049025344,
      "grad_norm": 0.49328237771987915,
      "learning_rate": 4.7960815099947465e-06,
      "loss": 0.1992,
      "step": 37340
    },
    {
      "epoch": 0.674297084335435,
      "grad_norm": 0.17381010949611664,
      "learning_rate": 4.793659773401805e-06,
      "loss": 0.2372,
      "step": 37345
    },
    {
      "epoch": 0.6743873637683357,
      "grad_norm": 0.2661042809486389,
      "learning_rate": 4.7912384556210854e-06,
      "loss": 0.2309,
      "step": 37350
    },
    {
      "epoch": 0.6744776432012365,
      "grad_norm": 0.3867124021053314,
      "learning_rate": 4.788817556847357e-06,
      "loss": 0.2342,
      "step": 37355
    },
    {
      "epoch": 0.6745679226341372,
      "grad_norm": 0.4215829074382782,
      "learning_rate": 4.786397077275371e-06,
      "loss": 0.2607,
      "step": 37360
    },
    {
      "epoch": 0.6746582020670379,
      "grad_norm": 0.3529093563556671,
      "learning_rate": 4.783977017099826e-06,
      "loss": 0.1426,
      "step": 37365
    },
    {
      "epoch": 0.6747484814999386,
      "grad_norm": 0.773078978061676,
      "learning_rate": 4.781557376515413e-06,
      "loss": 0.3066,
      "step": 37370
    },
    {
      "epoch": 0.6748387609328393,
      "grad_norm": 0.6937637329101562,
      "learning_rate": 4.779138155716764e-06,
      "loss": 0.2038,
      "step": 37375
    },
    {
      "epoch": 0.6749290403657401,
      "grad_norm": 0.653456449508667,
      "learning_rate": 4.7767193548984945e-06,
      "loss": 0.2444,
      "step": 37380
    },
    {
      "epoch": 0.6750193197986407,
      "grad_norm": 0.41567978262901306,
      "learning_rate": 4.774300974255173e-06,
      "loss": 0.24,
      "step": 37385
    },
    {
      "epoch": 0.6751095992315415,
      "grad_norm": 0.6049378514289856,
      "learning_rate": 4.771883013981346e-06,
      "loss": 0.2051,
      "step": 37390
    },
    {
      "epoch": 0.6751998786644422,
      "grad_norm": 0.5627599954605103,
      "learning_rate": 4.769465474271515e-06,
      "loss": 0.2942,
      "step": 37395
    },
    {
      "epoch": 0.6752901580973429,
      "grad_norm": 0.27880987524986267,
      "learning_rate": 4.767048355320158e-06,
      "loss": 0.1388,
      "step": 37400
    },
    {
      "epoch": 0.6753804375302436,
      "grad_norm": 0.5502472519874573,
      "learning_rate": 4.764631657321717e-06,
      "loss": 0.2598,
      "step": 37405
    },
    {
      "epoch": 0.6754707169631443,
      "grad_norm": 0.8498625159263611,
      "learning_rate": 4.76221538047059e-06,
      "loss": 0.2192,
      "step": 37410
    },
    {
      "epoch": 0.675560996396045,
      "grad_norm": 0.3717988133430481,
      "learning_rate": 4.759799524961157e-06,
      "loss": 0.1724,
      "step": 37415
    },
    {
      "epoch": 0.6756512758289458,
      "grad_norm": 0.4655216932296753,
      "learning_rate": 4.757384090987749e-06,
      "loss": 0.2329,
      "step": 37420
    },
    {
      "epoch": 0.6757415552618464,
      "grad_norm": 0.4961567521095276,
      "learning_rate": 4.754969078744677e-06,
      "loss": 0.1794,
      "step": 37425
    },
    {
      "epoch": 0.6758318346947472,
      "grad_norm": 0.49937573075294495,
      "learning_rate": 4.752554488426203e-06,
      "loss": 0.2079,
      "step": 37430
    },
    {
      "epoch": 0.6759221141276479,
      "grad_norm": 0.4563522934913635,
      "learning_rate": 4.750140320226568e-06,
      "loss": 0.2004,
      "step": 37435
    },
    {
      "epoch": 0.6760123935605487,
      "grad_norm": 0.41807126998901367,
      "learning_rate": 4.747726574339975e-06,
      "loss": 0.275,
      "step": 37440
    },
    {
      "epoch": 0.6761026729934493,
      "grad_norm": 0.3730819523334503,
      "learning_rate": 4.745313250960595e-06,
      "loss": 0.2351,
      "step": 37445
    },
    {
      "epoch": 0.67619295242635,
      "grad_norm": 2.460869550704956,
      "learning_rate": 4.742900350282555e-06,
      "loss": 0.18,
      "step": 37450
    },
    {
      "epoch": 0.6762832318592508,
      "grad_norm": 0.4683798551559448,
      "learning_rate": 4.740487872499962e-06,
      "loss": 0.2404,
      "step": 37455
    },
    {
      "epoch": 0.6763735112921515,
      "grad_norm": 0.6973472833633423,
      "learning_rate": 4.738075817806875e-06,
      "loss": 0.1958,
      "step": 37460
    },
    {
      "epoch": 0.6764637907250521,
      "grad_norm": 0.7393599152565002,
      "learning_rate": 4.735664186397332e-06,
      "loss": 0.2334,
      "step": 37465
    },
    {
      "epoch": 0.6765540701579529,
      "grad_norm": 0.6769455671310425,
      "learning_rate": 4.733252978465329e-06,
      "loss": 0.3239,
      "step": 37470
    },
    {
      "epoch": 0.6766443495908536,
      "grad_norm": 0.46761155128479004,
      "learning_rate": 4.7308421942048355e-06,
      "loss": 0.1858,
      "step": 37475
    },
    {
      "epoch": 0.6767346290237544,
      "grad_norm": 0.3539060652256012,
      "learning_rate": 4.728431833809774e-06,
      "loss": 0.2737,
      "step": 37480
    },
    {
      "epoch": 0.676824908456655,
      "grad_norm": 0.5235952734947205,
      "learning_rate": 4.726021897474045e-06,
      "loss": 0.2115,
      "step": 37485
    },
    {
      "epoch": 0.6769151878895557,
      "grad_norm": 0.9684334397315979,
      "learning_rate": 4.723612385391509e-06,
      "loss": 0.2367,
      "step": 37490
    },
    {
      "epoch": 0.6770054673224565,
      "grad_norm": 0.37715333700180054,
      "learning_rate": 4.721203297755991e-06,
      "loss": 0.2294,
      "step": 37495
    },
    {
      "epoch": 0.6770957467553572,
      "grad_norm": 0.5215760469436646,
      "learning_rate": 4.718794634761286e-06,
      "loss": 0.161,
      "step": 37500
    },
    {
      "epoch": 0.6771860261882579,
      "grad_norm": 0.34836265444755554,
      "learning_rate": 4.716386396601156e-06,
      "loss": 0.2247,
      "step": 37505
    },
    {
      "epoch": 0.6772763056211586,
      "grad_norm": 0.34557798504829407,
      "learning_rate": 4.713978583469327e-06,
      "loss": 0.1931,
      "step": 37510
    },
    {
      "epoch": 0.6773665850540593,
      "grad_norm": 0.7941352128982544,
      "learning_rate": 4.711571195559485e-06,
      "loss": 0.3337,
      "step": 37515
    },
    {
      "epoch": 0.6774568644869601,
      "grad_norm": 0.5474073886871338,
      "learning_rate": 4.709164233065294e-06,
      "loss": 0.1931,
      "step": 37520
    },
    {
      "epoch": 0.6775471439198607,
      "grad_norm": 0.39072561264038086,
      "learning_rate": 4.706757696180368e-06,
      "loss": 0.1887,
      "step": 37525
    },
    {
      "epoch": 0.6776374233527614,
      "grad_norm": 0.4135580360889435,
      "learning_rate": 4.7043515850983e-06,
      "loss": 0.2794,
      "step": 37530
    },
    {
      "epoch": 0.6777277027856622,
      "grad_norm": 0.3942650258541107,
      "learning_rate": 4.7019459000126435e-06,
      "loss": 0.1952,
      "step": 37535
    },
    {
      "epoch": 0.6778179822185629,
      "grad_norm": 0.6654621362686157,
      "learning_rate": 4.699540641116923e-06,
      "loss": 0.2369,
      "step": 37540
    },
    {
      "epoch": 0.6779082616514636,
      "grad_norm": 0.5702198147773743,
      "learning_rate": 4.697135808604616e-06,
      "loss": 0.3061,
      "step": 37545
    },
    {
      "epoch": 0.6779985410843643,
      "grad_norm": 0.4783928692340851,
      "learning_rate": 4.694731402669182e-06,
      "loss": 0.2044,
      "step": 37550
    },
    {
      "epoch": 0.678088820517265,
      "grad_norm": 0.409105122089386,
      "learning_rate": 4.692327423504029e-06,
      "loss": 0.2278,
      "step": 37555
    },
    {
      "epoch": 0.6781790999501658,
      "grad_norm": 0.4907389283180237,
      "learning_rate": 4.689923871302547e-06,
      "loss": 0.2065,
      "step": 37560
    },
    {
      "epoch": 0.6782693793830664,
      "grad_norm": 0.5981885194778442,
      "learning_rate": 4.687520746258075e-06,
      "loss": 0.2759,
      "step": 37565
    },
    {
      "epoch": 0.6783596588159672,
      "grad_norm": 0.300359308719635,
      "learning_rate": 4.6851180485639415e-06,
      "loss": 0.2398,
      "step": 37570
    },
    {
      "epoch": 0.6784499382488679,
      "grad_norm": 0.4736993610858917,
      "learning_rate": 4.682715778413418e-06,
      "loss": 0.2492,
      "step": 37575
    },
    {
      "epoch": 0.6785402176817686,
      "grad_norm": 0.54407799243927,
      "learning_rate": 4.6803139359997455e-06,
      "loss": 0.1323,
      "step": 37580
    },
    {
      "epoch": 0.6786304971146693,
      "grad_norm": 1.4211688041687012,
      "learning_rate": 4.677912521516143e-06,
      "loss": 0.1437,
      "step": 37585
    },
    {
      "epoch": 0.67872077654757,
      "grad_norm": 0.20921234786510468,
      "learning_rate": 4.675511535155779e-06,
      "loss": 0.2183,
      "step": 37590
    },
    {
      "epoch": 0.6788110559804708,
      "grad_norm": 0.33046719431877136,
      "learning_rate": 4.673110977111799e-06,
      "loss": 0.1872,
      "step": 37595
    },
    {
      "epoch": 0.6789013354133715,
      "grad_norm": 0.40264737606048584,
      "learning_rate": 4.6707108475773115e-06,
      "loss": 0.1985,
      "step": 37600
    },
    {
      "epoch": 0.6789916148462721,
      "grad_norm": 0.5488992929458618,
      "learning_rate": 4.668311146745391e-06,
      "loss": 0.2889,
      "step": 37605
    },
    {
      "epoch": 0.6790818942791729,
      "grad_norm": 0.9485848546028137,
      "learning_rate": 4.665911874809072e-06,
      "loss": 0.3017,
      "step": 37610
    },
    {
      "epoch": 0.6791721737120736,
      "grad_norm": 0.7803026437759399,
      "learning_rate": 4.663513031961362e-06,
      "loss": 0.2445,
      "step": 37615
    },
    {
      "epoch": 0.6792624531449744,
      "grad_norm": 0.7331886291503906,
      "learning_rate": 4.661114618395225e-06,
      "loss": 0.3379,
      "step": 37620
    },
    {
      "epoch": 0.679352732577875,
      "grad_norm": 0.8646613359451294,
      "learning_rate": 4.658716634303604e-06,
      "loss": 0.1735,
      "step": 37625
    },
    {
      "epoch": 0.6794430120107757,
      "grad_norm": 0.4974726438522339,
      "learning_rate": 4.656319079879386e-06,
      "loss": 0.1663,
      "step": 37630
    },
    {
      "epoch": 0.6795332914436765,
      "grad_norm": 0.34508395195007324,
      "learning_rate": 4.653921955315454e-06,
      "loss": 0.2258,
      "step": 37635
    },
    {
      "epoch": 0.6796235708765772,
      "grad_norm": 0.45985549688339233,
      "learning_rate": 4.651525260804628e-06,
      "loss": 0.2097,
      "step": 37640
    },
    {
      "epoch": 0.6797138503094778,
      "grad_norm": 0.4692579209804535,
      "learning_rate": 4.6491289965397114e-06,
      "loss": 0.1885,
      "step": 37645
    },
    {
      "epoch": 0.6798041297423786,
      "grad_norm": 0.1568463295698166,
      "learning_rate": 4.646733162713457e-06,
      "loss": 0.2665,
      "step": 37650
    },
    {
      "epoch": 0.6798944091752793,
      "grad_norm": 0.2690006494522095,
      "learning_rate": 4.6443377595186036e-06,
      "loss": 0.2539,
      "step": 37655
    },
    {
      "epoch": 0.6799846886081801,
      "grad_norm": 0.4124183654785156,
      "learning_rate": 4.6419427871478325e-06,
      "loss": 0.2961,
      "step": 37660
    },
    {
      "epoch": 0.6800749680410808,
      "grad_norm": 0.4410317540168762,
      "learning_rate": 4.639548245793808e-06,
      "loss": 0.2054,
      "step": 37665
    },
    {
      "epoch": 0.6801652474739814,
      "grad_norm": 0.2998250722885132,
      "learning_rate": 4.637154135649156e-06,
      "loss": 0.1674,
      "step": 37670
    },
    {
      "epoch": 0.6802555269068822,
      "grad_norm": 0.3939516246318817,
      "learning_rate": 4.63476045690646e-06,
      "loss": 0.2564,
      "step": 37675
    },
    {
      "epoch": 0.6803458063397829,
      "grad_norm": 0.7001118659973145,
      "learning_rate": 4.632367209758278e-06,
      "loss": 0.2158,
      "step": 37680
    },
    {
      "epoch": 0.6804360857726837,
      "grad_norm": 0.4548497498035431,
      "learning_rate": 4.629974394397125e-06,
      "loss": 0.2227,
      "step": 37685
    },
    {
      "epoch": 0.6805263652055843,
      "grad_norm": 0.6921202540397644,
      "learning_rate": 4.627582011015491e-06,
      "loss": 0.2923,
      "step": 37690
    },
    {
      "epoch": 0.680616644638485,
      "grad_norm": 0.6251453757286072,
      "learning_rate": 4.625190059805816e-06,
      "loss": 0.2418,
      "step": 37695
    },
    {
      "epoch": 0.6807069240713858,
      "grad_norm": 0.32615020871162415,
      "learning_rate": 4.6227985409605305e-06,
      "loss": 0.1913,
      "step": 37700
    },
    {
      "epoch": 0.6807972035042865,
      "grad_norm": 0.8246029615402222,
      "learning_rate": 4.620407454672001e-06,
      "loss": 0.1956,
      "step": 37705
    },
    {
      "epoch": 0.6808874829371871,
      "grad_norm": 0.6869685053825378,
      "learning_rate": 4.618016801132582e-06,
      "loss": 0.2555,
      "step": 37710
    },
    {
      "epoch": 0.6809777623700879,
      "grad_norm": 0.5275892019271851,
      "learning_rate": 4.6156265805345775e-06,
      "loss": 0.1851,
      "step": 37715
    },
    {
      "epoch": 0.6810680418029886,
      "grad_norm": 0.5740939378738403,
      "learning_rate": 4.613236793070271e-06,
      "loss": 0.3119,
      "step": 37720
    },
    {
      "epoch": 0.6811583212358894,
      "grad_norm": 0.5859683752059937,
      "learning_rate": 4.610847438931895e-06,
      "loss": 0.3205,
      "step": 37725
    },
    {
      "epoch": 0.68124860066879,
      "grad_norm": 0.3429293632507324,
      "learning_rate": 4.608458518311658e-06,
      "loss": 0.2084,
      "step": 37730
    },
    {
      "epoch": 0.6813388801016907,
      "grad_norm": 0.25039711594581604,
      "learning_rate": 4.606070031401733e-06,
      "loss": 0.2725,
      "step": 37735
    },
    {
      "epoch": 0.6814291595345915,
      "grad_norm": 0.27948713302612305,
      "learning_rate": 4.603681978394262e-06,
      "loss": 0.1722,
      "step": 37740
    },
    {
      "epoch": 0.6815194389674922,
      "grad_norm": 0.32593435049057007,
      "learning_rate": 4.6012943594813365e-06,
      "loss": 0.2556,
      "step": 37745
    },
    {
      "epoch": 0.6816097184003929,
      "grad_norm": 0.3133193850517273,
      "learning_rate": 4.59890717485503e-06,
      "loss": 0.2198,
      "step": 37750
    },
    {
      "epoch": 0.6816999978332936,
      "grad_norm": 1.4969346523284912,
      "learning_rate": 4.596520424707368e-06,
      "loss": 0.2234,
      "step": 37755
    },
    {
      "epoch": 0.6817902772661943,
      "grad_norm": 0.3820660412311554,
      "learning_rate": 4.594134109230351e-06,
      "loss": 0.2918,
      "step": 37760
    },
    {
      "epoch": 0.6818805566990951,
      "grad_norm": 0.5165436267852783,
      "learning_rate": 4.591748228615945e-06,
      "loss": 0.2041,
      "step": 37765
    },
    {
      "epoch": 0.6819708361319957,
      "grad_norm": 0.31727850437164307,
      "learning_rate": 4.589362783056068e-06,
      "loss": 0.2259,
      "step": 37770
    },
    {
      "epoch": 0.6820611155648965,
      "grad_norm": 0.4642702639102936,
      "learning_rate": 4.58697777274262e-06,
      "loss": 0.1235,
      "step": 37775
    },
    {
      "epoch": 0.6821513949977972,
      "grad_norm": 0.4593847692012787,
      "learning_rate": 4.58459319786745e-06,
      "loss": 0.3199,
      "step": 37780
    },
    {
      "epoch": 0.6822416744306979,
      "grad_norm": 0.31745219230651855,
      "learning_rate": 4.582209058622386e-06,
      "loss": 0.27,
      "step": 37785
    },
    {
      "epoch": 0.6823319538635986,
      "grad_norm": 0.4456140398979187,
      "learning_rate": 4.579825355199209e-06,
      "loss": 0.2096,
      "step": 37790
    },
    {
      "epoch": 0.6824222332964993,
      "grad_norm": 0.4190645217895508,
      "learning_rate": 4.577442087789674e-06,
      "loss": 0.2529,
      "step": 37795
    },
    {
      "epoch": 0.6825125127294001,
      "grad_norm": 0.32241523265838623,
      "learning_rate": 4.575059256585497e-06,
      "loss": 0.1457,
      "step": 37800
    },
    {
      "epoch": 0.6826027921623008,
      "grad_norm": 0.4470575451850891,
      "learning_rate": 4.572676861778363e-06,
      "loss": 0.1714,
      "step": 37805
    },
    {
      "epoch": 0.6826930715952014,
      "grad_norm": 0.315279096364975,
      "learning_rate": 4.5702949035599124e-06,
      "loss": 0.1288,
      "step": 37810
    },
    {
      "epoch": 0.6827833510281022,
      "grad_norm": 0.7390419244766235,
      "learning_rate": 4.567913382121761e-06,
      "loss": 0.1483,
      "step": 37815
    },
    {
      "epoch": 0.6828736304610029,
      "grad_norm": 0.5478441119194031,
      "learning_rate": 4.56553229765548e-06,
      "loss": 0.187,
      "step": 37820
    },
    {
      "epoch": 0.6829639098939037,
      "grad_norm": 0.4399346113204956,
      "learning_rate": 4.563151650352614e-06,
      "loss": 0.1779,
      "step": 37825
    },
    {
      "epoch": 0.6830541893268043,
      "grad_norm": 0.9848801493644714,
      "learning_rate": 4.560771440404668e-06,
      "loss": 0.2331,
      "step": 37830
    },
    {
      "epoch": 0.683144468759705,
      "grad_norm": 0.3818892240524292,
      "learning_rate": 4.558391668003116e-06,
      "loss": 0.1728,
      "step": 37835
    },
    {
      "epoch": 0.6832347481926058,
      "grad_norm": 0.5127613544464111,
      "learning_rate": 4.556012333339386e-06,
      "loss": 0.2138,
      "step": 37840
    },
    {
      "epoch": 0.6833250276255065,
      "grad_norm": 0.4102364182472229,
      "learning_rate": 4.553633436604885e-06,
      "loss": 0.287,
      "step": 37845
    },
    {
      "epoch": 0.6834153070584071,
      "grad_norm": 0.3781411349773407,
      "learning_rate": 4.5512549779909765e-06,
      "loss": 0.2346,
      "step": 37850
    },
    {
      "epoch": 0.6835055864913079,
      "grad_norm": 0.3069896101951599,
      "learning_rate": 4.548876957688985e-06,
      "loss": 0.2677,
      "step": 37855
    },
    {
      "epoch": 0.6835958659242086,
      "grad_norm": 0.351437509059906,
      "learning_rate": 4.546499375890209e-06,
      "loss": 0.2813,
      "step": 37860
    },
    {
      "epoch": 0.6836861453571094,
      "grad_norm": 0.3506467640399933,
      "learning_rate": 4.5441222327859055e-06,
      "loss": 0.194,
      "step": 37865
    },
    {
      "epoch": 0.68377642479001,
      "grad_norm": 0.45503199100494385,
      "learning_rate": 4.541745528567306e-06,
      "loss": 0.1705,
      "step": 37870
    },
    {
      "epoch": 0.6838667042229107,
      "grad_norm": 0.5410659313201904,
      "learning_rate": 4.539369263425589e-06,
      "loss": 0.1441,
      "step": 37875
    },
    {
      "epoch": 0.6839569836558115,
      "grad_norm": 0.5261062979698181,
      "learning_rate": 4.536993437551916e-06,
      "loss": 0.2447,
      "step": 37880
    },
    {
      "epoch": 0.6840472630887122,
      "grad_norm": 0.08837950229644775,
      "learning_rate": 4.534618051137397e-06,
      "loss": 0.1915,
      "step": 37885
    },
    {
      "epoch": 0.6841375425216129,
      "grad_norm": 0.5404590964317322,
      "learning_rate": 4.532243104373123e-06,
      "loss": 0.1768,
      "step": 37890
    },
    {
      "epoch": 0.6842278219545136,
      "grad_norm": 0.4018935263156891,
      "learning_rate": 4.5298685974501285e-06,
      "loss": 0.2204,
      "step": 37895
    },
    {
      "epoch": 0.6843181013874143,
      "grad_norm": 0.42148804664611816,
      "learning_rate": 4.527494530559442e-06,
      "loss": 0.2727,
      "step": 37900
    },
    {
      "epoch": 0.6844083808203151,
      "grad_norm": 0.36024728417396545,
      "learning_rate": 4.5251209038920276e-06,
      "loss": 0.274,
      "step": 37905
    },
    {
      "epoch": 0.6844986602532157,
      "grad_norm": 0.40288153290748596,
      "learning_rate": 4.522747717638833e-06,
      "loss": 0.285,
      "step": 37910
    },
    {
      "epoch": 0.6845889396861164,
      "grad_norm": 0.3067623972892761,
      "learning_rate": 4.520374971990758e-06,
      "loss": 0.2437,
      "step": 37915
    },
    {
      "epoch": 0.6846792191190172,
      "grad_norm": 0.25448545813560486,
      "learning_rate": 4.51800266713868e-06,
      "loss": 0.1896,
      "step": 37920
    },
    {
      "epoch": 0.6847694985519179,
      "grad_norm": 0.5942654609680176,
      "learning_rate": 4.515630803273425e-06,
      "loss": 0.3519,
      "step": 37925
    },
    {
      "epoch": 0.6848597779848186,
      "grad_norm": 0.4476010501384735,
      "learning_rate": 4.513259380585798e-06,
      "loss": 0.2339,
      "step": 37930
    },
    {
      "epoch": 0.6849500574177193,
      "grad_norm": 0.25513124465942383,
      "learning_rate": 4.51088839926656e-06,
      "loss": 0.285,
      "step": 37935
    },
    {
      "epoch": 0.68504033685062,
      "grad_norm": 0.8240607380867004,
      "learning_rate": 4.508517859506445e-06,
      "loss": 0.2533,
      "step": 37940
    },
    {
      "epoch": 0.6851306162835208,
      "grad_norm": 0.4471612572669983,
      "learning_rate": 4.506147761496141e-06,
      "loss": 0.2492,
      "step": 37945
    },
    {
      "epoch": 0.6852208957164214,
      "grad_norm": 0.4679805338382721,
      "learning_rate": 4.503778105426302e-06,
      "loss": 0.1738,
      "step": 37950
    },
    {
      "epoch": 0.6853111751493222,
      "grad_norm": 0.31638088822364807,
      "learning_rate": 4.5014088914875555e-06,
      "loss": 0.2002,
      "step": 37955
    },
    {
      "epoch": 0.6854014545822229,
      "grad_norm": 1.1712069511413574,
      "learning_rate": 4.499040119870479e-06,
      "loss": 0.2783,
      "step": 37960
    },
    {
      "epoch": 0.6854917340151236,
      "grad_norm": 0.5667579770088196,
      "learning_rate": 4.496671790765636e-06,
      "loss": 0.1797,
      "step": 37965
    },
    {
      "epoch": 0.6855820134480243,
      "grad_norm": 0.9462890028953552,
      "learning_rate": 4.4943039043635305e-06,
      "loss": 0.1809,
      "step": 37970
    },
    {
      "epoch": 0.685672292880925,
      "grad_norm": 0.6696999073028564,
      "learning_rate": 4.491936460854649e-06,
      "loss": 0.1621,
      "step": 37975
    },
    {
      "epoch": 0.6857625723138258,
      "grad_norm": 0.4125041663646698,
      "learning_rate": 4.4895694604294295e-06,
      "loss": 0.2684,
      "step": 37980
    },
    {
      "epoch": 0.6858528517467265,
      "grad_norm": 0.38330793380737305,
      "learning_rate": 4.487202903278283e-06,
      "loss": 0.2668,
      "step": 37985
    },
    {
      "epoch": 0.6859431311796271,
      "grad_norm": 0.4018458127975464,
      "learning_rate": 4.48483678959158e-06,
      "loss": 0.2052,
      "step": 37990
    },
    {
      "epoch": 0.6860334106125279,
      "grad_norm": 0.416517972946167,
      "learning_rate": 4.482471119559658e-06,
      "loss": 0.2099,
      "step": 37995
    },
    {
      "epoch": 0.6861236900454286,
      "grad_norm": 0.3194110095500946,
      "learning_rate": 4.480105893372817e-06,
      "loss": 0.213,
      "step": 38000
    },
    {
      "epoch": 0.6862139694783294,
      "grad_norm": 0.39987972378730774,
      "learning_rate": 4.477741111221327e-06,
      "loss": 0.2826,
      "step": 38005
    },
    {
      "epoch": 0.68630424891123,
      "grad_norm": 0.5515651702880859,
      "learning_rate": 4.475376773295411e-06,
      "loss": 0.2616,
      "step": 38010
    },
    {
      "epoch": 0.6863945283441307,
      "grad_norm": 0.42721080780029297,
      "learning_rate": 4.47301287978527e-06,
      "loss": 0.2316,
      "step": 38015
    },
    {
      "epoch": 0.6864848077770315,
      "grad_norm": 0.5164182782173157,
      "learning_rate": 4.470649430881053e-06,
      "loss": 0.166,
      "step": 38020
    },
    {
      "epoch": 0.6865750872099322,
      "grad_norm": 0.4573683738708496,
      "learning_rate": 4.468286426772887e-06,
      "loss": 0.2536,
      "step": 38025
    },
    {
      "epoch": 0.6866653666428328,
      "grad_norm": 0.40096044540405273,
      "learning_rate": 4.465923867650859e-06,
      "loss": 0.2237,
      "step": 38030
    },
    {
      "epoch": 0.6867556460757336,
      "grad_norm": 0.4793723225593567,
      "learning_rate": 4.463561753705022e-06,
      "loss": 0.1924,
      "step": 38035
    },
    {
      "epoch": 0.6868459255086343,
      "grad_norm": 1.0447996854782104,
      "learning_rate": 4.461200085125389e-06,
      "loss": 0.2464,
      "step": 38040
    },
    {
      "epoch": 0.6869362049415351,
      "grad_norm": 0.4933064877986908,
      "learning_rate": 4.458838862101934e-06,
      "loss": 0.2805,
      "step": 38045
    },
    {
      "epoch": 0.6870264843744358,
      "grad_norm": 0.44689321517944336,
      "learning_rate": 4.45647808482461e-06,
      "loss": 0.299,
      "step": 38050
    },
    {
      "epoch": 0.6871167638073364,
      "grad_norm": 0.5152558088302612,
      "learning_rate": 4.4541177534833125e-06,
      "loss": 0.2808,
      "step": 38055
    },
    {
      "epoch": 0.6872070432402372,
      "grad_norm": 0.2078225165605545,
      "learning_rate": 4.451757868267919e-06,
      "loss": 0.1442,
      "step": 38060
    },
    {
      "epoch": 0.6872973226731379,
      "grad_norm": 0.26201045513153076,
      "learning_rate": 4.449398429368268e-06,
      "loss": 0.26,
      "step": 38065
    },
    {
      "epoch": 0.6873876021060387,
      "grad_norm": 0.5290224552154541,
      "learning_rate": 4.4470394369741585e-06,
      "loss": 0.2903,
      "step": 38070
    },
    {
      "epoch": 0.6874778815389393,
      "grad_norm": 0.2802322506904602,
      "learning_rate": 4.4446808912753495e-06,
      "loss": 0.1756,
      "step": 38075
    },
    {
      "epoch": 0.68756816097184,
      "grad_norm": 0.5011495351791382,
      "learning_rate": 4.442322792461574e-06,
      "loss": 0.2399,
      "step": 38080
    },
    {
      "epoch": 0.6876584404047408,
      "grad_norm": 0.3852989673614502,
      "learning_rate": 4.43996514072252e-06,
      "loss": 0.2165,
      "step": 38085
    },
    {
      "epoch": 0.6877487198376415,
      "grad_norm": 0.5957157611846924,
      "learning_rate": 4.437607936247843e-06,
      "loss": 0.2796,
      "step": 38090
    },
    {
      "epoch": 0.6878389992705422,
      "grad_norm": 0.31345441937446594,
      "learning_rate": 4.4352511792271655e-06,
      "loss": 0.1827,
      "step": 38095
    },
    {
      "epoch": 0.6879292787034429,
      "grad_norm": 0.4975501298904419,
      "learning_rate": 4.432894869850077e-06,
      "loss": 0.2044,
      "step": 38100
    },
    {
      "epoch": 0.6880195581363436,
      "grad_norm": 1.0897471904754639,
      "learning_rate": 4.430539008306113e-06,
      "loss": 0.1926,
      "step": 38105
    },
    {
      "epoch": 0.6881098375692444,
      "grad_norm": 0.48652711510658264,
      "learning_rate": 4.428183594784796e-06,
      "loss": 0.2263,
      "step": 38110
    },
    {
      "epoch": 0.688200117002145,
      "grad_norm": 0.41459593176841736,
      "learning_rate": 4.425828629475595e-06,
      "loss": 0.2908,
      "step": 38115
    },
    {
      "epoch": 0.6882903964350457,
      "grad_norm": 0.4874315559864044,
      "learning_rate": 4.423474112567956e-06,
      "loss": 0.2287,
      "step": 38120
    },
    {
      "epoch": 0.6883806758679465,
      "grad_norm": 0.5218778252601624,
      "learning_rate": 4.421120044251276e-06,
      "loss": 0.2283,
      "step": 38125
    },
    {
      "epoch": 0.6884709553008472,
      "grad_norm": 0.4416961371898651,
      "learning_rate": 4.4187664247149256e-06,
      "loss": 0.2895,
      "step": 38130
    },
    {
      "epoch": 0.6885612347337479,
      "grad_norm": 0.18363463878631592,
      "learning_rate": 4.416413254148241e-06,
      "loss": 0.1656,
      "step": 38135
    },
    {
      "epoch": 0.6886515141666486,
      "grad_norm": 0.45894527435302734,
      "learning_rate": 4.414060532740509e-06,
      "loss": 0.3297,
      "step": 38140
    },
    {
      "epoch": 0.6887417935995493,
      "grad_norm": 1.545738697052002,
      "learning_rate": 4.411708260680997e-06,
      "loss": 0.1385,
      "step": 38145
    },
    {
      "epoch": 0.6888320730324501,
      "grad_norm": 0.5047971606254578,
      "learning_rate": 4.409356438158921e-06,
      "loss": 0.2842,
      "step": 38150
    },
    {
      "epoch": 0.6889223524653507,
      "grad_norm": 0.8523099422454834,
      "learning_rate": 4.407005065363473e-06,
      "loss": 0.2644,
      "step": 38155
    },
    {
      "epoch": 0.6890126318982515,
      "grad_norm": 0.7765586972236633,
      "learning_rate": 4.404654142483802e-06,
      "loss": 0.1474,
      "step": 38160
    },
    {
      "epoch": 0.6891029113311522,
      "grad_norm": 0.507250964641571,
      "learning_rate": 4.402303669709026e-06,
      "loss": 0.2036,
      "step": 38165
    },
    {
      "epoch": 0.6891931907640529,
      "grad_norm": 0.8221535682678223,
      "learning_rate": 4.399953647228216e-06,
      "loss": 0.2544,
      "step": 38170
    },
    {
      "epoch": 0.6892834701969536,
      "grad_norm": 0.44962579011917114,
      "learning_rate": 4.397604075230425e-06,
      "loss": 0.3094,
      "step": 38175
    },
    {
      "epoch": 0.6893737496298543,
      "grad_norm": 0.5254794955253601,
      "learning_rate": 4.395254953904648e-06,
      "loss": 0.2317,
      "step": 38180
    },
    {
      "epoch": 0.6894640290627551,
      "grad_norm": 0.5492463707923889,
      "learning_rate": 4.3929062834398624e-06,
      "loss": 0.2245,
      "step": 38185
    },
    {
      "epoch": 0.6895543084956558,
      "grad_norm": 0.5577696561813354,
      "learning_rate": 4.390558064024997e-06,
      "loss": 0.21,
      "step": 38190
    },
    {
      "epoch": 0.6896445879285564,
      "grad_norm": 0.3024185299873352,
      "learning_rate": 4.388210295848949e-06,
      "loss": 0.1505,
      "step": 38195
    },
    {
      "epoch": 0.6897348673614572,
      "grad_norm": 0.42269188165664673,
      "learning_rate": 4.385862979100581e-06,
      "loss": 0.2802,
      "step": 38200
    },
    {
      "epoch": 0.6898251467943579,
      "grad_norm": 0.7994741797447205,
      "learning_rate": 4.3835161139687236e-06,
      "loss": 0.2289,
      "step": 38205
    },
    {
      "epoch": 0.6899154262272587,
      "grad_norm": 0.25307267904281616,
      "learning_rate": 4.381169700642153e-06,
      "loss": 0.1675,
      "step": 38210
    },
    {
      "epoch": 0.6900057056601593,
      "grad_norm": 0.5518661141395569,
      "learning_rate": 4.378823739309631e-06,
      "loss": 0.1861,
      "step": 38215
    },
    {
      "epoch": 0.69009598509306,
      "grad_norm": 0.4862739145755768,
      "learning_rate": 4.376478230159866e-06,
      "loss": 0.2629,
      "step": 38220
    },
    {
      "epoch": 0.6901862645259608,
      "grad_norm": 0.3612036406993866,
      "learning_rate": 4.374133173381541e-06,
      "loss": 0.2857,
      "step": 38225
    },
    {
      "epoch": 0.6902765439588615,
      "grad_norm": 0.4374309182167053,
      "learning_rate": 4.371788569163301e-06,
      "loss": 0.2152,
      "step": 38230
    },
    {
      "epoch": 0.6903668233917621,
      "grad_norm": 0.8651866316795349,
      "learning_rate": 4.369444417693747e-06,
      "loss": 0.2508,
      "step": 38235
    },
    {
      "epoch": 0.6904571028246629,
      "grad_norm": 0.7004550695419312,
      "learning_rate": 4.3671007191614545e-06,
      "loss": 0.2494,
      "step": 38240
    },
    {
      "epoch": 0.6905473822575636,
      "grad_norm": 0.4672585725784302,
      "learning_rate": 4.364757473754951e-06,
      "loss": 0.1394,
      "step": 38245
    },
    {
      "epoch": 0.6906376616904644,
      "grad_norm": 0.36849546432495117,
      "learning_rate": 4.362414681662739e-06,
      "loss": 0.1905,
      "step": 38250
    },
    {
      "epoch": 0.690727941123365,
      "grad_norm": 0.3537253737449646,
      "learning_rate": 4.360072343073273e-06,
      "loss": 0.2032,
      "step": 38255
    },
    {
      "epoch": 0.6908182205562657,
      "grad_norm": 0.3273913860321045,
      "learning_rate": 4.357730458174982e-06,
      "loss": 0.1979,
      "step": 38260
    },
    {
      "epoch": 0.6909084999891665,
      "grad_norm": 0.36351215839385986,
      "learning_rate": 4.355389027156251e-06,
      "loss": 0.3141,
      "step": 38265
    },
    {
      "epoch": 0.6909987794220672,
      "grad_norm": 0.316074937582016,
      "learning_rate": 4.353048050205437e-06,
      "loss": 0.2884,
      "step": 38270
    },
    {
      "epoch": 0.6910890588549679,
      "grad_norm": 0.7243396639823914,
      "learning_rate": 4.350707527510847e-06,
      "loss": 0.246,
      "step": 38275
    },
    {
      "epoch": 0.6911793382878686,
      "grad_norm": 0.5573142170906067,
      "learning_rate": 4.348367459260764e-06,
      "loss": 0.2159,
      "step": 38280
    },
    {
      "epoch": 0.6912696177207693,
      "grad_norm": 0.43125444650650024,
      "learning_rate": 4.3460278456434245e-06,
      "loss": 0.1907,
      "step": 38285
    },
    {
      "epoch": 0.6913598971536701,
      "grad_norm": 0.5265786051750183,
      "learning_rate": 4.3436886868470375e-06,
      "loss": 0.2275,
      "step": 38290
    },
    {
      "epoch": 0.6914501765865707,
      "grad_norm": 0.3848966658115387,
      "learning_rate": 4.341349983059769e-06,
      "loss": 0.2373,
      "step": 38295
    },
    {
      "epoch": 0.6915404560194715,
      "grad_norm": 0.3046269118785858,
      "learning_rate": 4.339011734469757e-06,
      "loss": 0.1633,
      "step": 38300
    },
    {
      "epoch": 0.6916307354523722,
      "grad_norm": 0.4017146825790405,
      "learning_rate": 4.336673941265087e-06,
      "loss": 0.1801,
      "step": 38305
    },
    {
      "epoch": 0.6917210148852729,
      "grad_norm": 0.6081216335296631,
      "learning_rate": 4.334336603633826e-06,
      "loss": 0.2944,
      "step": 38310
    },
    {
      "epoch": 0.6918112943181736,
      "grad_norm": 0.34811705350875854,
      "learning_rate": 4.331999721763992e-06,
      "loss": 0.2143,
      "step": 38315
    },
    {
      "epoch": 0.6919015737510743,
      "grad_norm": 0.41205698251724243,
      "learning_rate": 4.329663295843566e-06,
      "loss": 0.2577,
      "step": 38320
    },
    {
      "epoch": 0.691991853183975,
      "grad_norm": 0.3803647756576538,
      "learning_rate": 4.3273273260605005e-06,
      "loss": 0.271,
      "step": 38325
    },
    {
      "epoch": 0.6920821326168758,
      "grad_norm": 0.5279304385185242,
      "learning_rate": 4.324991812602708e-06,
      "loss": 0.3284,
      "step": 38330
    },
    {
      "epoch": 0.6921724120497764,
      "grad_norm": 0.47987908124923706,
      "learning_rate": 4.322656755658066e-06,
      "loss": 0.2659,
      "step": 38335
    },
    {
      "epoch": 0.6922626914826772,
      "grad_norm": 0.45216530561447144,
      "learning_rate": 4.320322155414406e-06,
      "loss": 0.2845,
      "step": 38340
    },
    {
      "epoch": 0.6923529709155779,
      "grad_norm": 0.575176477432251,
      "learning_rate": 4.317988012059537e-06,
      "loss": 0.1986,
      "step": 38345
    },
    {
      "epoch": 0.6924432503484786,
      "grad_norm": 0.49986568093299866,
      "learning_rate": 4.315654325781217e-06,
      "loss": 0.27,
      "step": 38350
    },
    {
      "epoch": 0.6925335297813793,
      "grad_norm": 0.6504265069961548,
      "learning_rate": 4.3133210967671765e-06,
      "loss": 0.2235,
      "step": 38355
    },
    {
      "epoch": 0.69262380921428,
      "grad_norm": 0.23776890337467194,
      "learning_rate": 4.310988325205109e-06,
      "loss": 0.211,
      "step": 38360
    },
    {
      "epoch": 0.6927140886471808,
      "grad_norm": 0.40864163637161255,
      "learning_rate": 4.308656011282669e-06,
      "loss": 0.2471,
      "step": 38365
    },
    {
      "epoch": 0.6928043680800815,
      "grad_norm": 0.3818129599094391,
      "learning_rate": 4.3063241551874715e-06,
      "loss": 0.2566,
      "step": 38370
    },
    {
      "epoch": 0.6928946475129821,
      "grad_norm": 0.27496612071990967,
      "learning_rate": 4.303992757107102e-06,
      "loss": 0.2344,
      "step": 38375
    },
    {
      "epoch": 0.6929849269458829,
      "grad_norm": 0.6847406029701233,
      "learning_rate": 4.301661817229096e-06,
      "loss": 0.2341,
      "step": 38380
    },
    {
      "epoch": 0.6930752063787836,
      "grad_norm": 0.2835419178009033,
      "learning_rate": 4.29933133574097e-06,
      "loss": 0.2065,
      "step": 38385
    },
    {
      "epoch": 0.6931654858116844,
      "grad_norm": 0.5512233972549438,
      "learning_rate": 4.297001312830188e-06,
      "loss": 0.2103,
      "step": 38390
    },
    {
      "epoch": 0.693255765244585,
      "grad_norm": 0.32876721024513245,
      "learning_rate": 4.294671748684183e-06,
      "loss": 0.2412,
      "step": 38395
    },
    {
      "epoch": 0.6933460446774857,
      "grad_norm": 0.36407238245010376,
      "learning_rate": 4.292342643490357e-06,
      "loss": 0.1719,
      "step": 38400
    },
    {
      "epoch": 0.6934363241103865,
      "grad_norm": 0.44209739565849304,
      "learning_rate": 4.290013997436069e-06,
      "loss": 0.2445,
      "step": 38405
    },
    {
      "epoch": 0.6935266035432872,
      "grad_norm": 0.6222474575042725,
      "learning_rate": 4.287685810708639e-06,
      "loss": 0.2812,
      "step": 38410
    },
    {
      "epoch": 0.6936168829761878,
      "grad_norm": 0.06723568588495255,
      "learning_rate": 4.28535808349535e-06,
      "loss": 0.173,
      "step": 38415
    },
    {
      "epoch": 0.6937071624090886,
      "grad_norm": 0.3727383613586426,
      "learning_rate": 4.2830308159834535e-06,
      "loss": 0.2345,
      "step": 38420
    },
    {
      "epoch": 0.6937974418419893,
      "grad_norm": 0.36739301681518555,
      "learning_rate": 4.280704008360161e-06,
      "loss": 0.2195,
      "step": 38425
    },
    {
      "epoch": 0.6938877212748901,
      "grad_norm": 0.524774968624115,
      "learning_rate": 4.278377660812651e-06,
      "loss": 0.2525,
      "step": 38430
    },
    {
      "epoch": 0.6939780007077907,
      "grad_norm": 0.5487887859344482,
      "learning_rate": 4.276051773528056e-06,
      "loss": 0.2036,
      "step": 38435
    },
    {
      "epoch": 0.6940682801406914,
      "grad_norm": 0.4065292477607727,
      "learning_rate": 4.273726346693481e-06,
      "loss": 0.3146,
      "step": 38440
    },
    {
      "epoch": 0.6941585595735922,
      "grad_norm": 0.3451765179634094,
      "learning_rate": 4.271401380495984e-06,
      "loss": 0.2192,
      "step": 38445
    },
    {
      "epoch": 0.6942488390064929,
      "grad_norm": 0.456271231174469,
      "learning_rate": 4.269076875122598e-06,
      "loss": 0.2119,
      "step": 38450
    },
    {
      "epoch": 0.6943391184393937,
      "grad_norm": 0.45741716027259827,
      "learning_rate": 4.266752830760307e-06,
      "loss": 0.2886,
      "step": 38455
    },
    {
      "epoch": 0.6944293978722943,
      "grad_norm": 0.5923446416854858,
      "learning_rate": 4.264429247596066e-06,
      "loss": 0.2288,
      "step": 38460
    },
    {
      "epoch": 0.694519677305195,
      "grad_norm": 0.4566524922847748,
      "learning_rate": 4.262106125816789e-06,
      "loss": 0.299,
      "step": 38465
    },
    {
      "epoch": 0.6946099567380958,
      "grad_norm": 0.6255629658699036,
      "learning_rate": 4.259783465609361e-06,
      "loss": 0.1874,
      "step": 38470
    },
    {
      "epoch": 0.6947002361709965,
      "grad_norm": 0.5031535625457764,
      "learning_rate": 4.257461267160614e-06,
      "loss": 0.2611,
      "step": 38475
    },
    {
      "epoch": 0.6947905156038972,
      "grad_norm": 1.097783088684082,
      "learning_rate": 4.255139530657359e-06,
      "loss": 0.1741,
      "step": 38480
    },
    {
      "epoch": 0.6948807950367979,
      "grad_norm": 0.5664388537406921,
      "learning_rate": 4.252818256286355e-06,
      "loss": 0.3781,
      "step": 38485
    },
    {
      "epoch": 0.6949710744696986,
      "grad_norm": 0.7404657006263733,
      "learning_rate": 4.2504974442343374e-06,
      "loss": 0.2243,
      "step": 38490
    },
    {
      "epoch": 0.6950613539025994,
      "grad_norm": 0.37026259303092957,
      "learning_rate": 4.248177094688e-06,
      "loss": 0.1814,
      "step": 38495
    },
    {
      "epoch": 0.6951516333355,
      "grad_norm": 1.072288155555725,
      "learning_rate": 4.2458572078339935e-06,
      "loss": 0.256,
      "step": 38500
    },
    {
      "epoch": 0.6952419127684007,
      "grad_norm": 0.369281142950058,
      "learning_rate": 4.243537783858941e-06,
      "loss": 0.1408,
      "step": 38505
    },
    {
      "epoch": 0.6953321922013015,
      "grad_norm": 0.35067546367645264,
      "learning_rate": 4.241218822949416e-06,
      "loss": 0.2327,
      "step": 38510
    },
    {
      "epoch": 0.6954224716342022,
      "grad_norm": 0.5609583258628845,
      "learning_rate": 4.238900325291971e-06,
      "loss": 0.2028,
      "step": 38515
    },
    {
      "epoch": 0.6955127510671029,
      "grad_norm": 0.2797073721885681,
      "learning_rate": 4.236582291073104e-06,
      "loss": 0.2367,
      "step": 38520
    },
    {
      "epoch": 0.6956030305000036,
      "grad_norm": 0.3344927728176117,
      "learning_rate": 4.234264720479286e-06,
      "loss": 0.2538,
      "step": 38525
    },
    {
      "epoch": 0.6956933099329043,
      "grad_norm": 0.7331802845001221,
      "learning_rate": 4.231947613696951e-06,
      "loss": 0.1726,
      "step": 38530
    },
    {
      "epoch": 0.6957835893658051,
      "grad_norm": 0.6007274389266968,
      "learning_rate": 4.229630970912497e-06,
      "loss": 0.2451,
      "step": 38535
    },
    {
      "epoch": 0.6958738687987057,
      "grad_norm": 0.32419684529304504,
      "learning_rate": 4.227314792312273e-06,
      "loss": 0.267,
      "step": 38540
    },
    {
      "epoch": 0.6959641482316065,
      "grad_norm": 0.6841797828674316,
      "learning_rate": 4.224999078082606e-06,
      "loss": 0.2077,
      "step": 38545
    },
    {
      "epoch": 0.6960544276645072,
      "grad_norm": 0.37495937943458557,
      "learning_rate": 4.2226838284097704e-06,
      "loss": 0.2689,
      "step": 38550
    },
    {
      "epoch": 0.6961447070974079,
      "grad_norm": 0.4198375642299652,
      "learning_rate": 4.220369043480016e-06,
      "loss": 0.1817,
      "step": 38555
    },
    {
      "epoch": 0.6962349865303086,
      "grad_norm": 0.26349231600761414,
      "learning_rate": 4.21805472347955e-06,
      "loss": 0.1551,
      "step": 38560
    },
    {
      "epoch": 0.6963252659632093,
      "grad_norm": 0.4598008096218109,
      "learning_rate": 4.2157408685945466e-06,
      "loss": 0.2686,
      "step": 38565
    },
    {
      "epoch": 0.6964155453961101,
      "grad_norm": 0.2681581377983093,
      "learning_rate": 4.213427479011132e-06,
      "loss": 0.2999,
      "step": 38570
    },
    {
      "epoch": 0.6965058248290108,
      "grad_norm": 0.556578516960144,
      "learning_rate": 4.211114554915407e-06,
      "loss": 0.2522,
      "step": 38575
    },
    {
      "epoch": 0.6965961042619114,
      "grad_norm": 0.4224514067173004,
      "learning_rate": 4.208802096493423e-06,
      "loss": 0.2756,
      "step": 38580
    },
    {
      "epoch": 0.6966863836948122,
      "grad_norm": 0.4554044306278229,
      "learning_rate": 4.206490103931208e-06,
      "loss": 0.29,
      "step": 38585
    },
    {
      "epoch": 0.6967766631277129,
      "grad_norm": 0.6609202027320862,
      "learning_rate": 4.204178577414737e-06,
      "loss": 0.2144,
      "step": 38590
    },
    {
      "epoch": 0.6968669425606137,
      "grad_norm": 0.4529894292354584,
      "learning_rate": 4.2018675171299604e-06,
      "loss": 0.1861,
      "step": 38595
    },
    {
      "epoch": 0.6969572219935143,
      "grad_norm": 0.32098129391670227,
      "learning_rate": 4.1995569232627895e-06,
      "loss": 0.2179,
      "step": 38600
    },
    {
      "epoch": 0.697047501426415,
      "grad_norm": 0.5516554713249207,
      "learning_rate": 4.197246795999087e-06,
      "loss": 0.2387,
      "step": 38605
    },
    {
      "epoch": 0.6971377808593158,
      "grad_norm": 0.4384140968322754,
      "learning_rate": 4.194937135524694e-06,
      "loss": 0.1449,
      "step": 38610
    },
    {
      "epoch": 0.6972280602922165,
      "grad_norm": 0.5471542477607727,
      "learning_rate": 4.192627942025397e-06,
      "loss": 0.3431,
      "step": 38615
    },
    {
      "epoch": 0.6973183397251171,
      "grad_norm": 0.340838223695755,
      "learning_rate": 4.1903192156869576e-06,
      "loss": 0.2758,
      "step": 38620
    },
    {
      "epoch": 0.6974086191580179,
      "grad_norm": 0.4325224757194519,
      "learning_rate": 4.188010956695098e-06,
      "loss": 0.2174,
      "step": 38625
    },
    {
      "epoch": 0.6974988985909186,
      "grad_norm": 0.3328706920146942,
      "learning_rate": 4.185703165235503e-06,
      "loss": 0.2745,
      "step": 38630
    },
    {
      "epoch": 0.6975891780238194,
      "grad_norm": 0.394483745098114,
      "learning_rate": 4.183395841493811e-06,
      "loss": 0.2074,
      "step": 38635
    },
    {
      "epoch": 0.69767945745672,
      "grad_norm": 0.34542450308799744,
      "learning_rate": 4.1810889856556355e-06,
      "loss": 0.1403,
      "step": 38640
    },
    {
      "epoch": 0.6977697368896207,
      "grad_norm": 0.26991739869117737,
      "learning_rate": 4.178782597906541e-06,
      "loss": 0.174,
      "step": 38645
    },
    {
      "epoch": 0.6978600163225215,
      "grad_norm": 0.38894426822662354,
      "learning_rate": 4.176476678432066e-06,
      "loss": 0.2506,
      "step": 38650
    },
    {
      "epoch": 0.6979502957554222,
      "grad_norm": 0.5923386216163635,
      "learning_rate": 4.174171227417696e-06,
      "loss": 0.2821,
      "step": 38655
    },
    {
      "epoch": 0.6980405751883229,
      "grad_norm": 0.3713725507259369,
      "learning_rate": 4.171866245048894e-06,
      "loss": 0.1851,
      "step": 38660
    },
    {
      "epoch": 0.6981308546212236,
      "grad_norm": 0.4045335054397583,
      "learning_rate": 4.169561731511078e-06,
      "loss": 0.2421,
      "step": 38665
    },
    {
      "epoch": 0.6982211340541243,
      "grad_norm": 0.2156153917312622,
      "learning_rate": 4.167257686989632e-06,
      "loss": 0.2216,
      "step": 38670
    },
    {
      "epoch": 0.6983114134870251,
      "grad_norm": 0.35402876138687134,
      "learning_rate": 4.164954111669895e-06,
      "loss": 0.324,
      "step": 38675
    },
    {
      "epoch": 0.6984016929199257,
      "grad_norm": 0.639041543006897,
      "learning_rate": 4.162651005737177e-06,
      "loss": 0.1975,
      "step": 38680
    },
    {
      "epoch": 0.6984919723528265,
      "grad_norm": 0.38943997025489807,
      "learning_rate": 4.16034836937674e-06,
      "loss": 0.2254,
      "step": 38685
    },
    {
      "epoch": 0.6985822517857272,
      "grad_norm": 0.44801610708236694,
      "learning_rate": 4.158046202773819e-06,
      "loss": 0.2389,
      "step": 38690
    },
    {
      "epoch": 0.6986725312186279,
      "grad_norm": 0.5811364054679871,
      "learning_rate": 4.15574450611361e-06,
      "loss": 0.1561,
      "step": 38695
    },
    {
      "epoch": 0.6987628106515286,
      "grad_norm": 0.22529548406600952,
      "learning_rate": 4.153443279581259e-06,
      "loss": 0.2403,
      "step": 38700
    },
    {
      "epoch": 0.6988530900844293,
      "grad_norm": 0.9179227352142334,
      "learning_rate": 4.151142523361892e-06,
      "loss": 0.1893,
      "step": 38705
    },
    {
      "epoch": 0.69894336951733,
      "grad_norm": 0.6835107207298279,
      "learning_rate": 4.14884223764058e-06,
      "loss": 0.2053,
      "step": 38710
    },
    {
      "epoch": 0.6990336489502308,
      "grad_norm": 0.18289843201637268,
      "learning_rate": 4.146542422602372e-06,
      "loss": 0.1904,
      "step": 38715
    },
    {
      "epoch": 0.6991239283831314,
      "grad_norm": 1.0548094511032104,
      "learning_rate": 4.144243078432263e-06,
      "loss": 0.2543,
      "step": 38720
    },
    {
      "epoch": 0.6992142078160322,
      "grad_norm": 0.4621219336986542,
      "learning_rate": 4.141944205315223e-06,
      "loss": 0.1281,
      "step": 38725
    },
    {
      "epoch": 0.6993044872489329,
      "grad_norm": 0.3635997474193573,
      "learning_rate": 4.13964580343618e-06,
      "loss": 0.346,
      "step": 38730
    },
    {
      "epoch": 0.6993947666818336,
      "grad_norm": 0.5849667191505432,
      "learning_rate": 4.137347872980026e-06,
      "loss": 0.2862,
      "step": 38735
    },
    {
      "epoch": 0.6994850461147343,
      "grad_norm": 0.32722166180610657,
      "learning_rate": 4.135050414131606e-06,
      "loss": 0.2472,
      "step": 38740
    },
    {
      "epoch": 0.699575325547635,
      "grad_norm": 0.39641356468200684,
      "learning_rate": 4.132753427075743e-06,
      "loss": 0.1414,
      "step": 38745
    },
    {
      "epoch": 0.6996656049805358,
      "grad_norm": 0.8139769434928894,
      "learning_rate": 4.130456911997202e-06,
      "loss": 0.3283,
      "step": 38750
    },
    {
      "epoch": 0.6997558844134365,
      "grad_norm": 0.7956682443618774,
      "learning_rate": 4.12816086908073e-06,
      "loss": 0.1844,
      "step": 38755
    },
    {
      "epoch": 0.6998461638463371,
      "grad_norm": 0.4433048963546753,
      "learning_rate": 4.125865298511022e-06,
      "loss": 0.391,
      "step": 38760
    },
    {
      "epoch": 0.6999364432792379,
      "grad_norm": 0.4708251655101776,
      "learning_rate": 4.1235702004727444e-06,
      "loss": 0.2378,
      "step": 38765
    },
    {
      "epoch": 0.7000267227121386,
      "grad_norm": 0.6387839317321777,
      "learning_rate": 4.1212755751505184e-06,
      "loss": 0.2308,
      "step": 38770
    },
    {
      "epoch": 0.7001170021450394,
      "grad_norm": 0.4458686113357544,
      "learning_rate": 4.118981422728927e-06,
      "loss": 0.2072,
      "step": 38775
    },
    {
      "epoch": 0.70020728157794,
      "grad_norm": 0.6436731219291687,
      "learning_rate": 4.116687743392524e-06,
      "loss": 0.3366,
      "step": 38780
    },
    {
      "epoch": 0.7002975610108407,
      "grad_norm": 0.5403485894203186,
      "learning_rate": 4.1143945373258145e-06,
      "loss": 0.1223,
      "step": 38785
    },
    {
      "epoch": 0.7003878404437415,
      "grad_norm": 0.49621668457984924,
      "learning_rate": 4.11210180471327e-06,
      "loss": 0.2061,
      "step": 38790
    },
    {
      "epoch": 0.7004781198766422,
      "grad_norm": 0.2502276301383972,
      "learning_rate": 4.109809545739327e-06,
      "loss": 0.3227,
      "step": 38795
    },
    {
      "epoch": 0.7005683993095428,
      "grad_norm": 0.34249821305274963,
      "learning_rate": 4.107517760588384e-06,
      "loss": 0.2294,
      "step": 38800
    },
    {
      "epoch": 0.7006586787424436,
      "grad_norm": 0.3874669671058655,
      "learning_rate": 4.10522644944479e-06,
      "loss": 0.3392,
      "step": 38805
    },
    {
      "epoch": 0.7007489581753443,
      "grad_norm": 0.48335301876068115,
      "learning_rate": 4.102935612492874e-06,
      "loss": 0.2238,
      "step": 38810
    },
    {
      "epoch": 0.7008392376082451,
      "grad_norm": 0.12804335355758667,
      "learning_rate": 4.100645249916908e-06,
      "loss": 0.1596,
      "step": 38815
    },
    {
      "epoch": 0.7009295170411457,
      "grad_norm": 0.4075047969818115,
      "learning_rate": 4.0983553619011395e-06,
      "loss": 0.1921,
      "step": 38820
    },
    {
      "epoch": 0.7010197964740464,
      "grad_norm": 0.7569358348846436,
      "learning_rate": 4.096065948629773e-06,
      "loss": 0.3234,
      "step": 38825
    },
    {
      "epoch": 0.7011100759069472,
      "grad_norm": 0.375936359167099,
      "learning_rate": 4.093777010286978e-06,
      "loss": 0.3345,
      "step": 38830
    },
    {
      "epoch": 0.7012003553398479,
      "grad_norm": 0.6364753842353821,
      "learning_rate": 4.091488547056876e-06,
      "loss": 0.2591,
      "step": 38835
    },
    {
      "epoch": 0.7012906347727487,
      "grad_norm": 0.3923836648464203,
      "learning_rate": 4.089200559123566e-06,
      "loss": 0.1587,
      "step": 38840
    },
    {
      "epoch": 0.7013809142056493,
      "grad_norm": 0.4723087251186371,
      "learning_rate": 4.086913046671093e-06,
      "loss": 0.1805,
      "step": 38845
    },
    {
      "epoch": 0.70147119363855,
      "grad_norm": 0.41670605540275574,
      "learning_rate": 4.0846260098834745e-06,
      "loss": 0.1714,
      "step": 38850
    },
    {
      "epoch": 0.7015614730714508,
      "grad_norm": 0.17346037924289703,
      "learning_rate": 4.082339448944682e-06,
      "loss": 0.2007,
      "step": 38855
    },
    {
      "epoch": 0.7016517525043515,
      "grad_norm": 0.3650917410850525,
      "learning_rate": 4.080053364038655e-06,
      "loss": 0.2103,
      "step": 38860
    },
    {
      "epoch": 0.7017420319372522,
      "grad_norm": 0.33513301610946655,
      "learning_rate": 4.077767755349297e-06,
      "loss": 0.3319,
      "step": 38865
    },
    {
      "epoch": 0.7018323113701529,
      "grad_norm": 0.3749558627605438,
      "learning_rate": 4.075482623060459e-06,
      "loss": 0.2826,
      "step": 38870
    },
    {
      "epoch": 0.7019225908030536,
      "grad_norm": 0.3748433589935303,
      "learning_rate": 4.073197967355974e-06,
      "loss": 0.2341,
      "step": 38875
    },
    {
      "epoch": 0.7020128702359544,
      "grad_norm": 0.3942178189754486,
      "learning_rate": 4.070913788419616e-06,
      "loss": 0.2216,
      "step": 38880
    },
    {
      "epoch": 0.702103149668855,
      "grad_norm": 1.1604423522949219,
      "learning_rate": 4.068630086435136e-06,
      "loss": 0.2147,
      "step": 38885
    },
    {
      "epoch": 0.7021934291017558,
      "grad_norm": 0.6453258991241455,
      "learning_rate": 4.066346861586239e-06,
      "loss": 0.2069,
      "step": 38890
    },
    {
      "epoch": 0.7022837085346565,
      "grad_norm": 0.4149574637413025,
      "learning_rate": 4.064064114056599e-06,
      "loss": 0.1968,
      "step": 38895
    },
    {
      "epoch": 0.7023739879675572,
      "grad_norm": 0.23456250131130219,
      "learning_rate": 4.0617818440298406e-06,
      "loss": 0.246,
      "step": 38900
    },
    {
      "epoch": 0.7024642674004579,
      "grad_norm": 0.20911572873592377,
      "learning_rate": 4.05950005168956e-06,
      "loss": 0.169,
      "step": 38905
    },
    {
      "epoch": 0.7025545468333586,
      "grad_norm": 0.36427533626556396,
      "learning_rate": 4.057218737219305e-06,
      "loss": 0.191,
      "step": 38910
    },
    {
      "epoch": 0.7026448262662593,
      "grad_norm": 0.47539886832237244,
      "learning_rate": 4.054937900802598e-06,
      "loss": 0.1408,
      "step": 38915
    },
    {
      "epoch": 0.7027351056991601,
      "grad_norm": 0.3398011326789856,
      "learning_rate": 4.0526575426229086e-06,
      "loss": 0.1917,
      "step": 38920
    },
    {
      "epoch": 0.7028253851320607,
      "grad_norm": 0.3850838840007782,
      "learning_rate": 4.0503776628636796e-06,
      "loss": 0.1996,
      "step": 38925
    },
    {
      "epoch": 0.7029156645649615,
      "grad_norm": 0.49528321623802185,
      "learning_rate": 4.048098261708309e-06,
      "loss": 0.2035,
      "step": 38930
    },
    {
      "epoch": 0.7030059439978622,
      "grad_norm": 0.5522703528404236,
      "learning_rate": 4.045819339340161e-06,
      "loss": 0.2571,
      "step": 38935
    },
    {
      "epoch": 0.703096223430763,
      "grad_norm": 0.5673417448997498,
      "learning_rate": 4.043540895942554e-06,
      "loss": 0.292,
      "step": 38940
    },
    {
      "epoch": 0.7031865028636636,
      "grad_norm": 0.2567455470561981,
      "learning_rate": 4.041262931698778e-06,
      "loss": 0.2587,
      "step": 38945
    },
    {
      "epoch": 0.7032767822965643,
      "grad_norm": 0.20778410136699677,
      "learning_rate": 4.0389854467920695e-06,
      "loss": 0.258,
      "step": 38950
    },
    {
      "epoch": 0.7033670617294651,
      "grad_norm": 0.3900172710418701,
      "learning_rate": 4.0367084414056435e-06,
      "loss": 0.1686,
      "step": 38955
    },
    {
      "epoch": 0.7034573411623658,
      "grad_norm": 0.5585681200027466,
      "learning_rate": 4.034431915722669e-06,
      "loss": 0.2299,
      "step": 38960
    },
    {
      "epoch": 0.7035476205952664,
      "grad_norm": 0.2830726206302643,
      "learning_rate": 4.032155869926268e-06,
      "loss": 0.2741,
      "step": 38965
    },
    {
      "epoch": 0.7036379000281672,
      "grad_norm": 0.4869314432144165,
      "learning_rate": 4.029880304199541e-06,
      "loss": 0.2591,
      "step": 38970
    },
    {
      "epoch": 0.7037281794610679,
      "grad_norm": 0.4794871211051941,
      "learning_rate": 4.027605218725533e-06,
      "loss": 0.247,
      "step": 38975
    },
    {
      "epoch": 0.7038184588939687,
      "grad_norm": 0.6267117857933044,
      "learning_rate": 4.025330613687266e-06,
      "loss": 0.2555,
      "step": 38980
    },
    {
      "epoch": 0.7039087383268693,
      "grad_norm": 0.26632171869277954,
      "learning_rate": 4.023056489267707e-06,
      "loss": 0.2156,
      "step": 38985
    },
    {
      "epoch": 0.70399901775977,
      "grad_norm": 0.33814334869384766,
      "learning_rate": 4.020782845649798e-06,
      "loss": 0.1978,
      "step": 38990
    },
    {
      "epoch": 0.7040892971926708,
      "grad_norm": 0.21557562053203583,
      "learning_rate": 4.0185096830164335e-06,
      "loss": 0.1851,
      "step": 38995
    },
    {
      "epoch": 0.7041795766255715,
      "grad_norm": 0.5119245052337646,
      "learning_rate": 4.01623700155048e-06,
      "loss": 0.24,
      "step": 39000
    },
    {
      "epoch": 0.7042698560584721,
      "grad_norm": 0.2194567173719406,
      "learning_rate": 4.013964801434751e-06,
      "loss": 0.1603,
      "step": 39005
    },
    {
      "epoch": 0.7043601354913729,
      "grad_norm": 0.6126335859298706,
      "learning_rate": 4.011693082852034e-06,
      "loss": 0.2107,
      "step": 39010
    },
    {
      "epoch": 0.7044504149242736,
      "grad_norm": 0.5091403126716614,
      "learning_rate": 4.009421845985065e-06,
      "loss": 0.1837,
      "step": 39015
    },
    {
      "epoch": 0.7045406943571744,
      "grad_norm": 0.4852842092514038,
      "learning_rate": 4.0071510910165545e-06,
      "loss": 0.2375,
      "step": 39020
    },
    {
      "epoch": 0.704630973790075,
      "grad_norm": 0.5118070244789124,
      "learning_rate": 4.004880818129166e-06,
      "loss": 0.1604,
      "step": 39025
    },
    {
      "epoch": 0.7047212532229757,
      "grad_norm": 0.6322970986366272,
      "learning_rate": 4.002611027505529e-06,
      "loss": 0.286,
      "step": 39030
    },
    {
      "epoch": 0.7048115326558765,
      "grad_norm": 0.3018209934234619,
      "learning_rate": 4.000341719328227e-06,
      "loss": 0.2361,
      "step": 39035
    },
    {
      "epoch": 0.7049018120887772,
      "grad_norm": 0.5549687147140503,
      "learning_rate": 3.9980728937798155e-06,
      "loss": 0.224,
      "step": 39040
    },
    {
      "epoch": 0.7049920915216779,
      "grad_norm": 0.34325161576271057,
      "learning_rate": 3.995804551042798e-06,
      "loss": 0.2376,
      "step": 39045
    },
    {
      "epoch": 0.7050823709545786,
      "grad_norm": 0.31609299778938293,
      "learning_rate": 3.993536691299653e-06,
      "loss": 0.2662,
      "step": 39050
    },
    {
      "epoch": 0.7051726503874793,
      "grad_norm": 0.4124249815940857,
      "learning_rate": 3.991269314732806e-06,
      "loss": 0.217,
      "step": 39055
    },
    {
      "epoch": 0.7052629298203801,
      "grad_norm": 0.27731990814208984,
      "learning_rate": 3.989002421524656e-06,
      "loss": 0.2298,
      "step": 39060
    },
    {
      "epoch": 0.7053532092532807,
      "grad_norm": 0.6760087013244629,
      "learning_rate": 3.9867360118575605e-06,
      "loss": 0.3089,
      "step": 39065
    },
    {
      "epoch": 0.7054434886861815,
      "grad_norm": 0.6529771685600281,
      "learning_rate": 3.984470085913828e-06,
      "loss": 0.3373,
      "step": 39070
    },
    {
      "epoch": 0.7055337681190822,
      "grad_norm": 0.3386461138725281,
      "learning_rate": 3.982204643875744e-06,
      "loss": 0.2443,
      "step": 39075
    },
    {
      "epoch": 0.7056240475519829,
      "grad_norm": 0.22765114903450012,
      "learning_rate": 3.979939685925538e-06,
      "loss": 0.219,
      "step": 39080
    },
    {
      "epoch": 0.7057143269848836,
      "grad_norm": 0.307463675737381,
      "learning_rate": 3.9776752122454154e-06,
      "loss": 0.2476,
      "step": 39085
    },
    {
      "epoch": 0.7058046064177843,
      "grad_norm": 0.7050151228904724,
      "learning_rate": 3.975411223017536e-06,
      "loss": 0.2473,
      "step": 39090
    },
    {
      "epoch": 0.705894885850685,
      "grad_norm": 0.636820912361145,
      "learning_rate": 3.973147718424023e-06,
      "loss": 0.1865,
      "step": 39095
    },
    {
      "epoch": 0.7059851652835858,
      "grad_norm": 0.4979107975959778,
      "learning_rate": 3.970884698646953e-06,
      "loss": 0.1859,
      "step": 39100
    },
    {
      "epoch": 0.7060754447164864,
      "grad_norm": 0.2656373381614685,
      "learning_rate": 3.968622163868378e-06,
      "loss": 0.1779,
      "step": 39105
    },
    {
      "epoch": 0.7061657241493872,
      "grad_norm": 0.43536001443862915,
      "learning_rate": 3.966360114270292e-06,
      "loss": 0.2214,
      "step": 39110
    },
    {
      "epoch": 0.7062560035822879,
      "grad_norm": 0.49916452169418335,
      "learning_rate": 3.964098550034671e-06,
      "loss": 0.1997,
      "step": 39115
    },
    {
      "epoch": 0.7063462830151886,
      "grad_norm": 0.29856258630752563,
      "learning_rate": 3.9618374713434295e-06,
      "loss": 0.1341,
      "step": 39120
    },
    {
      "epoch": 0.7064365624480893,
      "grad_norm": 0.35987964272499084,
      "learning_rate": 3.959576878378468e-06,
      "loss": 0.2526,
      "step": 39125
    },
    {
      "epoch": 0.70652684188099,
      "grad_norm": 0.25284314155578613,
      "learning_rate": 3.957316771321625e-06,
      "loss": 0.1639,
      "step": 39130
    },
    {
      "epoch": 0.7066171213138908,
      "grad_norm": 0.536403238773346,
      "learning_rate": 3.955057150354717e-06,
      "loss": 0.2413,
      "step": 39135
    },
    {
      "epoch": 0.7067074007467915,
      "grad_norm": 0.4459882378578186,
      "learning_rate": 3.952798015659511e-06,
      "loss": 0.281,
      "step": 39140
    },
    {
      "epoch": 0.7067976801796921,
      "grad_norm": 0.4276878237724304,
      "learning_rate": 3.950539367417733e-06,
      "loss": 0.2388,
      "step": 39145
    },
    {
      "epoch": 0.7068879596125929,
      "grad_norm": 0.3131605386734009,
      "learning_rate": 3.948281205811078e-06,
      "loss": 0.1389,
      "step": 39150
    },
    {
      "epoch": 0.7069782390454936,
      "grad_norm": 0.5553649067878723,
      "learning_rate": 3.946023531021201e-06,
      "loss": 0.2079,
      "step": 39155
    },
    {
      "epoch": 0.7070685184783944,
      "grad_norm": 0.7444565892219543,
      "learning_rate": 3.943766343229718e-06,
      "loss": 0.2878,
      "step": 39160
    },
    {
      "epoch": 0.707158797911295,
      "grad_norm": 0.6670951843261719,
      "learning_rate": 3.9415096426181955e-06,
      "loss": 0.1765,
      "step": 39165
    },
    {
      "epoch": 0.7072490773441957,
      "grad_norm": 0.5706706643104553,
      "learning_rate": 3.939253429368177e-06,
      "loss": 0.1805,
      "step": 39170
    },
    {
      "epoch": 0.7073393567770965,
      "grad_norm": 0.34905120730400085,
      "learning_rate": 3.9369977036611494e-06,
      "loss": 0.1743,
      "step": 39175
    },
    {
      "epoch": 0.7074296362099972,
      "grad_norm": 0.5026421546936035,
      "learning_rate": 3.934742465678579e-06,
      "loss": 0.2911,
      "step": 39180
    },
    {
      "epoch": 0.7075199156428978,
      "grad_norm": 0.44103628396987915,
      "learning_rate": 3.932487715601875e-06,
      "loss": 0.2584,
      "step": 39185
    },
    {
      "epoch": 0.7076101950757986,
      "grad_norm": 0.16476301848888397,
      "learning_rate": 3.9302334536124186e-06,
      "loss": 0.25,
      "step": 39190
    },
    {
      "epoch": 0.7077004745086993,
      "grad_norm": 0.5779272317886353,
      "learning_rate": 3.92797967989155e-06,
      "loss": 0.1826,
      "step": 39195
    },
    {
      "epoch": 0.7077907539416001,
      "grad_norm": 0.3234092593193054,
      "learning_rate": 3.92572639462057e-06,
      "loss": 0.2257,
      "step": 39200
    },
    {
      "epoch": 0.7078810333745007,
      "grad_norm": 0.40954142808914185,
      "learning_rate": 3.923473597980736e-06,
      "loss": 0.2394,
      "step": 39205
    },
    {
      "epoch": 0.7079713128074014,
      "grad_norm": 0.41970279812812805,
      "learning_rate": 3.921221290153273e-06,
      "loss": 0.1453,
      "step": 39210
    },
    {
      "epoch": 0.7080615922403022,
      "grad_norm": 0.5298666954040527,
      "learning_rate": 3.918969471319357e-06,
      "loss": 0.2304,
      "step": 39215
    },
    {
      "epoch": 0.7081518716732029,
      "grad_norm": 0.40422743558883667,
      "learning_rate": 3.916718141660133e-06,
      "loss": 0.28,
      "step": 39220
    },
    {
      "epoch": 0.7082421511061036,
      "grad_norm": 0.517111599445343,
      "learning_rate": 3.914467301356706e-06,
      "loss": 0.2784,
      "step": 39225
    },
    {
      "epoch": 0.7083324305390043,
      "grad_norm": 0.36019113659858704,
      "learning_rate": 3.912216950590141e-06,
      "loss": 0.1932,
      "step": 39230
    },
    {
      "epoch": 0.708422709971905,
      "grad_norm": 0.46564367413520813,
      "learning_rate": 3.90996708954146e-06,
      "loss": 0.1829,
      "step": 39235
    },
    {
      "epoch": 0.7085129894048058,
      "grad_norm": 0.3894612193107605,
      "learning_rate": 3.907717718391645e-06,
      "loss": 0.159,
      "step": 39240
    },
    {
      "epoch": 0.7086032688377065,
      "grad_norm": 0.4303584098815918,
      "learning_rate": 3.905468837321648e-06,
      "loss": 0.261,
      "step": 39245
    },
    {
      "epoch": 0.7086935482706072,
      "grad_norm": 0.5752771496772766,
      "learning_rate": 3.903220446512367e-06,
      "loss": 0.2723,
      "step": 39250
    },
    {
      "epoch": 0.7087838277035079,
      "grad_norm": 0.5472204089164734,
      "learning_rate": 3.900972546144674e-06,
      "loss": 0.1747,
      "step": 39255
    },
    {
      "epoch": 0.7088741071364086,
      "grad_norm": 0.5496330261230469,
      "learning_rate": 3.898725136399394e-06,
      "loss": 0.2243,
      "step": 39260
    },
    {
      "epoch": 0.7089643865693094,
      "grad_norm": 0.8196171522140503,
      "learning_rate": 3.8964782174573215e-06,
      "loss": 0.2567,
      "step": 39265
    },
    {
      "epoch": 0.70905466600221,
      "grad_norm": 0.5102502107620239,
      "learning_rate": 3.8942317894991946e-06,
      "loss": 0.2294,
      "step": 39270
    },
    {
      "epoch": 0.7091449454351108,
      "grad_norm": 0.3395414352416992,
      "learning_rate": 3.891985852705731e-06,
      "loss": 0.1976,
      "step": 39275
    },
    {
      "epoch": 0.7092352248680115,
      "grad_norm": 0.5208386778831482,
      "learning_rate": 3.889740407257593e-06,
      "loss": 0.2594,
      "step": 39280
    },
    {
      "epoch": 0.7093255043009122,
      "grad_norm": 0.6493802666664124,
      "learning_rate": 3.887495453335412e-06,
      "loss": 0.2917,
      "step": 39285
    },
    {
      "epoch": 0.7094157837338129,
      "grad_norm": 0.864579439163208,
      "learning_rate": 3.885250991119781e-06,
      "loss": 0.2464,
      "step": 39290
    },
    {
      "epoch": 0.7095060631667136,
      "grad_norm": 0.7394172549247742,
      "learning_rate": 3.883007020791253e-06,
      "loss": 0.1616,
      "step": 39295
    },
    {
      "epoch": 0.7095963425996143,
      "grad_norm": 0.341521292924881,
      "learning_rate": 3.880763542530331e-06,
      "loss": 0.135,
      "step": 39300
    },
    {
      "epoch": 0.7096866220325151,
      "grad_norm": 0.31069761514663696,
      "learning_rate": 3.878520556517495e-06,
      "loss": 0.1188,
      "step": 39305
    },
    {
      "epoch": 0.7097769014654157,
      "grad_norm": 0.2305673062801361,
      "learning_rate": 3.8762780629331685e-06,
      "loss": 0.1908,
      "step": 39310
    },
    {
      "epoch": 0.7098671808983165,
      "grad_norm": 0.6119630336761475,
      "learning_rate": 3.874036061957752e-06,
      "loss": 0.2356,
      "step": 39315
    },
    {
      "epoch": 0.7099574603312172,
      "grad_norm": 0.3898882567882538,
      "learning_rate": 3.871794553771588e-06,
      "loss": 0.2726,
      "step": 39320
    },
    {
      "epoch": 0.710047739764118,
      "grad_norm": 0.7064270973205566,
      "learning_rate": 3.8695535385550005e-06,
      "loss": 0.2798,
      "step": 39325
    },
    {
      "epoch": 0.7101380191970186,
      "grad_norm": 0.5756570100784302,
      "learning_rate": 3.86731301648826e-06,
      "loss": 0.2828,
      "step": 39330
    },
    {
      "epoch": 0.7102282986299193,
      "grad_norm": 0.35779905319213867,
      "learning_rate": 3.865072987751594e-06,
      "loss": 0.1907,
      "step": 39335
    },
    {
      "epoch": 0.7103185780628201,
      "grad_norm": 0.31914791464805603,
      "learning_rate": 3.862833452525206e-06,
      "loss": 0.2182,
      "step": 39340
    },
    {
      "epoch": 0.7104088574957208,
      "grad_norm": 0.7411733269691467,
      "learning_rate": 3.86059441098924e-06,
      "loss": 0.2162,
      "step": 39345
    },
    {
      "epoch": 0.7104991369286214,
      "grad_norm": 0.4457027018070221,
      "learning_rate": 3.858355863323816e-06,
      "loss": 0.2336,
      "step": 39350
    },
    {
      "epoch": 0.7105894163615222,
      "grad_norm": 0.4871555268764496,
      "learning_rate": 3.856117809709009e-06,
      "loss": 0.2216,
      "step": 39355
    },
    {
      "epoch": 0.7106796957944229,
      "grad_norm": 0.725056529045105,
      "learning_rate": 3.853880250324858e-06,
      "loss": 0.2181,
      "step": 39360
    },
    {
      "epoch": 0.7107699752273237,
      "grad_norm": 0.24869728088378906,
      "learning_rate": 3.85164318535135e-06,
      "loss": 0.188,
      "step": 39365
    },
    {
      "epoch": 0.7108602546602243,
      "grad_norm": 0.34141483902931213,
      "learning_rate": 3.84940661496845e-06,
      "loss": 0.2649,
      "step": 39370
    },
    {
      "epoch": 0.710950534093125,
      "grad_norm": 0.6450366973876953,
      "learning_rate": 3.847170539356063e-06,
      "loss": 0.3282,
      "step": 39375
    },
    {
      "epoch": 0.7110408135260258,
      "grad_norm": 0.5957616567611694,
      "learning_rate": 3.844934958694075e-06,
      "loss": 0.2584,
      "step": 39380
    },
    {
      "epoch": 0.7111310929589265,
      "grad_norm": 0.5537654757499695,
      "learning_rate": 3.842699873162311e-06,
      "loss": 0.2418,
      "step": 39385
    },
    {
      "epoch": 0.7112213723918271,
      "grad_norm": 0.1669919788837433,
      "learning_rate": 3.840465282940581e-06,
      "loss": 0.1438,
      "step": 39390
    },
    {
      "epoch": 0.7113116518247279,
      "grad_norm": 0.2690705358982086,
      "learning_rate": 3.838231188208632e-06,
      "loss": 0.2292,
      "step": 39395
    },
    {
      "epoch": 0.7114019312576286,
      "grad_norm": 0.2518727779388428,
      "learning_rate": 3.835997589146187e-06,
      "loss": 0.1737,
      "step": 39400
    },
    {
      "epoch": 0.7114922106905294,
      "grad_norm": 0.5274950861930847,
      "learning_rate": 3.833764485932915e-06,
      "loss": 0.252,
      "step": 39405
    },
    {
      "epoch": 0.71158249012343,
      "grad_norm": 0.6367514133453369,
      "learning_rate": 3.8315318787484615e-06,
      "loss": 0.2363,
      "step": 39410
    },
    {
      "epoch": 0.7116727695563307,
      "grad_norm": 0.32732975482940674,
      "learning_rate": 3.829299767772416e-06,
      "loss": 0.226,
      "step": 39415
    },
    {
      "epoch": 0.7117630489892315,
      "grad_norm": 0.338902086019516,
      "learning_rate": 3.827068153184338e-06,
      "loss": 0.1735,
      "step": 39420
    },
    {
      "epoch": 0.7118533284221322,
      "grad_norm": 0.7373274564743042,
      "learning_rate": 3.824837035163748e-06,
      "loss": 0.2093,
      "step": 39425
    },
    {
      "epoch": 0.7119436078550329,
      "grad_norm": 0.36015719175338745,
      "learning_rate": 3.82260641389012e-06,
      "loss": 0.1947,
      "step": 39430
    },
    {
      "epoch": 0.7120338872879336,
      "grad_norm": 0.6400065422058105,
      "learning_rate": 3.820376289542893e-06,
      "loss": 0.143,
      "step": 39435
    },
    {
      "epoch": 0.7121241667208343,
      "grad_norm": 0.4007337987422943,
      "learning_rate": 3.8181466623014616e-06,
      "loss": 0.3211,
      "step": 39440
    },
    {
      "epoch": 0.7122144461537351,
      "grad_norm": 0.3060721457004547,
      "learning_rate": 3.815917532345188e-06,
      "loss": 0.1998,
      "step": 39445
    },
    {
      "epoch": 0.7123047255866357,
      "grad_norm": 0.5131520628929138,
      "learning_rate": 3.813688899853378e-06,
      "loss": 0.2637,
      "step": 39450
    },
    {
      "epoch": 0.7123950050195365,
      "grad_norm": 0.6646853089332581,
      "learning_rate": 3.811460765005326e-06,
      "loss": 0.336,
      "step": 39455
    },
    {
      "epoch": 0.7124852844524372,
      "grad_norm": 0.5253194570541382,
      "learning_rate": 3.8092331279802554e-06,
      "loss": 0.2323,
      "step": 39460
    },
    {
      "epoch": 0.7125755638853379,
      "grad_norm": 0.59714674949646,
      "learning_rate": 3.807005988957373e-06,
      "loss": 0.1693,
      "step": 39465
    },
    {
      "epoch": 0.7126658433182386,
      "grad_norm": 0.3074837028980255,
      "learning_rate": 3.8047793481158267e-06,
      "loss": 0.1404,
      "step": 39470
    },
    {
      "epoch": 0.7127561227511393,
      "grad_norm": 0.5420733094215393,
      "learning_rate": 3.8025532056347426e-06,
      "loss": 0.3196,
      "step": 39475
    },
    {
      "epoch": 0.71284640218404,
      "grad_norm": 0.423129677772522,
      "learning_rate": 3.80032756169319e-06,
      "loss": 0.248,
      "step": 39480
    },
    {
      "epoch": 0.7129366816169408,
      "grad_norm": 0.4089096486568451,
      "learning_rate": 3.7981024164702083e-06,
      "loss": 0.2264,
      "step": 39485
    },
    {
      "epoch": 0.7130269610498414,
      "grad_norm": 0.3753942847251892,
      "learning_rate": 3.7958777701447956e-06,
      "loss": 0.2404,
      "step": 39490
    },
    {
      "epoch": 0.7131172404827422,
      "grad_norm": 0.4123792052268982,
      "learning_rate": 3.793653622895912e-06,
      "loss": 0.2162,
      "step": 39495
    },
    {
      "epoch": 0.7132075199156429,
      "grad_norm": 0.4545404613018036,
      "learning_rate": 3.7914299749024652e-06,
      "loss": 0.2731,
      "step": 39500
    },
    {
      "epoch": 0.7132977993485436,
      "grad_norm": 0.5128106474876404,
      "learning_rate": 3.789206826343341e-06,
      "loss": 0.2263,
      "step": 39505
    },
    {
      "epoch": 0.7133880787814443,
      "grad_norm": 0.5913051962852478,
      "learning_rate": 3.7869841773973703e-06,
      "loss": 0.2951,
      "step": 39510
    },
    {
      "epoch": 0.713478358214345,
      "grad_norm": 0.429202675819397,
      "learning_rate": 3.7847620282433416e-06,
      "loss": 0.2091,
      "step": 39515
    },
    {
      "epoch": 0.7135686376472458,
      "grad_norm": 0.45054930448532104,
      "learning_rate": 3.782540379060027e-06,
      "loss": 0.2771,
      "step": 39520
    },
    {
      "epoch": 0.7136589170801465,
      "grad_norm": 0.29461824893951416,
      "learning_rate": 3.7803192300261294e-06,
      "loss": 0.238,
      "step": 39525
    },
    {
      "epoch": 0.7137491965130471,
      "grad_norm": 0.4052782952785492,
      "learning_rate": 3.778098581320332e-06,
      "loss": 0.2751,
      "step": 39530
    },
    {
      "epoch": 0.7138394759459479,
      "grad_norm": 0.4605269134044647,
      "learning_rate": 3.775878433121262e-06,
      "loss": 0.223,
      "step": 39535
    },
    {
      "epoch": 0.7139297553788486,
      "grad_norm": 0.5917302966117859,
      "learning_rate": 3.7736587856075224e-06,
      "loss": 0.2594,
      "step": 39540
    },
    {
      "epoch": 0.7140200348117494,
      "grad_norm": 0.4540460705757141,
      "learning_rate": 3.77143963895766e-06,
      "loss": 0.2928,
      "step": 39545
    },
    {
      "epoch": 0.71411031424465,
      "grad_norm": 0.8969790935516357,
      "learning_rate": 3.769220993350193e-06,
      "loss": 0.1789,
      "step": 39550
    },
    {
      "epoch": 0.7142005936775507,
      "grad_norm": 0.4711844027042389,
      "learning_rate": 3.7670028489635947e-06,
      "loss": 0.2453,
      "step": 39555
    },
    {
      "epoch": 0.7142908731104515,
      "grad_norm": 0.7896589040756226,
      "learning_rate": 3.7647852059763024e-06,
      "loss": 0.2987,
      "step": 39560
    },
    {
      "epoch": 0.7143811525433522,
      "grad_norm": 0.6333085298538208,
      "learning_rate": 3.762568064566702e-06,
      "loss": 0.295,
      "step": 39565
    },
    {
      "epoch": 0.7144714319762528,
      "grad_norm": 0.3821159303188324,
      "learning_rate": 3.7603514249131545e-06,
      "loss": 0.1993,
      "step": 39570
    },
    {
      "epoch": 0.7145617114091536,
      "grad_norm": 0.3645440340042114,
      "learning_rate": 3.7581352871939636e-06,
      "loss": 0.2361,
      "step": 39575
    },
    {
      "epoch": 0.7146519908420543,
      "grad_norm": 0.7931193113327026,
      "learning_rate": 3.7559196515874073e-06,
      "loss": 0.2554,
      "step": 39580
    },
    {
      "epoch": 0.7147422702749551,
      "grad_norm": 0.6085796356201172,
      "learning_rate": 3.7537045182717157e-06,
      "loss": 0.3057,
      "step": 39585
    },
    {
      "epoch": 0.7148325497078557,
      "grad_norm": 0.4631308913230896,
      "learning_rate": 3.7514898874250837e-06,
      "loss": 0.2324,
      "step": 39590
    },
    {
      "epoch": 0.7149228291407564,
      "grad_norm": 0.4935637414455414,
      "learning_rate": 3.749275759225657e-06,
      "loss": 0.2384,
      "step": 39595
    },
    {
      "epoch": 0.7150131085736572,
      "grad_norm": 0.6024258136749268,
      "learning_rate": 3.7470621338515513e-06,
      "loss": 0.3232,
      "step": 39600
    },
    {
      "epoch": 0.7151033880065579,
      "grad_norm": 0.41318437457084656,
      "learning_rate": 3.7448490114808332e-06,
      "loss": 0.2129,
      "step": 39605
    },
    {
      "epoch": 0.7151936674394586,
      "grad_norm": 0.6920496821403503,
      "learning_rate": 3.7426363922915287e-06,
      "loss": 0.2365,
      "step": 39610
    },
    {
      "epoch": 0.7152839468723593,
      "grad_norm": 0.6129166483879089,
      "learning_rate": 3.7404242764616315e-06,
      "loss": 0.1805,
      "step": 39615
    },
    {
      "epoch": 0.71537422630526,
      "grad_norm": 0.28114911913871765,
      "learning_rate": 3.738212664169091e-06,
      "loss": 0.261,
      "step": 39620
    },
    {
      "epoch": 0.7154645057381608,
      "grad_norm": 0.34122586250305176,
      "learning_rate": 3.736001555591816e-06,
      "loss": 0.2624,
      "step": 39625
    },
    {
      "epoch": 0.7155547851710615,
      "grad_norm": 0.5609122514724731,
      "learning_rate": 3.7337909509076696e-06,
      "loss": 0.1916,
      "step": 39630
    },
    {
      "epoch": 0.7156450646039622,
      "grad_norm": 0.45785167813301086,
      "learning_rate": 3.731580850294485e-06,
      "loss": 0.2253,
      "step": 39635
    },
    {
      "epoch": 0.7157353440368629,
      "grad_norm": 0.32274383306503296,
      "learning_rate": 3.729371253930041e-06,
      "loss": 0.2932,
      "step": 39640
    },
    {
      "epoch": 0.7158256234697636,
      "grad_norm": 0.6230811476707458,
      "learning_rate": 3.7271621619920915e-06,
      "loss": 0.2504,
      "step": 39645
    },
    {
      "epoch": 0.7159159029026644,
      "grad_norm": 0.7151699662208557,
      "learning_rate": 3.724953574658331e-06,
      "loss": 0.1631,
      "step": 39650
    },
    {
      "epoch": 0.716006182335565,
      "grad_norm": 0.5752794146537781,
      "learning_rate": 3.722745492106439e-06,
      "loss": 0.231,
      "step": 39655
    },
    {
      "epoch": 0.7160964617684658,
      "grad_norm": 0.6475176215171814,
      "learning_rate": 3.7205379145140286e-06,
      "loss": 0.2439,
      "step": 39660
    },
    {
      "epoch": 0.7161867412013665,
      "grad_norm": 0.6292634606361389,
      "learning_rate": 3.7183308420586895e-06,
      "loss": 0.2328,
      "step": 39665
    },
    {
      "epoch": 0.7162770206342672,
      "grad_norm": 0.34139055013656616,
      "learning_rate": 3.7161242749179593e-06,
      "loss": 0.1496,
      "step": 39670
    },
    {
      "epoch": 0.7163673000671679,
      "grad_norm": 0.6831566691398621,
      "learning_rate": 3.7139182132693463e-06,
      "loss": 0.1922,
      "step": 39675
    },
    {
      "epoch": 0.7164575795000686,
      "grad_norm": 0.26857325434684753,
      "learning_rate": 3.711712657290305e-06,
      "loss": 0.2363,
      "step": 39680
    },
    {
      "epoch": 0.7165478589329694,
      "grad_norm": 0.24632221460342407,
      "learning_rate": 3.7095076071582605e-06,
      "loss": 0.1825,
      "step": 39685
    },
    {
      "epoch": 0.7166381383658701,
      "grad_norm": 0.556376039981842,
      "learning_rate": 3.7073030630505923e-06,
      "loss": 0.1967,
      "step": 39690
    },
    {
      "epoch": 0.7167284177987707,
      "grad_norm": 0.26366910338401794,
      "learning_rate": 3.705099025144644e-06,
      "loss": 0.2337,
      "step": 39695
    },
    {
      "epoch": 0.7168186972316715,
      "grad_norm": 0.37259313464164734,
      "learning_rate": 3.702895493617711e-06,
      "loss": 0.2366,
      "step": 39700
    },
    {
      "epoch": 0.7169089766645722,
      "grad_norm": 0.4399190843105316,
      "learning_rate": 3.700692468647047e-06,
      "loss": 0.2512,
      "step": 39705
    },
    {
      "epoch": 0.716999256097473,
      "grad_norm": 0.5776998400688171,
      "learning_rate": 3.698489950409878e-06,
      "loss": 0.2542,
      "step": 39710
    },
    {
      "epoch": 0.7170895355303736,
      "grad_norm": 0.7476668953895569,
      "learning_rate": 3.6962879390833684e-06,
      "loss": 0.2446,
      "step": 39715
    },
    {
      "epoch": 0.7171798149632743,
      "grad_norm": 0.6083337664604187,
      "learning_rate": 3.6940864348446693e-06,
      "loss": 0.2465,
      "step": 39720
    },
    {
      "epoch": 0.7172700943961751,
      "grad_norm": 0.3944800794124603,
      "learning_rate": 3.691885437870865e-06,
      "loss": 0.1738,
      "step": 39725
    },
    {
      "epoch": 0.7173603738290758,
      "grad_norm": 0.7190540432929993,
      "learning_rate": 3.689684948339014e-06,
      "loss": 0.2022,
      "step": 39730
    },
    {
      "epoch": 0.7174506532619764,
      "grad_norm": 0.7269082069396973,
      "learning_rate": 3.6874849664261268e-06,
      "loss": 0.279,
      "step": 39735
    },
    {
      "epoch": 0.7175409326948772,
      "grad_norm": 0.4217861294746399,
      "learning_rate": 3.6852854923091806e-06,
      "loss": 0.1765,
      "step": 39740
    },
    {
      "epoch": 0.7176312121277779,
      "grad_norm": 0.5064873695373535,
      "learning_rate": 3.6830865261651015e-06,
      "loss": 0.2081,
      "step": 39745
    },
    {
      "epoch": 0.7177214915606787,
      "grad_norm": 0.626589298248291,
      "learning_rate": 3.6808880681707827e-06,
      "loss": 0.2431,
      "step": 39750
    },
    {
      "epoch": 0.7178117709935793,
      "grad_norm": 0.3502759337425232,
      "learning_rate": 3.678690118503073e-06,
      "loss": 0.1681,
      "step": 39755
    },
    {
      "epoch": 0.71790205042648,
      "grad_norm": 0.661049485206604,
      "learning_rate": 3.6764926773387876e-06,
      "loss": 0.2551,
      "step": 39760
    },
    {
      "epoch": 0.7179923298593808,
      "grad_norm": 0.3535442054271698,
      "learning_rate": 3.6742957448546867e-06,
      "loss": 0.2815,
      "step": 39765
    },
    {
      "epoch": 0.7180826092922815,
      "grad_norm": 0.47060221433639526,
      "learning_rate": 3.6720993212275037e-06,
      "loss": 0.3189,
      "step": 39770
    },
    {
      "epoch": 0.7181728887251821,
      "grad_norm": 0.8172930479049683,
      "learning_rate": 3.669903406633919e-06,
      "loss": 0.2129,
      "step": 39775
    },
    {
      "epoch": 0.7182631681580829,
      "grad_norm": 0.23240995407104492,
      "learning_rate": 3.6677080012505805e-06,
      "loss": 0.2062,
      "step": 39780
    },
    {
      "epoch": 0.7183534475909836,
      "grad_norm": 0.8371350169181824,
      "learning_rate": 3.665513105254097e-06,
      "loss": 0.2468,
      "step": 39785
    },
    {
      "epoch": 0.7184437270238844,
      "grad_norm": 0.5324345827102661,
      "learning_rate": 3.663318718821025e-06,
      "loss": 0.2794,
      "step": 39790
    },
    {
      "epoch": 0.718534006456785,
      "grad_norm": 0.2858458161354065,
      "learning_rate": 3.6611248421278923e-06,
      "loss": 0.2306,
      "step": 39795
    },
    {
      "epoch": 0.7186242858896857,
      "grad_norm": 0.38325032591819763,
      "learning_rate": 3.658931475351175e-06,
      "loss": 0.233,
      "step": 39800
    },
    {
      "epoch": 0.7187145653225865,
      "grad_norm": 0.5278452038764954,
      "learning_rate": 3.6567386186673203e-06,
      "loss": 0.1667,
      "step": 39805
    },
    {
      "epoch": 0.7188048447554872,
      "grad_norm": 0.47225314378738403,
      "learning_rate": 3.654546272252719e-06,
      "loss": 0.1887,
      "step": 39810
    },
    {
      "epoch": 0.7188951241883879,
      "grad_norm": 0.35314756631851196,
      "learning_rate": 3.652354436283736e-06,
      "loss": 0.1849,
      "step": 39815
    },
    {
      "epoch": 0.7189854036212886,
      "grad_norm": 0.5568872094154358,
      "learning_rate": 3.6501631109366855e-06,
      "loss": 0.1449,
      "step": 39820
    },
    {
      "epoch": 0.7190756830541893,
      "grad_norm": 0.6113433241844177,
      "learning_rate": 3.6479722963878483e-06,
      "loss": 0.2002,
      "step": 39825
    },
    {
      "epoch": 0.7191659624870901,
      "grad_norm": 0.5883828997612,
      "learning_rate": 3.6457819928134542e-06,
      "loss": 0.304,
      "step": 39830
    },
    {
      "epoch": 0.7192562419199907,
      "grad_norm": 0.5464839339256287,
      "learning_rate": 3.6435922003897006e-06,
      "loss": 0.2504,
      "step": 39835
    },
    {
      "epoch": 0.7193465213528915,
      "grad_norm": 0.36738574504852295,
      "learning_rate": 3.641402919292737e-06,
      "loss": 0.1838,
      "step": 39840
    },
    {
      "epoch": 0.7194368007857922,
      "grad_norm": 0.3718404769897461,
      "learning_rate": 3.6392141496986766e-06,
      "loss": 0.2372,
      "step": 39845
    },
    {
      "epoch": 0.7195270802186929,
      "grad_norm": 0.3584044277667999,
      "learning_rate": 3.63702589178359e-06,
      "loss": 0.1749,
      "step": 39850
    },
    {
      "epoch": 0.7196173596515936,
      "grad_norm": 0.5096855163574219,
      "learning_rate": 3.6348381457235115e-06,
      "loss": 0.247,
      "step": 39855
    },
    {
      "epoch": 0.7197076390844943,
      "grad_norm": 0.4133935272693634,
      "learning_rate": 3.6326509116944208e-06,
      "loss": 0.2273,
      "step": 39860
    },
    {
      "epoch": 0.719797918517395,
      "grad_norm": 0.6275957226753235,
      "learning_rate": 3.6304641898722726e-06,
      "loss": 0.2563,
      "step": 39865
    },
    {
      "epoch": 0.7198881979502958,
      "grad_norm": 0.480309396982193,
      "learning_rate": 3.6282779804329658e-06,
      "loss": 0.1821,
      "step": 39870
    },
    {
      "epoch": 0.7199784773831964,
      "grad_norm": 0.4104933738708496,
      "learning_rate": 3.6260922835523726e-06,
      "loss": 0.2553,
      "step": 39875
    },
    {
      "epoch": 0.7200687568160972,
      "grad_norm": 0.5115457773208618,
      "learning_rate": 3.6239070994063087e-06,
      "loss": 0.2359,
      "step": 39880
    },
    {
      "epoch": 0.7201590362489979,
      "grad_norm": 0.15795692801475525,
      "learning_rate": 3.62172242817056e-06,
      "loss": 0.1431,
      "step": 39885
    },
    {
      "epoch": 0.7202493156818986,
      "grad_norm": 0.3097035586833954,
      "learning_rate": 3.6195382700208704e-06,
      "loss": 0.2475,
      "step": 39890
    },
    {
      "epoch": 0.7203395951147993,
      "grad_norm": 0.5861602425575256,
      "learning_rate": 3.6173546251329337e-06,
      "loss": 0.2818,
      "step": 39895
    },
    {
      "epoch": 0.7204298745477,
      "grad_norm": 0.5597209334373474,
      "learning_rate": 3.6151714936824146e-06,
      "loss": 0.1739,
      "step": 39900
    },
    {
      "epoch": 0.7205201539806008,
      "grad_norm": 0.32762089371681213,
      "learning_rate": 3.612988875844923e-06,
      "loss": 0.1982,
      "step": 39905
    },
    {
      "epoch": 0.7206104334135015,
      "grad_norm": 0.4768185615539551,
      "learning_rate": 3.610806771796038e-06,
      "loss": 0.255,
      "step": 39910
    },
    {
      "epoch": 0.7207007128464021,
      "grad_norm": 0.6743978261947632,
      "learning_rate": 3.608625181711294e-06,
      "loss": 0.2388,
      "step": 39915
    },
    {
      "epoch": 0.7207909922793029,
      "grad_norm": 0.570248007774353,
      "learning_rate": 3.606444105766188e-06,
      "loss": 0.2112,
      "step": 39920
    },
    {
      "epoch": 0.7208812717122036,
      "grad_norm": 0.47991472482681274,
      "learning_rate": 3.604263544136165e-06,
      "loss": 0.2188,
      "step": 39925
    },
    {
      "epoch": 0.7209715511451044,
      "grad_norm": 0.3377138078212738,
      "learning_rate": 3.6020834969966423e-06,
      "loss": 0.192,
      "step": 39930
    },
    {
      "epoch": 0.721061830578005,
      "grad_norm": 0.2957462668418884,
      "learning_rate": 3.5999039645229806e-06,
      "loss": 0.2086,
      "step": 39935
    },
    {
      "epoch": 0.7211521100109057,
      "grad_norm": 0.3006899058818817,
      "learning_rate": 3.597724946890515e-06,
      "loss": 0.1798,
      "step": 39940
    },
    {
      "epoch": 0.7212423894438065,
      "grad_norm": 0.3699197471141815,
      "learning_rate": 3.595546444274526e-06,
      "loss": 0.244,
      "step": 39945
    },
    {
      "epoch": 0.7213326688767072,
      "grad_norm": 0.21937566995620728,
      "learning_rate": 3.5933684568502603e-06,
      "loss": 0.2092,
      "step": 39950
    },
    {
      "epoch": 0.7214229483096078,
      "grad_norm": 0.6600390672683716,
      "learning_rate": 3.5911909847929206e-06,
      "loss": 0.254,
      "step": 39955
    },
    {
      "epoch": 0.7215132277425086,
      "grad_norm": 0.9043104648590088,
      "learning_rate": 3.5890140282776743e-06,
      "loss": 0.2628,
      "step": 39960
    },
    {
      "epoch": 0.7216035071754093,
      "grad_norm": 0.4741593897342682,
      "learning_rate": 3.5868375874796326e-06,
      "loss": 0.2633,
      "step": 39965
    },
    {
      "epoch": 0.7216937866083101,
      "grad_norm": 0.40252596139907837,
      "learning_rate": 3.584661662573883e-06,
      "loss": 0.2537,
      "step": 39970
    },
    {
      "epoch": 0.7217840660412107,
      "grad_norm": 0.3973434865474701,
      "learning_rate": 3.582486253735454e-06,
      "loss": 0.1877,
      "step": 39975
    },
    {
      "epoch": 0.7218743454741114,
      "grad_norm": 0.42567405104637146,
      "learning_rate": 3.5803113611393482e-06,
      "loss": 0.1551,
      "step": 39980
    },
    {
      "epoch": 0.7219646249070122,
      "grad_norm": 0.7808625102043152,
      "learning_rate": 3.5781369849605197e-06,
      "loss": 0.287,
      "step": 39985
    },
    {
      "epoch": 0.7220549043399129,
      "grad_norm": 0.6534682512283325,
      "learning_rate": 3.575963125373876e-06,
      "loss": 0.2023,
      "step": 39990
    },
    {
      "epoch": 0.7221451837728136,
      "grad_norm": 0.3858139216899872,
      "learning_rate": 3.5737897825542954e-06,
      "loss": 0.1907,
      "step": 39995
    },
    {
      "epoch": 0.7222354632057143,
      "grad_norm": 0.2853589355945587,
      "learning_rate": 3.5716169566766e-06,
      "loss": 0.1961,
      "step": 40000
    },
    {
      "epoch": 0.722325742638615,
      "grad_norm": 0.5486742854118347,
      "learning_rate": 3.569444647915584e-06,
      "loss": 0.2662,
      "step": 40005
    },
    {
      "epoch": 0.7224160220715158,
      "grad_norm": 0.7488285899162292,
      "learning_rate": 3.5672728564459903e-06,
      "loss": 0.246,
      "step": 40010
    },
    {
      "epoch": 0.7225063015044164,
      "grad_norm": 0.45054709911346436,
      "learning_rate": 3.5651015824425226e-06,
      "loss": 0.199,
      "step": 40015
    },
    {
      "epoch": 0.7225965809373172,
      "grad_norm": 0.3069791793823242,
      "learning_rate": 3.5629308260798477e-06,
      "loss": 0.1964,
      "step": 40020
    },
    {
      "epoch": 0.7226868603702179,
      "grad_norm": 0.7028543949127197,
      "learning_rate": 3.560760587532589e-06,
      "loss": 0.2688,
      "step": 40025
    },
    {
      "epoch": 0.7227771398031186,
      "grad_norm": 0.3445713520050049,
      "learning_rate": 3.55859086697532e-06,
      "loss": 0.2132,
      "step": 40030
    },
    {
      "epoch": 0.7228674192360194,
      "grad_norm": 0.3838486671447754,
      "learning_rate": 3.5564216645825856e-06,
      "loss": 0.2215,
      "step": 40035
    },
    {
      "epoch": 0.72295769866892,
      "grad_norm": 0.5086146593093872,
      "learning_rate": 3.554252980528875e-06,
      "loss": 0.151,
      "step": 40040
    },
    {
      "epoch": 0.7230479781018208,
      "grad_norm": 0.4288047254085541,
      "learning_rate": 3.5520848149886467e-06,
      "loss": 0.183,
      "step": 40045
    },
    {
      "epoch": 0.7231382575347215,
      "grad_norm": 0.5946467518806458,
      "learning_rate": 3.5499171681363152e-06,
      "loss": 0.1896,
      "step": 40050
    },
    {
      "epoch": 0.7232285369676222,
      "grad_norm": 0.9705565571784973,
      "learning_rate": 3.547750040146254e-06,
      "loss": 0.2119,
      "step": 40055
    },
    {
      "epoch": 0.7233188164005229,
      "grad_norm": 0.2487785518169403,
      "learning_rate": 3.5455834311927895e-06,
      "loss": 0.2979,
      "step": 40060
    },
    {
      "epoch": 0.7234090958334236,
      "grad_norm": 0.5289493203163147,
      "learning_rate": 3.543417341450206e-06,
      "loss": 0.2717,
      "step": 40065
    },
    {
      "epoch": 0.7234993752663244,
      "grad_norm": 0.48151740431785583,
      "learning_rate": 3.5412517710927572e-06,
      "loss": 0.2648,
      "step": 40070
    },
    {
      "epoch": 0.7235896546992251,
      "grad_norm": 0.2873721420764923,
      "learning_rate": 3.53908672029464e-06,
      "loss": 0.2116,
      "step": 40075
    },
    {
      "epoch": 0.7236799341321257,
      "grad_norm": 0.44618523120880127,
      "learning_rate": 3.5369221892300223e-06,
      "loss": 0.2741,
      "step": 40080
    },
    {
      "epoch": 0.7237702135650265,
      "grad_norm": 0.2919872999191284,
      "learning_rate": 3.534758178073021e-06,
      "loss": 0.3187,
      "step": 40085
    },
    {
      "epoch": 0.7238604929979272,
      "grad_norm": 0.6174708604812622,
      "learning_rate": 3.5325946869977225e-06,
      "loss": 0.2481,
      "step": 40090
    },
    {
      "epoch": 0.723950772430828,
      "grad_norm": 0.3890153765678406,
      "learning_rate": 3.5304317161781556e-06,
      "loss": 0.3056,
      "step": 40095
    },
    {
      "epoch": 0.7240410518637286,
      "grad_norm": 0.5483882427215576,
      "learning_rate": 3.5282692657883212e-06,
      "loss": 0.222,
      "step": 40100
    },
    {
      "epoch": 0.7241313312966293,
      "grad_norm": 0.4131814241409302,
      "learning_rate": 3.526107336002168e-06,
      "loss": 0.314,
      "step": 40105
    },
    {
      "epoch": 0.7242216107295301,
      "grad_norm": 0.38478797674179077,
      "learning_rate": 3.5239459269936092e-06,
      "loss": 0.3439,
      "step": 40110
    },
    {
      "epoch": 0.7243118901624308,
      "grad_norm": 0.3770439028739929,
      "learning_rate": 3.521785038936515e-06,
      "loss": 0.1963,
      "step": 40115
    },
    {
      "epoch": 0.7244021695953314,
      "grad_norm": 0.649714469909668,
      "learning_rate": 3.5196246720047166e-06,
      "loss": 0.1478,
      "step": 40120
    },
    {
      "epoch": 0.7244924490282322,
      "grad_norm": 0.20353087782859802,
      "learning_rate": 3.5174648263719935e-06,
      "loss": 0.244,
      "step": 40125
    },
    {
      "epoch": 0.7245827284611329,
      "grad_norm": 0.47791555523872375,
      "learning_rate": 3.5153055022120952e-06,
      "loss": 0.1752,
      "step": 40130
    },
    {
      "epoch": 0.7246730078940337,
      "grad_norm": 0.31488651037216187,
      "learning_rate": 3.5131466996987183e-06,
      "loss": 0.1311,
      "step": 40135
    },
    {
      "epoch": 0.7247632873269343,
      "grad_norm": 0.35288169980049133,
      "learning_rate": 3.510988419005529e-06,
      "loss": 0.2397,
      "step": 40140
    },
    {
      "epoch": 0.724853566759835,
      "grad_norm": 0.4234815537929535,
      "learning_rate": 3.5088306603061385e-06,
      "loss": 0.1695,
      "step": 40145
    },
    {
      "epoch": 0.7249438461927358,
      "grad_norm": 0.6899263262748718,
      "learning_rate": 3.506673423774125e-06,
      "loss": 0.2204,
      "step": 40150
    },
    {
      "epoch": 0.7250341256256365,
      "grad_norm": 0.3410181701183319,
      "learning_rate": 3.5045167095830268e-06,
      "loss": 0.3102,
      "step": 40155
    },
    {
      "epoch": 0.7251244050585371,
      "grad_norm": 0.6206501722335815,
      "learning_rate": 3.5023605179063303e-06,
      "loss": 0.2645,
      "step": 40160
    },
    {
      "epoch": 0.7252146844914379,
      "grad_norm": 0.6185318231582642,
      "learning_rate": 3.50020484891749e-06,
      "loss": 0.2006,
      "step": 40165
    },
    {
      "epoch": 0.7253049639243386,
      "grad_norm": 0.2537311017513275,
      "learning_rate": 3.4980497027899098e-06,
      "loss": 0.2067,
      "step": 40170
    },
    {
      "epoch": 0.7253952433572394,
      "grad_norm": 0.6332979798316956,
      "learning_rate": 3.495895079696956e-06,
      "loss": 0.2341,
      "step": 40175
    },
    {
      "epoch": 0.72548552279014,
      "grad_norm": 0.4089825749397278,
      "learning_rate": 3.4937409798119537e-06,
      "loss": 0.2999,
      "step": 40180
    },
    {
      "epoch": 0.7255758022230407,
      "grad_norm": 0.4892581105232239,
      "learning_rate": 3.4915874033081877e-06,
      "loss": 0.2095,
      "step": 40185
    },
    {
      "epoch": 0.7256660816559415,
      "grad_norm": 0.4545522630214691,
      "learning_rate": 3.489434350358891e-06,
      "loss": 0.2304,
      "step": 40190
    },
    {
      "epoch": 0.7257563610888422,
      "grad_norm": 0.5012370944023132,
      "learning_rate": 3.487281821137268e-06,
      "loss": 0.2239,
      "step": 40195
    },
    {
      "epoch": 0.7258466405217429,
      "grad_norm": 0.7334614992141724,
      "learning_rate": 3.485129815816467e-06,
      "loss": 0.2329,
      "step": 40200
    },
    {
      "epoch": 0.7259369199546436,
      "grad_norm": 0.2023577094078064,
      "learning_rate": 3.482978334569609e-06,
      "loss": 0.1647,
      "step": 40205
    },
    {
      "epoch": 0.7260271993875443,
      "grad_norm": 0.3257547914981842,
      "learning_rate": 3.4808273775697566e-06,
      "loss": 0.1549,
      "step": 40210
    },
    {
      "epoch": 0.7261174788204451,
      "grad_norm": 0.44293710589408875,
      "learning_rate": 3.4786769449899414e-06,
      "loss": 0.2437,
      "step": 40215
    },
    {
      "epoch": 0.7262077582533457,
      "grad_norm": 0.4812375009059906,
      "learning_rate": 3.4765270370031533e-06,
      "loss": 0.1852,
      "step": 40220
    },
    {
      "epoch": 0.7262980376862465,
      "grad_norm": 0.49362730979919434,
      "learning_rate": 3.474377653782337e-06,
      "loss": 0.2395,
      "step": 40225
    },
    {
      "epoch": 0.7263883171191472,
      "grad_norm": 0.5562480688095093,
      "learning_rate": 3.4722287955003897e-06,
      "loss": 0.2293,
      "step": 40230
    },
    {
      "epoch": 0.7264785965520479,
      "grad_norm": 0.3648661971092224,
      "learning_rate": 3.4700804623301778e-06,
      "loss": 0.1545,
      "step": 40235
    },
    {
      "epoch": 0.7265688759849486,
      "grad_norm": 0.6169965267181396,
      "learning_rate": 3.4679326544445122e-06,
      "loss": 0.2442,
      "step": 40240
    },
    {
      "epoch": 0.7266591554178493,
      "grad_norm": 0.3303978443145752,
      "learning_rate": 3.4657853720161718e-06,
      "loss": 0.1603,
      "step": 40245
    },
    {
      "epoch": 0.72674943485075,
      "grad_norm": 0.3425135314464569,
      "learning_rate": 3.4636386152178935e-06,
      "loss": 0.2981,
      "step": 40250
    },
    {
      "epoch": 0.7268397142836508,
      "grad_norm": 0.48774453997612,
      "learning_rate": 3.4614923842223603e-06,
      "loss": 0.22,
      "step": 40255
    },
    {
      "epoch": 0.7269299937165514,
      "grad_norm": 0.5561987161636353,
      "learning_rate": 3.4593466792022288e-06,
      "loss": 0.2558,
      "step": 40260
    },
    {
      "epoch": 0.7270202731494522,
      "grad_norm": 0.2400379329919815,
      "learning_rate": 3.4572015003301e-06,
      "loss": 0.1867,
      "step": 40265
    },
    {
      "epoch": 0.7271105525823529,
      "grad_norm": 0.5912076234817505,
      "learning_rate": 3.455056847778542e-06,
      "loss": 0.1759,
      "step": 40270
    },
    {
      "epoch": 0.7272008320152537,
      "grad_norm": 0.38932791352272034,
      "learning_rate": 3.452912721720071e-06,
      "loss": 0.1969,
      "step": 40275
    },
    {
      "epoch": 0.7272911114481543,
      "grad_norm": 0.38241592049598694,
      "learning_rate": 3.4507691223271687e-06,
      "loss": 0.2328,
      "step": 40280
    },
    {
      "epoch": 0.727381390881055,
      "grad_norm": 0.525311291217804,
      "learning_rate": 3.4486260497722745e-06,
      "loss": 0.2274,
      "step": 40285
    },
    {
      "epoch": 0.7274716703139558,
      "grad_norm": 0.40287983417510986,
      "learning_rate": 3.4464835042277844e-06,
      "loss": 0.1836,
      "step": 40290
    },
    {
      "epoch": 0.7275619497468565,
      "grad_norm": 0.44902732968330383,
      "learning_rate": 3.444341485866044e-06,
      "loss": 0.2354,
      "step": 40295
    },
    {
      "epoch": 0.7276522291797571,
      "grad_norm": 0.6186977624893188,
      "learning_rate": 3.44219999485937e-06,
      "loss": 0.1643,
      "step": 40300
    },
    {
      "epoch": 0.7277425086126579,
      "grad_norm": 0.22817963361740112,
      "learning_rate": 3.4400590313800243e-06,
      "loss": 0.1861,
      "step": 40305
    },
    {
      "epoch": 0.7278327880455586,
      "grad_norm": 0.6198724508285522,
      "learning_rate": 3.4379185956002325e-06,
      "loss": 0.2969,
      "step": 40310
    },
    {
      "epoch": 0.7279230674784594,
      "grad_norm": 0.4639147222042084,
      "learning_rate": 3.4357786876921806e-06,
      "loss": 0.1743,
      "step": 40315
    },
    {
      "epoch": 0.72801334691136,
      "grad_norm": 0.2476786971092224,
      "learning_rate": 3.433639307828008e-06,
      "loss": 0.2198,
      "step": 40320
    },
    {
      "epoch": 0.7281036263442607,
      "grad_norm": 0.7034710645675659,
      "learning_rate": 3.4315004561798095e-06,
      "loss": 0.2146,
      "step": 40325
    },
    {
      "epoch": 0.7281939057771615,
      "grad_norm": 0.29537856578826904,
      "learning_rate": 3.429362132919645e-06,
      "loss": 0.2649,
      "step": 40330
    },
    {
      "epoch": 0.7282841852100622,
      "grad_norm": 0.5975651144981384,
      "learning_rate": 3.427224338219519e-06,
      "loss": 0.1971,
      "step": 40335
    },
    {
      "epoch": 0.7283744646429628,
      "grad_norm": 0.36948174238204956,
      "learning_rate": 3.4250870722514105e-06,
      "loss": 0.1718,
      "step": 40340
    },
    {
      "epoch": 0.7284647440758636,
      "grad_norm": 0.7214956283569336,
      "learning_rate": 3.4229503351872396e-06,
      "loss": 0.3039,
      "step": 40345
    },
    {
      "epoch": 0.7285550235087643,
      "grad_norm": 0.5402421355247498,
      "learning_rate": 3.420814127198895e-06,
      "loss": 0.2184,
      "step": 40350
    },
    {
      "epoch": 0.7286453029416651,
      "grad_norm": 0.6028202176094055,
      "learning_rate": 3.4186784484582204e-06,
      "loss": 0.336,
      "step": 40355
    },
    {
      "epoch": 0.7287355823745657,
      "grad_norm": 0.43091797828674316,
      "learning_rate": 3.4165432991370117e-06,
      "loss": 0.2733,
      "step": 40360
    },
    {
      "epoch": 0.7288258618074664,
      "grad_norm": 0.2607382535934448,
      "learning_rate": 3.414408679407031e-06,
      "loss": 0.2299,
      "step": 40365
    },
    {
      "epoch": 0.7289161412403672,
      "grad_norm": 0.27409112453460693,
      "learning_rate": 3.4122745894399866e-06,
      "loss": 0.2266,
      "step": 40370
    },
    {
      "epoch": 0.7290064206732679,
      "grad_norm": 0.2401573210954666,
      "learning_rate": 3.4101410294075544e-06,
      "loss": 0.1303,
      "step": 40375
    },
    {
      "epoch": 0.7290967001061686,
      "grad_norm": 1.1108605861663818,
      "learning_rate": 3.4080079994813643e-06,
      "loss": 0.282,
      "step": 40380
    },
    {
      "epoch": 0.7291869795390693,
      "grad_norm": 0.5349571704864502,
      "learning_rate": 3.4058754998330044e-06,
      "loss": 0.2915,
      "step": 40385
    },
    {
      "epoch": 0.72927725897197,
      "grad_norm": 0.646102786064148,
      "learning_rate": 3.4037435306340137e-06,
      "loss": 0.1459,
      "step": 40390
    },
    {
      "epoch": 0.7293675384048708,
      "grad_norm": 0.3781536817550659,
      "learning_rate": 3.401612092055899e-06,
      "loss": 0.2009,
      "step": 40395
    },
    {
      "epoch": 0.7294578178377714,
      "grad_norm": 0.44515880942344666,
      "learning_rate": 3.3994811842701147e-06,
      "loss": 0.2621,
      "step": 40400
    },
    {
      "epoch": 0.7295480972706722,
      "grad_norm": 0.4086102545261383,
      "learning_rate": 3.397350807448081e-06,
      "loss": 0.2204,
      "step": 40405
    },
    {
      "epoch": 0.7296383767035729,
      "grad_norm": 0.3024944067001343,
      "learning_rate": 3.3952209617611665e-06,
      "loss": 0.2066,
      "step": 40410
    },
    {
      "epoch": 0.7297286561364736,
      "grad_norm": 0.28862929344177246,
      "learning_rate": 3.393091647380704e-06,
      "loss": 0.2145,
      "step": 40415
    },
    {
      "epoch": 0.7298189355693743,
      "grad_norm": 0.41927874088287354,
      "learning_rate": 3.3909628644779813e-06,
      "loss": 0.229,
      "step": 40420
    },
    {
      "epoch": 0.729909215002275,
      "grad_norm": 0.6619019508361816,
      "learning_rate": 3.3888346132242466e-06,
      "loss": 0.2426,
      "step": 40425
    },
    {
      "epoch": 0.7299994944351758,
      "grad_norm": 0.7908322215080261,
      "learning_rate": 3.3867068937906998e-06,
      "loss": 0.2189,
      "step": 40430
    },
    {
      "epoch": 0.7300897738680765,
      "grad_norm": 0.5456673502922058,
      "learning_rate": 3.3845797063484973e-06,
      "loss": 0.1597,
      "step": 40435
    },
    {
      "epoch": 0.7301800533009772,
      "grad_norm": 0.39288800954818726,
      "learning_rate": 3.382453051068759e-06,
      "loss": 0.1385,
      "step": 40440
    },
    {
      "epoch": 0.7302703327338779,
      "grad_norm": 0.45614194869995117,
      "learning_rate": 3.380326928122558e-06,
      "loss": 0.1652,
      "step": 40445
    },
    {
      "epoch": 0.7303606121667786,
      "grad_norm": 0.3323318064212799,
      "learning_rate": 3.3782013376809287e-06,
      "loss": 0.2654,
      "step": 40450
    },
    {
      "epoch": 0.7304508915996794,
      "grad_norm": 0.2699224054813385,
      "learning_rate": 3.376076279914854e-06,
      "loss": 0.2126,
      "step": 40455
    },
    {
      "epoch": 0.7305411710325801,
      "grad_norm": 0.7620146870613098,
      "learning_rate": 3.373951754995285e-06,
      "loss": 0.2983,
      "step": 40460
    },
    {
      "epoch": 0.7306314504654807,
      "grad_norm": 0.4448334574699402,
      "learning_rate": 3.3718277630931186e-06,
      "loss": 0.1626,
      "step": 40465
    },
    {
      "epoch": 0.7307217298983815,
      "grad_norm": 0.3676196038722992,
      "learning_rate": 3.36970430437922e-06,
      "loss": 0.192,
      "step": 40470
    },
    {
      "epoch": 0.7308120093312822,
      "grad_norm": 0.39222726225852966,
      "learning_rate": 3.3675813790243993e-06,
      "loss": 0.2017,
      "step": 40475
    },
    {
      "epoch": 0.730902288764183,
      "grad_norm": 0.3409358561038971,
      "learning_rate": 3.365458987199436e-06,
      "loss": 0.145,
      "step": 40480
    },
    {
      "epoch": 0.7309925681970836,
      "grad_norm": 0.4742923676967621,
      "learning_rate": 3.363337129075058e-06,
      "loss": 0.1871,
      "step": 40485
    },
    {
      "epoch": 0.7310828476299843,
      "grad_norm": 0.4943235516548157,
      "learning_rate": 3.361215804821958e-06,
      "loss": 0.2017,
      "step": 40490
    },
    {
      "epoch": 0.7311731270628851,
      "grad_norm": 0.5879418849945068,
      "learning_rate": 3.3590950146107748e-06,
      "loss": 0.2369,
      "step": 40495
    },
    {
      "epoch": 0.7312634064957858,
      "grad_norm": 0.3419423997402191,
      "learning_rate": 3.356974758612117e-06,
      "loss": 0.1675,
      "step": 40500
    },
    {
      "epoch": 0.7313536859286864,
      "grad_norm": 0.5356691479682922,
      "learning_rate": 3.3548550369965383e-06,
      "loss": 0.1861,
      "step": 40505
    },
    {
      "epoch": 0.7314439653615872,
      "grad_norm": 0.5443263649940491,
      "learning_rate": 3.352735849934555e-06,
      "loss": 0.2761,
      "step": 40510
    },
    {
      "epoch": 0.7315342447944879,
      "grad_norm": 0.36747556924819946,
      "learning_rate": 3.350617197596644e-06,
      "loss": 0.1448,
      "step": 40515
    },
    {
      "epoch": 0.7316245242273887,
      "grad_norm": 0.44932252168655396,
      "learning_rate": 3.3484990801532357e-06,
      "loss": 0.2697,
      "step": 40520
    },
    {
      "epoch": 0.7317148036602893,
      "grad_norm": 0.37425053119659424,
      "learning_rate": 3.3463814977747153e-06,
      "loss": 0.2599,
      "step": 40525
    },
    {
      "epoch": 0.73180508309319,
      "grad_norm": 0.4571848511695862,
      "learning_rate": 3.344264450631425e-06,
      "loss": 0.249,
      "step": 40530
    },
    {
      "epoch": 0.7318953625260908,
      "grad_norm": 0.3737792372703552,
      "learning_rate": 3.3421479388936708e-06,
      "loss": 0.2467,
      "step": 40535
    },
    {
      "epoch": 0.7319856419589915,
      "grad_norm": 0.3214986026287079,
      "learning_rate": 3.3400319627317034e-06,
      "loss": 0.1831,
      "step": 40540
    },
    {
      "epoch": 0.7320759213918921,
      "grad_norm": 0.35493019223213196,
      "learning_rate": 3.337916522315743e-06,
      "loss": 0.1996,
      "step": 40545
    },
    {
      "epoch": 0.7321662008247929,
      "grad_norm": 0.4135398864746094,
      "learning_rate": 3.3358016178159593e-06,
      "loss": 0.1702,
      "step": 40550
    },
    {
      "epoch": 0.7322564802576936,
      "grad_norm": 0.8642098307609558,
      "learning_rate": 3.3336872494024852e-06,
      "loss": 0.2168,
      "step": 40555
    },
    {
      "epoch": 0.7323467596905944,
      "grad_norm": 0.3693483769893646,
      "learning_rate": 3.3315734172454007e-06,
      "loss": 0.2346,
      "step": 40560
    },
    {
      "epoch": 0.732437039123495,
      "grad_norm": 0.3232543468475342,
      "learning_rate": 3.329460121514754e-06,
      "loss": 0.1765,
      "step": 40565
    },
    {
      "epoch": 0.7325273185563957,
      "grad_norm": 0.46592220664024353,
      "learning_rate": 3.3273473623805376e-06,
      "loss": 0.2903,
      "step": 40570
    },
    {
      "epoch": 0.7326175979892965,
      "grad_norm": 0.4236691892147064,
      "learning_rate": 3.3252351400127105e-06,
      "loss": 0.3021,
      "step": 40575
    },
    {
      "epoch": 0.7327078774221972,
      "grad_norm": 0.3750850260257721,
      "learning_rate": 3.3231234545811863e-06,
      "loss": 0.2136,
      "step": 40580
    },
    {
      "epoch": 0.7327981568550979,
      "grad_norm": 0.3952043950557709,
      "learning_rate": 3.3210123062558374e-06,
      "loss": 0.285,
      "step": 40585
    },
    {
      "epoch": 0.7328884362879986,
      "grad_norm": 0.25859534740448,
      "learning_rate": 3.3189016952064855e-06,
      "loss": 0.2055,
      "step": 40590
    },
    {
      "epoch": 0.7329787157208993,
      "grad_norm": 0.34520620107650757,
      "learning_rate": 3.316791621602918e-06,
      "loss": 0.2823,
      "step": 40595
    },
    {
      "epoch": 0.7330689951538001,
      "grad_norm": 0.5526063442230225,
      "learning_rate": 3.314682085614871e-06,
      "loss": 0.0921,
      "step": 40600
    },
    {
      "epoch": 0.7331592745867007,
      "grad_norm": 0.3728067874908447,
      "learning_rate": 3.3125730874120453e-06,
      "loss": 0.2365,
      "step": 40605
    },
    {
      "epoch": 0.7332495540196015,
      "grad_norm": 0.49789491295814514,
      "learning_rate": 3.3104646271640894e-06,
      "loss": 0.2342,
      "step": 40610
    },
    {
      "epoch": 0.7333398334525022,
      "grad_norm": 0.6995531916618347,
      "learning_rate": 3.3083567050406173e-06,
      "loss": 0.3228,
      "step": 40615
    },
    {
      "epoch": 0.7334301128854029,
      "grad_norm": 0.394559383392334,
      "learning_rate": 3.306249321211198e-06,
      "loss": 0.3184,
      "step": 40620
    },
    {
      "epoch": 0.7335203923183036,
      "grad_norm": 0.48034948110580444,
      "learning_rate": 3.304142475845349e-06,
      "loss": 0.1532,
      "step": 40625
    },
    {
      "epoch": 0.7336106717512043,
      "grad_norm": 0.5348523259162903,
      "learning_rate": 3.302036169112558e-06,
      "loss": 0.1852,
      "step": 40630
    },
    {
      "epoch": 0.733700951184105,
      "grad_norm": 0.4209555685520172,
      "learning_rate": 3.2999304011822552e-06,
      "loss": 0.2359,
      "step": 40635
    },
    {
      "epoch": 0.7337912306170058,
      "grad_norm": 1.5587431192398071,
      "learning_rate": 3.2978251722238376e-06,
      "loss": 0.2462,
      "step": 40640
    },
    {
      "epoch": 0.7338815100499064,
      "grad_norm": 0.3898302912712097,
      "learning_rate": 3.2957204824066546e-06,
      "loss": 0.2945,
      "step": 40645
    },
    {
      "epoch": 0.7339717894828072,
      "grad_norm": 0.42389562726020813,
      "learning_rate": 3.293616331900017e-06,
      "loss": 0.1962,
      "step": 40650
    },
    {
      "epoch": 0.7340620689157079,
      "grad_norm": 0.44758307933807373,
      "learning_rate": 3.2915127208731835e-06,
      "loss": 0.1647,
      "step": 40655
    },
    {
      "epoch": 0.7341523483486087,
      "grad_norm": 0.45123669505119324,
      "learning_rate": 3.2894096494953774e-06,
      "loss": 0.2992,
      "step": 40660
    },
    {
      "epoch": 0.7342426277815093,
      "grad_norm": 0.4321818947792053,
      "learning_rate": 3.287307117935773e-06,
      "loss": 0.1862,
      "step": 40665
    },
    {
      "epoch": 0.73433290721441,
      "grad_norm": 0.4173588454723358,
      "learning_rate": 3.2852051263635077e-06,
      "loss": 0.2836,
      "step": 40670
    },
    {
      "epoch": 0.7344231866473108,
      "grad_norm": 0.6080979704856873,
      "learning_rate": 3.283103674947665e-06,
      "loss": 0.2901,
      "step": 40675
    },
    {
      "epoch": 0.7345134660802115,
      "grad_norm": 0.6286442875862122,
      "learning_rate": 3.281002763857295e-06,
      "loss": 0.2525,
      "step": 40680
    },
    {
      "epoch": 0.7346037455131121,
      "grad_norm": 0.3144855797290802,
      "learning_rate": 3.2789023932614017e-06,
      "loss": 0.1769,
      "step": 40685
    },
    {
      "epoch": 0.7346940249460129,
      "grad_norm": 0.6229161620140076,
      "learning_rate": 3.276802563328946e-06,
      "loss": 0.2706,
      "step": 40690
    },
    {
      "epoch": 0.7347843043789136,
      "grad_norm": 0.6532561182975769,
      "learning_rate": 3.2747032742288386e-06,
      "loss": 0.2707,
      "step": 40695
    },
    {
      "epoch": 0.7348745838118144,
      "grad_norm": 0.264279305934906,
      "learning_rate": 3.272604526129959e-06,
      "loss": 0.2004,
      "step": 40700
    },
    {
      "epoch": 0.734964863244715,
      "grad_norm": 0.43015000224113464,
      "learning_rate": 3.270506319201129e-06,
      "loss": 0.1812,
      "step": 40705
    },
    {
      "epoch": 0.7350551426776157,
      "grad_norm": 0.5014654994010925,
      "learning_rate": 3.268408653611137e-06,
      "loss": 0.2445,
      "step": 40710
    },
    {
      "epoch": 0.7351454221105165,
      "grad_norm": 0.36967700719833374,
      "learning_rate": 3.2663115295287296e-06,
      "loss": 0.1601,
      "step": 40715
    },
    {
      "epoch": 0.7352357015434172,
      "grad_norm": 0.5202431678771973,
      "learning_rate": 3.264214947122597e-06,
      "loss": 0.1835,
      "step": 40720
    },
    {
      "epoch": 0.7353259809763179,
      "grad_norm": 0.6178275942802429,
      "learning_rate": 3.262118906561401e-06,
      "loss": 0.3235,
      "step": 40725
    },
    {
      "epoch": 0.7354162604092186,
      "grad_norm": 0.4469394385814667,
      "learning_rate": 3.2600234080137473e-06,
      "loss": 0.175,
      "step": 40730
    },
    {
      "epoch": 0.7355065398421193,
      "grad_norm": 0.3345348536968231,
      "learning_rate": 3.257928451648209e-06,
      "loss": 0.2191,
      "step": 40735
    },
    {
      "epoch": 0.7355968192750201,
      "grad_norm": 0.567681074142456,
      "learning_rate": 3.255834037633303e-06,
      "loss": 0.218,
      "step": 40740
    },
    {
      "epoch": 0.7356870987079207,
      "grad_norm": 0.3793160617351532,
      "learning_rate": 3.2537401661375155e-06,
      "loss": 0.2864,
      "step": 40745
    },
    {
      "epoch": 0.7357773781408214,
      "grad_norm": 0.3733515441417694,
      "learning_rate": 3.251646837329281e-06,
      "loss": 0.2865,
      "step": 40750
    },
    {
      "epoch": 0.7358676575737222,
      "grad_norm": 0.5380032658576965,
      "learning_rate": 3.2495540513769962e-06,
      "loss": 0.2304,
      "step": 40755
    },
    {
      "epoch": 0.7359579370066229,
      "grad_norm": 0.24043571949005127,
      "learning_rate": 3.247461808449004e-06,
      "loss": 0.214,
      "step": 40760
    },
    {
      "epoch": 0.7360482164395236,
      "grad_norm": 0.5552172064781189,
      "learning_rate": 3.245370108713617e-06,
      "loss": 0.2595,
      "step": 40765
    },
    {
      "epoch": 0.7361384958724243,
      "grad_norm": 0.5105094909667969,
      "learning_rate": 3.243278952339091e-06,
      "loss": 0.1932,
      "step": 40770
    },
    {
      "epoch": 0.736228775305325,
      "grad_norm": 0.3858696520328522,
      "learning_rate": 3.2411883394936463e-06,
      "loss": 0.2166,
      "step": 40775
    },
    {
      "epoch": 0.7363190547382258,
      "grad_norm": 0.6489410996437073,
      "learning_rate": 3.239098270345459e-06,
      "loss": 0.2125,
      "step": 40780
    },
    {
      "epoch": 0.7364093341711264,
      "grad_norm": 0.5374500751495361,
      "learning_rate": 3.237008745062663e-06,
      "loss": 0.2938,
      "step": 40785
    },
    {
      "epoch": 0.7364996136040272,
      "grad_norm": 0.8062470555305481,
      "learning_rate": 3.234919763813338e-06,
      "loss": 0.2541,
      "step": 40790
    },
    {
      "epoch": 0.7365898930369279,
      "grad_norm": 0.7905746698379517,
      "learning_rate": 3.232831326765534e-06,
      "loss": 0.3326,
      "step": 40795
    },
    {
      "epoch": 0.7366801724698286,
      "grad_norm": 0.812006413936615,
      "learning_rate": 3.230743434087249e-06,
      "loss": 0.1899,
      "step": 40800
    },
    {
      "epoch": 0.7367704519027293,
      "grad_norm": 0.46172842383384705,
      "learning_rate": 3.2286560859464335e-06,
      "loss": 0.2729,
      "step": 40805
    },
    {
      "epoch": 0.73686073133563,
      "grad_norm": 0.41269052028656006,
      "learning_rate": 3.226569282511003e-06,
      "loss": 0.2156,
      "step": 40810
    },
    {
      "epoch": 0.7369510107685308,
      "grad_norm": 0.4248383045196533,
      "learning_rate": 3.2244830239488265e-06,
      "loss": 0.2346,
      "step": 40815
    },
    {
      "epoch": 0.7370412902014315,
      "grad_norm": 0.39304444193840027,
      "learning_rate": 3.2223973104277305e-06,
      "loss": 0.2313,
      "step": 40820
    },
    {
      "epoch": 0.7371315696343322,
      "grad_norm": 0.3386858403682709,
      "learning_rate": 3.2203121421154905e-06,
      "loss": 0.198,
      "step": 40825
    },
    {
      "epoch": 0.7372218490672329,
      "grad_norm": 0.8953399062156677,
      "learning_rate": 3.2182275191798483e-06,
      "loss": 0.223,
      "step": 40830
    },
    {
      "epoch": 0.7373121285001336,
      "grad_norm": 0.3431897759437561,
      "learning_rate": 3.216143441788492e-06,
      "loss": 0.2881,
      "step": 40835
    },
    {
      "epoch": 0.7374024079330344,
      "grad_norm": 0.5578610301017761,
      "learning_rate": 3.2140599101090706e-06,
      "loss": 0.2263,
      "step": 40840
    },
    {
      "epoch": 0.7374926873659351,
      "grad_norm": 0.4074469804763794,
      "learning_rate": 3.2119769243091925e-06,
      "loss": 0.2254,
      "step": 40845
    },
    {
      "epoch": 0.7375829667988357,
      "grad_norm": 0.3962375521659851,
      "learning_rate": 3.2098944845564196e-06,
      "loss": 0.2721,
      "step": 40850
    },
    {
      "epoch": 0.7376732462317365,
      "grad_norm": 0.37833014130592346,
      "learning_rate": 3.2078125910182634e-06,
      "loss": 0.158,
      "step": 40855
    },
    {
      "epoch": 0.7377635256646372,
      "grad_norm": 0.5450956225395203,
      "learning_rate": 3.2057312438622033e-06,
      "loss": 0.1927,
      "step": 40860
    },
    {
      "epoch": 0.737853805097538,
      "grad_norm": 0.42087388038635254,
      "learning_rate": 3.2036504432556627e-06,
      "loss": 0.2228,
      "step": 40865
    },
    {
      "epoch": 0.7379440845304386,
      "grad_norm": 0.47205621004104614,
      "learning_rate": 3.2015701893660323e-06,
      "loss": 0.2313,
      "step": 40870
    },
    {
      "epoch": 0.7380343639633393,
      "grad_norm": 0.36260485649108887,
      "learning_rate": 3.199490482360644e-06,
      "loss": 0.2527,
      "step": 40875
    },
    {
      "epoch": 0.7381246433962401,
      "grad_norm": 0.5770822763442993,
      "learning_rate": 3.1974113224068095e-06,
      "loss": 0.25,
      "step": 40880
    },
    {
      "epoch": 0.7382149228291408,
      "grad_norm": 0.23731869459152222,
      "learning_rate": 3.1953327096717713e-06,
      "loss": 0.1618,
      "step": 40885
    },
    {
      "epoch": 0.7383052022620414,
      "grad_norm": 0.33724477887153625,
      "learning_rate": 3.1932546443227443e-06,
      "loss": 0.2892,
      "step": 40890
    },
    {
      "epoch": 0.7383954816949422,
      "grad_norm": 0.46661415696144104,
      "learning_rate": 3.191177126526892e-06,
      "loss": 0.2314,
      "step": 40895
    },
    {
      "epoch": 0.7384857611278429,
      "grad_norm": 0.4015856981277466,
      "learning_rate": 3.189100156451331e-06,
      "loss": 0.2721,
      "step": 40900
    },
    {
      "epoch": 0.7385760405607437,
      "grad_norm": 0.4695940911769867,
      "learning_rate": 3.187023734263144e-06,
      "loss": 0.231,
      "step": 40905
    },
    {
      "epoch": 0.7386663199936443,
      "grad_norm": 0.46987631916999817,
      "learning_rate": 3.1849478601293603e-06,
      "loss": 0.2431,
      "step": 40910
    },
    {
      "epoch": 0.738756599426545,
      "grad_norm": 0.2321973294019699,
      "learning_rate": 3.1828725342169753e-06,
      "loss": 0.1568,
      "step": 40915
    },
    {
      "epoch": 0.7388468788594458,
      "grad_norm": 0.37069231271743774,
      "learning_rate": 3.180797756692927e-06,
      "loss": 0.1882,
      "step": 40920
    },
    {
      "epoch": 0.7389371582923465,
      "grad_norm": 0.3208655118942261,
      "learning_rate": 3.1787235277241213e-06,
      "loss": 0.2385,
      "step": 40925
    },
    {
      "epoch": 0.7390274377252471,
      "grad_norm": 0.3854255676269531,
      "learning_rate": 3.176649847477409e-06,
      "loss": 0.186,
      "step": 40930
    },
    {
      "epoch": 0.7391177171581479,
      "grad_norm": 0.4488772749900818,
      "learning_rate": 3.1745767161196094e-06,
      "loss": 0.2726,
      "step": 40935
    },
    {
      "epoch": 0.7392079965910486,
      "grad_norm": 0.6385711431503296,
      "learning_rate": 3.1725041338174844e-06,
      "loss": 0.2031,
      "step": 40940
    },
    {
      "epoch": 0.7392982760239494,
      "grad_norm": 0.6290650367736816,
      "learning_rate": 3.1704321007377603e-06,
      "loss": 0.2538,
      "step": 40945
    },
    {
      "epoch": 0.73938855545685,
      "grad_norm": 0.3569958806037903,
      "learning_rate": 3.168360617047118e-06,
      "loss": 0.4083,
      "step": 40950
    },
    {
      "epoch": 0.7394788348897507,
      "grad_norm": 0.7097567319869995,
      "learning_rate": 3.166289682912197e-06,
      "loss": 0.2972,
      "step": 40955
    },
    {
      "epoch": 0.7395691143226515,
      "grad_norm": 0.35611993074417114,
      "learning_rate": 3.164219298499582e-06,
      "loss": 0.2326,
      "step": 40960
    },
    {
      "epoch": 0.7396593937555522,
      "grad_norm": 0.4563051462173462,
      "learning_rate": 3.1621494639758266e-06,
      "loss": 0.2146,
      "step": 40965
    },
    {
      "epoch": 0.7397496731884529,
      "grad_norm": 0.3569638729095459,
      "learning_rate": 3.1600801795074276e-06,
      "loss": 0.3027,
      "step": 40970
    },
    {
      "epoch": 0.7398399526213536,
      "grad_norm": 0.4573405683040619,
      "learning_rate": 3.1580114452608468e-06,
      "loss": 0.2431,
      "step": 40975
    },
    {
      "epoch": 0.7399302320542543,
      "grad_norm": 0.3699833154678345,
      "learning_rate": 3.1559432614024997e-06,
      "loss": 0.1661,
      "step": 40980
    },
    {
      "epoch": 0.7400205114871551,
      "grad_norm": 0.4052470922470093,
      "learning_rate": 3.153875628098759e-06,
      "loss": 0.2584,
      "step": 40985
    },
    {
      "epoch": 0.7401107909200557,
      "grad_norm": 0.4142111837863922,
      "learning_rate": 3.1518085455159485e-06,
      "loss": 0.2181,
      "step": 40990
    },
    {
      "epoch": 0.7402010703529565,
      "grad_norm": 0.46312415599823,
      "learning_rate": 3.149742013820345e-06,
      "loss": 0.1124,
      "step": 40995
    },
    {
      "epoch": 0.7402913497858572,
      "grad_norm": 0.346750944852829,
      "learning_rate": 3.1476760331781943e-06,
      "loss": 0.2567,
      "step": 41000
    },
    {
      "epoch": 0.7403816292187579,
      "grad_norm": 0.40103939175605774,
      "learning_rate": 3.145610603755682e-06,
      "loss": 0.1935,
      "step": 41005
    },
    {
      "epoch": 0.7404719086516586,
      "grad_norm": 0.36617350578308105,
      "learning_rate": 3.1435457257189605e-06,
      "loss": 0.2619,
      "step": 41010
    },
    {
      "epoch": 0.7405621880845593,
      "grad_norm": 0.4069492518901825,
      "learning_rate": 3.141481399234134e-06,
      "loss": 0.1913,
      "step": 41015
    },
    {
      "epoch": 0.7406524675174601,
      "grad_norm": 0.41923683881759644,
      "learning_rate": 3.139417624467267e-06,
      "loss": 0.2532,
      "step": 41020
    },
    {
      "epoch": 0.7407427469503608,
      "grad_norm": 0.4082649052143097,
      "learning_rate": 3.137354401584366e-06,
      "loss": 0.1974,
      "step": 41025
    },
    {
      "epoch": 0.7408330263832614,
      "grad_norm": 0.406902551651001,
      "learning_rate": 3.1352917307514107e-06,
      "loss": 0.1547,
      "step": 41030
    },
    {
      "epoch": 0.7409233058161622,
      "grad_norm": 0.3015364706516266,
      "learning_rate": 3.133229612134322e-06,
      "loss": 0.2613,
      "step": 41035
    },
    {
      "epoch": 0.7410135852490629,
      "grad_norm": 0.23433537781238556,
      "learning_rate": 3.1311680458989847e-06,
      "loss": 0.1593,
      "step": 41040
    },
    {
      "epoch": 0.7411038646819637,
      "grad_norm": 0.694824755191803,
      "learning_rate": 3.129107032211236e-06,
      "loss": 0.2223,
      "step": 41045
    },
    {
      "epoch": 0.7411941441148643,
      "grad_norm": 0.8749660849571228,
      "learning_rate": 3.127046571236875e-06,
      "loss": 0.3951,
      "step": 41050
    },
    {
      "epoch": 0.741284423547765,
      "grad_norm": 0.5870570540428162,
      "learning_rate": 3.1249866631416414e-06,
      "loss": 0.2568,
      "step": 41055
    },
    {
      "epoch": 0.7413747029806658,
      "grad_norm": 0.6222681999206543,
      "learning_rate": 3.122927308091248e-06,
      "loss": 0.2059,
      "step": 41060
    },
    {
      "epoch": 0.7414649824135665,
      "grad_norm": 0.38352474570274353,
      "learning_rate": 3.120868506251349e-06,
      "loss": 0.2695,
      "step": 41065
    },
    {
      "epoch": 0.7415552618464671,
      "grad_norm": 0.5242384672164917,
      "learning_rate": 3.1188102577875657e-06,
      "loss": 0.181,
      "step": 41070
    },
    {
      "epoch": 0.7416455412793679,
      "grad_norm": 0.44780153036117554,
      "learning_rate": 3.1167525628654615e-06,
      "loss": 0.2373,
      "step": 41075
    },
    {
      "epoch": 0.7417358207122686,
      "grad_norm": 0.6195186972618103,
      "learning_rate": 3.1146954216505685e-06,
      "loss": 0.3313,
      "step": 41080
    },
    {
      "epoch": 0.7418261001451694,
      "grad_norm": 0.37485525012016296,
      "learning_rate": 3.1126388343083714e-06,
      "loss": 0.2259,
      "step": 41085
    },
    {
      "epoch": 0.74191637957807,
      "grad_norm": 0.5623783469200134,
      "learning_rate": 3.1105828010043013e-06,
      "loss": 0.2085,
      "step": 41090
    },
    {
      "epoch": 0.7420066590109707,
      "grad_norm": 0.540580153465271,
      "learning_rate": 3.108527321903756e-06,
      "loss": 0.1922,
      "step": 41095
    },
    {
      "epoch": 0.7420969384438715,
      "grad_norm": 0.3516944944858551,
      "learning_rate": 3.10647239717208e-06,
      "loss": 0.2026,
      "step": 41100
    },
    {
      "epoch": 0.7421872178767722,
      "grad_norm": 0.37779971957206726,
      "learning_rate": 3.1044180269745773e-06,
      "loss": 0.2623,
      "step": 41105
    },
    {
      "epoch": 0.7422774973096729,
      "grad_norm": 0.4179331660270691,
      "learning_rate": 3.1023642114765094e-06,
      "loss": 0.2363,
      "step": 41110
    },
    {
      "epoch": 0.7423677767425736,
      "grad_norm": 0.35786205530166626,
      "learning_rate": 3.100310950843093e-06,
      "loss": 0.2671,
      "step": 41115
    },
    {
      "epoch": 0.7424580561754743,
      "grad_norm": 0.35624074935913086,
      "learning_rate": 3.0982582452394927e-06,
      "loss": 0.3233,
      "step": 41120
    },
    {
      "epoch": 0.7425483356083751,
      "grad_norm": 0.29139894247055054,
      "learning_rate": 3.096206094830838e-06,
      "loss": 0.1685,
      "step": 41125
    },
    {
      "epoch": 0.7426386150412757,
      "grad_norm": 0.547553300857544,
      "learning_rate": 3.094154499782205e-06,
      "loss": 0.1926,
      "step": 41130
    },
    {
      "epoch": 0.7427288944741764,
      "grad_norm": 0.4268980622291565,
      "learning_rate": 3.0921034602586354e-06,
      "loss": 0.1671,
      "step": 41135
    },
    {
      "epoch": 0.7428191739070772,
      "grad_norm": 0.23809778690338135,
      "learning_rate": 3.090052976425111e-06,
      "loss": 0.1866,
      "step": 41140
    },
    {
      "epoch": 0.7429094533399779,
      "grad_norm": 0.4401637613773346,
      "learning_rate": 3.0880030484465896e-06,
      "loss": 0.2484,
      "step": 41145
    },
    {
      "epoch": 0.7429997327728786,
      "grad_norm": 0.6036712527275085,
      "learning_rate": 3.0859536764879662e-06,
      "loss": 0.3163,
      "step": 41150
    },
    {
      "epoch": 0.7430900122057793,
      "grad_norm": 0.3984891176223755,
      "learning_rate": 3.083904860714102e-06,
      "loss": 0.1377,
      "step": 41155
    },
    {
      "epoch": 0.74318029163868,
      "grad_norm": 0.47382864356040955,
      "learning_rate": 3.081856601289803e-06,
      "loss": 0.2435,
      "step": 41160
    },
    {
      "epoch": 0.7432705710715808,
      "grad_norm": 0.5753681659698486,
      "learning_rate": 3.0798088983798447e-06,
      "loss": 0.2523,
      "step": 41165
    },
    {
      "epoch": 0.7433608505044814,
      "grad_norm": 0.37586915493011475,
      "learning_rate": 3.0777617521489424e-06,
      "loss": 0.1963,
      "step": 41170
    },
    {
      "epoch": 0.7434511299373822,
      "grad_norm": 0.1946060210466385,
      "learning_rate": 3.0757151627617776e-06,
      "loss": 0.1702,
      "step": 41175
    },
    {
      "epoch": 0.7435414093702829,
      "grad_norm": 0.4404691755771637,
      "learning_rate": 3.073669130382988e-06,
      "loss": 0.2525,
      "step": 41180
    },
    {
      "epoch": 0.7436316888031836,
      "grad_norm": 0.639215350151062,
      "learning_rate": 3.0716236551771526e-06,
      "loss": 0.2088,
      "step": 41185
    },
    {
      "epoch": 0.7437219682360843,
      "grad_norm": 0.5188442468643188,
      "learning_rate": 3.0695787373088236e-06,
      "loss": 0.2945,
      "step": 41190
    },
    {
      "epoch": 0.743812247668985,
      "grad_norm": 0.2617374360561371,
      "learning_rate": 3.0675343769424936e-06,
      "loss": 0.2652,
      "step": 41195
    },
    {
      "epoch": 0.7439025271018858,
      "grad_norm": 0.4658057689666748,
      "learning_rate": 3.065490574242622e-06,
      "loss": 0.2763,
      "step": 41200
    },
    {
      "epoch": 0.7439928065347865,
      "grad_norm": 0.3671187162399292,
      "learning_rate": 3.0634473293736087e-06,
      "loss": 0.2873,
      "step": 41205
    },
    {
      "epoch": 0.7440830859676871,
      "grad_norm": 0.43951717019081116,
      "learning_rate": 3.06140464249983e-06,
      "loss": 0.1686,
      "step": 41210
    },
    {
      "epoch": 0.7441733654005879,
      "grad_norm": 0.5031009912490845,
      "learning_rate": 3.0593625137855965e-06,
      "loss": 0.1683,
      "step": 41215
    },
    {
      "epoch": 0.7442636448334886,
      "grad_norm": 0.4004661738872528,
      "learning_rate": 3.057320943395189e-06,
      "loss": 0.2477,
      "step": 41220
    },
    {
      "epoch": 0.7443539242663894,
      "grad_norm": 0.568417489528656,
      "learning_rate": 3.0552799314928295e-06,
      "loss": 0.2861,
      "step": 41225
    },
    {
      "epoch": 0.7444442036992901,
      "grad_norm": 0.3609898090362549,
      "learning_rate": 3.0532394782427098e-06,
      "loss": 0.2234,
      "step": 41230
    },
    {
      "epoch": 0.7445344831321907,
      "grad_norm": 0.30836221575737,
      "learning_rate": 3.0511995838089634e-06,
      "loss": 0.2409,
      "step": 41235
    },
    {
      "epoch": 0.7446247625650915,
      "grad_norm": 0.5751001834869385,
      "learning_rate": 3.049160248355688e-06,
      "loss": 0.1996,
      "step": 41240
    },
    {
      "epoch": 0.7447150419979922,
      "grad_norm": 0.24405145645141602,
      "learning_rate": 3.0471214720469324e-06,
      "loss": 0.1719,
      "step": 41245
    },
    {
      "epoch": 0.744805321430893,
      "grad_norm": 0.3759881556034088,
      "learning_rate": 3.0450832550467056e-06,
      "loss": 0.1612,
      "step": 41250
    },
    {
      "epoch": 0.7448956008637936,
      "grad_norm": 0.5241115093231201,
      "learning_rate": 3.043045597518961e-06,
      "loss": 0.2724,
      "step": 41255
    },
    {
      "epoch": 0.7449858802966943,
      "grad_norm": 0.49353301525115967,
      "learning_rate": 3.041008499627619e-06,
      "loss": 0.215,
      "step": 41260
    },
    {
      "epoch": 0.7450761597295951,
      "grad_norm": 0.22127854824066162,
      "learning_rate": 3.0389719615365465e-06,
      "loss": 0.224,
      "step": 41265
    },
    {
      "epoch": 0.7451664391624958,
      "grad_norm": 0.6090020537376404,
      "learning_rate": 3.0369359834095615e-06,
      "loss": 0.218,
      "step": 41270
    },
    {
      "epoch": 0.7452567185953964,
      "grad_norm": 0.9602436423301697,
      "learning_rate": 3.0349005654104567e-06,
      "loss": 0.2295,
      "step": 41275
    },
    {
      "epoch": 0.7453469980282972,
      "grad_norm": 0.3734925091266632,
      "learning_rate": 3.0328657077029577e-06,
      "loss": 0.1783,
      "step": 41280
    },
    {
      "epoch": 0.7454372774611979,
      "grad_norm": 0.35964828729629517,
      "learning_rate": 3.0308314104507597e-06,
      "loss": 0.1745,
      "step": 41285
    },
    {
      "epoch": 0.7455275568940987,
      "grad_norm": 0.44735097885131836,
      "learning_rate": 3.0287976738174995e-06,
      "loss": 0.2827,
      "step": 41290
    },
    {
      "epoch": 0.7456178363269993,
      "grad_norm": 0.3094846308231354,
      "learning_rate": 3.0267644979667854e-06,
      "loss": 0.2236,
      "step": 41295
    },
    {
      "epoch": 0.7457081157599,
      "grad_norm": 0.2760784924030304,
      "learning_rate": 3.024731883062163e-06,
      "loss": 0.2225,
      "step": 41300
    },
    {
      "epoch": 0.7457983951928008,
      "grad_norm": 0.49946433305740356,
      "learning_rate": 3.022699829267146e-06,
      "loss": 0.1663,
      "step": 41305
    },
    {
      "epoch": 0.7458886746257015,
      "grad_norm": 0.19271667301654816,
      "learning_rate": 3.0206683367451985e-06,
      "loss": 0.0996,
      "step": 41310
    },
    {
      "epoch": 0.7459789540586022,
      "grad_norm": 0.41054147481918335,
      "learning_rate": 3.018637405659741e-06,
      "loss": 0.2102,
      "step": 41315
    },
    {
      "epoch": 0.7460692334915029,
      "grad_norm": 0.49323397874832153,
      "learning_rate": 3.0166070361741417e-06,
      "loss": 0.2042,
      "step": 41320
    },
    {
      "epoch": 0.7461595129244036,
      "grad_norm": 0.3955391049385071,
      "learning_rate": 3.014577228451736e-06,
      "loss": 0.2037,
      "step": 41325
    },
    {
      "epoch": 0.7462497923573044,
      "grad_norm": 1.6140589714050293,
      "learning_rate": 3.012547982655799e-06,
      "loss": 0.2341,
      "step": 41330
    },
    {
      "epoch": 0.746340071790205,
      "grad_norm": 0.285859078168869,
      "learning_rate": 3.0105192989495736e-06,
      "loss": 0.2124,
      "step": 41335
    },
    {
      "epoch": 0.7464303512231057,
      "grad_norm": 0.41897910833358765,
      "learning_rate": 3.008491177496252e-06,
      "loss": 0.2757,
      "step": 41340
    },
    {
      "epoch": 0.7465206306560065,
      "grad_norm": 0.4707546532154083,
      "learning_rate": 3.006463618458986e-06,
      "loss": 0.2367,
      "step": 41345
    },
    {
      "epoch": 0.7466109100889072,
      "grad_norm": 0.5632999539375305,
      "learning_rate": 3.0044366220008738e-06,
      "loss": 0.28,
      "step": 41350
    },
    {
      "epoch": 0.7467011895218079,
      "grad_norm": 0.2610125243663788,
      "learning_rate": 3.002410188284969e-06,
      "loss": 0.1549,
      "step": 41355
    },
    {
      "epoch": 0.7467914689547086,
      "grad_norm": 0.5858786702156067,
      "learning_rate": 3.000384317474291e-06,
      "loss": 0.2081,
      "step": 41360
    },
    {
      "epoch": 0.7468817483876093,
      "grad_norm": 0.24929028749465942,
      "learning_rate": 2.998359009731798e-06,
      "loss": 0.2276,
      "step": 41365
    },
    {
      "epoch": 0.7469720278205101,
      "grad_norm": 0.3487606346607208,
      "learning_rate": 2.9963342652204163e-06,
      "loss": 0.1951,
      "step": 41370
    },
    {
      "epoch": 0.7470623072534107,
      "grad_norm": 0.5782601237297058,
      "learning_rate": 2.9943100841030225e-06,
      "loss": 0.1691,
      "step": 41375
    },
    {
      "epoch": 0.7471525866863115,
      "grad_norm": 0.4717338979244232,
      "learning_rate": 2.9922864665424487e-06,
      "loss": 0.2257,
      "step": 41380
    },
    {
      "epoch": 0.7472428661192122,
      "grad_norm": 0.22838842868804932,
      "learning_rate": 2.990263412701474e-06,
      "loss": 0.2086,
      "step": 41385
    },
    {
      "epoch": 0.7473331455521129,
      "grad_norm": 0.35909873247146606,
      "learning_rate": 2.9882409227428454e-06,
      "loss": 0.1488,
      "step": 41390
    },
    {
      "epoch": 0.7474234249850136,
      "grad_norm": 0.7137178778648376,
      "learning_rate": 2.986218996829251e-06,
      "loss": 0.3026,
      "step": 41395
    },
    {
      "epoch": 0.7475137044179143,
      "grad_norm": 0.4564216732978821,
      "learning_rate": 2.984197635123344e-06,
      "loss": 0.2638,
      "step": 41400
    },
    {
      "epoch": 0.7476039838508151,
      "grad_norm": 0.43131279945373535,
      "learning_rate": 2.982176837787727e-06,
      "loss": 0.1893,
      "step": 41405
    },
    {
      "epoch": 0.7476942632837158,
      "grad_norm": 0.41147229075431824,
      "learning_rate": 2.9801566049849617e-06,
      "loss": 0.2538,
      "step": 41410
    },
    {
      "epoch": 0.7477845427166164,
      "grad_norm": 0.5008199214935303,
      "learning_rate": 2.9781369368775546e-06,
      "loss": 0.3361,
      "step": 41415
    },
    {
      "epoch": 0.7478748221495172,
      "grad_norm": 0.4060599207878113,
      "learning_rate": 2.9761178336279817e-06,
      "loss": 0.2744,
      "step": 41420
    },
    {
      "epoch": 0.7479651015824179,
      "grad_norm": 0.46722060441970825,
      "learning_rate": 2.9740992953986558e-06,
      "loss": 0.2239,
      "step": 41425
    },
    {
      "epoch": 0.7480553810153187,
      "grad_norm": 0.2274867445230484,
      "learning_rate": 2.972081322351963e-06,
      "loss": 0.2617,
      "step": 41430
    },
    {
      "epoch": 0.7481456604482193,
      "grad_norm": 0.5291921496391296,
      "learning_rate": 2.9700639146502264e-06,
      "loss": 0.2695,
      "step": 41435
    },
    {
      "epoch": 0.74823593988112,
      "grad_norm": 0.5454273819923401,
      "learning_rate": 2.968047072455734e-06,
      "loss": 0.1875,
      "step": 41440
    },
    {
      "epoch": 0.7483262193140208,
      "grad_norm": 0.3729475736618042,
      "learning_rate": 2.966030795930732e-06,
      "loss": 0.1642,
      "step": 41445
    },
    {
      "epoch": 0.7484164987469215,
      "grad_norm": 0.45739078521728516,
      "learning_rate": 2.964015085237407e-06,
      "loss": 0.2755,
      "step": 41450
    },
    {
      "epoch": 0.7485067781798221,
      "grad_norm": 0.2924373149871826,
      "learning_rate": 2.961999940537915e-06,
      "loss": 0.2015,
      "step": 41455
    },
    {
      "epoch": 0.7485970576127229,
      "grad_norm": 0.296959787607193,
      "learning_rate": 2.959985361994353e-06,
      "loss": 0.2333,
      "step": 41460
    },
    {
      "epoch": 0.7486873370456236,
      "grad_norm": 0.37825894355773926,
      "learning_rate": 2.9579713497687866e-06,
      "loss": 0.2877,
      "step": 41465
    },
    {
      "epoch": 0.7487776164785244,
      "grad_norm": 0.5115907788276672,
      "learning_rate": 2.955957904023218e-06,
      "loss": 0.3644,
      "step": 41470
    },
    {
      "epoch": 0.748867895911425,
      "grad_norm": 0.48065969347953796,
      "learning_rate": 2.9539450249196267e-06,
      "loss": 0.2066,
      "step": 41475
    },
    {
      "epoch": 0.7489581753443257,
      "grad_norm": 0.5366142392158508,
      "learning_rate": 2.9519327126199258e-06,
      "loss": 0.1929,
      "step": 41480
    },
    {
      "epoch": 0.7490484547772265,
      "grad_norm": 0.597087562084198,
      "learning_rate": 2.9499209672859963e-06,
      "loss": 0.2874,
      "step": 41485
    },
    {
      "epoch": 0.7491387342101272,
      "grad_norm": 0.35067883133888245,
      "learning_rate": 2.947909789079664e-06,
      "loss": 0.1681,
      "step": 41490
    },
    {
      "epoch": 0.7492290136430279,
      "grad_norm": 0.4782887101173401,
      "learning_rate": 2.945899178162718e-06,
      "loss": 0.269,
      "step": 41495
    },
    {
      "epoch": 0.7493192930759286,
      "grad_norm": 0.4069378674030304,
      "learning_rate": 2.9438891346968913e-06,
      "loss": 0.2355,
      "step": 41500
    },
    {
      "epoch": 0.7494095725088293,
      "grad_norm": 0.2894822955131531,
      "learning_rate": 2.9418796588438814e-06,
      "loss": 0.2093,
      "step": 41505
    },
    {
      "epoch": 0.7494998519417301,
      "grad_norm": 0.45250555872917175,
      "learning_rate": 2.9398707507653345e-06,
      "loss": 0.2883,
      "step": 41510
    },
    {
      "epoch": 0.7495901313746307,
      "grad_norm": 0.3528308868408203,
      "learning_rate": 2.937862410622858e-06,
      "loss": 0.1596,
      "step": 41515
    },
    {
      "epoch": 0.7496804108075314,
      "grad_norm": 0.3167407214641571,
      "learning_rate": 2.9358546385780006e-06,
      "loss": 0.2779,
      "step": 41520
    },
    {
      "epoch": 0.7497706902404322,
      "grad_norm": 0.5783354043960571,
      "learning_rate": 2.9338474347922787e-06,
      "loss": 0.1904,
      "step": 41525
    },
    {
      "epoch": 0.7498609696733329,
      "grad_norm": 0.35650262236595154,
      "learning_rate": 2.9318407994271514e-06,
      "loss": 0.2287,
      "step": 41530
    },
    {
      "epoch": 0.7499512491062336,
      "grad_norm": 0.5059142112731934,
      "learning_rate": 2.9298347326440424e-06,
      "loss": 0.2607,
      "step": 41535
    },
    {
      "epoch": 0.7500415285391343,
      "grad_norm": 0.4670112729072571,
      "learning_rate": 2.9278292346043268e-06,
      "loss": 0.2889,
      "step": 41540
    },
    {
      "epoch": 0.750131807972035,
      "grad_norm": 0.3478320837020874,
      "learning_rate": 2.9258243054693257e-06,
      "loss": 0.2606,
      "step": 41545
    },
    {
      "epoch": 0.7502220874049358,
      "grad_norm": 0.5915669798851013,
      "learning_rate": 2.923819945400329e-06,
      "loss": 0.2795,
      "step": 41550
    },
    {
      "epoch": 0.7503123668378364,
      "grad_norm": 0.37255313992500305,
      "learning_rate": 2.9218161545585654e-06,
      "loss": 0.1787,
      "step": 41555
    },
    {
      "epoch": 0.7504026462707372,
      "grad_norm": 0.3664399981498718,
      "learning_rate": 2.9198129331052318e-06,
      "loss": 0.1552,
      "step": 41560
    },
    {
      "epoch": 0.7504929257036379,
      "grad_norm": 0.32734349370002747,
      "learning_rate": 2.9178102812014674e-06,
      "loss": 0.1657,
      "step": 41565
    },
    {
      "epoch": 0.7505832051365386,
      "grad_norm": 0.3999686539173126,
      "learning_rate": 2.9158081990083732e-06,
      "loss": 0.1633,
      "step": 41570
    },
    {
      "epoch": 0.7506734845694393,
      "grad_norm": 0.4873049557209015,
      "learning_rate": 2.913806686687002e-06,
      "loss": 0.2261,
      "step": 41575
    },
    {
      "epoch": 0.75076376400234,
      "grad_norm": 0.4038124978542328,
      "learning_rate": 2.9118057443983637e-06,
      "loss": 0.2136,
      "step": 41580
    },
    {
      "epoch": 0.7508540434352408,
      "grad_norm": 0.5271873474121094,
      "learning_rate": 2.909805372303415e-06,
      "loss": 0.2043,
      "step": 41585
    },
    {
      "epoch": 0.7509443228681415,
      "grad_norm": 0.3286977708339691,
      "learning_rate": 2.9078055705630758e-06,
      "loss": 0.2283,
      "step": 41590
    },
    {
      "epoch": 0.7510346023010421,
      "grad_norm": 0.6366809010505676,
      "learning_rate": 2.905806339338211e-06,
      "loss": 0.2817,
      "step": 41595
    },
    {
      "epoch": 0.7511248817339429,
      "grad_norm": 0.24309015274047852,
      "learning_rate": 2.903807678789646e-06,
      "loss": 0.2005,
      "step": 41600
    },
    {
      "epoch": 0.7512151611668436,
      "grad_norm": 0.23062177002429962,
      "learning_rate": 2.9018095890781585e-06,
      "loss": 0.266,
      "step": 41605
    },
    {
      "epoch": 0.7513054405997444,
      "grad_norm": 0.533152163028717,
      "learning_rate": 2.899812070364484e-06,
      "loss": 0.2617,
      "step": 41610
    },
    {
      "epoch": 0.7513957200326451,
      "grad_norm": 0.4984569847583771,
      "learning_rate": 2.897815122809302e-06,
      "loss": 0.2927,
      "step": 41615
    },
    {
      "epoch": 0.7514859994655457,
      "grad_norm": 0.4147002398967743,
      "learning_rate": 2.895818746573259e-06,
      "loss": 0.2625,
      "step": 41620
    },
    {
      "epoch": 0.7515762788984465,
      "grad_norm": 0.515230655670166,
      "learning_rate": 2.893822941816945e-06,
      "loss": 0.168,
      "step": 41625
    },
    {
      "epoch": 0.7516665583313472,
      "grad_norm": 0.4705766439437866,
      "learning_rate": 2.891827708700906e-06,
      "loss": 0.2495,
      "step": 41630
    },
    {
      "epoch": 0.751756837764248,
      "grad_norm": 0.2987738847732544,
      "learning_rate": 2.889833047385646e-06,
      "loss": 0.2963,
      "step": 41635
    },
    {
      "epoch": 0.7518471171971486,
      "grad_norm": 0.6711522340774536,
      "learning_rate": 2.887838958031621e-06,
      "loss": 0.1937,
      "step": 41640
    },
    {
      "epoch": 0.7519373966300493,
      "grad_norm": 0.5347040891647339,
      "learning_rate": 2.8858454407992455e-06,
      "loss": 0.2265,
      "step": 41645
    },
    {
      "epoch": 0.7520276760629501,
      "grad_norm": 0.8567517399787903,
      "learning_rate": 2.8838524958488754e-06,
      "loss": 0.2047,
      "step": 41650
    },
    {
      "epoch": 0.7521179554958508,
      "grad_norm": 0.3282148838043213,
      "learning_rate": 2.881860123340836e-06,
      "loss": 0.1848,
      "step": 41655
    },
    {
      "epoch": 0.7522082349287514,
      "grad_norm": 0.42307010293006897,
      "learning_rate": 2.879868323435393e-06,
      "loss": 0.1674,
      "step": 41660
    },
    {
      "epoch": 0.7522985143616522,
      "grad_norm": 0.5557116270065308,
      "learning_rate": 2.8778770962927748e-06,
      "loss": 0.1023,
      "step": 41665
    },
    {
      "epoch": 0.7523887937945529,
      "grad_norm": 0.6449236273765564,
      "learning_rate": 2.8758864420731612e-06,
      "loss": 0.2737,
      "step": 41670
    },
    {
      "epoch": 0.7524790732274537,
      "grad_norm": 0.24184074997901917,
      "learning_rate": 2.873896360936689e-06,
      "loss": 0.1679,
      "step": 41675
    },
    {
      "epoch": 0.7525693526603543,
      "grad_norm": 0.7325660586357117,
      "learning_rate": 2.8719068530434403e-06,
      "loss": 0.2481,
      "step": 41680
    },
    {
      "epoch": 0.752659632093255,
      "grad_norm": 0.3087085485458374,
      "learning_rate": 2.869917918553462e-06,
      "loss": 0.1987,
      "step": 41685
    },
    {
      "epoch": 0.7527499115261558,
      "grad_norm": 0.5672470927238464,
      "learning_rate": 2.867929557626743e-06,
      "loss": 0.1882,
      "step": 41690
    },
    {
      "epoch": 0.7528401909590565,
      "grad_norm": 0.6781942248344421,
      "learning_rate": 2.8659417704232383e-06,
      "loss": 0.1773,
      "step": 41695
    },
    {
      "epoch": 0.7529304703919572,
      "grad_norm": 0.3354477882385254,
      "learning_rate": 2.863954557102847e-06,
      "loss": 0.1943,
      "step": 41700
    },
    {
      "epoch": 0.7530207498248579,
      "grad_norm": 0.34309548139572144,
      "learning_rate": 2.8619679178254267e-06,
      "loss": 0.208,
      "step": 41705
    },
    {
      "epoch": 0.7531110292577586,
      "grad_norm": 0.4000491201877594,
      "learning_rate": 2.8599818527507896e-06,
      "loss": 0.1707,
      "step": 41710
    },
    {
      "epoch": 0.7532013086906594,
      "grad_norm": 0.5786740183830261,
      "learning_rate": 2.8579963620387007e-06,
      "loss": 0.2568,
      "step": 41715
    },
    {
      "epoch": 0.75329158812356,
      "grad_norm": 0.4471411406993866,
      "learning_rate": 2.8560114458488784e-06,
      "loss": 0.181,
      "step": 41720
    },
    {
      "epoch": 0.7533818675564607,
      "grad_norm": 0.5158851742744446,
      "learning_rate": 2.8540271043409907e-06,
      "loss": 0.2814,
      "step": 41725
    },
    {
      "epoch": 0.7534721469893615,
      "grad_norm": 0.2408069670200348,
      "learning_rate": 2.852043337674666e-06,
      "loss": 0.2056,
      "step": 41730
    },
    {
      "epoch": 0.7535624264222622,
      "grad_norm": 0.33224543929100037,
      "learning_rate": 2.8500601460094847e-06,
      "loss": 0.2299,
      "step": 41735
    },
    {
      "epoch": 0.7536527058551629,
      "grad_norm": 1.0748296976089478,
      "learning_rate": 2.848077529504981e-06,
      "loss": 0.2426,
      "step": 41740
    },
    {
      "epoch": 0.7537429852880636,
      "grad_norm": 0.5358622670173645,
      "learning_rate": 2.8460954883206383e-06,
      "loss": 0.2196,
      "step": 41745
    },
    {
      "epoch": 0.7538332647209643,
      "grad_norm": 0.6004823446273804,
      "learning_rate": 2.8441140226159038e-06,
      "loss": 0.289,
      "step": 41750
    },
    {
      "epoch": 0.7539235441538651,
      "grad_norm": 0.4540747106075287,
      "learning_rate": 2.8421331325501633e-06,
      "loss": 0.2017,
      "step": 41755
    },
    {
      "epoch": 0.7540138235867657,
      "grad_norm": 0.2679804563522339,
      "learning_rate": 2.840152818282773e-06,
      "loss": 0.2331,
      "step": 41760
    },
    {
      "epoch": 0.7541041030196665,
      "grad_norm": 0.39254021644592285,
      "learning_rate": 2.8381730799730287e-06,
      "loss": 0.1945,
      "step": 41765
    },
    {
      "epoch": 0.7541943824525672,
      "grad_norm": 0.7232952117919922,
      "learning_rate": 2.836193917780188e-06,
      "loss": 0.2052,
      "step": 41770
    },
    {
      "epoch": 0.7542846618854679,
      "grad_norm": 0.43765270709991455,
      "learning_rate": 2.8342153318634613e-06,
      "loss": 0.2896,
      "step": 41775
    },
    {
      "epoch": 0.7543749413183686,
      "grad_norm": 0.88607257604599,
      "learning_rate": 2.8322373223820134e-06,
      "loss": 0.2851,
      "step": 41780
    },
    {
      "epoch": 0.7544652207512693,
      "grad_norm": 0.5815815329551697,
      "learning_rate": 2.830259889494955e-06,
      "loss": 0.2658,
      "step": 41785
    },
    {
      "epoch": 0.7545555001841701,
      "grad_norm": 0.666411280632019,
      "learning_rate": 2.828283033361363e-06,
      "loss": 0.1633,
      "step": 41790
    },
    {
      "epoch": 0.7546457796170708,
      "grad_norm": 0.5315019488334656,
      "learning_rate": 2.8263067541402543e-06,
      "loss": 0.1446,
      "step": 41795
    },
    {
      "epoch": 0.7547360590499714,
      "grad_norm": 0.44192075729370117,
      "learning_rate": 2.8243310519906096e-06,
      "loss": 0.1463,
      "step": 41800
    },
    {
      "epoch": 0.7548263384828722,
      "grad_norm": 0.3611176609992981,
      "learning_rate": 2.8223559270713597e-06,
      "loss": 0.2946,
      "step": 41805
    },
    {
      "epoch": 0.7549166179157729,
      "grad_norm": 0.6399714350700378,
      "learning_rate": 2.8203813795413926e-06,
      "loss": 0.1436,
      "step": 41810
    },
    {
      "epoch": 0.7550068973486737,
      "grad_norm": 0.25619006156921387,
      "learning_rate": 2.8184074095595416e-06,
      "loss": 0.2205,
      "step": 41815
    },
    {
      "epoch": 0.7550971767815743,
      "grad_norm": 0.4509105384349823,
      "learning_rate": 2.816434017284596e-06,
      "loss": 0.2073,
      "step": 41820
    },
    {
      "epoch": 0.755187456214475,
      "grad_norm": 0.6260651350021362,
      "learning_rate": 2.814461202875308e-06,
      "loss": 0.1968,
      "step": 41825
    },
    {
      "epoch": 0.7552777356473758,
      "grad_norm": 0.45519912242889404,
      "learning_rate": 2.81248896649037e-06,
      "loss": 0.2017,
      "step": 41830
    },
    {
      "epoch": 0.7553680150802765,
      "grad_norm": 0.5440272688865662,
      "learning_rate": 2.810517308288435e-06,
      "loss": 0.2817,
      "step": 41835
    },
    {
      "epoch": 0.7554582945131771,
      "grad_norm": 0.5860805511474609,
      "learning_rate": 2.8085462284281094e-06,
      "loss": 0.1735,
      "step": 41840
    },
    {
      "epoch": 0.7555485739460779,
      "grad_norm": 0.38434216380119324,
      "learning_rate": 2.806575727067957e-06,
      "loss": 0.2317,
      "step": 41845
    },
    {
      "epoch": 0.7556388533789786,
      "grad_norm": 0.345718115568161,
      "learning_rate": 2.8046058043664837e-06,
      "loss": 0.2065,
      "step": 41850
    },
    {
      "epoch": 0.7557291328118794,
      "grad_norm": 0.6720041632652283,
      "learning_rate": 2.8026364604821597e-06,
      "loss": 0.2969,
      "step": 41855
    },
    {
      "epoch": 0.75581941224478,
      "grad_norm": 0.36170029640197754,
      "learning_rate": 2.8006676955734e-06,
      "loss": 0.1878,
      "step": 41860
    },
    {
      "epoch": 0.7559096916776807,
      "grad_norm": 0.4976137578487396,
      "learning_rate": 2.798699509798579e-06,
      "loss": 0.2749,
      "step": 41865
    },
    {
      "epoch": 0.7559999711105815,
      "grad_norm": 0.23476167023181915,
      "learning_rate": 2.7967319033160255e-06,
      "loss": 0.278,
      "step": 41870
    },
    {
      "epoch": 0.7560902505434822,
      "grad_norm": 0.7942454218864441,
      "learning_rate": 2.7947648762840194e-06,
      "loss": 0.2802,
      "step": 41875
    },
    {
      "epoch": 0.7561805299763829,
      "grad_norm": 0.32001301646232605,
      "learning_rate": 2.79279842886079e-06,
      "loss": 0.2193,
      "step": 41880
    },
    {
      "epoch": 0.7562708094092836,
      "grad_norm": 0.44344019889831543,
      "learning_rate": 2.7908325612045283e-06,
      "loss": 0.1445,
      "step": 41885
    },
    {
      "epoch": 0.7563610888421843,
      "grad_norm": 0.4995500445365906,
      "learning_rate": 2.788867273473368e-06,
      "loss": 0.266,
      "step": 41890
    },
    {
      "epoch": 0.7564513682750851,
      "grad_norm": 0.4097002148628235,
      "learning_rate": 2.786902565825408e-06,
      "loss": 0.2314,
      "step": 41895
    },
    {
      "epoch": 0.7565416477079857,
      "grad_norm": 0.5643100142478943,
      "learning_rate": 2.7849384384186893e-06,
      "loss": 0.2094,
      "step": 41900
    },
    {
      "epoch": 0.7566319271408865,
      "grad_norm": 0.3898535668849945,
      "learning_rate": 2.7829748914112154e-06,
      "loss": 0.3218,
      "step": 41905
    },
    {
      "epoch": 0.7567222065737872,
      "grad_norm": 0.5345604419708252,
      "learning_rate": 2.7810119249609413e-06,
      "loss": 0.1929,
      "step": 41910
    },
    {
      "epoch": 0.7568124860066879,
      "grad_norm": 0.46164649724960327,
      "learning_rate": 2.7790495392257675e-06,
      "loss": 0.1711,
      "step": 41915
    },
    {
      "epoch": 0.7569027654395886,
      "grad_norm": 0.1710173636674881,
      "learning_rate": 2.7770877343635603e-06,
      "loss": 0.227,
      "step": 41920
    },
    {
      "epoch": 0.7569930448724893,
      "grad_norm": 0.7308053970336914,
      "learning_rate": 2.7751265105321266e-06,
      "loss": 0.1467,
      "step": 41925
    },
    {
      "epoch": 0.75708332430539,
      "grad_norm": 0.5215123295783997,
      "learning_rate": 2.7731658678892346e-06,
      "loss": 0.2196,
      "step": 41930
    },
    {
      "epoch": 0.7571736037382908,
      "grad_norm": 0.29790088534355164,
      "learning_rate": 2.771205806592605e-06,
      "loss": 0.1777,
      "step": 41935
    },
    {
      "epoch": 0.7572638831711914,
      "grad_norm": 0.5989565253257751,
      "learning_rate": 2.7692463267999127e-06,
      "loss": 0.2497,
      "step": 41940
    },
    {
      "epoch": 0.7573541626040922,
      "grad_norm": 0.5519168972969055,
      "learning_rate": 2.767287428668777e-06,
      "loss": 0.216,
      "step": 41945
    },
    {
      "epoch": 0.7574444420369929,
      "grad_norm": 0.7208122611045837,
      "learning_rate": 2.765329112356784e-06,
      "loss": 0.2529,
      "step": 41950
    },
    {
      "epoch": 0.7575347214698936,
      "grad_norm": 0.5940227508544922,
      "learning_rate": 2.76337137802146e-06,
      "loss": 0.2009,
      "step": 41955
    },
    {
      "epoch": 0.7576250009027943,
      "grad_norm": 0.49036723375320435,
      "learning_rate": 2.7614142258202957e-06,
      "loss": 0.1871,
      "step": 41960
    },
    {
      "epoch": 0.757715280335695,
      "grad_norm": 0.37940332293510437,
      "learning_rate": 2.7594576559107254e-06,
      "loss": 0.1681,
      "step": 41965
    },
    {
      "epoch": 0.7578055597685958,
      "grad_norm": 0.8369633555412292,
      "learning_rate": 2.757501668450142e-06,
      "loss": 0.1977,
      "step": 41970
    },
    {
      "epoch": 0.7578958392014965,
      "grad_norm": 0.5507857203483582,
      "learning_rate": 2.755546263595892e-06,
      "loss": 0.2498,
      "step": 41975
    },
    {
      "epoch": 0.7579861186343971,
      "grad_norm": 0.38975781202316284,
      "learning_rate": 2.7535914415052754e-06,
      "loss": 0.1758,
      "step": 41980
    },
    {
      "epoch": 0.7580763980672979,
      "grad_norm": 0.5902056694030762,
      "learning_rate": 2.7516372023355374e-06,
      "loss": 0.1492,
      "step": 41985
    },
    {
      "epoch": 0.7581666775001986,
      "grad_norm": 0.5487368702888489,
      "learning_rate": 2.74968354624389e-06,
      "loss": 0.1325,
      "step": 41990
    },
    {
      "epoch": 0.7582569569330994,
      "grad_norm": 0.41757887601852417,
      "learning_rate": 2.7477304733874823e-06,
      "loss": 0.0984,
      "step": 41995
    },
    {
      "epoch": 0.758347236366,
      "grad_norm": 1.0859988927841187,
      "learning_rate": 2.74577798392343e-06,
      "loss": 0.2213,
      "step": 42000
    },
    {
      "epoch": 0.7584375157989007,
      "grad_norm": 0.35026293992996216,
      "learning_rate": 2.7438260780087975e-06,
      "loss": 0.1343,
      "step": 42005
    },
    {
      "epoch": 0.7585277952318015,
      "grad_norm": 0.48994943499565125,
      "learning_rate": 2.7418747558005964e-06,
      "loss": 0.2131,
      "step": 42010
    },
    {
      "epoch": 0.7586180746647022,
      "grad_norm": 0.44541123509407043,
      "learning_rate": 2.739924017455804e-06,
      "loss": 0.2064,
      "step": 42015
    },
    {
      "epoch": 0.758708354097603,
      "grad_norm": 0.34418150782585144,
      "learning_rate": 2.7379738631313346e-06,
      "loss": 0.1252,
      "step": 42020
    },
    {
      "epoch": 0.7587986335305036,
      "grad_norm": 0.3524698317050934,
      "learning_rate": 2.7360242929840708e-06,
      "loss": 0.2012,
      "step": 42025
    },
    {
      "epoch": 0.7588889129634043,
      "grad_norm": 0.907588541507721,
      "learning_rate": 2.734075307170835e-06,
      "loss": 0.2177,
      "step": 42030
    },
    {
      "epoch": 0.7589791923963051,
      "grad_norm": 0.4300106167793274,
      "learning_rate": 2.7321269058484133e-06,
      "loss": 0.2384,
      "step": 42035
    },
    {
      "epoch": 0.7590694718292058,
      "grad_norm": 0.2557997405529022,
      "learning_rate": 2.7301790891735392e-06,
      "loss": 0.1753,
      "step": 42040
    },
    {
      "epoch": 0.7591597512621064,
      "grad_norm": 0.38408249616622925,
      "learning_rate": 2.728231857302903e-06,
      "loss": 0.1843,
      "step": 42045
    },
    {
      "epoch": 0.7592500306950072,
      "grad_norm": 0.5381377935409546,
      "learning_rate": 2.72628521039314e-06,
      "loss": 0.1846,
      "step": 42050
    },
    {
      "epoch": 0.7593403101279079,
      "grad_norm": 0.4469461739063263,
      "learning_rate": 2.724339148600851e-06,
      "loss": 0.2237,
      "step": 42055
    },
    {
      "epoch": 0.7594305895608087,
      "grad_norm": 0.5588510036468506,
      "learning_rate": 2.722393672082575e-06,
      "loss": 0.2207,
      "step": 42060
    },
    {
      "epoch": 0.7595208689937093,
      "grad_norm": 0.5202145576477051,
      "learning_rate": 2.7204487809948145e-06,
      "loss": 0.1477,
      "step": 42065
    },
    {
      "epoch": 0.75961114842661,
      "grad_norm": 1.5417118072509766,
      "learning_rate": 2.7185044754940215e-06,
      "loss": 0.2629,
      "step": 42070
    },
    {
      "epoch": 0.7597014278595108,
      "grad_norm": 0.6991029977798462,
      "learning_rate": 2.7165607557366057e-06,
      "loss": 0.2387,
      "step": 42075
    },
    {
      "epoch": 0.7597917072924115,
      "grad_norm": 0.650897204875946,
      "learning_rate": 2.714617621878918e-06,
      "loss": 0.195,
      "step": 42080
    },
    {
      "epoch": 0.7598819867253122,
      "grad_norm": 0.5020387768745422,
      "learning_rate": 2.7126750740772755e-06,
      "loss": 0.1731,
      "step": 42085
    },
    {
      "epoch": 0.7599722661582129,
      "grad_norm": 0.4681573510169983,
      "learning_rate": 2.71073311248794e-06,
      "loss": 0.1819,
      "step": 42090
    },
    {
      "epoch": 0.7600625455911136,
      "grad_norm": 0.42313507199287415,
      "learning_rate": 2.7087917372671248e-06,
      "loss": 0.2363,
      "step": 42095
    },
    {
      "epoch": 0.7601528250240144,
      "grad_norm": 0.2385631799697876,
      "learning_rate": 2.7068509485710004e-06,
      "loss": 0.2184,
      "step": 42100
    },
    {
      "epoch": 0.760243104456915,
      "grad_norm": 0.3899441957473755,
      "learning_rate": 2.704910746555692e-06,
      "loss": 0.2377,
      "step": 42105
    },
    {
      "epoch": 0.7603333838898158,
      "grad_norm": 0.5357471704483032,
      "learning_rate": 2.7029711313772767e-06,
      "loss": 0.2082,
      "step": 42110
    },
    {
      "epoch": 0.7604236633227165,
      "grad_norm": 0.5904387831687927,
      "learning_rate": 2.7010321031917765e-06,
      "loss": 0.2638,
      "step": 42115
    },
    {
      "epoch": 0.7605139427556172,
      "grad_norm": 0.6223477125167847,
      "learning_rate": 2.699093662155178e-06,
      "loss": 0.2409,
      "step": 42120
    },
    {
      "epoch": 0.7606042221885179,
      "grad_norm": 0.468454509973526,
      "learning_rate": 2.6971558084234085e-06,
      "loss": 0.1239,
      "step": 42125
    },
    {
      "epoch": 0.7606945016214186,
      "grad_norm": 0.5519012808799744,
      "learning_rate": 2.695218542152357e-06,
      "loss": 0.2196,
      "step": 42130
    },
    {
      "epoch": 0.7607847810543193,
      "grad_norm": 0.5590959787368774,
      "learning_rate": 2.693281863497863e-06,
      "loss": 0.218,
      "step": 42135
    },
    {
      "epoch": 0.7608750604872201,
      "grad_norm": 0.6511601805686951,
      "learning_rate": 2.691345772615721e-06,
      "loss": 0.263,
      "step": 42140
    },
    {
      "epoch": 0.7609653399201207,
      "grad_norm": 0.33294880390167236,
      "learning_rate": 2.6894102696616707e-06,
      "loss": 0.2152,
      "step": 42145
    },
    {
      "epoch": 0.7610556193530215,
      "grad_norm": 0.2821709215641022,
      "learning_rate": 2.6874753547914132e-06,
      "loss": 0.1989,
      "step": 42150
    },
    {
      "epoch": 0.7611458987859222,
      "grad_norm": 0.4125169515609741,
      "learning_rate": 2.6855410281605933e-06,
      "loss": 0.1377,
      "step": 42155
    },
    {
      "epoch": 0.761236178218823,
      "grad_norm": 0.7620353102684021,
      "learning_rate": 2.683607289924819e-06,
      "loss": 0.2553,
      "step": 42160
    },
    {
      "epoch": 0.7613264576517236,
      "grad_norm": 0.40569189190864563,
      "learning_rate": 2.6816741402396407e-06,
      "loss": 0.2181,
      "step": 42165
    },
    {
      "epoch": 0.7614167370846243,
      "grad_norm": 0.4498937427997589,
      "learning_rate": 2.6797415792605673e-06,
      "loss": 0.3055,
      "step": 42170
    },
    {
      "epoch": 0.7615070165175251,
      "grad_norm": 0.4452691674232483,
      "learning_rate": 2.6778096071430616e-06,
      "loss": 0.349,
      "step": 42175
    },
    {
      "epoch": 0.7615972959504258,
      "grad_norm": 0.4897069036960602,
      "learning_rate": 2.6758782240425374e-06,
      "loss": 0.1811,
      "step": 42180
    },
    {
      "epoch": 0.7616875753833264,
      "grad_norm": 0.7254247665405273,
      "learning_rate": 2.67394743011436e-06,
      "loss": 0.2535,
      "step": 42185
    },
    {
      "epoch": 0.7617778548162272,
      "grad_norm": 0.5436791181564331,
      "learning_rate": 2.672017225513842e-06,
      "loss": 0.1872,
      "step": 42190
    },
    {
      "epoch": 0.7618681342491279,
      "grad_norm": 0.5768334269523621,
      "learning_rate": 2.670087610396259e-06,
      "loss": 0.3027,
      "step": 42195
    },
    {
      "epoch": 0.7619584136820287,
      "grad_norm": 0.5182393193244934,
      "learning_rate": 2.6681585849168346e-06,
      "loss": 0.2969,
      "step": 42200
    },
    {
      "epoch": 0.7620486931149293,
      "grad_norm": 0.3409898281097412,
      "learning_rate": 2.6662301492307474e-06,
      "loss": 0.2224,
      "step": 42205
    },
    {
      "epoch": 0.76213897254783,
      "grad_norm": 0.39048269391059875,
      "learning_rate": 2.6643023034931193e-06,
      "loss": 0.262,
      "step": 42210
    },
    {
      "epoch": 0.7622292519807308,
      "grad_norm": 0.28336530923843384,
      "learning_rate": 2.662375047859039e-06,
      "loss": 0.2134,
      "step": 42215
    },
    {
      "epoch": 0.7623195314136315,
      "grad_norm": 0.663398265838623,
      "learning_rate": 2.6604483824835325e-06,
      "loss": 0.2351,
      "step": 42220
    },
    {
      "epoch": 0.7624098108465321,
      "grad_norm": 0.13834026455879211,
      "learning_rate": 2.6585223075215937e-06,
      "loss": 0.1265,
      "step": 42225
    },
    {
      "epoch": 0.7625000902794329,
      "grad_norm": 0.4230329692363739,
      "learning_rate": 2.6565968231281545e-06,
      "loss": 0.203,
      "step": 42230
    },
    {
      "epoch": 0.7625903697123336,
      "grad_norm": 0.3502117395401001,
      "learning_rate": 2.654671929458109e-06,
      "loss": 0.1546,
      "step": 42235
    },
    {
      "epoch": 0.7626806491452344,
      "grad_norm": 0.5848438739776611,
      "learning_rate": 2.652747626666301e-06,
      "loss": 0.2122,
      "step": 42240
    },
    {
      "epoch": 0.762770928578135,
      "grad_norm": 0.2812301516532898,
      "learning_rate": 2.65082391490753e-06,
      "loss": 0.2429,
      "step": 42245
    },
    {
      "epoch": 0.7628612080110357,
      "grad_norm": 0.6477394700050354,
      "learning_rate": 2.6489007943365388e-06,
      "loss": 0.2137,
      "step": 42250
    },
    {
      "epoch": 0.7629514874439365,
      "grad_norm": 0.09040392935276031,
      "learning_rate": 2.6469782651080323e-06,
      "loss": 0.1736,
      "step": 42255
    },
    {
      "epoch": 0.7630417668768372,
      "grad_norm": 0.4947648048400879,
      "learning_rate": 2.6450563273766615e-06,
      "loss": 0.2682,
      "step": 42260
    },
    {
      "epoch": 0.7631320463097379,
      "grad_norm": 0.48899325728416443,
      "learning_rate": 2.643134981297033e-06,
      "loss": 0.2245,
      "step": 42265
    },
    {
      "epoch": 0.7632223257426386,
      "grad_norm": 0.3385583162307739,
      "learning_rate": 2.6412142270237083e-06,
      "loss": 0.1647,
      "step": 42270
    },
    {
      "epoch": 0.7633126051755393,
      "grad_norm": 0.4164818525314331,
      "learning_rate": 2.639294064711193e-06,
      "loss": 0.2203,
      "step": 42275
    },
    {
      "epoch": 0.7634028846084401,
      "grad_norm": 0.7585594058036804,
      "learning_rate": 2.637374494513956e-06,
      "loss": 0.3093,
      "step": 42280
    },
    {
      "epoch": 0.7634931640413407,
      "grad_norm": 0.6199455261230469,
      "learning_rate": 2.6354555165864047e-06,
      "loss": 0.1823,
      "step": 42285
    },
    {
      "epoch": 0.7635834434742415,
      "grad_norm": 0.626918613910675,
      "learning_rate": 2.633537131082916e-06,
      "loss": 0.1591,
      "step": 42290
    },
    {
      "epoch": 0.7636737229071422,
      "grad_norm": 0.6108911037445068,
      "learning_rate": 2.6316193381578035e-06,
      "loss": 0.3038,
      "step": 42295
    },
    {
      "epoch": 0.7637640023400429,
      "grad_norm": 0.5488325953483582,
      "learning_rate": 2.629702137965341e-06,
      "loss": 0.1507,
      "step": 42300
    },
    {
      "epoch": 0.7638542817729436,
      "grad_norm": 0.25596383213996887,
      "learning_rate": 2.6277855306597543e-06,
      "loss": 0.1498,
      "step": 42305
    },
    {
      "epoch": 0.7639445612058443,
      "grad_norm": 0.7998791933059692,
      "learning_rate": 2.625869516395223e-06,
      "loss": 0.2152,
      "step": 42310
    },
    {
      "epoch": 0.764034840638745,
      "grad_norm": 0.4003254473209381,
      "learning_rate": 2.6239540953258724e-06,
      "loss": 0.2383,
      "step": 42315
    },
    {
      "epoch": 0.7641251200716458,
      "grad_norm": 0.38832858204841614,
      "learning_rate": 2.6220392676057873e-06,
      "loss": 0.1456,
      "step": 42320
    },
    {
      "epoch": 0.7642153995045464,
      "grad_norm": 0.5294615626335144,
      "learning_rate": 2.6201250333889984e-06,
      "loss": 0.2623,
      "step": 42325
    },
    {
      "epoch": 0.7643056789374472,
      "grad_norm": 0.38336092233657837,
      "learning_rate": 2.6182113928294926e-06,
      "loss": 0.2714,
      "step": 42330
    },
    {
      "epoch": 0.7643959583703479,
      "grad_norm": 0.44061025977134705,
      "learning_rate": 2.6162983460812097e-06,
      "loss": 0.1364,
      "step": 42335
    },
    {
      "epoch": 0.7644862378032486,
      "grad_norm": 0.4322599768638611,
      "learning_rate": 2.6143858932980437e-06,
      "loss": 0.1758,
      "step": 42340
    },
    {
      "epoch": 0.7645765172361493,
      "grad_norm": 0.5041878819465637,
      "learning_rate": 2.6124740346338305e-06,
      "loss": 0.1915,
      "step": 42345
    },
    {
      "epoch": 0.76466679666905,
      "grad_norm": 0.36872994899749756,
      "learning_rate": 2.610562770242372e-06,
      "loss": 0.24,
      "step": 42350
    },
    {
      "epoch": 0.7647570761019508,
      "grad_norm": 0.4666585326194763,
      "learning_rate": 2.6086521002774092e-06,
      "loss": 0.2747,
      "step": 42355
    },
    {
      "epoch": 0.7648473555348515,
      "grad_norm": 0.48667874932289124,
      "learning_rate": 2.6067420248926477e-06,
      "loss": 0.2137,
      "step": 42360
    },
    {
      "epoch": 0.7649376349677521,
      "grad_norm": 0.34638234972953796,
      "learning_rate": 2.6048325442417334e-06,
      "loss": 0.2217,
      "step": 42365
    },
    {
      "epoch": 0.7650279144006529,
      "grad_norm": 0.5076397657394409,
      "learning_rate": 2.6029236584782727e-06,
      "loss": 0.1999,
      "step": 42370
    },
    {
      "epoch": 0.7651181938335536,
      "grad_norm": 0.5190479159355164,
      "learning_rate": 2.6010153677558237e-06,
      "loss": 0.1606,
      "step": 42375
    },
    {
      "epoch": 0.7652084732664544,
      "grad_norm": 0.2692440152168274,
      "learning_rate": 2.599107672227891e-06,
      "loss": 0.182,
      "step": 42380
    },
    {
      "epoch": 0.765298752699355,
      "grad_norm": 0.4179719090461731,
      "learning_rate": 2.5972005720479377e-06,
      "loss": 0.2905,
      "step": 42385
    },
    {
      "epoch": 0.7653890321322557,
      "grad_norm": 0.493459016084671,
      "learning_rate": 2.5952940673693726e-06,
      "loss": 0.1793,
      "step": 42390
    },
    {
      "epoch": 0.7654793115651565,
      "grad_norm": 0.45662614703178406,
      "learning_rate": 2.5933881583455632e-06,
      "loss": 0.2157,
      "step": 42395
    },
    {
      "epoch": 0.7655695909980572,
      "grad_norm": 0.8813372254371643,
      "learning_rate": 2.5914828451298248e-06,
      "loss": 0.2523,
      "step": 42400
    },
    {
      "epoch": 0.765659870430958,
      "grad_norm": 0.6023253798484802,
      "learning_rate": 2.5895781278754286e-06,
      "loss": 0.253,
      "step": 42405
    },
    {
      "epoch": 0.7657501498638586,
      "grad_norm": 0.546947717666626,
      "learning_rate": 2.587674006735591e-06,
      "loss": 0.2306,
      "step": 42410
    },
    {
      "epoch": 0.7658404292967593,
      "grad_norm": 0.48942071199417114,
      "learning_rate": 2.585770481863489e-06,
      "loss": 0.3523,
      "step": 42415
    },
    {
      "epoch": 0.7659307087296601,
      "grad_norm": 0.20931553840637207,
      "learning_rate": 2.5838675534122428e-06,
      "loss": 0.2231,
      "step": 42420
    },
    {
      "epoch": 0.7660209881625608,
      "grad_norm": 1.144301414489746,
      "learning_rate": 2.5819652215349334e-06,
      "loss": 0.303,
      "step": 42425
    },
    {
      "epoch": 0.7661112675954614,
      "grad_norm": 0.4387619197368622,
      "learning_rate": 2.580063486384585e-06,
      "loss": 0.2749,
      "step": 42430
    },
    {
      "epoch": 0.7662015470283622,
      "grad_norm": 0.5536942481994629,
      "learning_rate": 2.5781623481141817e-06,
      "loss": 0.2277,
      "step": 42435
    },
    {
      "epoch": 0.7662918264612629,
      "grad_norm": 0.32391244173049927,
      "learning_rate": 2.5762618068766553e-06,
      "loss": 0.1775,
      "step": 42440
    },
    {
      "epoch": 0.7663821058941637,
      "grad_norm": 0.2672235369682312,
      "learning_rate": 2.5743618628248934e-06,
      "loss": 0.2203,
      "step": 42445
    },
    {
      "epoch": 0.7664723853270643,
      "grad_norm": 0.4057731628417969,
      "learning_rate": 2.5724625161117265e-06,
      "loss": 0.2012,
      "step": 42450
    },
    {
      "epoch": 0.766562664759965,
      "grad_norm": 0.410159170627594,
      "learning_rate": 2.570563766889951e-06,
      "loss": 0.254,
      "step": 42455
    },
    {
      "epoch": 0.7666529441928658,
      "grad_norm": 0.5925469398498535,
      "learning_rate": 2.568665615312299e-06,
      "loss": 0.2442,
      "step": 42460
    },
    {
      "epoch": 0.7667432236257665,
      "grad_norm": 1.049332857131958,
      "learning_rate": 2.5667680615314685e-06,
      "loss": 0.2323,
      "step": 42465
    },
    {
      "epoch": 0.7668335030586672,
      "grad_norm": 0.5431271195411682,
      "learning_rate": 2.5648711057001052e-06,
      "loss": 0.1807,
      "step": 42470
    },
    {
      "epoch": 0.7669237824915679,
      "grad_norm": 0.33623602986335754,
      "learning_rate": 2.562974747970799e-06,
      "loss": 0.248,
      "step": 42475
    },
    {
      "epoch": 0.7670140619244686,
      "grad_norm": 0.5575045347213745,
      "learning_rate": 2.5610789884961063e-06,
      "loss": 0.1661,
      "step": 42480
    },
    {
      "epoch": 0.7671043413573694,
      "grad_norm": 0.46629127860069275,
      "learning_rate": 2.5591838274285185e-06,
      "loss": 0.2999,
      "step": 42485
    },
    {
      "epoch": 0.76719462079027,
      "grad_norm": 0.3633790612220764,
      "learning_rate": 2.557289264920495e-06,
      "loss": 0.2391,
      "step": 42490
    },
    {
      "epoch": 0.7672849002231708,
      "grad_norm": 0.2716864049434662,
      "learning_rate": 2.5553953011244337e-06,
      "loss": 0.2262,
      "step": 42495
    },
    {
      "epoch": 0.7673751796560715,
      "grad_norm": 0.5711254477500916,
      "learning_rate": 2.5535019361926937e-06,
      "loss": 0.1842,
      "step": 42500
    },
    {
      "epoch": 0.7674654590889722,
      "grad_norm": 0.5874897241592407,
      "learning_rate": 2.55160917027758e-06,
      "loss": 0.2661,
      "step": 42505
    },
    {
      "epoch": 0.7675557385218729,
      "grad_norm": 0.30902114510536194,
      "learning_rate": 2.549717003531358e-06,
      "loss": 0.2317,
      "step": 42510
    },
    {
      "epoch": 0.7676460179547736,
      "grad_norm": 0.47627681493759155,
      "learning_rate": 2.5478254361062306e-06,
      "loss": 0.1844,
      "step": 42515
    },
    {
      "epoch": 0.7677362973876743,
      "grad_norm": 0.8980509638786316,
      "learning_rate": 2.5459344681543672e-06,
      "loss": 0.2205,
      "step": 42520
    },
    {
      "epoch": 0.7678265768205751,
      "grad_norm": 0.516670286655426,
      "learning_rate": 2.544044099827877e-06,
      "loss": 0.2158,
      "step": 42525
    },
    {
      "epoch": 0.7679168562534757,
      "grad_norm": 0.4446510970592499,
      "learning_rate": 2.542154331278829e-06,
      "loss": 0.2529,
      "step": 42530
    },
    {
      "epoch": 0.7680071356863765,
      "grad_norm": 0.17722104489803314,
      "learning_rate": 2.540265162659241e-06,
      "loss": 0.1482,
      "step": 42535
    },
    {
      "epoch": 0.7680974151192772,
      "grad_norm": 0.4235355257987976,
      "learning_rate": 2.538376594121086e-06,
      "loss": 0.211,
      "step": 42540
    },
    {
      "epoch": 0.768187694552178,
      "grad_norm": 0.38903820514678955,
      "learning_rate": 2.53648862581628e-06,
      "loss": 0.2324,
      "step": 42545
    },
    {
      "epoch": 0.7682779739850786,
      "grad_norm": 0.2501767873764038,
      "learning_rate": 2.5346012578967017e-06,
      "loss": 0.3024,
      "step": 42550
    },
    {
      "epoch": 0.7683682534179793,
      "grad_norm": 0.5181525945663452,
      "learning_rate": 2.5327144905141733e-06,
      "loss": 0.272,
      "step": 42555
    },
    {
      "epoch": 0.7684585328508801,
      "grad_norm": 0.5097023844718933,
      "learning_rate": 2.530828323820469e-06,
      "loss": 0.1868,
      "step": 42560
    },
    {
      "epoch": 0.7685488122837808,
      "grad_norm": 0.8557437658309937,
      "learning_rate": 2.528942757967321e-06,
      "loss": 0.225,
      "step": 42565
    },
    {
      "epoch": 0.7686390917166814,
      "grad_norm": 0.49657967686653137,
      "learning_rate": 2.527057793106407e-06,
      "loss": 0.1764,
      "step": 42570
    },
    {
      "epoch": 0.7687293711495822,
      "grad_norm": 0.6104001998901367,
      "learning_rate": 2.5251734293893637e-06,
      "loss": 0.2147,
      "step": 42575
    },
    {
      "epoch": 0.7688196505824829,
      "grad_norm": 0.5121384859085083,
      "learning_rate": 2.5232896669677674e-06,
      "loss": 0.2092,
      "step": 42580
    },
    {
      "epoch": 0.7689099300153837,
      "grad_norm": 0.276767760515213,
      "learning_rate": 2.521406505993159e-06,
      "loss": 0.2346,
      "step": 42585
    },
    {
      "epoch": 0.7690002094482843,
      "grad_norm": 0.20188720524311066,
      "learning_rate": 2.5195239466170207e-06,
      "loss": 0.1299,
      "step": 42590
    },
    {
      "epoch": 0.769090488881185,
      "grad_norm": 0.5935357213020325,
      "learning_rate": 2.517641988990791e-06,
      "loss": 0.2915,
      "step": 42595
    },
    {
      "epoch": 0.7691807683140858,
      "grad_norm": 0.4150652587413788,
      "learning_rate": 2.5157606332658624e-06,
      "loss": 0.232,
      "step": 42600
    },
    {
      "epoch": 0.7692710477469865,
      "grad_norm": 0.5176143050193787,
      "learning_rate": 2.513879879593578e-06,
      "loss": 0.2345,
      "step": 42605
    },
    {
      "epoch": 0.7693613271798871,
      "grad_norm": 0.4309931695461273,
      "learning_rate": 2.511999728125225e-06,
      "loss": 0.1915,
      "step": 42610
    },
    {
      "epoch": 0.7694516066127879,
      "grad_norm": 0.45065101981163025,
      "learning_rate": 2.5101201790120545e-06,
      "loss": 0.2002,
      "step": 42615
    },
    {
      "epoch": 0.7695418860456886,
      "grad_norm": 0.49554944038391113,
      "learning_rate": 2.5082412324052553e-06,
      "loss": 0.1543,
      "step": 42620
    },
    {
      "epoch": 0.7696321654785894,
      "grad_norm": 0.48984622955322266,
      "learning_rate": 2.506362888455981e-06,
      "loss": 0.2093,
      "step": 42625
    },
    {
      "epoch": 0.76972244491149,
      "grad_norm": 0.5073429942131042,
      "learning_rate": 2.504485147315323e-06,
      "loss": 0.2177,
      "step": 42630
    },
    {
      "epoch": 0.7698127243443907,
      "grad_norm": 0.4508723020553589,
      "learning_rate": 2.502608009134343e-06,
      "loss": 0.1841,
      "step": 42635
    },
    {
      "epoch": 0.7699030037772915,
      "grad_norm": 0.2888708710670471,
      "learning_rate": 2.5007314740640375e-06,
      "loss": 0.2321,
      "step": 42640
    },
    {
      "epoch": 0.7699932832101922,
      "grad_norm": 0.5804747343063354,
      "learning_rate": 2.498855542255356e-06,
      "loss": 0.2339,
      "step": 42645
    },
    {
      "epoch": 0.7700835626430929,
      "grad_norm": 0.742763102054596,
      "learning_rate": 2.4969802138592125e-06,
      "loss": 0.1643,
      "step": 42650
    },
    {
      "epoch": 0.7701738420759936,
      "grad_norm": 0.4516551196575165,
      "learning_rate": 2.4951054890264536e-06,
      "loss": 0.2385,
      "step": 42655
    },
    {
      "epoch": 0.7702641215088943,
      "grad_norm": 0.30727094411849976,
      "learning_rate": 2.4932313679078936e-06,
      "loss": 0.2728,
      "step": 42660
    },
    {
      "epoch": 0.7703544009417951,
      "grad_norm": 0.5855993032455444,
      "learning_rate": 2.49135785065429e-06,
      "loss": 0.2481,
      "step": 42665
    },
    {
      "epoch": 0.7704446803746957,
      "grad_norm": 0.23302792012691498,
      "learning_rate": 2.4894849374163577e-06,
      "loss": 0.2134,
      "step": 42670
    },
    {
      "epoch": 0.7705349598075965,
      "grad_norm": 0.5293843150138855,
      "learning_rate": 2.4876126283447523e-06,
      "loss": 0.2389,
      "step": 42675
    },
    {
      "epoch": 0.7706252392404972,
      "grad_norm": 0.49191197752952576,
      "learning_rate": 2.4857409235900932e-06,
      "loss": 0.2098,
      "step": 42680
    },
    {
      "epoch": 0.7707155186733979,
      "grad_norm": 0.7379592657089233,
      "learning_rate": 2.48386982330294e-06,
      "loss": 0.2333,
      "step": 42685
    },
    {
      "epoch": 0.7708057981062986,
      "grad_norm": 0.483277827501297,
      "learning_rate": 2.481999327633815e-06,
      "loss": 0.1944,
      "step": 42690
    },
    {
      "epoch": 0.7708960775391993,
      "grad_norm": 0.425761342048645,
      "learning_rate": 2.4801294367331763e-06,
      "loss": 0.2939,
      "step": 42695
    },
    {
      "epoch": 0.7709863569721,
      "grad_norm": 0.3765316307544708,
      "learning_rate": 2.478260150751457e-06,
      "loss": 0.2718,
      "step": 42700
    },
    {
      "epoch": 0.7710766364050008,
      "grad_norm": 0.3427625894546509,
      "learning_rate": 2.4763914698390166e-06,
      "loss": 0.2891,
      "step": 42705
    },
    {
      "epoch": 0.7711669158379014,
      "grad_norm": 0.5614849925041199,
      "learning_rate": 2.474523394146183e-06,
      "loss": 0.2864,
      "step": 42710
    },
    {
      "epoch": 0.7712571952708022,
      "grad_norm": 0.5848512649536133,
      "learning_rate": 2.472655923823225e-06,
      "loss": 0.2946,
      "step": 42715
    },
    {
      "epoch": 0.7713474747037029,
      "grad_norm": 1.3261486291885376,
      "learning_rate": 2.470789059020371e-06,
      "loss": 0.2106,
      "step": 42720
    },
    {
      "epoch": 0.7714377541366036,
      "grad_norm": 0.36234983801841736,
      "learning_rate": 2.4689227998877917e-06,
      "loss": 0.1517,
      "step": 42725
    },
    {
      "epoch": 0.7715280335695043,
      "grad_norm": 0.3489014804363251,
      "learning_rate": 2.4670571465756167e-06,
      "loss": 0.2127,
      "step": 42730
    },
    {
      "epoch": 0.771618313002405,
      "grad_norm": 0.49692797660827637,
      "learning_rate": 2.465192099233927e-06,
      "loss": 0.2127,
      "step": 42735
    },
    {
      "epoch": 0.7717085924353058,
      "grad_norm": 1.3462014198303223,
      "learning_rate": 2.4633276580127474e-06,
      "loss": 0.3143,
      "step": 42740
    },
    {
      "epoch": 0.7717988718682065,
      "grad_norm": 0.47123292088508606,
      "learning_rate": 2.4614638230620645e-06,
      "loss": 0.2834,
      "step": 42745
    },
    {
      "epoch": 0.7718891513011071,
      "grad_norm": 0.6388853788375854,
      "learning_rate": 2.4596005945318024e-06,
      "loss": 0.1828,
      "step": 42750
    },
    {
      "epoch": 0.7719794307340079,
      "grad_norm": 0.3754936456680298,
      "learning_rate": 2.457737972571852e-06,
      "loss": 0.1666,
      "step": 42755
    },
    {
      "epoch": 0.7720697101669086,
      "grad_norm": 0.5376711487770081,
      "learning_rate": 2.455875957332041e-06,
      "loss": 0.2344,
      "step": 42760
    },
    {
      "epoch": 0.7721599895998094,
      "grad_norm": 0.2916709780693054,
      "learning_rate": 2.4540145489621582e-06,
      "loss": 0.2351,
      "step": 42765
    },
    {
      "epoch": 0.77225026903271,
      "grad_norm": 0.28260645270347595,
      "learning_rate": 2.4521537476119407e-06,
      "loss": 0.2395,
      "step": 42770
    },
    {
      "epoch": 0.7723405484656107,
      "grad_norm": 0.5174070596694946,
      "learning_rate": 2.450293553431078e-06,
      "loss": 0.2205,
      "step": 42775
    },
    {
      "epoch": 0.7724308278985115,
      "grad_norm": 0.46815547347068787,
      "learning_rate": 2.4484339665692048e-06,
      "loss": 0.1975,
      "step": 42780
    },
    {
      "epoch": 0.7725211073314122,
      "grad_norm": 0.5334517359733582,
      "learning_rate": 2.446574987175917e-06,
      "loss": 0.2686,
      "step": 42785
    },
    {
      "epoch": 0.7726113867643128,
      "grad_norm": 0.4952988028526306,
      "learning_rate": 2.4447166154007494e-06,
      "loss": 0.3388,
      "step": 42790
    },
    {
      "epoch": 0.7727016661972136,
      "grad_norm": 0.4773997366428375,
      "learning_rate": 2.4428588513931985e-06,
      "loss": 0.314,
      "step": 42795
    },
    {
      "epoch": 0.7727919456301143,
      "grad_norm": 0.40349653363227844,
      "learning_rate": 2.441001695302707e-06,
      "loss": 0.2495,
      "step": 42800
    },
    {
      "epoch": 0.7728822250630151,
      "grad_norm": 0.4642648994922638,
      "learning_rate": 2.439145147278673e-06,
      "loss": 0.2607,
      "step": 42805
    },
    {
      "epoch": 0.7729725044959158,
      "grad_norm": 0.5170674324035645,
      "learning_rate": 2.4372892074704356e-06,
      "loss": 0.1919,
      "step": 42810
    },
    {
      "epoch": 0.7730627839288164,
      "grad_norm": 0.46825897693634033,
      "learning_rate": 2.4354338760272987e-06,
      "loss": 0.1711,
      "step": 42815
    },
    {
      "epoch": 0.7731530633617172,
      "grad_norm": 0.6066874861717224,
      "learning_rate": 2.4335791530985044e-06,
      "loss": 0.2502,
      "step": 42820
    },
    {
      "epoch": 0.7732433427946179,
      "grad_norm": 0.37474513053894043,
      "learning_rate": 2.431725038833256e-06,
      "loss": 0.3085,
      "step": 42825
    },
    {
      "epoch": 0.7733336222275187,
      "grad_norm": 0.27002474665641785,
      "learning_rate": 2.4298715333806987e-06,
      "loss": 0.2874,
      "step": 42830
    },
    {
      "epoch": 0.7734239016604193,
      "grad_norm": 0.2877873182296753,
      "learning_rate": 2.428018636889936e-06,
      "loss": 0.1283,
      "step": 42835
    },
    {
      "epoch": 0.77351418109332,
      "grad_norm": 0.5184369683265686,
      "learning_rate": 2.426166349510024e-06,
      "loss": 0.2457,
      "step": 42840
    },
    {
      "epoch": 0.7736044605262208,
      "grad_norm": 0.706448495388031,
      "learning_rate": 2.424314671389959e-06,
      "loss": 0.239,
      "step": 42845
    },
    {
      "epoch": 0.7736947399591215,
      "grad_norm": 0.27848494052886963,
      "learning_rate": 2.422463602678701e-06,
      "loss": 0.188,
      "step": 42850
    },
    {
      "epoch": 0.7737850193920222,
      "grad_norm": 0.711946964263916,
      "learning_rate": 2.420613143525149e-06,
      "loss": 0.2974,
      "step": 42855
    },
    {
      "epoch": 0.7738752988249229,
      "grad_norm": 0.28272512555122375,
      "learning_rate": 2.4187632940781626e-06,
      "loss": 0.1606,
      "step": 42860
    },
    {
      "epoch": 0.7739655782578236,
      "grad_norm": 0.5454052090644836,
      "learning_rate": 2.416914054486549e-06,
      "loss": 0.2318,
      "step": 42865
    },
    {
      "epoch": 0.7740558576907244,
      "grad_norm": 0.4498244524002075,
      "learning_rate": 2.415065424899067e-06,
      "loss": 0.2162,
      "step": 42870
    },
    {
      "epoch": 0.774146137123625,
      "grad_norm": 0.4151298999786377,
      "learning_rate": 2.413217405464421e-06,
      "loss": 0.2594,
      "step": 42875
    },
    {
      "epoch": 0.7742364165565258,
      "grad_norm": 0.40834593772888184,
      "learning_rate": 2.4113699963312766e-06,
      "loss": 0.221,
      "step": 42880
    },
    {
      "epoch": 0.7743266959894265,
      "grad_norm": 0.7377216815948486,
      "learning_rate": 2.409523197648238e-06,
      "loss": 0.204,
      "step": 42885
    },
    {
      "epoch": 0.7744169754223272,
      "grad_norm": 0.7599356770515442,
      "learning_rate": 2.4076770095638734e-06,
      "loss": 0.2129,
      "step": 42890
    },
    {
      "epoch": 0.7745072548552279,
      "grad_norm": 0.5863339900970459,
      "learning_rate": 2.4058314322266863e-06,
      "loss": 0.1681,
      "step": 42895
    },
    {
      "epoch": 0.7745975342881286,
      "grad_norm": 0.4008416533470154,
      "learning_rate": 2.403986465785151e-06,
      "loss": 0.3034,
      "step": 42900
    },
    {
      "epoch": 0.7746878137210294,
      "grad_norm": 0.3215121626853943,
      "learning_rate": 2.402142110387673e-06,
      "loss": 0.249,
      "step": 42905
    },
    {
      "epoch": 0.7747780931539301,
      "grad_norm": 0.4487439692020416,
      "learning_rate": 2.400298366182624e-06,
      "loss": 0.2336,
      "step": 42910
    },
    {
      "epoch": 0.7748683725868307,
      "grad_norm": 0.7740156054496765,
      "learning_rate": 2.3984552333183164e-06,
      "loss": 0.3046,
      "step": 42915
    },
    {
      "epoch": 0.7749586520197315,
      "grad_norm": 0.3982470631599426,
      "learning_rate": 2.396612711943013e-06,
      "loss": 0.2324,
      "step": 42920
    },
    {
      "epoch": 0.7750489314526322,
      "grad_norm": 0.9876501560211182,
      "learning_rate": 2.3947708022049355e-06,
      "loss": 0.2852,
      "step": 42925
    },
    {
      "epoch": 0.775139210885533,
      "grad_norm": 0.3493773639202118,
      "learning_rate": 2.3929295042522514e-06,
      "loss": 0.1799,
      "step": 42930
    },
    {
      "epoch": 0.7752294903184336,
      "grad_norm": 0.7048976421356201,
      "learning_rate": 2.391088818233083e-06,
      "loss": 0.2382,
      "step": 42935
    },
    {
      "epoch": 0.7753197697513343,
      "grad_norm": 1.0359954833984375,
      "learning_rate": 2.3892487442954935e-06,
      "loss": 0.192,
      "step": 42940
    },
    {
      "epoch": 0.7754100491842351,
      "grad_norm": 0.30253368616104126,
      "learning_rate": 2.3874092825875107e-06,
      "loss": 0.2552,
      "step": 42945
    },
    {
      "epoch": 0.7755003286171358,
      "grad_norm": 0.3940982222557068,
      "learning_rate": 2.3855704332570994e-06,
      "loss": 0.2129,
      "step": 42950
    },
    {
      "epoch": 0.7755906080500364,
      "grad_norm": 0.3559393286705017,
      "learning_rate": 2.3837321964521865e-06,
      "loss": 0.2135,
      "step": 42955
    },
    {
      "epoch": 0.7756808874829372,
      "grad_norm": 0.730930745601654,
      "learning_rate": 2.3818945723206378e-06,
      "loss": 0.1775,
      "step": 42960
    },
    {
      "epoch": 0.7757711669158379,
      "grad_norm": 0.4467575252056122,
      "learning_rate": 2.380057561010287e-06,
      "loss": 0.3191,
      "step": 42965
    },
    {
      "epoch": 0.7758614463487387,
      "grad_norm": 0.36840131878852844,
      "learning_rate": 2.3782211626689013e-06,
      "loss": 0.3227,
      "step": 42970
    },
    {
      "epoch": 0.7759517257816393,
      "grad_norm": 0.47732090950012207,
      "learning_rate": 2.376385377444209e-06,
      "loss": 0.2419,
      "step": 42975
    },
    {
      "epoch": 0.77604200521454,
      "grad_norm": 0.32633882761001587,
      "learning_rate": 2.374550205483881e-06,
      "loss": 0.2945,
      "step": 42980
    },
    {
      "epoch": 0.7761322846474408,
      "grad_norm": 0.6281416416168213,
      "learning_rate": 2.3727156469355504e-06,
      "loss": 0.304,
      "step": 42985
    },
    {
      "epoch": 0.7762225640803415,
      "grad_norm": 0.38120004534721375,
      "learning_rate": 2.3708817019467868e-06,
      "loss": 0.1746,
      "step": 42990
    },
    {
      "epoch": 0.7763128435132421,
      "grad_norm": 0.39160090684890747,
      "learning_rate": 2.3690483706651225e-06,
      "loss": 0.2276,
      "step": 42995
    },
    {
      "epoch": 0.7764031229461429,
      "grad_norm": 0.40250253677368164,
      "learning_rate": 2.367215653238033e-06,
      "loss": 0.2698,
      "step": 43000
    },
    {
      "epoch": 0.7764934023790436,
      "grad_norm": 0.3191860318183899,
      "learning_rate": 2.365383549812952e-06,
      "loss": 0.2482,
      "step": 43005
    },
    {
      "epoch": 0.7765836818119444,
      "grad_norm": 0.5613829493522644,
      "learning_rate": 2.363552060537255e-06,
      "loss": 0.1981,
      "step": 43010
    },
    {
      "epoch": 0.776673961244845,
      "grad_norm": 0.38162267208099365,
      "learning_rate": 2.36172118555827e-06,
      "loss": 0.2371,
      "step": 43015
    },
    {
      "epoch": 0.7767642406777457,
      "grad_norm": 0.3718111515045166,
      "learning_rate": 2.3598909250232816e-06,
      "loss": 0.1705,
      "step": 43020
    },
    {
      "epoch": 0.7768545201106465,
      "grad_norm": 0.7747634649276733,
      "learning_rate": 2.3580612790795154e-06,
      "loss": 0.2347,
      "step": 43025
    },
    {
      "epoch": 0.7769447995435472,
      "grad_norm": 0.41002294421195984,
      "learning_rate": 2.356232247874162e-06,
      "loss": 0.2013,
      "step": 43030
    },
    {
      "epoch": 0.7770350789764479,
      "grad_norm": 0.4790644347667694,
      "learning_rate": 2.354403831554345e-06,
      "loss": 0.1949,
      "step": 43035
    },
    {
      "epoch": 0.7771253584093486,
      "grad_norm": 0.23889146745204926,
      "learning_rate": 2.352576030267154e-06,
      "loss": 0.2301,
      "step": 43040
    },
    {
      "epoch": 0.7772156378422493,
      "grad_norm": 0.3431209325790405,
      "learning_rate": 2.350748844159617e-06,
      "loss": 0.227,
      "step": 43045
    },
    {
      "epoch": 0.7773059172751501,
      "grad_norm": 0.34881192445755005,
      "learning_rate": 2.348922273378723e-06,
      "loss": 0.2195,
      "step": 43050
    },
    {
      "epoch": 0.7773961967080507,
      "grad_norm": 0.4788563549518585,
      "learning_rate": 2.3470963180714e-06,
      "loss": 0.1821,
      "step": 43055
    },
    {
      "epoch": 0.7774864761409515,
      "grad_norm": 0.488158255815506,
      "learning_rate": 2.3452709783845375e-06,
      "loss": 0.1377,
      "step": 43060
    },
    {
      "epoch": 0.7775767555738522,
      "grad_norm": 0.5387958884239197,
      "learning_rate": 2.3434462544649683e-06,
      "loss": 0.255,
      "step": 43065
    },
    {
      "epoch": 0.7776670350067529,
      "grad_norm": 0.4991535246372223,
      "learning_rate": 2.3416221464594826e-06,
      "loss": 0.3212,
      "step": 43070
    },
    {
      "epoch": 0.7777573144396536,
      "grad_norm": 0.4338509738445282,
      "learning_rate": 2.33979865451481e-06,
      "loss": 0.166,
      "step": 43075
    },
    {
      "epoch": 0.7778475938725543,
      "grad_norm": 0.4558165669441223,
      "learning_rate": 2.3379757787776435e-06,
      "loss": 0.2285,
      "step": 43080
    },
    {
      "epoch": 0.777937873305455,
      "grad_norm": 0.42054474353790283,
      "learning_rate": 2.336153519394615e-06,
      "loss": 0.2826,
      "step": 43085
    },
    {
      "epoch": 0.7780281527383558,
      "grad_norm": 0.4690871834754944,
      "learning_rate": 2.3343318765123146e-06,
      "loss": 0.1927,
      "step": 43090
    },
    {
      "epoch": 0.7781184321712564,
      "grad_norm": 0.22620423138141632,
      "learning_rate": 2.33251085027728e-06,
      "loss": 0.2258,
      "step": 43095
    },
    {
      "epoch": 0.7782087116041572,
      "grad_norm": 0.343610554933548,
      "learning_rate": 2.330690440836001e-06,
      "loss": 0.2006,
      "step": 43100
    },
    {
      "epoch": 0.7782989910370579,
      "grad_norm": 0.2786353528499603,
      "learning_rate": 2.3288706483349157e-06,
      "loss": 0.2464,
      "step": 43105
    },
    {
      "epoch": 0.7783892704699586,
      "grad_norm": 0.22044533491134644,
      "learning_rate": 2.3270514729204096e-06,
      "loss": 0.2396,
      "step": 43110
    },
    {
      "epoch": 0.7784795499028593,
      "grad_norm": 0.46998295187950134,
      "learning_rate": 2.3252329147388274e-06,
      "loss": 0.2202,
      "step": 43115
    },
    {
      "epoch": 0.77856982933576,
      "grad_norm": 0.4731232821941376,
      "learning_rate": 2.323414973936453e-06,
      "loss": 0.2087,
      "step": 43120
    },
    {
      "epoch": 0.7786601087686608,
      "grad_norm": 0.45158910751342773,
      "learning_rate": 2.3215976506595296e-06,
      "loss": 0.2806,
      "step": 43125
    },
    {
      "epoch": 0.7787503882015615,
      "grad_norm": 0.4839245676994324,
      "learning_rate": 2.3197809450542487e-06,
      "loss": 0.1251,
      "step": 43130
    },
    {
      "epoch": 0.7788406676344621,
      "grad_norm": 1.5681735277175903,
      "learning_rate": 2.3179648572667522e-06,
      "loss": 0.2691,
      "step": 43135
    },
    {
      "epoch": 0.7789309470673629,
      "grad_norm": 0.4585834741592407,
      "learning_rate": 2.3161493874431272e-06,
      "loss": 0.1773,
      "step": 43140
    },
    {
      "epoch": 0.7790212265002636,
      "grad_norm": 0.29608702659606934,
      "learning_rate": 2.314334535729419e-06,
      "loss": 0.1347,
      "step": 43145
    },
    {
      "epoch": 0.7791115059331644,
      "grad_norm": 0.5196985602378845,
      "learning_rate": 2.3125203022716147e-06,
      "loss": 0.1933,
      "step": 43150
    },
    {
      "epoch": 0.779201785366065,
      "grad_norm": 0.4159768223762512,
      "learning_rate": 2.3107066872156592e-06,
      "loss": 0.2573,
      "step": 43155
    },
    {
      "epoch": 0.7792920647989657,
      "grad_norm": 0.6043077707290649,
      "learning_rate": 2.3088936907074434e-06,
      "loss": 0.3016,
      "step": 43160
    },
    {
      "epoch": 0.7793823442318665,
      "grad_norm": 0.3892230689525604,
      "learning_rate": 2.3070813128928137e-06,
      "loss": 0.179,
      "step": 43165
    },
    {
      "epoch": 0.7794726236647672,
      "grad_norm": 0.35871732234954834,
      "learning_rate": 2.3052695539175562e-06,
      "loss": 0.2789,
      "step": 43170
    },
    {
      "epoch": 0.7795629030976678,
      "grad_norm": 0.5863191485404968,
      "learning_rate": 2.303458413927421e-06,
      "loss": 0.2052,
      "step": 43175
    },
    {
      "epoch": 0.7796531825305686,
      "grad_norm": 0.6840908527374268,
      "learning_rate": 2.301647893068093e-06,
      "loss": 0.1901,
      "step": 43180
    },
    {
      "epoch": 0.7797434619634693,
      "grad_norm": 0.6739901304244995,
      "learning_rate": 2.299837991485223e-06,
      "loss": 0.3192,
      "step": 43185
    },
    {
      "epoch": 0.7798337413963701,
      "grad_norm": 0.6026039719581604,
      "learning_rate": 2.298028709324398e-06,
      "loss": 0.1811,
      "step": 43190
    },
    {
      "epoch": 0.7799240208292708,
      "grad_norm": 0.7208412885665894,
      "learning_rate": 2.2962200467311646e-06,
      "loss": 0.2616,
      "step": 43195
    },
    {
      "epoch": 0.7800143002621714,
      "grad_norm": 0.7069392204284668,
      "learning_rate": 2.2944120038510187e-06,
      "loss": 0.1668,
      "step": 43200
    },
    {
      "epoch": 0.7801045796950722,
      "grad_norm": 0.94843989610672,
      "learning_rate": 2.2926045808294007e-06,
      "loss": 0.2501,
      "step": 43205
    },
    {
      "epoch": 0.7801948591279729,
      "grad_norm": 0.47515806555747986,
      "learning_rate": 2.2907977778117075e-06,
      "loss": 0.3316,
      "step": 43210
    },
    {
      "epoch": 0.7802851385608737,
      "grad_norm": 0.11298256367444992,
      "learning_rate": 2.2889915949432794e-06,
      "loss": 0.2556,
      "step": 43215
    },
    {
      "epoch": 0.7803754179937743,
      "grad_norm": 0.2375980019569397,
      "learning_rate": 2.287186032369416e-06,
      "loss": 0.2101,
      "step": 43220
    },
    {
      "epoch": 0.780465697426675,
      "grad_norm": 0.4574558138847351,
      "learning_rate": 2.2853810902353523e-06,
      "loss": 0.234,
      "step": 43225
    },
    {
      "epoch": 0.7805559768595758,
      "grad_norm": 0.4236000180244446,
      "learning_rate": 2.283576768686296e-06,
      "loss": 0.2537,
      "step": 43230
    },
    {
      "epoch": 0.7806462562924765,
      "grad_norm": 0.26832252740859985,
      "learning_rate": 2.281773067867381e-06,
      "loss": 0.1987,
      "step": 43235
    },
    {
      "epoch": 0.7807365357253772,
      "grad_norm": 0.7041072249412537,
      "learning_rate": 2.279969987923709e-06,
      "loss": 0.208,
      "step": 43240
    },
    {
      "epoch": 0.7808268151582779,
      "grad_norm": 0.41026878356933594,
      "learning_rate": 2.2781675290003192e-06,
      "loss": 0.176,
      "step": 43245
    },
    {
      "epoch": 0.7809170945911786,
      "grad_norm": 0.22822080552577972,
      "learning_rate": 2.2763656912422104e-06,
      "loss": 0.2001,
      "step": 43250
    },
    {
      "epoch": 0.7810073740240794,
      "grad_norm": 0.5400646328926086,
      "learning_rate": 2.274564474794324e-06,
      "loss": 0.2427,
      "step": 43255
    },
    {
      "epoch": 0.78109765345698,
      "grad_norm": 0.5550546646118164,
      "learning_rate": 2.272763879801555e-06,
      "loss": 0.194,
      "step": 43260
    },
    {
      "epoch": 0.7811879328898808,
      "grad_norm": 0.6339682340621948,
      "learning_rate": 2.27096390640875e-06,
      "loss": 0.3327,
      "step": 43265
    },
    {
      "epoch": 0.7812782123227815,
      "grad_norm": 0.2583569586277008,
      "learning_rate": 2.2691645547607057e-06,
      "loss": 0.2188,
      "step": 43270
    },
    {
      "epoch": 0.7813684917556822,
      "grad_norm": 0.6766750812530518,
      "learning_rate": 2.2673658250021613e-06,
      "loss": 0.215,
      "step": 43275
    },
    {
      "epoch": 0.7814587711885829,
      "grad_norm": 0.5420390367507935,
      "learning_rate": 2.265567717277818e-06,
      "loss": 0.3132,
      "step": 43280
    },
    {
      "epoch": 0.7815490506214836,
      "grad_norm": 0.6376538872718811,
      "learning_rate": 2.2637702317323175e-06,
      "loss": 0.1995,
      "step": 43285
    },
    {
      "epoch": 0.7816393300543844,
      "grad_norm": 0.5242908596992493,
      "learning_rate": 2.261973368510247e-06,
      "loss": 0.2295,
      "step": 43290
    },
    {
      "epoch": 0.7817296094872851,
      "grad_norm": 0.3954088091850281,
      "learning_rate": 2.2601771277561646e-06,
      "loss": 0.246,
      "step": 43295
    },
    {
      "epoch": 0.7818198889201857,
      "grad_norm": 0.5970197319984436,
      "learning_rate": 2.2583815096145554e-06,
      "loss": 0.1949,
      "step": 43300
    },
    {
      "epoch": 0.7819101683530865,
      "grad_norm": 0.4973948001861572,
      "learning_rate": 2.2565865142298705e-06,
      "loss": 0.2085,
      "step": 43305
    },
    {
      "epoch": 0.7820004477859872,
      "grad_norm": 0.40069374442100525,
      "learning_rate": 2.254792141746496e-06,
      "loss": 0.1581,
      "step": 43310
    },
    {
      "epoch": 0.782090727218888,
      "grad_norm": 0.4839843213558197,
      "learning_rate": 2.252998392308784e-06,
      "loss": 0.2285,
      "step": 43315
    },
    {
      "epoch": 0.7821810066517886,
      "grad_norm": 0.3791256546974182,
      "learning_rate": 2.2512052660610228e-06,
      "loss": 0.198,
      "step": 43320
    },
    {
      "epoch": 0.7822712860846893,
      "grad_norm": 0.17784009873867035,
      "learning_rate": 2.249412763147458e-06,
      "loss": 0.2166,
      "step": 43325
    },
    {
      "epoch": 0.7823615655175901,
      "grad_norm": 0.8475596308708191,
      "learning_rate": 2.2476208837122848e-06,
      "loss": 0.2128,
      "step": 43330
    },
    {
      "epoch": 0.7824518449504908,
      "grad_norm": 0.27115878462791443,
      "learning_rate": 2.2458296278996495e-06,
      "loss": 0.2006,
      "step": 43335
    },
    {
      "epoch": 0.7825421243833914,
      "grad_norm": 0.7656043767929077,
      "learning_rate": 2.244038995853639e-06,
      "loss": 0.2685,
      "step": 43340
    },
    {
      "epoch": 0.7826324038162922,
      "grad_norm": 0.6903694868087769,
      "learning_rate": 2.242248987718303e-06,
      "loss": 0.2675,
      "step": 43345
    },
    {
      "epoch": 0.7827226832491929,
      "grad_norm": 0.5449568033218384,
      "learning_rate": 2.240459603637628e-06,
      "loss": 0.241,
      "step": 43350
    },
    {
      "epoch": 0.7828129626820937,
      "grad_norm": 0.3075452148914337,
      "learning_rate": 2.2386708437555615e-06,
      "loss": 0.2468,
      "step": 43355
    },
    {
      "epoch": 0.7829032421149943,
      "grad_norm": 0.6332199573516846,
      "learning_rate": 2.2368827082159948e-06,
      "loss": 0.2719,
      "step": 43360
    },
    {
      "epoch": 0.782993521547895,
      "grad_norm": 0.7227048277854919,
      "learning_rate": 2.235095197162774e-06,
      "loss": 0.2386,
      "step": 43365
    },
    {
      "epoch": 0.7830838009807958,
      "grad_norm": 0.4199582040309906,
      "learning_rate": 2.2333083107396856e-06,
      "loss": 0.2172,
      "step": 43370
    },
    {
      "epoch": 0.7831740804136965,
      "grad_norm": 0.5018370747566223,
      "learning_rate": 2.2315220490904775e-06,
      "loss": 0.2905,
      "step": 43375
    },
    {
      "epoch": 0.7832643598465971,
      "grad_norm": 0.5064613223075867,
      "learning_rate": 2.229736412358837e-06,
      "loss": 0.1765,
      "step": 43380
    },
    {
      "epoch": 0.7833546392794979,
      "grad_norm": 0.2941718101501465,
      "learning_rate": 2.2279514006884053e-06,
      "loss": 0.1625,
      "step": 43385
    },
    {
      "epoch": 0.7834449187123986,
      "grad_norm": 0.7269048690795898,
      "learning_rate": 2.2261670142227743e-06,
      "loss": 0.2063,
      "step": 43390
    },
    {
      "epoch": 0.7835351981452994,
      "grad_norm": 0.3319731056690216,
      "learning_rate": 2.2243832531054865e-06,
      "loss": 0.1869,
      "step": 43395
    },
    {
      "epoch": 0.7836254775782,
      "grad_norm": 0.7761351466178894,
      "learning_rate": 2.222600117480034e-06,
      "loss": 0.2278,
      "step": 43400
    },
    {
      "epoch": 0.7837157570111007,
      "grad_norm": 0.1965448260307312,
      "learning_rate": 2.2208176074898523e-06,
      "loss": 0.1648,
      "step": 43405
    },
    {
      "epoch": 0.7838060364440015,
      "grad_norm": 0.26565489172935486,
      "learning_rate": 2.2190357232783367e-06,
      "loss": 0.2174,
      "step": 43410
    },
    {
      "epoch": 0.7838963158769022,
      "grad_norm": 0.537438690662384,
      "learning_rate": 2.2172544649888207e-06,
      "loss": 0.2583,
      "step": 43415
    },
    {
      "epoch": 0.7839865953098029,
      "grad_norm": 0.49586981534957886,
      "learning_rate": 2.2154738327645973e-06,
      "loss": 0.2747,
      "step": 43420
    },
    {
      "epoch": 0.7840768747427036,
      "grad_norm": 0.38649171590805054,
      "learning_rate": 2.2136938267489062e-06,
      "loss": 0.2154,
      "step": 43425
    },
    {
      "epoch": 0.7841671541756043,
      "grad_norm": 0.6896102428436279,
      "learning_rate": 2.211914447084936e-06,
      "loss": 0.2633,
      "step": 43430
    },
    {
      "epoch": 0.7842574336085051,
      "grad_norm": 0.4301951825618744,
      "learning_rate": 2.2101356939158226e-06,
      "loss": 0.2399,
      "step": 43435
    },
    {
      "epoch": 0.7843477130414057,
      "grad_norm": 0.18921710550785065,
      "learning_rate": 2.2083575673846557e-06,
      "loss": 0.1989,
      "step": 43440
    },
    {
      "epoch": 0.7844379924743065,
      "grad_norm": 0.2111215591430664,
      "learning_rate": 2.2065800676344704e-06,
      "loss": 0.1991,
      "step": 43445
    },
    {
      "epoch": 0.7845282719072072,
      "grad_norm": 0.5096626281738281,
      "learning_rate": 2.204803194808257e-06,
      "loss": 0.2655,
      "step": 43450
    },
    {
      "epoch": 0.7846185513401079,
      "grad_norm": 0.6938595175743103,
      "learning_rate": 2.2030269490489476e-06,
      "loss": 0.2023,
      "step": 43455
    },
    {
      "epoch": 0.7847088307730086,
      "grad_norm": 0.26685193181037903,
      "learning_rate": 2.201251330499431e-06,
      "loss": 0.2811,
      "step": 43460
    },
    {
      "epoch": 0.7847991102059093,
      "grad_norm": 0.47773346304893494,
      "learning_rate": 2.199476339302542e-06,
      "loss": 0.2015,
      "step": 43465
    },
    {
      "epoch": 0.78488938963881,
      "grad_norm": 0.3084557354450226,
      "learning_rate": 2.1977019756010688e-06,
      "loss": 0.169,
      "step": 43470
    },
    {
      "epoch": 0.7849796690717108,
      "grad_norm": 0.5808871984481812,
      "learning_rate": 2.1959282395377444e-06,
      "loss": 0.2921,
      "step": 43475
    },
    {
      "epoch": 0.7850699485046114,
      "grad_norm": 0.400627076625824,
      "learning_rate": 2.194155131255249e-06,
      "loss": 0.1515,
      "step": 43480
    },
    {
      "epoch": 0.7851602279375122,
      "grad_norm": 0.489260733127594,
      "learning_rate": 2.192382650896219e-06,
      "loss": 0.224,
      "step": 43485
    },
    {
      "epoch": 0.7852505073704129,
      "grad_norm": 0.36692380905151367,
      "learning_rate": 2.1906107986032376e-06,
      "loss": 0.2647,
      "step": 43490
    },
    {
      "epoch": 0.7853407868033137,
      "grad_norm": 0.3548275828361511,
      "learning_rate": 2.1888395745188406e-06,
      "loss": 0.1723,
      "step": 43495
    },
    {
      "epoch": 0.7854310662362143,
      "grad_norm": 0.6279739141464233,
      "learning_rate": 2.187068978785505e-06,
      "loss": 0.1765,
      "step": 43500
    },
    {
      "epoch": 0.785521345669115,
      "grad_norm": 0.3587413728237152,
      "learning_rate": 2.1852990115456662e-06,
      "loss": 0.1683,
      "step": 43505
    },
    {
      "epoch": 0.7856116251020158,
      "grad_norm": 0.5421385169029236,
      "learning_rate": 2.183529672941701e-06,
      "loss": 0.2902,
      "step": 43510
    },
    {
      "epoch": 0.7857019045349165,
      "grad_norm": 0.5938059091567993,
      "learning_rate": 2.1817609631159465e-06,
      "loss": 0.2155,
      "step": 43515
    },
    {
      "epoch": 0.7857921839678171,
      "grad_norm": 0.6020984053611755,
      "learning_rate": 2.179992882210674e-06,
      "loss": 0.3028,
      "step": 43520
    },
    {
      "epoch": 0.7858824634007179,
      "grad_norm": 0.26961055397987366,
      "learning_rate": 2.1782254303681183e-06,
      "loss": 0.278,
      "step": 43525
    },
    {
      "epoch": 0.7859727428336186,
      "grad_norm": 0.6371787190437317,
      "learning_rate": 2.176458607730457e-06,
      "loss": 0.2311,
      "step": 43530
    },
    {
      "epoch": 0.7860630222665194,
      "grad_norm": 0.25425082445144653,
      "learning_rate": 2.1746924144398205e-06,
      "loss": 0.1772,
      "step": 43535
    },
    {
      "epoch": 0.78615330169942,
      "grad_norm": 0.23963913321495056,
      "learning_rate": 2.1729268506382806e-06,
      "loss": 0.2656,
      "step": 43540
    },
    {
      "epoch": 0.7862435811323207,
      "grad_norm": 0.7477115988731384,
      "learning_rate": 2.17116191646787e-06,
      "loss": 0.2777,
      "step": 43545
    },
    {
      "epoch": 0.7863338605652215,
      "grad_norm": 0.4997061789035797,
      "learning_rate": 2.1693976120705597e-06,
      "loss": 0.2512,
      "step": 43550
    },
    {
      "epoch": 0.7864241399981222,
      "grad_norm": 0.7205522656440735,
      "learning_rate": 2.1676339375882783e-06,
      "loss": 0.1962,
      "step": 43555
    },
    {
      "epoch": 0.7865144194310228,
      "grad_norm": 0.529256284236908,
      "learning_rate": 2.1658708931629014e-06,
      "loss": 0.1789,
      "step": 43560
    },
    {
      "epoch": 0.7866046988639236,
      "grad_norm": 0.5814422965049744,
      "learning_rate": 2.1641084789362486e-06,
      "loss": 0.1929,
      "step": 43565
    },
    {
      "epoch": 0.7866949782968243,
      "grad_norm": 0.6365538835525513,
      "learning_rate": 2.1623466950501004e-06,
      "loss": 0.2114,
      "step": 43570
    },
    {
      "epoch": 0.7867852577297251,
      "grad_norm": 0.8101227879524231,
      "learning_rate": 2.1605855416461706e-06,
      "loss": 0.2193,
      "step": 43575
    },
    {
      "epoch": 0.7868755371626257,
      "grad_norm": 0.7182210087776184,
      "learning_rate": 2.1588250188661406e-06,
      "loss": 0.26,
      "step": 43580
    },
    {
      "epoch": 0.7869658165955264,
      "grad_norm": 0.6568745374679565,
      "learning_rate": 2.157065126851623e-06,
      "loss": 0.2076,
      "step": 43585
    },
    {
      "epoch": 0.7870560960284272,
      "grad_norm": 0.402112752199173,
      "learning_rate": 2.155305865744193e-06,
      "loss": 0.3074,
      "step": 43590
    },
    {
      "epoch": 0.7871463754613279,
      "grad_norm": 0.2906830608844757,
      "learning_rate": 2.1535472356853692e-06,
      "loss": 0.1138,
      "step": 43595
    },
    {
      "epoch": 0.7872366548942287,
      "grad_norm": 0.4955861568450928,
      "learning_rate": 2.151789236816624e-06,
      "loss": 0.1751,
      "step": 43600
    },
    {
      "epoch": 0.7873269343271293,
      "grad_norm": 0.4004211723804474,
      "learning_rate": 2.1500318692793697e-06,
      "loss": 0.1865,
      "step": 43605
    },
    {
      "epoch": 0.78741721376003,
      "grad_norm": 0.5741631984710693,
      "learning_rate": 2.1482751332149797e-06,
      "loss": 0.2453,
      "step": 43610
    },
    {
      "epoch": 0.7875074931929308,
      "grad_norm": 0.3867337703704834,
      "learning_rate": 2.1465190287647654e-06,
      "loss": 0.2174,
      "step": 43615
    },
    {
      "epoch": 0.7875977726258315,
      "grad_norm": 0.4493885636329651,
      "learning_rate": 2.1447635560699943e-06,
      "loss": 0.1979,
      "step": 43620
    },
    {
      "epoch": 0.7876880520587322,
      "grad_norm": 0.41886889934539795,
      "learning_rate": 2.1430087152718815e-06,
      "loss": 0.2179,
      "step": 43625
    },
    {
      "epoch": 0.7877783314916329,
      "grad_norm": 0.2565055787563324,
      "learning_rate": 2.1412545065115953e-06,
      "loss": 0.2582,
      "step": 43630
    },
    {
      "epoch": 0.7878686109245336,
      "grad_norm": 0.4966196119785309,
      "learning_rate": 2.139500929930243e-06,
      "loss": 0.2175,
      "step": 43635
    },
    {
      "epoch": 0.7879588903574344,
      "grad_norm": 0.2613748610019684,
      "learning_rate": 2.137747985668892e-06,
      "loss": 0.1366,
      "step": 43640
    },
    {
      "epoch": 0.788049169790335,
      "grad_norm": 0.5003235936164856,
      "learning_rate": 2.1359956738685483e-06,
      "loss": 0.2165,
      "step": 43645
    },
    {
      "epoch": 0.7881394492232358,
      "grad_norm": 0.5621291399002075,
      "learning_rate": 2.1342439946701788e-06,
      "loss": 0.2396,
      "step": 43650
    },
    {
      "epoch": 0.7882297286561365,
      "grad_norm": 0.24266287684440613,
      "learning_rate": 2.132492948214687e-06,
      "loss": 0.2551,
      "step": 43655
    },
    {
      "epoch": 0.7883200080890372,
      "grad_norm": 0.46053916215896606,
      "learning_rate": 2.1307425346429354e-06,
      "loss": 0.1598,
      "step": 43660
    },
    {
      "epoch": 0.7884102875219379,
      "grad_norm": 0.32211849093437195,
      "learning_rate": 2.1289927540957355e-06,
      "loss": 0.2583,
      "step": 43665
    },
    {
      "epoch": 0.7885005669548386,
      "grad_norm": 0.19784227013587952,
      "learning_rate": 2.127243606713838e-06,
      "loss": 0.1961,
      "step": 43670
    },
    {
      "epoch": 0.7885908463877394,
      "grad_norm": 0.2373937964439392,
      "learning_rate": 2.1254950926379537e-06,
      "loss": 0.1871,
      "step": 43675
    },
    {
      "epoch": 0.7886811258206401,
      "grad_norm": 0.5469317436218262,
      "learning_rate": 2.123747212008733e-06,
      "loss": 0.2107,
      "step": 43680
    },
    {
      "epoch": 0.7887714052535407,
      "grad_norm": 1.6750993728637695,
      "learning_rate": 2.1219999649667844e-06,
      "loss": 0.2939,
      "step": 43685
    },
    {
      "epoch": 0.7888616846864415,
      "grad_norm": 0.6998043656349182,
      "learning_rate": 2.12025335165266e-06,
      "loss": 0.2774,
      "step": 43690
    },
    {
      "epoch": 0.7889519641193422,
      "grad_norm": 0.37849679589271545,
      "learning_rate": 2.118507372206864e-06,
      "loss": 0.1752,
      "step": 43695
    },
    {
      "epoch": 0.789042243552243,
      "grad_norm": 0.48230743408203125,
      "learning_rate": 2.116762026769844e-06,
      "loss": 0.2855,
      "step": 43700
    },
    {
      "epoch": 0.7891325229851436,
      "grad_norm": 0.4847571849822998,
      "learning_rate": 2.115017315482004e-06,
      "loss": 0.1631,
      "step": 43705
    },
    {
      "epoch": 0.7892228024180443,
      "grad_norm": 0.6108053922653198,
      "learning_rate": 2.1132732384836896e-06,
      "loss": 0.2345,
      "step": 43710
    },
    {
      "epoch": 0.7893130818509451,
      "grad_norm": 0.46877530217170715,
      "learning_rate": 2.1115297959152036e-06,
      "loss": 0.1766,
      "step": 43715
    },
    {
      "epoch": 0.7894033612838458,
      "grad_norm": 0.24572497606277466,
      "learning_rate": 2.1097869879167887e-06,
      "loss": 0.2355,
      "step": 43720
    },
    {
      "epoch": 0.7894936407167464,
      "grad_norm": 0.35887739062309265,
      "learning_rate": 2.1080448146286426e-06,
      "loss": 0.2242,
      "step": 43725
    },
    {
      "epoch": 0.7895839201496472,
      "grad_norm": 0.5892414450645447,
      "learning_rate": 2.1063032761909107e-06,
      "loss": 0.2766,
      "step": 43730
    },
    {
      "epoch": 0.7896741995825479,
      "grad_norm": 0.2584325969219208,
      "learning_rate": 2.1045623727436916e-06,
      "loss": 0.1576,
      "step": 43735
    },
    {
      "epoch": 0.7897644790154487,
      "grad_norm": 0.5412663817405701,
      "learning_rate": 2.1028221044270204e-06,
      "loss": 0.2048,
      "step": 43740
    },
    {
      "epoch": 0.7898547584483493,
      "grad_norm": 0.567855715751648,
      "learning_rate": 2.101082471380895e-06,
      "loss": 0.248,
      "step": 43745
    },
    {
      "epoch": 0.78994503788125,
      "grad_norm": 0.33024871349334717,
      "learning_rate": 2.0993434737452535e-06,
      "loss": 0.2433,
      "step": 43750
    },
    {
      "epoch": 0.7900353173141508,
      "grad_norm": 0.1798364520072937,
      "learning_rate": 2.097605111659985e-06,
      "loss": 0.2805,
      "step": 43755
    },
    {
      "epoch": 0.7901255967470515,
      "grad_norm": 0.7021406292915344,
      "learning_rate": 2.0958673852649314e-06,
      "loss": 0.1943,
      "step": 43760
    },
    {
      "epoch": 0.7902158761799521,
      "grad_norm": 0.5602862238883972,
      "learning_rate": 2.0941302946998754e-06,
      "loss": 0.2384,
      "step": 43765
    },
    {
      "epoch": 0.7903061556128529,
      "grad_norm": 0.43217143416404724,
      "learning_rate": 2.0923938401045596e-06,
      "loss": 0.2101,
      "step": 43770
    },
    {
      "epoch": 0.7903964350457536,
      "grad_norm": 0.30548378825187683,
      "learning_rate": 2.090658021618662e-06,
      "loss": 0.2774,
      "step": 43775
    },
    {
      "epoch": 0.7904867144786544,
      "grad_norm": 0.5456411242485046,
      "learning_rate": 2.0889228393818238e-06,
      "loss": 0.3424,
      "step": 43780
    },
    {
      "epoch": 0.790576993911555,
      "grad_norm": 0.636239230632782,
      "learning_rate": 2.087188293533621e-06,
      "loss": 0.206,
      "step": 43785
    },
    {
      "epoch": 0.7906672733444557,
      "grad_norm": 0.3845883011817932,
      "learning_rate": 2.0854543842135877e-06,
      "loss": 0.1681,
      "step": 43790
    },
    {
      "epoch": 0.7907575527773565,
      "grad_norm": 0.5687711834907532,
      "learning_rate": 2.0837211115612056e-06,
      "loss": 0.24,
      "step": 43795
    },
    {
      "epoch": 0.7908478322102572,
      "grad_norm": 0.38964512944221497,
      "learning_rate": 2.081988475715905e-06,
      "loss": 0.2319,
      "step": 43800
    },
    {
      "epoch": 0.7909381116431579,
      "grad_norm": 0.6546294093132019,
      "learning_rate": 2.0802564768170597e-06,
      "loss": 0.2245,
      "step": 43805
    },
    {
      "epoch": 0.7910283910760586,
      "grad_norm": 1.599808931350708,
      "learning_rate": 2.078525115004002e-06,
      "loss": 0.2973,
      "step": 43810
    },
    {
      "epoch": 0.7911186705089593,
      "grad_norm": 0.3674757182598114,
      "learning_rate": 2.0767943904160003e-06,
      "loss": 0.162,
      "step": 43815
    },
    {
      "epoch": 0.7912089499418601,
      "grad_norm": 0.43936610221862793,
      "learning_rate": 2.0750643031922833e-06,
      "loss": 0.2228,
      "step": 43820
    },
    {
      "epoch": 0.7912992293747607,
      "grad_norm": 0.3835139572620392,
      "learning_rate": 2.0733348534720233e-06,
      "loss": 0.1925,
      "step": 43825
    },
    {
      "epoch": 0.7913895088076615,
      "grad_norm": 0.5503572821617126,
      "learning_rate": 2.071606041394345e-06,
      "loss": 0.1709,
      "step": 43830
    },
    {
      "epoch": 0.7914797882405622,
      "grad_norm": 0.4535362422466278,
      "learning_rate": 2.0698778670983144e-06,
      "loss": 0.3186,
      "step": 43835
    },
    {
      "epoch": 0.7915700676734629,
      "grad_norm": 0.2782318592071533,
      "learning_rate": 2.0681503307229502e-06,
      "loss": 0.2229,
      "step": 43840
    },
    {
      "epoch": 0.7916603471063636,
      "grad_norm": 0.5089423656463623,
      "learning_rate": 2.066423432407224e-06,
      "loss": 0.1557,
      "step": 43845
    },
    {
      "epoch": 0.7917506265392643,
      "grad_norm": 0.3967224657535553,
      "learning_rate": 2.0646971722900476e-06,
      "loss": 0.2635,
      "step": 43850
    },
    {
      "epoch": 0.791840905972165,
      "grad_norm": 0.20601853728294373,
      "learning_rate": 2.0629715505102886e-06,
      "loss": 0.2717,
      "step": 43855
    },
    {
      "epoch": 0.7919311854050658,
      "grad_norm": 0.5291382670402527,
      "learning_rate": 2.0612465672067606e-06,
      "loss": 0.373,
      "step": 43860
    },
    {
      "epoch": 0.7920214648379664,
      "grad_norm": 0.42321300506591797,
      "learning_rate": 2.0595222225182286e-06,
      "loss": 0.2309,
      "step": 43865
    },
    {
      "epoch": 0.7921117442708672,
      "grad_norm": 0.24983927607536316,
      "learning_rate": 2.057798516583398e-06,
      "loss": 0.1551,
      "step": 43870
    },
    {
      "epoch": 0.7922020237037679,
      "grad_norm": 0.4658087491989136,
      "learning_rate": 2.056075449540934e-06,
      "loss": 0.2469,
      "step": 43875
    },
    {
      "epoch": 0.7922923031366687,
      "grad_norm": 0.501194179058075,
      "learning_rate": 2.0543530215294396e-06,
      "loss": 0.2513,
      "step": 43880
    },
    {
      "epoch": 0.7923825825695693,
      "grad_norm": 0.21056383848190308,
      "learning_rate": 2.0526312326874742e-06,
      "loss": 0.161,
      "step": 43885
    },
    {
      "epoch": 0.79247286200247,
      "grad_norm": 0.3605361580848694,
      "learning_rate": 2.0509100831535433e-06,
      "loss": 0.1909,
      "step": 43890
    },
    {
      "epoch": 0.7925631414353708,
      "grad_norm": 0.8153850436210632,
      "learning_rate": 2.049189573066103e-06,
      "loss": 0.1772,
      "step": 43895
    },
    {
      "epoch": 0.7926534208682715,
      "grad_norm": 0.7240114212036133,
      "learning_rate": 2.0474697025635516e-06,
      "loss": 0.3147,
      "step": 43900
    },
    {
      "epoch": 0.7927437003011721,
      "grad_norm": 0.7593914270401001,
      "learning_rate": 2.0457504717842436e-06,
      "loss": 0.2647,
      "step": 43905
    },
    {
      "epoch": 0.7928339797340729,
      "grad_norm": 0.5896070599555969,
      "learning_rate": 2.044031880866475e-06,
      "loss": 0.2271,
      "step": 43910
    },
    {
      "epoch": 0.7929242591669736,
      "grad_norm": 0.7276526689529419,
      "learning_rate": 2.0423139299484994e-06,
      "loss": 0.1991,
      "step": 43915
    },
    {
      "epoch": 0.7930145385998744,
      "grad_norm": 0.7303504943847656,
      "learning_rate": 2.0405966191685067e-06,
      "loss": 0.1968,
      "step": 43920
    },
    {
      "epoch": 0.793104818032775,
      "grad_norm": 0.27771618962287903,
      "learning_rate": 2.038879948664646e-06,
      "loss": 0.2311,
      "step": 43925
    },
    {
      "epoch": 0.7931950974656757,
      "grad_norm": 0.6631217002868652,
      "learning_rate": 2.0371639185750138e-06,
      "loss": 0.1537,
      "step": 43930
    },
    {
      "epoch": 0.7932853768985765,
      "grad_norm": 0.2898246943950653,
      "learning_rate": 2.0354485290376457e-06,
      "loss": 0.2196,
      "step": 43935
    },
    {
      "epoch": 0.7933756563314772,
      "grad_norm": 0.22012820839881897,
      "learning_rate": 2.0337337801905378e-06,
      "loss": 0.1732,
      "step": 43940
    },
    {
      "epoch": 0.7934659357643778,
      "grad_norm": 0.4568951725959778,
      "learning_rate": 2.0320196721716245e-06,
      "loss": 0.2369,
      "step": 43945
    },
    {
      "epoch": 0.7935562151972786,
      "grad_norm": 0.9336597919464111,
      "learning_rate": 2.030306205118796e-06,
      "loss": 0.2489,
      "step": 43950
    },
    {
      "epoch": 0.7936464946301793,
      "grad_norm": 0.2505965828895569,
      "learning_rate": 2.0285933791698874e-06,
      "loss": 0.206,
      "step": 43955
    },
    {
      "epoch": 0.7937367740630801,
      "grad_norm": 0.4175946116447449,
      "learning_rate": 2.0268811944626877e-06,
      "loss": 0.17,
      "step": 43960
    },
    {
      "epoch": 0.7938270534959807,
      "grad_norm": 0.47757190465927124,
      "learning_rate": 2.0251696511349216e-06,
      "loss": 0.1638,
      "step": 43965
    },
    {
      "epoch": 0.7939173329288814,
      "grad_norm": 0.37691453099250793,
      "learning_rate": 2.0234587493242776e-06,
      "loss": 0.2345,
      "step": 43970
    },
    {
      "epoch": 0.7940076123617822,
      "grad_norm": 0.2829607129096985,
      "learning_rate": 2.0217484891683805e-06,
      "loss": 0.2027,
      "step": 43975
    },
    {
      "epoch": 0.7940978917946829,
      "grad_norm": 0.6008973121643066,
      "learning_rate": 2.0200388708048112e-06,
      "loss": 0.1772,
      "step": 43980
    },
    {
      "epoch": 0.7941881712275836,
      "grad_norm": 0.4292893707752228,
      "learning_rate": 2.018329894371093e-06,
      "loss": 0.2053,
      "step": 43985
    },
    {
      "epoch": 0.7942784506604843,
      "grad_norm": 0.47939789295196533,
      "learning_rate": 2.016621560004701e-06,
      "loss": 0.2613,
      "step": 43990
    },
    {
      "epoch": 0.794368730093385,
      "grad_norm": 0.4499230980873108,
      "learning_rate": 2.014913867843061e-06,
      "loss": 0.2316,
      "step": 43995
    },
    {
      "epoch": 0.7944590095262858,
      "grad_norm": 0.36144039034843445,
      "learning_rate": 2.0132068180235463e-06,
      "loss": 0.1648,
      "step": 44000
    },
    {
      "epoch": 0.7945492889591865,
      "grad_norm": 0.46915683150291443,
      "learning_rate": 2.01150041068347e-06,
      "loss": 0.2406,
      "step": 44005
    },
    {
      "epoch": 0.7946395683920872,
      "grad_norm": 0.3485051095485687,
      "learning_rate": 2.009794645960107e-06,
      "loss": 0.2579,
      "step": 44010
    },
    {
      "epoch": 0.7947298478249879,
      "grad_norm": 0.46306657791137695,
      "learning_rate": 2.0080895239906674e-06,
      "loss": 0.2315,
      "step": 44015
    },
    {
      "epoch": 0.7948201272578886,
      "grad_norm": 0.46812009811401367,
      "learning_rate": 2.006385044912318e-06,
      "loss": 0.198,
      "step": 44020
    },
    {
      "epoch": 0.7949104066907894,
      "grad_norm": 0.6879592537879944,
      "learning_rate": 2.004681208862176e-06,
      "loss": 0.3155,
      "step": 44025
    },
    {
      "epoch": 0.79500068612369,
      "grad_norm": 0.26458317041397095,
      "learning_rate": 2.002978015977296e-06,
      "loss": 0.1753,
      "step": 44030
    },
    {
      "epoch": 0.7950909655565908,
      "grad_norm": 0.5082259774208069,
      "learning_rate": 2.001275466394693e-06,
      "loss": 0.2284,
      "step": 44035
    },
    {
      "epoch": 0.7951812449894915,
      "grad_norm": 0.593024492263794,
      "learning_rate": 1.999573560251321e-06,
      "loss": 0.2442,
      "step": 44040
    },
    {
      "epoch": 0.7952715244223922,
      "grad_norm": 0.4718942940235138,
      "learning_rate": 1.9978722976840894e-06,
      "loss": 0.2392,
      "step": 44045
    },
    {
      "epoch": 0.7953618038552929,
      "grad_norm": 0.3856017291545868,
      "learning_rate": 1.9961716788298467e-06,
      "loss": 0.2105,
      "step": 44050
    },
    {
      "epoch": 0.7954520832881936,
      "grad_norm": 0.3091108500957489,
      "learning_rate": 1.9944717038254002e-06,
      "loss": 0.2276,
      "step": 44055
    },
    {
      "epoch": 0.7955423627210944,
      "grad_norm": 0.6325206160545349,
      "learning_rate": 1.9927723728074997e-06,
      "loss": 0.2424,
      "step": 44060
    },
    {
      "epoch": 0.7956326421539951,
      "grad_norm": 0.39874640107154846,
      "learning_rate": 1.9910736859128453e-06,
      "loss": 0.2064,
      "step": 44065
    },
    {
      "epoch": 0.7957229215868957,
      "grad_norm": 0.3896929919719696,
      "learning_rate": 1.98937564327808e-06,
      "loss": 0.2513,
      "step": 44070
    },
    {
      "epoch": 0.7958132010197965,
      "grad_norm": 0.5646119117736816,
      "learning_rate": 1.9876782450398036e-06,
      "loss": 0.2181,
      "step": 44075
    },
    {
      "epoch": 0.7959034804526972,
      "grad_norm": 0.3404809534549713,
      "learning_rate": 1.9859814913345556e-06,
      "loss": 0.1505,
      "step": 44080
    },
    {
      "epoch": 0.795993759885598,
      "grad_norm": 0.23280753195285797,
      "learning_rate": 1.9842853822988285e-06,
      "loss": 0.2858,
      "step": 44085
    },
    {
      "epoch": 0.7960840393184986,
      "grad_norm": 0.4387173354625702,
      "learning_rate": 1.9825899180690623e-06,
      "loss": 0.2395,
      "step": 44090
    },
    {
      "epoch": 0.7961743187513993,
      "grad_norm": 0.5827388167381287,
      "learning_rate": 1.9808950987816477e-06,
      "loss": 0.1909,
      "step": 44095
    },
    {
      "epoch": 0.7962645981843001,
      "grad_norm": 0.28081467747688293,
      "learning_rate": 1.979200924572915e-06,
      "loss": 0.2074,
      "step": 44100
    },
    {
      "epoch": 0.7963548776172008,
      "grad_norm": 0.36692309379577637,
      "learning_rate": 1.9775073955791536e-06,
      "loss": 0.2335,
      "step": 44105
    },
    {
      "epoch": 0.7964451570501014,
      "grad_norm": 0.5759391188621521,
      "learning_rate": 1.9758145119365914e-06,
      "loss": 0.175,
      "step": 44110
    },
    {
      "epoch": 0.7965354364830022,
      "grad_norm": 0.6587239503860474,
      "learning_rate": 1.9741222737814113e-06,
      "loss": 0.1967,
      "step": 44115
    },
    {
      "epoch": 0.7966257159159029,
      "grad_norm": 0.5736746788024902,
      "learning_rate": 1.9724306812497394e-06,
      "loss": 0.3098,
      "step": 44120
    },
    {
      "epoch": 0.7967159953488037,
      "grad_norm": 0.3512267768383026,
      "learning_rate": 1.9707397344776524e-06,
      "loss": 0.1847,
      "step": 44125
    },
    {
      "epoch": 0.7968062747817043,
      "grad_norm": 0.4105289876461029,
      "learning_rate": 1.969049433601178e-06,
      "loss": 0.1587,
      "step": 44130
    },
    {
      "epoch": 0.796896554214605,
      "grad_norm": 0.3709127604961395,
      "learning_rate": 1.9673597787562835e-06,
      "loss": 0.1709,
      "step": 44135
    },
    {
      "epoch": 0.7969868336475058,
      "grad_norm": 0.24260620772838593,
      "learning_rate": 1.9656707700788945e-06,
      "loss": 0.2637,
      "step": 44140
    },
    {
      "epoch": 0.7970771130804065,
      "grad_norm": 0.3361005485057831,
      "learning_rate": 1.9639824077048754e-06,
      "loss": 0.1844,
      "step": 44145
    },
    {
      "epoch": 0.7971673925133071,
      "grad_norm": 0.6023269295692444,
      "learning_rate": 1.9622946917700425e-06,
      "loss": 0.2392,
      "step": 44150
    },
    {
      "epoch": 0.7972576719462079,
      "grad_norm": 0.4582315981388092,
      "learning_rate": 1.9606076224101624e-06,
      "loss": 0.2976,
      "step": 44155
    },
    {
      "epoch": 0.7973479513791086,
      "grad_norm": 0.1931309700012207,
      "learning_rate": 1.9589211997609493e-06,
      "loss": 0.1326,
      "step": 44160
    },
    {
      "epoch": 0.7974382308120094,
      "grad_norm": 0.5437281131744385,
      "learning_rate": 1.9572354239580594e-06,
      "loss": 0.21,
      "step": 44165
    },
    {
      "epoch": 0.79752851024491,
      "grad_norm": 0.31165415048599243,
      "learning_rate": 1.955550295137105e-06,
      "loss": 0.2583,
      "step": 44170
    },
    {
      "epoch": 0.7976187896778107,
      "grad_norm": 0.3427805006504059,
      "learning_rate": 1.9538658134336385e-06,
      "loss": 0.2355,
      "step": 44175
    },
    {
      "epoch": 0.7977090691107115,
      "grad_norm": 0.30942103266716003,
      "learning_rate": 1.9521819789831687e-06,
      "loss": 0.1942,
      "step": 44180
    },
    {
      "epoch": 0.7977993485436122,
      "grad_norm": 0.4148818552494049,
      "learning_rate": 1.9504987919211426e-06,
      "loss": 0.299,
      "step": 44185
    },
    {
      "epoch": 0.7978896279765129,
      "grad_norm": 0.34355810284614563,
      "learning_rate": 1.9488162523829635e-06,
      "loss": 0.246,
      "step": 44190
    },
    {
      "epoch": 0.7979799074094136,
      "grad_norm": 0.5479265451431274,
      "learning_rate": 1.9471343605039796e-06,
      "loss": 0.1632,
      "step": 44195
    },
    {
      "epoch": 0.7980701868423143,
      "grad_norm": 0.43104642629623413,
      "learning_rate": 1.945453116419488e-06,
      "loss": 0.2268,
      "step": 44200
    },
    {
      "epoch": 0.7981604662752151,
      "grad_norm": 0.6481384038925171,
      "learning_rate": 1.943772520264732e-06,
      "loss": 0.231,
      "step": 44205
    },
    {
      "epoch": 0.7982507457081157,
      "grad_norm": 0.3647168278694153,
      "learning_rate": 1.9420925721748985e-06,
      "loss": 0.2154,
      "step": 44210
    },
    {
      "epoch": 0.7983410251410165,
      "grad_norm": 0.5041742920875549,
      "learning_rate": 1.940413272285131e-06,
      "loss": 0.1393,
      "step": 44215
    },
    {
      "epoch": 0.7984313045739172,
      "grad_norm": 0.16152489185333252,
      "learning_rate": 1.938734620730518e-06,
      "loss": 0.2195,
      "step": 44220
    },
    {
      "epoch": 0.7985215840068179,
      "grad_norm": 0.21917890012264252,
      "learning_rate": 1.9370566176460957e-06,
      "loss": 0.1921,
      "step": 44225
    },
    {
      "epoch": 0.7986118634397186,
      "grad_norm": 1.227859616279602,
      "learning_rate": 1.935379263166842e-06,
      "loss": 0.294,
      "step": 44230
    },
    {
      "epoch": 0.7987021428726193,
      "grad_norm": 0.507227897644043,
      "learning_rate": 1.9337025574276945e-06,
      "loss": 0.2109,
      "step": 44235
    },
    {
      "epoch": 0.7987924223055201,
      "grad_norm": 0.521886944770813,
      "learning_rate": 1.9320265005635262e-06,
      "loss": 0.2597,
      "step": 44240
    },
    {
      "epoch": 0.7988827017384208,
      "grad_norm": 0.1763850450515747,
      "learning_rate": 1.9303510927091684e-06,
      "loss": 0.2288,
      "step": 44245
    },
    {
      "epoch": 0.7989729811713214,
      "grad_norm": 0.7960894703865051,
      "learning_rate": 1.9286763339993907e-06,
      "loss": 0.3042,
      "step": 44250
    },
    {
      "epoch": 0.7990632606042222,
      "grad_norm": 0.39358705282211304,
      "learning_rate": 1.927002224568917e-06,
      "loss": 0.2801,
      "step": 44255
    },
    {
      "epoch": 0.7991535400371229,
      "grad_norm": 0.8040003776550293,
      "learning_rate": 1.925328764552419e-06,
      "loss": 0.2096,
      "step": 44260
    },
    {
      "epoch": 0.7992438194700237,
      "grad_norm": 0.48055535554885864,
      "learning_rate": 1.9236559540845157e-06,
      "loss": 0.3142,
      "step": 44265
    },
    {
      "epoch": 0.7993340989029243,
      "grad_norm": 0.38093963265419006,
      "learning_rate": 1.9219837932997666e-06,
      "loss": 0.2982,
      "step": 44270
    },
    {
      "epoch": 0.799424378335825,
      "grad_norm": 0.5091568827629089,
      "learning_rate": 1.9203122823326925e-06,
      "loss": 0.1926,
      "step": 44275
    },
    {
      "epoch": 0.7995146577687258,
      "grad_norm": 0.4233086109161377,
      "learning_rate": 1.918641421317746e-06,
      "loss": 0.2658,
      "step": 44280
    },
    {
      "epoch": 0.7996049372016265,
      "grad_norm": 0.3393450379371643,
      "learning_rate": 1.9169712103893412e-06,
      "loss": 0.2316,
      "step": 44285
    },
    {
      "epoch": 0.7996952166345271,
      "grad_norm": 0.47289741039276123,
      "learning_rate": 1.915301649681833e-06,
      "loss": 0.3163,
      "step": 44290
    },
    {
      "epoch": 0.7997854960674279,
      "grad_norm": 0.8122720122337341,
      "learning_rate": 1.9136327393295285e-06,
      "loss": 0.2344,
      "step": 44295
    },
    {
      "epoch": 0.7998757755003286,
      "grad_norm": 0.40111979842185974,
      "learning_rate": 1.911964479466677e-06,
      "loss": 0.2124,
      "step": 44300
    },
    {
      "epoch": 0.7999660549332294,
      "grad_norm": 0.3363553583621979,
      "learning_rate": 1.9102968702274737e-06,
      "loss": 0.1898,
      "step": 44305
    },
    {
      "epoch": 0.80005633436613,
      "grad_norm": 0.5055487751960754,
      "learning_rate": 1.9086299117460726e-06,
      "loss": 0.2962,
      "step": 44310
    },
    {
      "epoch": 0.8001466137990307,
      "grad_norm": 0.22720535099506378,
      "learning_rate": 1.906963604156562e-06,
      "loss": 0.1968,
      "step": 44315
    },
    {
      "epoch": 0.8002368932319315,
      "grad_norm": 0.4407348930835724,
      "learning_rate": 1.9052979475929877e-06,
      "loss": 0.2451,
      "step": 44320
    },
    {
      "epoch": 0.8003271726648322,
      "grad_norm": 0.40608954429626465,
      "learning_rate": 1.9036329421893395e-06,
      "loss": 0.1808,
      "step": 44325
    },
    {
      "epoch": 0.8004174520977329,
      "grad_norm": 0.34735071659088135,
      "learning_rate": 1.901968588079558e-06,
      "loss": 0.1944,
      "step": 44330
    },
    {
      "epoch": 0.8005077315306336,
      "grad_norm": 0.7078969478607178,
      "learning_rate": 1.9003048853975215e-06,
      "loss": 0.2149,
      "step": 44335
    },
    {
      "epoch": 0.8005980109635343,
      "grad_norm": 0.5943984389305115,
      "learning_rate": 1.8986418342770696e-06,
      "loss": 0.226,
      "step": 44340
    },
    {
      "epoch": 0.8006882903964351,
      "grad_norm": 0.5430403351783752,
      "learning_rate": 1.8969794348519777e-06,
      "loss": 0.2443,
      "step": 44345
    },
    {
      "epoch": 0.8007785698293357,
      "grad_norm": 0.38101640343666077,
      "learning_rate": 1.8953176872559742e-06,
      "loss": 0.2337,
      "step": 44350
    },
    {
      "epoch": 0.8008688492622364,
      "grad_norm": 0.30152231454849243,
      "learning_rate": 1.8936565916227367e-06,
      "loss": 0.2549,
      "step": 44355
    },
    {
      "epoch": 0.8009591286951372,
      "grad_norm": 0.6075379252433777,
      "learning_rate": 1.8919961480858907e-06,
      "loss": 0.1433,
      "step": 44360
    },
    {
      "epoch": 0.8010494081280379,
      "grad_norm": 0.4083125591278076,
      "learning_rate": 1.890336356779e-06,
      "loss": 0.2292,
      "step": 44365
    },
    {
      "epoch": 0.8011396875609386,
      "grad_norm": 0.44115856289863586,
      "learning_rate": 1.8886772178355894e-06,
      "loss": 0.1646,
      "step": 44370
    },
    {
      "epoch": 0.8012299669938393,
      "grad_norm": 0.4118017554283142,
      "learning_rate": 1.8870187313891197e-06,
      "loss": 0.1987,
      "step": 44375
    },
    {
      "epoch": 0.80132024642674,
      "grad_norm": 0.3881694972515106,
      "learning_rate": 1.8853608975730075e-06,
      "loss": 0.2035,
      "step": 44380
    },
    {
      "epoch": 0.8014105258596408,
      "grad_norm": 0.5062962770462036,
      "learning_rate": 1.8837037165206074e-06,
      "loss": 0.3047,
      "step": 44385
    },
    {
      "epoch": 0.8015008052925415,
      "grad_norm": 0.37755340337753296,
      "learning_rate": 1.882047188365238e-06,
      "loss": 0.1709,
      "step": 44390
    },
    {
      "epoch": 0.8015910847254422,
      "grad_norm": 0.19946175813674927,
      "learning_rate": 1.8803913132401474e-06,
      "loss": 0.2803,
      "step": 44395
    },
    {
      "epoch": 0.8016813641583429,
      "grad_norm": 0.5771229267120361,
      "learning_rate": 1.878736091278539e-06,
      "loss": 0.2037,
      "step": 44400
    },
    {
      "epoch": 0.8017716435912436,
      "grad_norm": 0.33102309703826904,
      "learning_rate": 1.8770815226135675e-06,
      "loss": 0.2233,
      "step": 44405
    },
    {
      "epoch": 0.8018619230241444,
      "grad_norm": 0.5204035043716431,
      "learning_rate": 1.875427607378325e-06,
      "loss": 0.3369,
      "step": 44410
    },
    {
      "epoch": 0.801952202457045,
      "grad_norm": 0.5146638751029968,
      "learning_rate": 1.8737743457058598e-06,
      "loss": 0.1338,
      "step": 44415
    },
    {
      "epoch": 0.8020424818899458,
      "grad_norm": 0.393114298582077,
      "learning_rate": 1.8721217377291656e-06,
      "loss": 0.1774,
      "step": 44420
    },
    {
      "epoch": 0.8021327613228465,
      "grad_norm": 0.5393953919410706,
      "learning_rate": 1.8704697835811858e-06,
      "loss": 0.2302,
      "step": 44425
    },
    {
      "epoch": 0.8022230407557472,
      "grad_norm": 0.33049318194389343,
      "learning_rate": 1.8688184833948008e-06,
      "loss": 0.2776,
      "step": 44430
    },
    {
      "epoch": 0.8023133201886479,
      "grad_norm": 0.4153202772140503,
      "learning_rate": 1.8671678373028524e-06,
      "loss": 0.1873,
      "step": 44435
    },
    {
      "epoch": 0.8024035996215486,
      "grad_norm": 0.5525389313697815,
      "learning_rate": 1.8655178454381162e-06,
      "loss": 0.1835,
      "step": 44440
    },
    {
      "epoch": 0.8024938790544494,
      "grad_norm": 0.4064835011959076,
      "learning_rate": 1.8638685079333296e-06,
      "loss": 0.1795,
      "step": 44445
    },
    {
      "epoch": 0.8025841584873501,
      "grad_norm": 0.39771416783332825,
      "learning_rate": 1.8622198249211609e-06,
      "loss": 0.1714,
      "step": 44450
    },
    {
      "epoch": 0.8026744379202507,
      "grad_norm": 0.9161157608032227,
      "learning_rate": 1.860571796534245e-06,
      "loss": 0.2378,
      "step": 44455
    },
    {
      "epoch": 0.8027647173531515,
      "grad_norm": 0.472883015871048,
      "learning_rate": 1.8589244229051472e-06,
      "loss": 0.2975,
      "step": 44460
    },
    {
      "epoch": 0.8028549967860522,
      "grad_norm": 0.3426239490509033,
      "learning_rate": 1.8572777041663892e-06,
      "loss": 0.1242,
      "step": 44465
    },
    {
      "epoch": 0.802945276218953,
      "grad_norm": 0.29185158014297485,
      "learning_rate": 1.8556316404504348e-06,
      "loss": 0.193,
      "step": 44470
    },
    {
      "epoch": 0.8030355556518536,
      "grad_norm": 0.40711694955825806,
      "learning_rate": 1.8539862318897027e-06,
      "loss": 0.2212,
      "step": 44475
    },
    {
      "epoch": 0.8031258350847543,
      "grad_norm": 0.8984672427177429,
      "learning_rate": 1.8523414786165472e-06,
      "loss": 0.2174,
      "step": 44480
    },
    {
      "epoch": 0.8032161145176551,
      "grad_norm": 0.47199028730392456,
      "learning_rate": 1.850697380763281e-06,
      "loss": 0.2688,
      "step": 44485
    },
    {
      "epoch": 0.8033063939505558,
      "grad_norm": 0.33139854669570923,
      "learning_rate": 1.849053938462163e-06,
      "loss": 0.2384,
      "step": 44490
    },
    {
      "epoch": 0.8033966733834564,
      "grad_norm": 0.39386871457099915,
      "learning_rate": 1.8474111518453886e-06,
      "loss": 0.2262,
      "step": 44495
    },
    {
      "epoch": 0.8034869528163572,
      "grad_norm": 0.5428234934806824,
      "learning_rate": 1.845769021045115e-06,
      "loss": 0.1837,
      "step": 44500
    },
    {
      "epoch": 0.8035772322492579,
      "grad_norm": 0.4081260561943054,
      "learning_rate": 1.8441275461934338e-06,
      "loss": 0.1902,
      "step": 44505
    },
    {
      "epoch": 0.8036675116821587,
      "grad_norm": 0.6265204548835754,
      "learning_rate": 1.8424867274223956e-06,
      "loss": 0.2034,
      "step": 44510
    },
    {
      "epoch": 0.8037577911150593,
      "grad_norm": 0.31426239013671875,
      "learning_rate": 1.8408465648639873e-06,
      "loss": 0.2762,
      "step": 44515
    },
    {
      "epoch": 0.80384807054796,
      "grad_norm": 0.5084913969039917,
      "learning_rate": 1.8392070586501487e-06,
      "loss": 0.2807,
      "step": 44520
    },
    {
      "epoch": 0.8039383499808608,
      "grad_norm": 0.5164384841918945,
      "learning_rate": 1.837568208912769e-06,
      "loss": 0.1994,
      "step": 44525
    },
    {
      "epoch": 0.8040286294137615,
      "grad_norm": 0.5218554139137268,
      "learning_rate": 1.8359300157836812e-06,
      "loss": 0.3358,
      "step": 44530
    },
    {
      "epoch": 0.8041189088466622,
      "grad_norm": 0.4245232343673706,
      "learning_rate": 1.8342924793946637e-06,
      "loss": 0.1924,
      "step": 44535
    },
    {
      "epoch": 0.8042091882795629,
      "grad_norm": 0.7326409220695496,
      "learning_rate": 1.8326555998774487e-06,
      "loss": 0.2482,
      "step": 44540
    },
    {
      "epoch": 0.8042994677124636,
      "grad_norm": 0.5915966033935547,
      "learning_rate": 1.831019377363704e-06,
      "loss": 0.2157,
      "step": 44545
    },
    {
      "epoch": 0.8043897471453644,
      "grad_norm": 0.4375337064266205,
      "learning_rate": 1.8293838119850583e-06,
      "loss": 0.192,
      "step": 44550
    },
    {
      "epoch": 0.804480026578265,
      "grad_norm": 1.0574471950531006,
      "learning_rate": 1.8277489038730779e-06,
      "loss": 0.3344,
      "step": 44555
    },
    {
      "epoch": 0.8045703060111657,
      "grad_norm": 0.3823387026786804,
      "learning_rate": 1.826114653159282e-06,
      "loss": 0.2604,
      "step": 44560
    },
    {
      "epoch": 0.8046605854440665,
      "grad_norm": 0.5871850252151489,
      "learning_rate": 1.8244810599751307e-06,
      "loss": 0.259,
      "step": 44565
    },
    {
      "epoch": 0.8047508648769672,
      "grad_norm": 0.3595975339412689,
      "learning_rate": 1.822848124452039e-06,
      "loss": 0.1577,
      "step": 44570
    },
    {
      "epoch": 0.8048411443098679,
      "grad_norm": 0.3500572144985199,
      "learning_rate": 1.8212158467213613e-06,
      "loss": 0.278,
      "step": 44575
    },
    {
      "epoch": 0.8049314237427686,
      "grad_norm": 2.607898235321045,
      "learning_rate": 1.8195842269144016e-06,
      "loss": 0.21,
      "step": 44580
    },
    {
      "epoch": 0.8050217031756693,
      "grad_norm": 0.2642761170864105,
      "learning_rate": 1.8179532651624133e-06,
      "loss": 0.1544,
      "step": 44585
    },
    {
      "epoch": 0.8051119826085701,
      "grad_norm": 0.7221905589103699,
      "learning_rate": 1.816322961596595e-06,
      "loss": 0.3092,
      "step": 44590
    },
    {
      "epoch": 0.8052022620414707,
      "grad_norm": 0.4336141049861908,
      "learning_rate": 1.8146933163480963e-06,
      "loss": 0.2354,
      "step": 44595
    },
    {
      "epoch": 0.8052925414743715,
      "grad_norm": 0.3331955075263977,
      "learning_rate": 1.8130643295480044e-06,
      "loss": 0.2931,
      "step": 44600
    },
    {
      "epoch": 0.8053828209072722,
      "grad_norm": 0.4939133822917938,
      "learning_rate": 1.8114360013273657e-06,
      "loss": 0.4071,
      "step": 44605
    },
    {
      "epoch": 0.8054731003401729,
      "grad_norm": 1.1969037055969238,
      "learning_rate": 1.8098083318171611e-06,
      "loss": 0.2334,
      "step": 44610
    },
    {
      "epoch": 0.8055633797730736,
      "grad_norm": 0.6354053020477295,
      "learning_rate": 1.808181321148328e-06,
      "loss": 0.2467,
      "step": 44615
    },
    {
      "epoch": 0.8056536592059743,
      "grad_norm": 0.8038125038146973,
      "learning_rate": 1.806554969451747e-06,
      "loss": 0.2662,
      "step": 44620
    },
    {
      "epoch": 0.8057439386388751,
      "grad_norm": 0.4923882484436035,
      "learning_rate": 1.8049292768582483e-06,
      "loss": 0.2144,
      "step": 44625
    },
    {
      "epoch": 0.8058342180717758,
      "grad_norm": 0.4922740161418915,
      "learning_rate": 1.8033042434986049e-06,
      "loss": 0.1656,
      "step": 44630
    },
    {
      "epoch": 0.8059244975046764,
      "grad_norm": 0.5181177854537964,
      "learning_rate": 1.8016798695035399e-06,
      "loss": 0.1638,
      "step": 44635
    },
    {
      "epoch": 0.8060147769375772,
      "grad_norm": 0.428590327501297,
      "learning_rate": 1.80005615500372e-06,
      "loss": 0.1832,
      "step": 44640
    },
    {
      "epoch": 0.8061050563704779,
      "grad_norm": 0.47034353017807007,
      "learning_rate": 1.798433100129766e-06,
      "loss": 0.2203,
      "step": 44645
    },
    {
      "epoch": 0.8061953358033787,
      "grad_norm": 0.33631256222724915,
      "learning_rate": 1.796810705012233e-06,
      "loss": 0.1783,
      "step": 44650
    },
    {
      "epoch": 0.8062856152362793,
      "grad_norm": 0.5317744612693787,
      "learning_rate": 1.7951889697816404e-06,
      "loss": 0.2507,
      "step": 44655
    },
    {
      "epoch": 0.80637589466918,
      "grad_norm": 0.3411064147949219,
      "learning_rate": 1.7935678945684375e-06,
      "loss": 0.1693,
      "step": 44660
    },
    {
      "epoch": 0.8064661741020808,
      "grad_norm": 0.42006516456604004,
      "learning_rate": 1.7919474795030346e-06,
      "loss": 0.283,
      "step": 44665
    },
    {
      "epoch": 0.8065564535349815,
      "grad_norm": 0.49459028244018555,
      "learning_rate": 1.790327724715778e-06,
      "loss": 0.1784,
      "step": 44670
    },
    {
      "epoch": 0.8066467329678821,
      "grad_norm": 0.3617062568664551,
      "learning_rate": 1.7887086303369638e-06,
      "loss": 0.1276,
      "step": 44675
    },
    {
      "epoch": 0.8067370124007829,
      "grad_norm": 0.4747965633869171,
      "learning_rate": 1.787090196496838e-06,
      "loss": 0.205,
      "step": 44680
    },
    {
      "epoch": 0.8068272918336836,
      "grad_norm": 0.25721991062164307,
      "learning_rate": 1.7854724233255926e-06,
      "loss": 0.1509,
      "step": 44685
    },
    {
      "epoch": 0.8069175712665844,
      "grad_norm": 0.997877299785614,
      "learning_rate": 1.7838553109533674e-06,
      "loss": 0.1909,
      "step": 44690
    },
    {
      "epoch": 0.807007850699485,
      "grad_norm": 0.38840773701667786,
      "learning_rate": 1.7822388595102424e-06,
      "loss": 0.2013,
      "step": 44695
    },
    {
      "epoch": 0.8070981301323857,
      "grad_norm": 0.5792832374572754,
      "learning_rate": 1.780623069126256e-06,
      "loss": 0.098,
      "step": 44700
    },
    {
      "epoch": 0.8071884095652865,
      "grad_norm": 0.22275789082050323,
      "learning_rate": 1.7790079399313798e-06,
      "loss": 0.195,
      "step": 44705
    },
    {
      "epoch": 0.8072786889981872,
      "grad_norm": 0.28965526819229126,
      "learning_rate": 1.7773934720555452e-06,
      "loss": 0.1749,
      "step": 44710
    },
    {
      "epoch": 0.8073689684310879,
      "grad_norm": 0.4989018440246582,
      "learning_rate": 1.7757796656286164e-06,
      "loss": 0.2092,
      "step": 44715
    },
    {
      "epoch": 0.8074592478639886,
      "grad_norm": 0.3674529194831848,
      "learning_rate": 1.7741665207804227e-06,
      "loss": 0.2806,
      "step": 44720
    },
    {
      "epoch": 0.8075495272968893,
      "grad_norm": 0.7431139945983887,
      "learning_rate": 1.7725540376407224e-06,
      "loss": 0.1866,
      "step": 44725
    },
    {
      "epoch": 0.8076398067297901,
      "grad_norm": 0.5370224714279175,
      "learning_rate": 1.7709422163392332e-06,
      "loss": 0.1992,
      "step": 44730
    },
    {
      "epoch": 0.8077300861626907,
      "grad_norm": 0.5872454047203064,
      "learning_rate": 1.7693310570056089e-06,
      "loss": 0.259,
      "step": 44735
    },
    {
      "epoch": 0.8078203655955914,
      "grad_norm": 0.42964139580726624,
      "learning_rate": 1.7677205597694602e-06,
      "loss": 0.1762,
      "step": 44740
    },
    {
      "epoch": 0.8079106450284922,
      "grad_norm": 0.3020368218421936,
      "learning_rate": 1.766110724760336e-06,
      "loss": 0.2001,
      "step": 44745
    },
    {
      "epoch": 0.8080009244613929,
      "grad_norm": 0.5135726928710938,
      "learning_rate": 1.764501552107738e-06,
      "loss": 0.2018,
      "step": 44750
    },
    {
      "epoch": 0.8080912038942936,
      "grad_norm": 0.4926067888736725,
      "learning_rate": 1.7628930419411128e-06,
      "loss": 0.2562,
      "step": 44755
    },
    {
      "epoch": 0.8081814833271943,
      "grad_norm": 0.37829169631004333,
      "learning_rate": 1.7612851943898546e-06,
      "loss": 0.1305,
      "step": 44760
    },
    {
      "epoch": 0.808271762760095,
      "grad_norm": 0.5007221698760986,
      "learning_rate": 1.759678009583301e-06,
      "loss": 0.2188,
      "step": 44765
    },
    {
      "epoch": 0.8083620421929958,
      "grad_norm": 0.34714531898498535,
      "learning_rate": 1.7580714876507366e-06,
      "loss": 0.2671,
      "step": 44770
    },
    {
      "epoch": 0.8084523216258964,
      "grad_norm": 0.3378044664859772,
      "learning_rate": 1.7564656287213988e-06,
      "loss": 0.2097,
      "step": 44775
    },
    {
      "epoch": 0.8085426010587972,
      "grad_norm": 0.5682040452957153,
      "learning_rate": 1.75486043292446e-06,
      "loss": 0.1759,
      "step": 44780
    },
    {
      "epoch": 0.8086328804916979,
      "grad_norm": 0.5743517279624939,
      "learning_rate": 1.753255900389056e-06,
      "loss": 0.237,
      "step": 44785
    },
    {
      "epoch": 0.8087231599245986,
      "grad_norm": 0.47140514850616455,
      "learning_rate": 1.751652031244253e-06,
      "loss": 0.2202,
      "step": 44790
    },
    {
      "epoch": 0.8088134393574994,
      "grad_norm": 0.628346860408783,
      "learning_rate": 1.7500488256190752e-06,
      "loss": 0.2611,
      "step": 44795
    },
    {
      "epoch": 0.8089037187904,
      "grad_norm": 0.2804016172885895,
      "learning_rate": 1.7484462836424842e-06,
      "loss": 0.2269,
      "step": 44800
    },
    {
      "epoch": 0.8089939982233008,
      "grad_norm": 0.4037613868713379,
      "learning_rate": 1.7468444054433976e-06,
      "loss": 0.1999,
      "step": 44805
    },
    {
      "epoch": 0.8090842776562015,
      "grad_norm": 0.7222532033920288,
      "learning_rate": 1.7452431911506695e-06,
      "loss": 0.3218,
      "step": 44810
    },
    {
      "epoch": 0.8091745570891022,
      "grad_norm": 0.23788173496723175,
      "learning_rate": 1.7436426408931084e-06,
      "loss": 0.1793,
      "step": 44815
    },
    {
      "epoch": 0.8092648365220029,
      "grad_norm": 0.47189345955848694,
      "learning_rate": 1.742042754799468e-06,
      "loss": 0.2618,
      "step": 44820
    },
    {
      "epoch": 0.8093551159549036,
      "grad_norm": 0.3974069654941559,
      "learning_rate": 1.7404435329984492e-06,
      "loss": 0.2938,
      "step": 44825
    },
    {
      "epoch": 0.8094453953878044,
      "grad_norm": 0.3743002116680145,
      "learning_rate": 1.7388449756186932e-06,
      "loss": 0.2527,
      "step": 44830
    },
    {
      "epoch": 0.8095356748207051,
      "grad_norm": 0.3871017396450043,
      "learning_rate": 1.7372470827887967e-06,
      "loss": 0.2407,
      "step": 44835
    },
    {
      "epoch": 0.8096259542536057,
      "grad_norm": 0.7488775849342346,
      "learning_rate": 1.7356498546372936e-06,
      "loss": 0.3174,
      "step": 44840
    },
    {
      "epoch": 0.8097162336865065,
      "grad_norm": 0.20142963528633118,
      "learning_rate": 1.7340532912926722e-06,
      "loss": 0.1753,
      "step": 44845
    },
    {
      "epoch": 0.8098065131194072,
      "grad_norm": 0.9276465177536011,
      "learning_rate": 1.7324573928833666e-06,
      "loss": 0.2644,
      "step": 44850
    },
    {
      "epoch": 0.809896792552308,
      "grad_norm": 0.43048661947250366,
      "learning_rate": 1.7308621595377517e-06,
      "loss": 0.2023,
      "step": 44855
    },
    {
      "epoch": 0.8099870719852086,
      "grad_norm": 0.4367915987968445,
      "learning_rate": 1.729267591384155e-06,
      "loss": 0.1995,
      "step": 44860
    },
    {
      "epoch": 0.8100773514181093,
      "grad_norm": 0.45970258116722107,
      "learning_rate": 1.7276736885508438e-06,
      "loss": 0.1826,
      "step": 44865
    },
    {
      "epoch": 0.8101676308510101,
      "grad_norm": 0.4545137286186218,
      "learning_rate": 1.7260804511660411e-06,
      "loss": 0.275,
      "step": 44870
    },
    {
      "epoch": 0.8102579102839108,
      "grad_norm": 0.6485731601715088,
      "learning_rate": 1.7244878793579067e-06,
      "loss": 0.1917,
      "step": 44875
    },
    {
      "epoch": 0.8103481897168114,
      "grad_norm": 0.3482241630554199,
      "learning_rate": 1.7228959732545547e-06,
      "loss": 0.1651,
      "step": 44880
    },
    {
      "epoch": 0.8104384691497122,
      "grad_norm": 0.1493377387523651,
      "learning_rate": 1.7213047329840405e-06,
      "loss": 0.191,
      "step": 44885
    },
    {
      "epoch": 0.8105287485826129,
      "grad_norm": 0.5001938939094543,
      "learning_rate": 1.7197141586743705e-06,
      "loss": 0.1329,
      "step": 44890
    },
    {
      "epoch": 0.8106190280155137,
      "grad_norm": 0.6227378249168396,
      "learning_rate": 1.7181242504534913e-06,
      "loss": 0.2342,
      "step": 44895
    },
    {
      "epoch": 0.8107093074484143,
      "grad_norm": 0.5441110730171204,
      "learning_rate": 1.7165350084493026e-06,
      "loss": 0.278,
      "step": 44900
    },
    {
      "epoch": 0.810799586881315,
      "grad_norm": 0.4558870196342468,
      "learning_rate": 1.7149464327896437e-06,
      "loss": 0.1451,
      "step": 44905
    },
    {
      "epoch": 0.8108898663142158,
      "grad_norm": 0.37264585494995117,
      "learning_rate": 1.7133585236023065e-06,
      "loss": 0.3025,
      "step": 44910
    },
    {
      "epoch": 0.8109801457471165,
      "grad_norm": 0.41475045680999756,
      "learning_rate": 1.7117712810150268e-06,
      "loss": 0.1962,
      "step": 44915
    },
    {
      "epoch": 0.8110704251800172,
      "grad_norm": 0.7094918489456177,
      "learning_rate": 1.7101847051554876e-06,
      "loss": 0.2787,
      "step": 44920
    },
    {
      "epoch": 0.8111607046129179,
      "grad_norm": 0.6133314967155457,
      "learning_rate": 1.7085987961513151e-06,
      "loss": 0.2263,
      "step": 44925
    },
    {
      "epoch": 0.8112509840458186,
      "grad_norm": 0.5395045280456543,
      "learning_rate": 1.7070135541300858e-06,
      "loss": 0.2635,
      "step": 44930
    },
    {
      "epoch": 0.8113412634787194,
      "grad_norm": 0.7307504415512085,
      "learning_rate": 1.705428979219319e-06,
      "loss": 0.2148,
      "step": 44935
    },
    {
      "epoch": 0.81143154291162,
      "grad_norm": 0.448566198348999,
      "learning_rate": 1.703845071546486e-06,
      "loss": 0.2212,
      "step": 44940
    },
    {
      "epoch": 0.8115218223445207,
      "grad_norm": 0.400714635848999,
      "learning_rate": 1.702261831238995e-06,
      "loss": 0.3076,
      "step": 44945
    },
    {
      "epoch": 0.8116121017774215,
      "grad_norm": 0.368564248085022,
      "learning_rate": 1.7006792584242094e-06,
      "loss": 0.1411,
      "step": 44950
    },
    {
      "epoch": 0.8117023812103222,
      "grad_norm": 0.3286515772342682,
      "learning_rate": 1.6990973532294386e-06,
      "loss": 0.2919,
      "step": 44955
    },
    {
      "epoch": 0.8117926606432229,
      "grad_norm": 0.6517858505249023,
      "learning_rate": 1.6975161157819297e-06,
      "loss": 0.2365,
      "step": 44960
    },
    {
      "epoch": 0.8118829400761236,
      "grad_norm": 0.4448356330394745,
      "learning_rate": 1.6959355462088867e-06,
      "loss": 0.2446,
      "step": 44965
    },
    {
      "epoch": 0.8119732195090243,
      "grad_norm": 0.35378003120422363,
      "learning_rate": 1.6943556446374497e-06,
      "loss": 0.2291,
      "step": 44970
    },
    {
      "epoch": 0.8120634989419251,
      "grad_norm": 0.8098388910293579,
      "learning_rate": 1.6927764111947142e-06,
      "loss": 0.2064,
      "step": 44975
    },
    {
      "epoch": 0.8121537783748257,
      "grad_norm": 0.6267945766448975,
      "learning_rate": 1.691197846007716e-06,
      "loss": 0.1499,
      "step": 44980
    },
    {
      "epoch": 0.8122440578077265,
      "grad_norm": 0.6006688475608826,
      "learning_rate": 1.6896199492034437e-06,
      "loss": 0.2227,
      "step": 44985
    },
    {
      "epoch": 0.8123343372406272,
      "grad_norm": 0.34231728315353394,
      "learning_rate": 1.6880427209088212e-06,
      "loss": 0.2521,
      "step": 44990
    },
    {
      "epoch": 0.8124246166735279,
      "grad_norm": 0.6057373285293579,
      "learning_rate": 1.6864661612507305e-06,
      "loss": 0.1788,
      "step": 44995
    },
    {
      "epoch": 0.8125148961064286,
      "grad_norm": 0.7820007801055908,
      "learning_rate": 1.684890270355989e-06,
      "loss": 0.1882,
      "step": 45000
    },
    {
      "epoch": 0.8126051755393293,
      "grad_norm": 0.6362524628639221,
      "learning_rate": 1.6833150483513717e-06,
      "loss": 0.2114,
      "step": 45005
    },
    {
      "epoch": 0.8126954549722301,
      "grad_norm": 0.4829091429710388,
      "learning_rate": 1.6817404953635874e-06,
      "loss": 0.2289,
      "step": 45010
    },
    {
      "epoch": 0.8127857344051308,
      "grad_norm": 0.486634224653244,
      "learning_rate": 1.6801666115193016e-06,
      "loss": 0.2444,
      "step": 45015
    },
    {
      "epoch": 0.8128760138380314,
      "grad_norm": 0.4345434308052063,
      "learning_rate": 1.6785933969451207e-06,
      "loss": 0.1903,
      "step": 45020
    },
    {
      "epoch": 0.8129662932709322,
      "grad_norm": 0.3699716031551361,
      "learning_rate": 1.6770208517676e-06,
      "loss": 0.2499,
      "step": 45025
    },
    {
      "epoch": 0.8130565727038329,
      "grad_norm": 0.15359029173851013,
      "learning_rate": 1.6754489761132365e-06,
      "loss": 0.1945,
      "step": 45030
    },
    {
      "epoch": 0.8131468521367337,
      "grad_norm": 0.3875373601913452,
      "learning_rate": 1.6738777701084795e-06,
      "loss": 0.2114,
      "step": 45035
    },
    {
      "epoch": 0.8132371315696343,
      "grad_norm": 0.42671942710876465,
      "learning_rate": 1.6723072338797197e-06,
      "loss": 0.2397,
      "step": 45040
    },
    {
      "epoch": 0.813327411002535,
      "grad_norm": 0.8030031323432922,
      "learning_rate": 1.670737367553289e-06,
      "loss": 0.2518,
      "step": 45045
    },
    {
      "epoch": 0.8134176904354358,
      "grad_norm": 0.8285560011863708,
      "learning_rate": 1.669168171255483e-06,
      "loss": 0.2263,
      "step": 45050
    },
    {
      "epoch": 0.8135079698683365,
      "grad_norm": 0.880510151386261,
      "learning_rate": 1.6675996451125243e-06,
      "loss": 0.2712,
      "step": 45055
    },
    {
      "epoch": 0.8135982493012371,
      "grad_norm": 0.4474562108516693,
      "learning_rate": 1.6660317892505939e-06,
      "loss": 0.2186,
      "step": 45060
    },
    {
      "epoch": 0.8136885287341379,
      "grad_norm": 0.5012331008911133,
      "learning_rate": 1.66446460379581e-06,
      "loss": 0.211,
      "step": 45065
    },
    {
      "epoch": 0.8137788081670386,
      "grad_norm": 0.5450423359870911,
      "learning_rate": 1.6628980888742452e-06,
      "loss": 0.2358,
      "step": 45070
    },
    {
      "epoch": 0.8138690875999394,
      "grad_norm": 0.704733669757843,
      "learning_rate": 1.6613322446119107e-06,
      "loss": 0.1618,
      "step": 45075
    },
    {
      "epoch": 0.81395936703284,
      "grad_norm": 0.22256499528884888,
      "learning_rate": 1.6597670711347691e-06,
      "loss": 0.2362,
      "step": 45080
    },
    {
      "epoch": 0.8140496464657407,
      "grad_norm": 0.3464551270008087,
      "learning_rate": 1.6582025685687274e-06,
      "loss": 0.2208,
      "step": 45085
    },
    {
      "epoch": 0.8141399258986415,
      "grad_norm": 0.35018548369407654,
      "learning_rate": 1.6566387370396409e-06,
      "loss": 0.1883,
      "step": 45090
    },
    {
      "epoch": 0.8142302053315422,
      "grad_norm": 0.4291853904724121,
      "learning_rate": 1.655075576673304e-06,
      "loss": 0.2775,
      "step": 45095
    },
    {
      "epoch": 0.8143204847644429,
      "grad_norm": 0.30712375044822693,
      "learning_rate": 1.6535130875954642e-06,
      "loss": 0.1741,
      "step": 45100
    },
    {
      "epoch": 0.8144107641973436,
      "grad_norm": 0.559040904045105,
      "learning_rate": 1.6519512699318107e-06,
      "loss": 0.2643,
      "step": 45105
    },
    {
      "epoch": 0.8145010436302443,
      "grad_norm": 0.48954933881759644,
      "learning_rate": 1.650390123807981e-06,
      "loss": 0.2021,
      "step": 45110
    },
    {
      "epoch": 0.8145913230631451,
      "grad_norm": 0.8701210618019104,
      "learning_rate": 1.648829649349558e-06,
      "loss": 0.131,
      "step": 45115
    },
    {
      "epoch": 0.8146816024960457,
      "grad_norm": 0.3505600094795227,
      "learning_rate": 1.6472698466820735e-06,
      "loss": 0.201,
      "step": 45120
    },
    {
      "epoch": 0.8147718819289465,
      "grad_norm": 0.2789531350135803,
      "learning_rate": 1.6457107159309993e-06,
      "loss": 0.2203,
      "step": 45125
    },
    {
      "epoch": 0.8148621613618472,
      "grad_norm": 1.8841131925582886,
      "learning_rate": 1.644152257221755e-06,
      "loss": 0.1857,
      "step": 45130
    },
    {
      "epoch": 0.8149524407947479,
      "grad_norm": 0.5127928256988525,
      "learning_rate": 1.6425944706797103e-06,
      "loss": 0.1711,
      "step": 45135
    },
    {
      "epoch": 0.8150427202276486,
      "grad_norm": 0.6695572733879089,
      "learning_rate": 1.641037356430174e-06,
      "loss": 0.1877,
      "step": 45140
    },
    {
      "epoch": 0.8151329996605493,
      "grad_norm": 0.6078423857688904,
      "learning_rate": 1.639480914598407e-06,
      "loss": 0.2369,
      "step": 45145
    },
    {
      "epoch": 0.81522327909345,
      "grad_norm": 0.4141404330730438,
      "learning_rate": 1.6379251453096124e-06,
      "loss": 0.3065,
      "step": 45150
    },
    {
      "epoch": 0.8153135585263508,
      "grad_norm": 0.4831113815307617,
      "learning_rate": 1.6363700486889456e-06,
      "loss": 0.1821,
      "step": 45155
    },
    {
      "epoch": 0.8154038379592514,
      "grad_norm": 0.5786094069480896,
      "learning_rate": 1.6348156248614955e-06,
      "loss": 0.23,
      "step": 45160
    },
    {
      "epoch": 0.8154941173921522,
      "grad_norm": 0.45996254682540894,
      "learning_rate": 1.6332618739523109e-06,
      "loss": 0.1924,
      "step": 45165
    },
    {
      "epoch": 0.8155843968250529,
      "grad_norm": 0.2889226973056793,
      "learning_rate": 1.6317087960863731e-06,
      "loss": 0.1009,
      "step": 45170
    },
    {
      "epoch": 0.8156746762579536,
      "grad_norm": 0.211906298995018,
      "learning_rate": 1.630156391388621e-06,
      "loss": 0.2564,
      "step": 45175
    },
    {
      "epoch": 0.8157649556908544,
      "grad_norm": 0.38543757796287537,
      "learning_rate": 1.628604659983931e-06,
      "loss": 0.2133,
      "step": 45180
    },
    {
      "epoch": 0.815855235123755,
      "grad_norm": 0.36902737617492676,
      "learning_rate": 1.6270536019971329e-06,
      "loss": 0.1993,
      "step": 45185
    },
    {
      "epoch": 0.8159455145566558,
      "grad_norm": 0.6300240755081177,
      "learning_rate": 1.6255032175529927e-06,
      "loss": 0.1647,
      "step": 45190
    },
    {
      "epoch": 0.8160357939895565,
      "grad_norm": 0.6084157228469849,
      "learning_rate": 1.6239535067762336e-06,
      "loss": 0.1406,
      "step": 45195
    },
    {
      "epoch": 0.8161260734224572,
      "grad_norm": 0.7221276760101318,
      "learning_rate": 1.622404469791512e-06,
      "loss": 0.1899,
      "step": 45200
    },
    {
      "epoch": 0.8162163528553579,
      "grad_norm": 0.44410908222198486,
      "learning_rate": 1.6208561067234418e-06,
      "loss": 0.2237,
      "step": 45205
    },
    {
      "epoch": 0.8163066322882586,
      "grad_norm": 0.6417168974876404,
      "learning_rate": 1.6193084176965735e-06,
      "loss": 0.2775,
      "step": 45210
    },
    {
      "epoch": 0.8163969117211594,
      "grad_norm": 0.5598950386047363,
      "learning_rate": 1.6177614028354095e-06,
      "loss": 0.2089,
      "step": 45215
    },
    {
      "epoch": 0.8164871911540601,
      "grad_norm": 0.5878446102142334,
      "learning_rate": 1.616215062264399e-06,
      "loss": 0.1731,
      "step": 45220
    },
    {
      "epoch": 0.8165774705869607,
      "grad_norm": 0.490138441324234,
      "learning_rate": 1.6146693961079274e-06,
      "loss": 0.2354,
      "step": 45225
    },
    {
      "epoch": 0.8166677500198615,
      "grad_norm": 0.36699575185775757,
      "learning_rate": 1.6131244044903383e-06,
      "loss": 0.2205,
      "step": 45230
    },
    {
      "epoch": 0.8167580294527622,
      "grad_norm": 0.4489833116531372,
      "learning_rate": 1.61158008753591e-06,
      "loss": 0.1959,
      "step": 45235
    },
    {
      "epoch": 0.816848308885663,
      "grad_norm": 0.47021374106407166,
      "learning_rate": 1.6100364453688733e-06,
      "loss": 0.2349,
      "step": 45240
    },
    {
      "epoch": 0.8169385883185636,
      "grad_norm": 0.3278280794620514,
      "learning_rate": 1.6084934781134044e-06,
      "loss": 0.2608,
      "step": 45245
    },
    {
      "epoch": 0.8170288677514643,
      "grad_norm": 0.3986305892467499,
      "learning_rate": 1.606951185893626e-06,
      "loss": 0.234,
      "step": 45250
    },
    {
      "epoch": 0.8171191471843651,
      "grad_norm": 0.364401251077652,
      "learning_rate": 1.605409568833598e-06,
      "loss": 0.2502,
      "step": 45255
    },
    {
      "epoch": 0.8172094266172658,
      "grad_norm": 0.5744989514350891,
      "learning_rate": 1.6038686270573378e-06,
      "loss": 0.1782,
      "step": 45260
    },
    {
      "epoch": 0.8172997060501664,
      "grad_norm": 0.25071555376052856,
      "learning_rate": 1.6023283606887996e-06,
      "loss": 0.2248,
      "step": 45265
    },
    {
      "epoch": 0.8173899854830672,
      "grad_norm": 0.21699373424053192,
      "learning_rate": 1.6007887698518888e-06,
      "loss": 0.1588,
      "step": 45270
    },
    {
      "epoch": 0.8174802649159679,
      "grad_norm": 0.3030523955821991,
      "learning_rate": 1.5992498546704516e-06,
      "loss": 0.2139,
      "step": 45275
    },
    {
      "epoch": 0.8175705443488687,
      "grad_norm": 0.3498544692993164,
      "learning_rate": 1.5977116152682848e-06,
      "loss": 0.2147,
      "step": 45280
    },
    {
      "epoch": 0.8176608237817693,
      "grad_norm": 0.4606175422668457,
      "learning_rate": 1.5961740517691272e-06,
      "loss": 0.1715,
      "step": 45285
    },
    {
      "epoch": 0.81775110321467,
      "grad_norm": 0.3159838914871216,
      "learning_rate": 1.5946371642966674e-06,
      "loss": 0.2202,
      "step": 45290
    },
    {
      "epoch": 0.8178413826475708,
      "grad_norm": 0.3965045213699341,
      "learning_rate": 1.5931009529745333e-06,
      "loss": 0.1971,
      "step": 45295
    },
    {
      "epoch": 0.8179316620804715,
      "grad_norm": 0.6597896218299866,
      "learning_rate": 1.5915654179263063e-06,
      "loss": 0.1405,
      "step": 45300
    },
    {
      "epoch": 0.8180219415133722,
      "grad_norm": 0.904430627822876,
      "learning_rate": 1.5900305592755028e-06,
      "loss": 0.2466,
      "step": 45305
    },
    {
      "epoch": 0.8181122209462729,
      "grad_norm": 0.4741705358028412,
      "learning_rate": 1.5884963771455942e-06,
      "loss": 0.2244,
      "step": 45310
    },
    {
      "epoch": 0.8182025003791736,
      "grad_norm": 0.44903695583343506,
      "learning_rate": 1.5869628716599982e-06,
      "loss": 0.2194,
      "step": 45315
    },
    {
      "epoch": 0.8182927798120744,
      "grad_norm": 0.4540101885795593,
      "learning_rate": 1.5854300429420667e-06,
      "loss": 0.3024,
      "step": 45320
    },
    {
      "epoch": 0.818383059244975,
      "grad_norm": 0.2935616374015808,
      "learning_rate": 1.5838978911151114e-06,
      "loss": 0.1556,
      "step": 45325
    },
    {
      "epoch": 0.8184733386778758,
      "grad_norm": 0.3232356011867523,
      "learning_rate": 1.5823664163023777e-06,
      "loss": 0.2741,
      "step": 45330
    },
    {
      "epoch": 0.8185636181107765,
      "grad_norm": 0.24531584978103638,
      "learning_rate": 1.5808356186270657e-06,
      "loss": 0.2053,
      "step": 45335
    },
    {
      "epoch": 0.8186538975436772,
      "grad_norm": 0.725897490978241,
      "learning_rate": 1.5793054982123123e-06,
      "loss": 0.2228,
      "step": 45340
    },
    {
      "epoch": 0.8187441769765779,
      "grad_norm": 0.49496546387672424,
      "learning_rate": 1.5777760551812083e-06,
      "loss": 0.1896,
      "step": 45345
    },
    {
      "epoch": 0.8188344564094786,
      "grad_norm": 0.5691260695457458,
      "learning_rate": 1.5762472896567838e-06,
      "loss": 0.1977,
      "step": 45350
    },
    {
      "epoch": 0.8189247358423793,
      "grad_norm": 0.3409838378429413,
      "learning_rate": 1.5747192017620216e-06,
      "loss": 0.2193,
      "step": 45355
    },
    {
      "epoch": 0.8190150152752801,
      "grad_norm": 0.6356446146965027,
      "learning_rate": 1.5731917916198392e-06,
      "loss": 0.3169,
      "step": 45360
    },
    {
      "epoch": 0.8191052947081807,
      "grad_norm": 0.30235686898231506,
      "learning_rate": 1.5716650593531112e-06,
      "loss": 0.2241,
      "step": 45365
    },
    {
      "epoch": 0.8191955741410815,
      "grad_norm": 0.4847787022590637,
      "learning_rate": 1.5701390050846465e-06,
      "loss": 0.292,
      "step": 45370
    },
    {
      "epoch": 0.8192858535739822,
      "grad_norm": 0.18954876065254211,
      "learning_rate": 1.5686136289372079e-06,
      "loss": 0.2705,
      "step": 45375
    },
    {
      "epoch": 0.819376133006883,
      "grad_norm": 0.5880841016769409,
      "learning_rate": 1.5670889310335003e-06,
      "loss": 0.1502,
      "step": 45380
    },
    {
      "epoch": 0.8194664124397836,
      "grad_norm": 0.3620661199092865,
      "learning_rate": 1.5655649114961768e-06,
      "loss": 0.2292,
      "step": 45385
    },
    {
      "epoch": 0.8195566918726843,
      "grad_norm": 0.7227039337158203,
      "learning_rate": 1.5640415704478308e-06,
      "loss": 0.252,
      "step": 45390
    },
    {
      "epoch": 0.8196469713055851,
      "grad_norm": 0.5891028046607971,
      "learning_rate": 1.562518908011007e-06,
      "loss": 0.1781,
      "step": 45395
    },
    {
      "epoch": 0.8197372507384858,
      "grad_norm": 0.7310418486595154,
      "learning_rate": 1.5609969243081892e-06,
      "loss": 0.3571,
      "step": 45400
    },
    {
      "epoch": 0.8198275301713864,
      "grad_norm": 0.7425979375839233,
      "learning_rate": 1.5594756194618121e-06,
      "loss": 0.2302,
      "step": 45405
    },
    {
      "epoch": 0.8199178096042872,
      "grad_norm": 0.49799680709838867,
      "learning_rate": 1.5579549935942517e-06,
      "loss": 0.1718,
      "step": 45410
    },
    {
      "epoch": 0.8200080890371879,
      "grad_norm": 0.41933315992355347,
      "learning_rate": 1.556435046827832e-06,
      "loss": 0.2294,
      "step": 45415
    },
    {
      "epoch": 0.8200983684700887,
      "grad_norm": 0.33986660838127136,
      "learning_rate": 1.5549157792848247e-06,
      "loss": 0.2222,
      "step": 45420
    },
    {
      "epoch": 0.8201886479029893,
      "grad_norm": 0.4626463055610657,
      "learning_rate": 1.5533971910874378e-06,
      "loss": 0.1882,
      "step": 45425
    },
    {
      "epoch": 0.82027892733589,
      "grad_norm": 0.4492209851741791,
      "learning_rate": 1.5518792823578377e-06,
      "loss": 0.2006,
      "step": 45430
    },
    {
      "epoch": 0.8203692067687908,
      "grad_norm": 0.33428868651390076,
      "learning_rate": 1.550362053218123e-06,
      "loss": 0.1741,
      "step": 45435
    },
    {
      "epoch": 0.8204594862016915,
      "grad_norm": 0.29296696186065674,
      "learning_rate": 1.548845503790346e-06,
      "loss": 0.1906,
      "step": 45440
    },
    {
      "epoch": 0.8205497656345921,
      "grad_norm": 0.5941862463951111,
      "learning_rate": 1.5473296341965027e-06,
      "loss": 0.2781,
      "step": 45445
    },
    {
      "epoch": 0.8206400450674929,
      "grad_norm": 0.4935421943664551,
      "learning_rate": 1.5458144445585367e-06,
      "loss": 0.2923,
      "step": 45450
    },
    {
      "epoch": 0.8207303245003936,
      "grad_norm": 0.5380502343177795,
      "learning_rate": 1.5442999349983267e-06,
      "loss": 0.3464,
      "step": 45455
    },
    {
      "epoch": 0.8208206039332944,
      "grad_norm": 0.4661569595336914,
      "learning_rate": 1.5427861056377114e-06,
      "loss": 0.1701,
      "step": 45460
    },
    {
      "epoch": 0.820910883366195,
      "grad_norm": 0.662245512008667,
      "learning_rate": 1.541272956598462e-06,
      "loss": 0.2531,
      "step": 45465
    },
    {
      "epoch": 0.8210011627990957,
      "grad_norm": 0.5433655977249146,
      "learning_rate": 1.5397604880023032e-06,
      "loss": 0.1601,
      "step": 45470
    },
    {
      "epoch": 0.8210914422319965,
      "grad_norm": 2.0113091468811035,
      "learning_rate": 1.5382486999709002e-06,
      "loss": 0.1519,
      "step": 45475
    },
    {
      "epoch": 0.8211817216648972,
      "grad_norm": 0.4511041045188904,
      "learning_rate": 1.5367375926258665e-06,
      "loss": 0.2001,
      "step": 45480
    },
    {
      "epoch": 0.8212720010977979,
      "grad_norm": 0.29495155811309814,
      "learning_rate": 1.5352271660887585e-06,
      "loss": 0.1836,
      "step": 45485
    },
    {
      "epoch": 0.8213622805306986,
      "grad_norm": 0.554208517074585,
      "learning_rate": 1.5337174204810835e-06,
      "loss": 0.2228,
      "step": 45490
    },
    {
      "epoch": 0.8214525599635993,
      "grad_norm": 0.5604204535484314,
      "learning_rate": 1.5322083559242862e-06,
      "loss": 0.191,
      "step": 45495
    },
    {
      "epoch": 0.8215428393965001,
      "grad_norm": 0.24129433929920197,
      "learning_rate": 1.5306999725397564e-06,
      "loss": 0.1646,
      "step": 45500
    },
    {
      "epoch": 0.8216331188294007,
      "grad_norm": 0.45954737067222595,
      "learning_rate": 1.529192270448837e-06,
      "loss": 0.1976,
      "step": 45505
    },
    {
      "epoch": 0.8217233982623015,
      "grad_norm": 0.27473676204681396,
      "learning_rate": 1.5276852497728112e-06,
      "loss": 0.2805,
      "step": 45510
    },
    {
      "epoch": 0.8218136776952022,
      "grad_norm": 0.3894435465335846,
      "learning_rate": 1.5261789106329084e-06,
      "loss": 0.253,
      "step": 45515
    },
    {
      "epoch": 0.8219039571281029,
      "grad_norm": 0.4440051019191742,
      "learning_rate": 1.5246732531503005e-06,
      "loss": 0.2896,
      "step": 45520
    },
    {
      "epoch": 0.8219942365610036,
      "grad_norm": 0.36047643423080444,
      "learning_rate": 1.52316827744611e-06,
      "loss": 0.2452,
      "step": 45525
    },
    {
      "epoch": 0.8220845159939043,
      "grad_norm": 0.6247464418411255,
      "learning_rate": 1.521663983641396e-06,
      "loss": 0.3099,
      "step": 45530
    },
    {
      "epoch": 0.822174795426805,
      "grad_norm": 0.866725742816925,
      "learning_rate": 1.5201603718571734e-06,
      "loss": 0.2416,
      "step": 45535
    },
    {
      "epoch": 0.8222650748597058,
      "grad_norm": 0.44059160351753235,
      "learning_rate": 1.5186574422143918e-06,
      "loss": 0.2011,
      "step": 45540
    },
    {
      "epoch": 0.8223553542926064,
      "grad_norm": 0.40573906898498535,
      "learning_rate": 1.5171551948339547e-06,
      "loss": 0.2119,
      "step": 45545
    },
    {
      "epoch": 0.8224456337255072,
      "grad_norm": 0.3320337235927582,
      "learning_rate": 1.5156536298367053e-06,
      "loss": 0.2905,
      "step": 45550
    },
    {
      "epoch": 0.8225359131584079,
      "grad_norm": 0.4631249010562897,
      "learning_rate": 1.514152747343437e-06,
      "loss": 0.2341,
      "step": 45555
    },
    {
      "epoch": 0.8226261925913086,
      "grad_norm": 0.6496574282646179,
      "learning_rate": 1.51265254747488e-06,
      "loss": 0.2163,
      "step": 45560
    },
    {
      "epoch": 0.8227164720242093,
      "grad_norm": 0.38601595163345337,
      "learning_rate": 1.5111530303517185e-06,
      "loss": 0.2474,
      "step": 45565
    },
    {
      "epoch": 0.82280675145711,
      "grad_norm": 0.7684838175773621,
      "learning_rate": 1.5096541960945731e-06,
      "loss": 0.2907,
      "step": 45570
    },
    {
      "epoch": 0.8228970308900108,
      "grad_norm": 0.29702773690223694,
      "learning_rate": 1.5081560448240174e-06,
      "loss": 0.2478,
      "step": 45575
    },
    {
      "epoch": 0.8229873103229115,
      "grad_norm": 0.36197206377983093,
      "learning_rate": 1.5066585766605656e-06,
      "loss": 0.1844,
      "step": 45580
    },
    {
      "epoch": 0.8230775897558122,
      "grad_norm": 0.37863409519195557,
      "learning_rate": 1.505161791724682e-06,
      "loss": 0.2684,
      "step": 45585
    },
    {
      "epoch": 0.8231678691887129,
      "grad_norm": 0.5644698143005371,
      "learning_rate": 1.5036656901367685e-06,
      "loss": 0.2234,
      "step": 45590
    },
    {
      "epoch": 0.8232581486216136,
      "grad_norm": 0.4333361089229584,
      "learning_rate": 1.5021702720171727e-06,
      "loss": 0.1902,
      "step": 45595
    },
    {
      "epoch": 0.8233484280545144,
      "grad_norm": 0.6529067158699036,
      "learning_rate": 1.5006755374861969e-06,
      "loss": 0.2258,
      "step": 45600
    },
    {
      "epoch": 0.8234387074874151,
      "grad_norm": 0.46586236357688904,
      "learning_rate": 1.4991814866640753e-06,
      "loss": 0.2391,
      "step": 45605
    },
    {
      "epoch": 0.8235289869203157,
      "grad_norm": 0.32549789547920227,
      "learning_rate": 1.4976881196709959e-06,
      "loss": 0.2115,
      "step": 45610
    },
    {
      "epoch": 0.8236192663532165,
      "grad_norm": 0.3597162961959839,
      "learning_rate": 1.4961954366270891e-06,
      "loss": 0.2647,
      "step": 45615
    },
    {
      "epoch": 0.8237095457861172,
      "grad_norm": 0.48388177156448364,
      "learning_rate": 1.4947034376524338e-06,
      "loss": 0.2034,
      "step": 45620
    },
    {
      "epoch": 0.823799825219018,
      "grad_norm": 0.4890066683292389,
      "learning_rate": 1.4932121228670448e-06,
      "loss": 0.2268,
      "step": 45625
    },
    {
      "epoch": 0.8238901046519186,
      "grad_norm": 0.5106359124183655,
      "learning_rate": 1.4917214923908929e-06,
      "loss": 0.1494,
      "step": 45630
    },
    {
      "epoch": 0.8239803840848193,
      "grad_norm": 0.5168476104736328,
      "learning_rate": 1.4902315463438833e-06,
      "loss": 0.2053,
      "step": 45635
    },
    {
      "epoch": 0.8240706635177201,
      "grad_norm": 0.36747848987579346,
      "learning_rate": 1.4887422848458733e-06,
      "loss": 0.2252,
      "step": 45640
    },
    {
      "epoch": 0.8241609429506208,
      "grad_norm": 0.6707513332366943,
      "learning_rate": 1.4872537080166649e-06,
      "loss": 0.2243,
      "step": 45645
    },
    {
      "epoch": 0.8242512223835214,
      "grad_norm": 0.7406870722770691,
      "learning_rate": 1.4857658159760035e-06,
      "loss": 0.2212,
      "step": 45650
    },
    {
      "epoch": 0.8243415018164222,
      "grad_norm": 0.1467631310224533,
      "learning_rate": 1.4842786088435745e-06,
      "loss": 0.1931,
      "step": 45655
    },
    {
      "epoch": 0.8244317812493229,
      "grad_norm": 0.31390658020973206,
      "learning_rate": 1.48279208673902e-06,
      "loss": 0.2757,
      "step": 45660
    },
    {
      "epoch": 0.8245220606822237,
      "grad_norm": 0.5410983562469482,
      "learning_rate": 1.481306249781913e-06,
      "loss": 0.2389,
      "step": 45665
    },
    {
      "epoch": 0.8246123401151243,
      "grad_norm": 0.48541054129600525,
      "learning_rate": 1.4798210980917837e-06,
      "loss": 0.2438,
      "step": 45670
    },
    {
      "epoch": 0.824702619548025,
      "grad_norm": 0.46386122703552246,
      "learning_rate": 1.4783366317880976e-06,
      "loss": 0.2263,
      "step": 45675
    },
    {
      "epoch": 0.8247928989809258,
      "grad_norm": 0.35350045561790466,
      "learning_rate": 1.4768528509902703e-06,
      "loss": 0.2448,
      "step": 45680
    },
    {
      "epoch": 0.8248831784138265,
      "grad_norm": 0.5350896120071411,
      "learning_rate": 1.4753697558176637e-06,
      "loss": 0.1816,
      "step": 45685
    },
    {
      "epoch": 0.8249734578467272,
      "grad_norm": 0.5602902770042419,
      "learning_rate": 1.4738873463895797e-06,
      "loss": 0.2954,
      "step": 45690
    },
    {
      "epoch": 0.8250637372796279,
      "grad_norm": 0.6952036619186401,
      "learning_rate": 1.4724056228252682e-06,
      "loss": 0.272,
      "step": 45695
    },
    {
      "epoch": 0.8251540167125286,
      "grad_norm": 0.55024653673172,
      "learning_rate": 1.4709245852439212e-06,
      "loss": 0.146,
      "step": 45700
    },
    {
      "epoch": 0.8252442961454294,
      "grad_norm": 0.26396316289901733,
      "learning_rate": 1.4694442337646797e-06,
      "loss": 0.1623,
      "step": 45705
    },
    {
      "epoch": 0.82533457557833,
      "grad_norm": 0.37291499972343445,
      "learning_rate": 1.4679645685066256e-06,
      "loss": 0.1808,
      "step": 45710
    },
    {
      "epoch": 0.8254248550112308,
      "grad_norm": 0.47473058104515076,
      "learning_rate": 1.4664855895887908e-06,
      "loss": 0.2854,
      "step": 45715
    },
    {
      "epoch": 0.8255151344441315,
      "grad_norm": 0.17294996976852417,
      "learning_rate": 1.4650072971301433e-06,
      "loss": 0.2381,
      "step": 45720
    },
    {
      "epoch": 0.8256054138770322,
      "grad_norm": 0.6652577519416809,
      "learning_rate": 1.4635296912496054e-06,
      "loss": 0.2251,
      "step": 45725
    },
    {
      "epoch": 0.8256956933099329,
      "grad_norm": 0.42790499329566956,
      "learning_rate": 1.4620527720660348e-06,
      "loss": 0.2164,
      "step": 45730
    },
    {
      "epoch": 0.8257859727428336,
      "grad_norm": 0.7221193313598633,
      "learning_rate": 1.4605765396982452e-06,
      "loss": 0.1605,
      "step": 45735
    },
    {
      "epoch": 0.8258762521757343,
      "grad_norm": 0.4172767698764801,
      "learning_rate": 1.4591009942649825e-06,
      "loss": 0.2616,
      "step": 45740
    },
    {
      "epoch": 0.8259665316086351,
      "grad_norm": 0.5771311521530151,
      "learning_rate": 1.4576261358849464e-06,
      "loss": 0.1407,
      "step": 45745
    },
    {
      "epoch": 0.8260568110415357,
      "grad_norm": 0.41090327501296997,
      "learning_rate": 1.4561519646767797e-06,
      "loss": 0.2271,
      "step": 45750
    },
    {
      "epoch": 0.8261470904744365,
      "grad_norm": 0.8600770831108093,
      "learning_rate": 1.4546784807590687e-06,
      "loss": 0.143,
      "step": 45755
    },
    {
      "epoch": 0.8262373699073372,
      "grad_norm": 0.4484533667564392,
      "learning_rate": 1.4532056842503416e-06,
      "loss": 0.2129,
      "step": 45760
    },
    {
      "epoch": 0.826327649340238,
      "grad_norm": 0.418782114982605,
      "learning_rate": 1.4517335752690786e-06,
      "loss": 0.1961,
      "step": 45765
    },
    {
      "epoch": 0.8264179287731386,
      "grad_norm": 0.3176212012767792,
      "learning_rate": 1.4502621539336947e-06,
      "loss": 0.2542,
      "step": 45770
    },
    {
      "epoch": 0.8265082082060393,
      "grad_norm": 0.5707917809486389,
      "learning_rate": 1.4487914203625574e-06,
      "loss": 0.2532,
      "step": 45775
    },
    {
      "epoch": 0.8265984876389401,
      "grad_norm": 0.8464087843894958,
      "learning_rate": 1.447321374673981e-06,
      "loss": 0.3759,
      "step": 45780
    },
    {
      "epoch": 0.8266887670718408,
      "grad_norm": 0.38225147128105164,
      "learning_rate": 1.4458520169862123e-06,
      "loss": 0.1762,
      "step": 45785
    },
    {
      "epoch": 0.8267790465047414,
      "grad_norm": 0.49505189061164856,
      "learning_rate": 1.4443833474174574e-06,
      "loss": 0.2672,
      "step": 45790
    },
    {
      "epoch": 0.8268693259376422,
      "grad_norm": 0.20402777194976807,
      "learning_rate": 1.4429153660858542e-06,
      "loss": 0.1732,
      "step": 45795
    },
    {
      "epoch": 0.8269596053705429,
      "grad_norm": 0.6084776520729065,
      "learning_rate": 1.441448073109496e-06,
      "loss": 0.2127,
      "step": 45800
    },
    {
      "epoch": 0.8270498848034437,
      "grad_norm": 1.3320783376693726,
      "learning_rate": 1.4399814686064107e-06,
      "loss": 0.228,
      "step": 45805
    },
    {
      "epoch": 0.8271401642363443,
      "grad_norm": 0.4035283029079437,
      "learning_rate": 1.4385155526945793e-06,
      "loss": 0.184,
      "step": 45810
    },
    {
      "epoch": 0.827230443669245,
      "grad_norm": 0.4012914001941681,
      "learning_rate": 1.4370503254919244e-06,
      "loss": 0.2614,
      "step": 45815
    },
    {
      "epoch": 0.8273207231021458,
      "grad_norm": 0.47161754965782166,
      "learning_rate": 1.4355857871163138e-06,
      "loss": 0.1704,
      "step": 45820
    },
    {
      "epoch": 0.8274110025350465,
      "grad_norm": 0.4330509305000305,
      "learning_rate": 1.4341219376855552e-06,
      "loss": 0.1205,
      "step": 45825
    },
    {
      "epoch": 0.8275012819679471,
      "grad_norm": 0.9258213043212891,
      "learning_rate": 1.4326587773174094e-06,
      "loss": 0.2305,
      "step": 45830
    },
    {
      "epoch": 0.8275915614008479,
      "grad_norm": 0.668340265750885,
      "learning_rate": 1.431196306129572e-06,
      "loss": 0.2348,
      "step": 45835
    },
    {
      "epoch": 0.8276818408337486,
      "grad_norm": 0.579482913017273,
      "learning_rate": 1.42973452423969e-06,
      "loss": 0.2361,
      "step": 45840
    },
    {
      "epoch": 0.8277721202666494,
      "grad_norm": 0.4980883002281189,
      "learning_rate": 1.4282734317653536e-06,
      "loss": 0.1864,
      "step": 45845
    },
    {
      "epoch": 0.82786239969955,
      "grad_norm": 0.3318484425544739,
      "learning_rate": 1.4268130288241e-06,
      "loss": 0.2001,
      "step": 45850
    },
    {
      "epoch": 0.8279526791324507,
      "grad_norm": 0.6138582825660706,
      "learning_rate": 1.4253533155334022e-06,
      "loss": 0.1656,
      "step": 45855
    },
    {
      "epoch": 0.8280429585653515,
      "grad_norm": 0.5384577512741089,
      "learning_rate": 1.423894292010689e-06,
      "loss": 0.1881,
      "step": 45860
    },
    {
      "epoch": 0.8281332379982522,
      "grad_norm": 0.47394129633903503,
      "learning_rate": 1.422435958373326e-06,
      "loss": 0.2215,
      "step": 45865
    },
    {
      "epoch": 0.8282235174311529,
      "grad_norm": 0.32517707347869873,
      "learning_rate": 1.4209783147386237e-06,
      "loss": 0.2363,
      "step": 45870
    },
    {
      "epoch": 0.8283137968640536,
      "grad_norm": 0.4851417541503906,
      "learning_rate": 1.419521361223839e-06,
      "loss": 0.2226,
      "step": 45875
    },
    {
      "epoch": 0.8284040762969543,
      "grad_norm": 0.411033570766449,
      "learning_rate": 1.4180650979461751e-06,
      "loss": 0.2029,
      "step": 45880
    },
    {
      "epoch": 0.8284943557298551,
      "grad_norm": 0.28907907009124756,
      "learning_rate": 1.4166095250227795e-06,
      "loss": 0.2287,
      "step": 45885
    },
    {
      "epoch": 0.8285846351627557,
      "grad_norm": 0.6083971858024597,
      "learning_rate": 1.4151546425707384e-06,
      "loss": 0.2267,
      "step": 45890
    },
    {
      "epoch": 0.8286749145956565,
      "grad_norm": 0.3592473566532135,
      "learning_rate": 1.4137004507070907e-06,
      "loss": 0.2908,
      "step": 45895
    },
    {
      "epoch": 0.8287651940285572,
      "grad_norm": 0.7764474749565125,
      "learning_rate": 1.4122469495488112e-06,
      "loss": 0.2336,
      "step": 45900
    },
    {
      "epoch": 0.8288554734614579,
      "grad_norm": 0.7354249954223633,
      "learning_rate": 1.4107941392128254e-06,
      "loss": 0.2546,
      "step": 45905
    },
    {
      "epoch": 0.8289457528943586,
      "grad_norm": 0.2935306131839752,
      "learning_rate": 1.4093420198160013e-06,
      "loss": 0.1447,
      "step": 45910
    },
    {
      "epoch": 0.8290360323272593,
      "grad_norm": 0.5847833156585693,
      "learning_rate": 1.407890591475155e-06,
      "loss": 0.1871,
      "step": 45915
    },
    {
      "epoch": 0.82912631176016,
      "grad_norm": 0.9674150347709656,
      "learning_rate": 1.4064398543070369e-06,
      "loss": 0.2378,
      "step": 45920
    },
    {
      "epoch": 0.8292165911930608,
      "grad_norm": 0.42791542410850525,
      "learning_rate": 1.404989808428353e-06,
      "loss": 0.1536,
      "step": 45925
    },
    {
      "epoch": 0.8293068706259614,
      "grad_norm": 0.43429914116859436,
      "learning_rate": 1.403540453955745e-06,
      "loss": 0.205,
      "step": 45930
    },
    {
      "epoch": 0.8293971500588622,
      "grad_norm": 0.5471498370170593,
      "learning_rate": 1.4020917910058086e-06,
      "loss": 0.2592,
      "step": 45935
    },
    {
      "epoch": 0.8294874294917629,
      "grad_norm": 0.28336235880851746,
      "learning_rate": 1.4006438196950712e-06,
      "loss": 0.3411,
      "step": 45940
    },
    {
      "epoch": 0.8295777089246636,
      "grad_norm": 0.5176075100898743,
      "learning_rate": 1.3991965401400154e-06,
      "loss": 0.1394,
      "step": 45945
    },
    {
      "epoch": 0.8296679883575643,
      "grad_norm": 0.48003312945365906,
      "learning_rate": 1.3977499524570648e-06,
      "loss": 0.2763,
      "step": 45950
    },
    {
      "epoch": 0.829758267790465,
      "grad_norm": 0.4888760447502136,
      "learning_rate": 1.3963040567625886e-06,
      "loss": 0.1987,
      "step": 45955
    },
    {
      "epoch": 0.8298485472233658,
      "grad_norm": 0.5276985168457031,
      "learning_rate": 1.3948588531728946e-06,
      "loss": 0.2529,
      "step": 45960
    },
    {
      "epoch": 0.8299388266562665,
      "grad_norm": 0.6013329029083252,
      "learning_rate": 1.393414341804239e-06,
      "loss": 0.1331,
      "step": 45965
    },
    {
      "epoch": 0.8300291060891672,
      "grad_norm": 0.6370763778686523,
      "learning_rate": 1.3919705227728242e-06,
      "loss": 0.2626,
      "step": 45970
    },
    {
      "epoch": 0.8301193855220679,
      "grad_norm": 0.3273886740207672,
      "learning_rate": 1.390527396194794e-06,
      "loss": 0.2427,
      "step": 45975
    },
    {
      "epoch": 0.8302096649549686,
      "grad_norm": 0.715945303440094,
      "learning_rate": 1.3890849621862401e-06,
      "loss": 0.3468,
      "step": 45980
    },
    {
      "epoch": 0.8302999443878694,
      "grad_norm": 0.5415103435516357,
      "learning_rate": 1.3876432208631917e-06,
      "loss": 0.3264,
      "step": 45985
    },
    {
      "epoch": 0.8303902238207701,
      "grad_norm": 0.4317578077316284,
      "learning_rate": 1.3862021723416298e-06,
      "loss": 0.2118,
      "step": 45990
    },
    {
      "epoch": 0.8304805032536707,
      "grad_norm": 0.23266218602657318,
      "learning_rate": 1.3847618167374733e-06,
      "loss": 0.1878,
      "step": 45995
    },
    {
      "epoch": 0.8305707826865715,
      "grad_norm": 0.19852973520755768,
      "learning_rate": 1.383322154166591e-06,
      "loss": 0.2825,
      "step": 46000
    },
    {
      "epoch": 0.8306610621194722,
      "grad_norm": 0.3784917891025543,
      "learning_rate": 1.381883184744791e-06,
      "loss": 0.2635,
      "step": 46005
    },
    {
      "epoch": 0.830751341552373,
      "grad_norm": 0.508661687374115,
      "learning_rate": 1.380444908587828e-06,
      "loss": 0.2338,
      "step": 46010
    },
    {
      "epoch": 0.8308416209852736,
      "grad_norm": 0.42893388867378235,
      "learning_rate": 1.3790073258114023e-06,
      "loss": 0.1343,
      "step": 46015
    },
    {
      "epoch": 0.8309319004181743,
      "grad_norm": 0.6285560727119446,
      "learning_rate": 1.3775704365311594e-06,
      "loss": 0.254,
      "step": 46020
    },
    {
      "epoch": 0.8310221798510751,
      "grad_norm": 0.6512768864631653,
      "learning_rate": 1.376134240862681e-06,
      "loss": 0.1963,
      "step": 46025
    },
    {
      "epoch": 0.8311124592839758,
      "grad_norm": 0.5742652416229248,
      "learning_rate": 1.374698738921504e-06,
      "loss": 0.2186,
      "step": 46030
    },
    {
      "epoch": 0.8312027387168764,
      "grad_norm": 0.470001757144928,
      "learning_rate": 1.3732639308230988e-06,
      "loss": 0.2349,
      "step": 46035
    },
    {
      "epoch": 0.8312930181497772,
      "grad_norm": 0.6508385539054871,
      "learning_rate": 1.371829816682888e-06,
      "loss": 0.2872,
      "step": 46040
    },
    {
      "epoch": 0.8313832975826779,
      "grad_norm": 0.49703094363212585,
      "learning_rate": 1.3703963966162348e-06,
      "loss": 0.2483,
      "step": 46045
    },
    {
      "epoch": 0.8314735770155787,
      "grad_norm": 0.5042121410369873,
      "learning_rate": 1.3689636707384512e-06,
      "loss": 0.2363,
      "step": 46050
    },
    {
      "epoch": 0.8315638564484793,
      "grad_norm": 0.37353435158729553,
      "learning_rate": 1.3675316391647874e-06,
      "loss": 0.2679,
      "step": 46055
    },
    {
      "epoch": 0.83165413588138,
      "grad_norm": 0.46420320868492126,
      "learning_rate": 1.3661003020104358e-06,
      "loss": 0.2406,
      "step": 46060
    },
    {
      "epoch": 0.8317444153142808,
      "grad_norm": 0.42775171995162964,
      "learning_rate": 1.364669659390543e-06,
      "loss": 0.2727,
      "step": 46065
    },
    {
      "epoch": 0.8318346947471815,
      "grad_norm": 0.36116909980773926,
      "learning_rate": 1.363239711420189e-06,
      "loss": 0.2324,
      "step": 46070
    },
    {
      "epoch": 0.8319249741800822,
      "grad_norm": 0.6921894550323486,
      "learning_rate": 1.3618104582144042e-06,
      "loss": 0.2463,
      "step": 46075
    },
    {
      "epoch": 0.8320152536129829,
      "grad_norm": 0.536548376083374,
      "learning_rate": 1.3603818998881634e-06,
      "loss": 0.2015,
      "step": 46080
    },
    {
      "epoch": 0.8321055330458836,
      "grad_norm": 0.4366627335548401,
      "learning_rate": 1.3589540365563835e-06,
      "loss": 0.251,
      "step": 46085
    },
    {
      "epoch": 0.8321958124787844,
      "grad_norm": 0.3272264897823334,
      "learning_rate": 1.3575268683339239e-06,
      "loss": 0.1897,
      "step": 46090
    },
    {
      "epoch": 0.832286091911685,
      "grad_norm": 0.5101941227912903,
      "learning_rate": 1.3561003953355933e-06,
      "loss": 0.2492,
      "step": 46095
    },
    {
      "epoch": 0.8323763713445858,
      "grad_norm": 0.5696504712104797,
      "learning_rate": 1.354674617676135e-06,
      "loss": 0.123,
      "step": 46100
    },
    {
      "epoch": 0.8324666507774865,
      "grad_norm": 0.5099152326583862,
      "learning_rate": 1.353249535470248e-06,
      "loss": 0.1961,
      "step": 46105
    },
    {
      "epoch": 0.8325569302103872,
      "grad_norm": 0.30795571208000183,
      "learning_rate": 1.3518251488325661e-06,
      "loss": 0.2404,
      "step": 46110
    },
    {
      "epoch": 0.8326472096432879,
      "grad_norm": 0.631792426109314,
      "learning_rate": 1.3504014578776758e-06,
      "loss": 0.2101,
      "step": 46115
    },
    {
      "epoch": 0.8327374890761886,
      "grad_norm": 1.0345816612243652,
      "learning_rate": 1.3489784627200976e-06,
      "loss": 0.2382,
      "step": 46120
    },
    {
      "epoch": 0.8328277685090894,
      "grad_norm": 0.2446281611919403,
      "learning_rate": 1.3475561634743051e-06,
      "loss": 0.1597,
      "step": 46125
    },
    {
      "epoch": 0.8329180479419901,
      "grad_norm": 0.5093153119087219,
      "learning_rate": 1.3461345602547072e-06,
      "loss": 0.2228,
      "step": 46130
    },
    {
      "epoch": 0.8330083273748907,
      "grad_norm": 0.2520841956138611,
      "learning_rate": 1.3447136531756667e-06,
      "loss": 0.1745,
      "step": 46135
    },
    {
      "epoch": 0.8330986068077915,
      "grad_norm": 0.3567710220813751,
      "learning_rate": 1.343293442351481e-06,
      "loss": 0.1607,
      "step": 46140
    },
    {
      "epoch": 0.8331888862406922,
      "grad_norm": 0.6307740807533264,
      "learning_rate": 1.3418739278963965e-06,
      "loss": 0.2686,
      "step": 46145
    },
    {
      "epoch": 0.833279165673593,
      "grad_norm": 0.391429603099823,
      "learning_rate": 1.3404551099246066e-06,
      "loss": 0.2448,
      "step": 46150
    },
    {
      "epoch": 0.8333694451064936,
      "grad_norm": 0.28969264030456543,
      "learning_rate": 1.3390369885502408e-06,
      "loss": 0.2472,
      "step": 46155
    },
    {
      "epoch": 0.8334597245393943,
      "grad_norm": 0.4014858603477478,
      "learning_rate": 1.337619563887379e-06,
      "loss": 0.2494,
      "step": 46160
    },
    {
      "epoch": 0.8335500039722951,
      "grad_norm": 0.522121787071228,
      "learning_rate": 1.33620283605004e-06,
      "loss": 0.2404,
      "step": 46165
    },
    {
      "epoch": 0.8336402834051958,
      "grad_norm": 0.42990896105766296,
      "learning_rate": 1.3347868051521907e-06,
      "loss": 0.3098,
      "step": 46170
    },
    {
      "epoch": 0.8337305628380964,
      "grad_norm": 0.4846160113811493,
      "learning_rate": 1.3333714713077417e-06,
      "loss": 0.3495,
      "step": 46175
    },
    {
      "epoch": 0.8338208422709972,
      "grad_norm": 0.3562118113040924,
      "learning_rate": 1.3319568346305466e-06,
      "loss": 0.2418,
      "step": 46180
    },
    {
      "epoch": 0.8339111217038979,
      "grad_norm": 0.5375434160232544,
      "learning_rate": 1.330542895234399e-06,
      "loss": 0.1946,
      "step": 46185
    },
    {
      "epoch": 0.8340014011367987,
      "grad_norm": 0.31592509150505066,
      "learning_rate": 1.3291296532330454e-06,
      "loss": 0.162,
      "step": 46190
    },
    {
      "epoch": 0.8340916805696993,
      "grad_norm": 0.39862120151519775,
      "learning_rate": 1.3277171087401652e-06,
      "loss": 0.1801,
      "step": 46195
    },
    {
      "epoch": 0.8341819600026,
      "grad_norm": 0.45258471369743347,
      "learning_rate": 1.3263052618693917e-06,
      "loss": 0.2124,
      "step": 46200
    },
    {
      "epoch": 0.8342722394355008,
      "grad_norm": 0.4387921988964081,
      "learning_rate": 1.324894112734293e-06,
      "loss": 0.126,
      "step": 46205
    },
    {
      "epoch": 0.8343625188684015,
      "grad_norm": 0.3463088274002075,
      "learning_rate": 1.3234836614483915e-06,
      "loss": 0.2149,
      "step": 46210
    },
    {
      "epoch": 0.8344527983013021,
      "grad_norm": 0.29718273878097534,
      "learning_rate": 1.3220739081251443e-06,
      "loss": 0.1239,
      "step": 46215
    },
    {
      "epoch": 0.8345430777342029,
      "grad_norm": 0.36515769362449646,
      "learning_rate": 1.3206648528779586e-06,
      "loss": 0.2371,
      "step": 46220
    },
    {
      "epoch": 0.8346333571671036,
      "grad_norm": 0.2504488229751587,
      "learning_rate": 1.319256495820178e-06,
      "loss": 0.205,
      "step": 46225
    },
    {
      "epoch": 0.8347236366000044,
      "grad_norm": 0.5337098240852356,
      "learning_rate": 1.3178488370650989e-06,
      "loss": 0.2764,
      "step": 46230
    },
    {
      "epoch": 0.834813916032905,
      "grad_norm": 0.47260913252830505,
      "learning_rate": 1.3164418767259545e-06,
      "loss": 0.2033,
      "step": 46235
    },
    {
      "epoch": 0.8349041954658057,
      "grad_norm": 0.40227386355400085,
      "learning_rate": 1.3150356149159248e-06,
      "loss": 0.2325,
      "step": 46240
    },
    {
      "epoch": 0.8349944748987065,
      "grad_norm": 0.4283299148082733,
      "learning_rate": 1.313630051748136e-06,
      "loss": 0.1534,
      "step": 46245
    },
    {
      "epoch": 0.8350847543316072,
      "grad_norm": 0.541343092918396,
      "learning_rate": 1.3122251873356517e-06,
      "loss": 0.2231,
      "step": 46250
    },
    {
      "epoch": 0.8351750337645079,
      "grad_norm": 0.4880170524120331,
      "learning_rate": 1.3108210217914874e-06,
      "loss": 0.2555,
      "step": 46255
    },
    {
      "epoch": 0.8352653131974086,
      "grad_norm": 0.5511779189109802,
      "learning_rate": 1.3094175552285925e-06,
      "loss": 0.2015,
      "step": 46260
    },
    {
      "epoch": 0.8353555926303093,
      "grad_norm": 0.43681982159614563,
      "learning_rate": 1.3080147877598715e-06,
      "loss": 0.2518,
      "step": 46265
    },
    {
      "epoch": 0.8354458720632101,
      "grad_norm": 0.4958720803260803,
      "learning_rate": 1.3066127194981593e-06,
      "loss": 0.1949,
      "step": 46270
    },
    {
      "epoch": 0.8355361514961107,
      "grad_norm": 0.6504906415939331,
      "learning_rate": 1.3052113505562513e-06,
      "loss": 0.2196,
      "step": 46275
    },
    {
      "epoch": 0.8356264309290115,
      "grad_norm": 0.6573824882507324,
      "learning_rate": 1.303810681046871e-06,
      "loss": 0.22,
      "step": 46280
    },
    {
      "epoch": 0.8357167103619122,
      "grad_norm": 0.4231870472431183,
      "learning_rate": 1.3024107110826956e-06,
      "loss": 0.3308,
      "step": 46285
    },
    {
      "epoch": 0.8358069897948129,
      "grad_norm": 0.4615139961242676,
      "learning_rate": 1.3010114407763397e-06,
      "loss": 0.2548,
      "step": 46290
    },
    {
      "epoch": 0.8358972692277136,
      "grad_norm": 0.5508519411087036,
      "learning_rate": 1.2996128702403665e-06,
      "loss": 0.2336,
      "step": 46295
    },
    {
      "epoch": 0.8359875486606143,
      "grad_norm": 0.5502642393112183,
      "learning_rate": 1.2982149995872784e-06,
      "loss": 0.2087,
      "step": 46300
    },
    {
      "epoch": 0.836077828093515,
      "grad_norm": 0.37094512581825256,
      "learning_rate": 1.2968178289295264e-06,
      "loss": 0.2083,
      "step": 46305
    },
    {
      "epoch": 0.8361681075264158,
      "grad_norm": 0.595604658126831,
      "learning_rate": 1.2954213583795005e-06,
      "loss": 0.2734,
      "step": 46310
    },
    {
      "epoch": 0.8362583869593164,
      "grad_norm": 0.41564124822616577,
      "learning_rate": 1.29402558804954e-06,
      "loss": 0.2285,
      "step": 46315
    },
    {
      "epoch": 0.8363486663922172,
      "grad_norm": 0.33614933490753174,
      "learning_rate": 1.2926305180519204e-06,
      "loss": 0.2481,
      "step": 46320
    },
    {
      "epoch": 0.8364389458251179,
      "grad_norm": 0.480729341506958,
      "learning_rate": 1.29123614849887e-06,
      "loss": 0.2043,
      "step": 46325
    },
    {
      "epoch": 0.8365292252580186,
      "grad_norm": 0.5558351278305054,
      "learning_rate": 1.2898424795025522e-06,
      "loss": 0.2212,
      "step": 46330
    },
    {
      "epoch": 0.8366195046909193,
      "grad_norm": 0.5643190145492554,
      "learning_rate": 1.2884495111750762e-06,
      "loss": 0.1975,
      "step": 46335
    },
    {
      "epoch": 0.83670978412382,
      "grad_norm": 0.42842450737953186,
      "learning_rate": 1.2870572436284967e-06,
      "loss": 0.1398,
      "step": 46340
    },
    {
      "epoch": 0.8368000635567208,
      "grad_norm": 0.5728448629379272,
      "learning_rate": 1.2856656769748132e-06,
      "loss": 0.2152,
      "step": 46345
    },
    {
      "epoch": 0.8368903429896215,
      "grad_norm": 0.4689022898674011,
      "learning_rate": 1.2842748113259695e-06,
      "loss": 0.1772,
      "step": 46350
    },
    {
      "epoch": 0.8369806224225221,
      "grad_norm": 0.4232998490333557,
      "learning_rate": 1.282884646793845e-06,
      "loss": 0.3376,
      "step": 46355
    },
    {
      "epoch": 0.8370709018554229,
      "grad_norm": 0.7367646098136902,
      "learning_rate": 1.2814951834902734e-06,
      "loss": 0.3007,
      "step": 46360
    },
    {
      "epoch": 0.8371611812883236,
      "grad_norm": 0.4986398220062256,
      "learning_rate": 1.280106421527022e-06,
      "loss": 0.2284,
      "step": 46365
    },
    {
      "epoch": 0.8372514607212244,
      "grad_norm": 0.5241086483001709,
      "learning_rate": 1.2787183610158104e-06,
      "loss": 0.2932,
      "step": 46370
    },
    {
      "epoch": 0.8373417401541251,
      "grad_norm": 0.6657364964485168,
      "learning_rate": 1.2773310020682961e-06,
      "loss": 0.1989,
      "step": 46375
    },
    {
      "epoch": 0.8374320195870257,
      "grad_norm": 0.5218197703361511,
      "learning_rate": 1.275944344796085e-06,
      "loss": 0.1435,
      "step": 46380
    },
    {
      "epoch": 0.8375222990199265,
      "grad_norm": 0.401005357503891,
      "learning_rate": 1.2745583893107183e-06,
      "loss": 0.236,
      "step": 46385
    },
    {
      "epoch": 0.8376125784528272,
      "grad_norm": 0.4521532356739044,
      "learning_rate": 1.273173135723692e-06,
      "loss": 0.2441,
      "step": 46390
    },
    {
      "epoch": 0.837702857885728,
      "grad_norm": 0.41541245579719543,
      "learning_rate": 1.271788584146435e-06,
      "loss": 0.2277,
      "step": 46395
    },
    {
      "epoch": 0.8377931373186286,
      "grad_norm": 0.33301645517349243,
      "learning_rate": 1.270404734690328e-06,
      "loss": 0.1948,
      "step": 46400
    },
    {
      "epoch": 0.8378834167515293,
      "grad_norm": 0.789784848690033,
      "learning_rate": 1.2690215874666856e-06,
      "loss": 0.2297,
      "step": 46405
    },
    {
      "epoch": 0.8379736961844301,
      "grad_norm": 0.3074299693107605,
      "learning_rate": 1.2676391425867806e-06,
      "loss": 0.2477,
      "step": 46410
    },
    {
      "epoch": 0.8380639756173308,
      "grad_norm": 0.4493200480937958,
      "learning_rate": 1.2662574001618156e-06,
      "loss": 0.1588,
      "step": 46415
    },
    {
      "epoch": 0.8381542550502314,
      "grad_norm": 0.7905411124229431,
      "learning_rate": 1.2648763603029413e-06,
      "loss": 0.2534,
      "step": 46420
    },
    {
      "epoch": 0.8382445344831322,
      "grad_norm": 0.3890785276889801,
      "learning_rate": 1.2634960231212546e-06,
      "loss": 0.2178,
      "step": 46425
    },
    {
      "epoch": 0.8383348139160329,
      "grad_norm": 0.7565959095954895,
      "learning_rate": 1.2621163887277899e-06,
      "loss": 0.1351,
      "step": 46430
    },
    {
      "epoch": 0.8384250933489337,
      "grad_norm": 0.4592171013355255,
      "learning_rate": 1.2607374572335317e-06,
      "loss": 0.2981,
      "step": 46435
    },
    {
      "epoch": 0.8385153727818343,
      "grad_norm": 0.22006268799304962,
      "learning_rate": 1.2593592287494038e-06,
      "loss": 0.2293,
      "step": 46440
    },
    {
      "epoch": 0.838605652214735,
      "grad_norm": 0.4412384331226349,
      "learning_rate": 1.2579817033862784e-06,
      "loss": 0.2287,
      "step": 46445
    },
    {
      "epoch": 0.8386959316476358,
      "grad_norm": 0.532873272895813,
      "learning_rate": 1.2566048812549614e-06,
      "loss": 0.19,
      "step": 46450
    },
    {
      "epoch": 0.8387862110805365,
      "grad_norm": 0.23210127651691437,
      "learning_rate": 1.2552287624662141e-06,
      "loss": 0.1869,
      "step": 46455
    },
    {
      "epoch": 0.8388764905134372,
      "grad_norm": 0.6327080130577087,
      "learning_rate": 1.253853347130729e-06,
      "loss": 0.1106,
      "step": 46460
    },
    {
      "epoch": 0.8389667699463379,
      "grad_norm": 0.44668781757354736,
      "learning_rate": 1.2524786353591544e-06,
      "loss": 0.1868,
      "step": 46465
    },
    {
      "epoch": 0.8390570493792386,
      "grad_norm": 0.3663041889667511,
      "learning_rate": 1.2511046272620685e-06,
      "loss": 0.1131,
      "step": 46470
    },
    {
      "epoch": 0.8391473288121394,
      "grad_norm": 0.12444600462913513,
      "learning_rate": 1.2497313229500086e-06,
      "loss": 0.2369,
      "step": 46475
    },
    {
      "epoch": 0.83923760824504,
      "grad_norm": 0.7890183329582214,
      "learning_rate": 1.2483587225334414e-06,
      "loss": 0.1307,
      "step": 46480
    },
    {
      "epoch": 0.8393278876779408,
      "grad_norm": 0.4431261420249939,
      "learning_rate": 1.246986826122787e-06,
      "loss": 0.2444,
      "step": 46485
    },
    {
      "epoch": 0.8394181671108415,
      "grad_norm": 0.39454951882362366,
      "learning_rate": 1.2456156338283998e-06,
      "loss": 0.1739,
      "step": 46490
    },
    {
      "epoch": 0.8395084465437422,
      "grad_norm": 0.3442912995815277,
      "learning_rate": 1.2442451457605852e-06,
      "loss": 0.2237,
      "step": 46495
    },
    {
      "epoch": 0.8395987259766429,
      "grad_norm": 0.35818588733673096,
      "learning_rate": 1.2428753620295875e-06,
      "loss": 0.2182,
      "step": 46500
    },
    {
      "epoch": 0.8396890054095436,
      "grad_norm": 0.4712678790092468,
      "learning_rate": 1.2415062827455948e-06,
      "loss": 0.2619,
      "step": 46505
    },
    {
      "epoch": 0.8397792848424444,
      "grad_norm": 0.41358664631843567,
      "learning_rate": 1.2401379080187447e-06,
      "loss": 0.2226,
      "step": 46510
    },
    {
      "epoch": 0.8398695642753451,
      "grad_norm": 1.274709939956665,
      "learning_rate": 1.238770237959107e-06,
      "loss": 0.1985,
      "step": 46515
    },
    {
      "epoch": 0.8399598437082457,
      "grad_norm": 0.3299989402294159,
      "learning_rate": 1.237403272676706e-06,
      "loss": 0.1947,
      "step": 46520
    },
    {
      "epoch": 0.8400501231411465,
      "grad_norm": 0.2533578872680664,
      "learning_rate": 1.236037012281499e-06,
      "loss": 0.2319,
      "step": 46525
    },
    {
      "epoch": 0.8401404025740472,
      "grad_norm": 0.845720648765564,
      "learning_rate": 1.234671456883396e-06,
      "loss": 0.2142,
      "step": 46530
    },
    {
      "epoch": 0.840230682006948,
      "grad_norm": 0.48371750116348267,
      "learning_rate": 1.2333066065922394e-06,
      "loss": 0.2569,
      "step": 46535
    },
    {
      "epoch": 0.8403209614398486,
      "grad_norm": 0.3321785628795624,
      "learning_rate": 1.231942461517832e-06,
      "loss": 0.1951,
      "step": 46540
    },
    {
      "epoch": 0.8404112408727493,
      "grad_norm": 0.3849267065525055,
      "learning_rate": 1.230579021769901e-06,
      "loss": 0.2217,
      "step": 46545
    },
    {
      "epoch": 0.8405015203056501,
      "grad_norm": 0.6842372417449951,
      "learning_rate": 1.2292162874581293e-06,
      "loss": 0.2251,
      "step": 46550
    },
    {
      "epoch": 0.8405917997385508,
      "grad_norm": 0.4850200116634369,
      "learning_rate": 1.227854258692136e-06,
      "loss": 0.2633,
      "step": 46555
    },
    {
      "epoch": 0.8406820791714514,
      "grad_norm": 0.28475072979927063,
      "learning_rate": 1.2264929355814904e-06,
      "loss": 0.1486,
      "step": 46560
    },
    {
      "epoch": 0.8407723586043522,
      "grad_norm": 0.5592582821846008,
      "learning_rate": 1.2251323182356957e-06,
      "loss": 0.256,
      "step": 46565
    },
    {
      "epoch": 0.8408626380372529,
      "grad_norm": 0.7354555726051331,
      "learning_rate": 1.2237724067642066e-06,
      "loss": 0.159,
      "step": 46570
    },
    {
      "epoch": 0.8409529174701537,
      "grad_norm": 0.25784942507743835,
      "learning_rate": 1.2224132012764178e-06,
      "loss": 0.1818,
      "step": 46575
    },
    {
      "epoch": 0.8410431969030543,
      "grad_norm": 0.9170084595680237,
      "learning_rate": 1.221054701881671e-06,
      "loss": 0.2719,
      "step": 46580
    },
    {
      "epoch": 0.841133476335955,
      "grad_norm": 0.21908967196941376,
      "learning_rate": 1.2196969086892408e-06,
      "loss": 0.1778,
      "step": 46585
    },
    {
      "epoch": 0.8412237557688558,
      "grad_norm": 0.7114096283912659,
      "learning_rate": 1.2183398218083587e-06,
      "loss": 0.235,
      "step": 46590
    },
    {
      "epoch": 0.8413140352017565,
      "grad_norm": 0.5532537698745728,
      "learning_rate": 1.2169834413481862e-06,
      "loss": 0.3203,
      "step": 46595
    },
    {
      "epoch": 0.8414043146346571,
      "grad_norm": 0.2997341752052307,
      "learning_rate": 1.2156277674178374e-06,
      "loss": 0.1601,
      "step": 46600
    },
    {
      "epoch": 0.8414945940675579,
      "grad_norm": 0.7469394207000732,
      "learning_rate": 1.2142728001263682e-06,
      "loss": 0.2101,
      "step": 46605
    },
    {
      "epoch": 0.8415848735004586,
      "grad_norm": 0.5516133308410645,
      "learning_rate": 1.2129185395827715e-06,
      "loss": 0.2367,
      "step": 46610
    },
    {
      "epoch": 0.8416751529333594,
      "grad_norm": 0.5415496826171875,
      "learning_rate": 1.2115649858959922e-06,
      "loss": 0.2088,
      "step": 46615
    },
    {
      "epoch": 0.84176543236626,
      "grad_norm": 0.4268716275691986,
      "learning_rate": 1.21021213917491e-06,
      "loss": 0.2514,
      "step": 46620
    },
    {
      "epoch": 0.8418557117991607,
      "grad_norm": 0.26933982968330383,
      "learning_rate": 1.2088599995283546e-06,
      "loss": 0.215,
      "step": 46625
    },
    {
      "epoch": 0.8419459912320615,
      "grad_norm": 0.4040341377258301,
      "learning_rate": 1.2075085670650931e-06,
      "loss": 0.2693,
      "step": 46630
    },
    {
      "epoch": 0.8420362706649622,
      "grad_norm": 0.39315953850746155,
      "learning_rate": 1.2061578418938403e-06,
      "loss": 0.203,
      "step": 46635
    },
    {
      "epoch": 0.8421265500978629,
      "grad_norm": 0.48658832907676697,
      "learning_rate": 1.2048078241232508e-06,
      "loss": 0.2236,
      "step": 46640
    },
    {
      "epoch": 0.8422168295307636,
      "grad_norm": 0.534344494342804,
      "learning_rate": 1.203458513861927e-06,
      "loss": 0.2558,
      "step": 46645
    },
    {
      "epoch": 0.8423071089636643,
      "grad_norm": 0.5080280303955078,
      "learning_rate": 1.2021099112184065e-06,
      "loss": 0.214,
      "step": 46650
    },
    {
      "epoch": 0.8423973883965651,
      "grad_norm": 0.2412056177854538,
      "learning_rate": 1.2007620163011792e-06,
      "loss": 0.2084,
      "step": 46655
    },
    {
      "epoch": 0.8424876678294657,
      "grad_norm": 0.5974761843681335,
      "learning_rate": 1.1994148292186681e-06,
      "loss": 0.1612,
      "step": 46660
    },
    {
      "epoch": 0.8425779472623665,
      "grad_norm": 1.1710773706436157,
      "learning_rate": 1.1980683500792477e-06,
      "loss": 0.227,
      "step": 46665
    },
    {
      "epoch": 0.8426682266952672,
      "grad_norm": 0.4503132998943329,
      "learning_rate": 1.196722578991233e-06,
      "loss": 0.2271,
      "step": 46670
    },
    {
      "epoch": 0.8427585061281679,
      "grad_norm": 0.4720381200313568,
      "learning_rate": 1.1953775160628822e-06,
      "loss": 0.2954,
      "step": 46675
    },
    {
      "epoch": 0.8428487855610686,
      "grad_norm": 0.432170569896698,
      "learning_rate": 1.1940331614023936e-06,
      "loss": 0.3108,
      "step": 46680
    },
    {
      "epoch": 0.8429390649939693,
      "grad_norm": 0.5633404850959778,
      "learning_rate": 1.192689515117913e-06,
      "loss": 0.2709,
      "step": 46685
    },
    {
      "epoch": 0.84302934442687,
      "grad_norm": 0.3888779282569885,
      "learning_rate": 1.1913465773175247e-06,
      "loss": 0.2879,
      "step": 46690
    },
    {
      "epoch": 0.8431196238597708,
      "grad_norm": 0.7158108949661255,
      "learning_rate": 1.1900043481092583e-06,
      "loss": 0.3079,
      "step": 46695
    },
    {
      "epoch": 0.8432099032926714,
      "grad_norm": 0.2776183784008026,
      "learning_rate": 1.188662827601087e-06,
      "loss": 0.1768,
      "step": 46700
    },
    {
      "epoch": 0.8433001827255722,
      "grad_norm": 0.39774903655052185,
      "learning_rate": 1.1873220159009258e-06,
      "loss": 0.156,
      "step": 46705
    },
    {
      "epoch": 0.8433904621584729,
      "grad_norm": 0.5344502925872803,
      "learning_rate": 1.1859819131166372e-06,
      "loss": 0.1581,
      "step": 46710
    },
    {
      "epoch": 0.8434807415913737,
      "grad_norm": 0.23680981993675232,
      "learning_rate": 1.1846425193560162e-06,
      "loss": 0.1056,
      "step": 46715
    },
    {
      "epoch": 0.8435710210242743,
      "grad_norm": 0.33252429962158203,
      "learning_rate": 1.1833038347268145e-06,
      "loss": 0.2275,
      "step": 46720
    },
    {
      "epoch": 0.843661300457175,
      "grad_norm": 0.558745801448822,
      "learning_rate": 1.1819658593367123e-06,
      "loss": 0.247,
      "step": 46725
    },
    {
      "epoch": 0.8437515798900758,
      "grad_norm": 0.19038057327270508,
      "learning_rate": 1.1806285932933436e-06,
      "loss": 0.208,
      "step": 46730
    },
    {
      "epoch": 0.8438418593229765,
      "grad_norm": 0.4949471652507782,
      "learning_rate": 1.1792920367042814e-06,
      "loss": 0.2876,
      "step": 46735
    },
    {
      "epoch": 0.8439321387558771,
      "grad_norm": 0.37117841839790344,
      "learning_rate": 1.1779561896770453e-06,
      "loss": 0.2373,
      "step": 46740
    },
    {
      "epoch": 0.8440224181887779,
      "grad_norm": 0.5064247846603394,
      "learning_rate": 1.176621052319088e-06,
      "loss": 0.2694,
      "step": 46745
    },
    {
      "epoch": 0.8441126976216786,
      "grad_norm": 0.46294933557510376,
      "learning_rate": 1.175286624737817e-06,
      "loss": 0.2378,
      "step": 46750
    },
    {
      "epoch": 0.8442029770545794,
      "grad_norm": 0.3186456263065338,
      "learning_rate": 1.1739529070405742e-06,
      "loss": 0.2229,
      "step": 46755
    },
    {
      "epoch": 0.84429325648748,
      "grad_norm": 0.7466938495635986,
      "learning_rate": 1.1726198993346495e-06,
      "loss": 0.2063,
      "step": 46760
    },
    {
      "epoch": 0.8443835359203807,
      "grad_norm": 0.64537113904953,
      "learning_rate": 1.171287601727271e-06,
      "loss": 0.2007,
      "step": 46765
    },
    {
      "epoch": 0.8444738153532815,
      "grad_norm": 0.4769459068775177,
      "learning_rate": 1.169956014325614e-06,
      "loss": 0.2044,
      "step": 46770
    },
    {
      "epoch": 0.8445640947861822,
      "grad_norm": 0.3023054599761963,
      "learning_rate": 1.1686251372367952e-06,
      "loss": 0.1671,
      "step": 46775
    },
    {
      "epoch": 0.844654374219083,
      "grad_norm": 0.3669189214706421,
      "learning_rate": 1.1672949705678748e-06,
      "loss": 0.0945,
      "step": 46780
    },
    {
      "epoch": 0.8447446536519836,
      "grad_norm": 0.2845577597618103,
      "learning_rate": 1.1659655144258553e-06,
      "loss": 0.2238,
      "step": 46785
    },
    {
      "epoch": 0.8448349330848843,
      "grad_norm": 0.3285166919231415,
      "learning_rate": 1.1646367689176774e-06,
      "loss": 0.2137,
      "step": 46790
    },
    {
      "epoch": 0.8449252125177851,
      "grad_norm": 0.5089724659919739,
      "learning_rate": 1.1633087341502347e-06,
      "loss": 0.2013,
      "step": 46795
    },
    {
      "epoch": 0.8450154919506858,
      "grad_norm": 0.24118827283382416,
      "learning_rate": 1.1619814102303516e-06,
      "loss": 0.2986,
      "step": 46800
    },
    {
      "epoch": 0.8451057713835864,
      "grad_norm": 0.40448451042175293,
      "learning_rate": 1.1606547972648096e-06,
      "loss": 0.2873,
      "step": 46805
    },
    {
      "epoch": 0.8451960508164872,
      "grad_norm": 0.5910768508911133,
      "learning_rate": 1.1593288953603189e-06,
      "loss": 0.2162,
      "step": 46810
    },
    {
      "epoch": 0.8452863302493879,
      "grad_norm": 0.4762016832828522,
      "learning_rate": 1.1580037046235425e-06,
      "loss": 0.1681,
      "step": 46815
    },
    {
      "epoch": 0.8453766096822887,
      "grad_norm": 0.653786301612854,
      "learning_rate": 1.1566792251610781e-06,
      "loss": 0.2653,
      "step": 46820
    },
    {
      "epoch": 0.8454668891151893,
      "grad_norm": 0.552449643611908,
      "learning_rate": 1.1553554570794757e-06,
      "loss": 0.2972,
      "step": 46825
    },
    {
      "epoch": 0.84555716854809,
      "grad_norm": 0.5560564994812012,
      "learning_rate": 1.1540324004852166e-06,
      "loss": 0.3347,
      "step": 46830
    },
    {
      "epoch": 0.8456474479809908,
      "grad_norm": 0.42363405227661133,
      "learning_rate": 1.152710055484736e-06,
      "loss": 0.2751,
      "step": 46835
    },
    {
      "epoch": 0.8457377274138915,
      "grad_norm": 0.20735247433185577,
      "learning_rate": 1.151388422184404e-06,
      "loss": 0.1928,
      "step": 46840
    },
    {
      "epoch": 0.8458280068467922,
      "grad_norm": 0.6323709487915039,
      "learning_rate": 1.1500675006905416e-06,
      "loss": 0.2188,
      "step": 46845
    },
    {
      "epoch": 0.8459182862796929,
      "grad_norm": 0.6633914113044739,
      "learning_rate": 1.1487472911094e-06,
      "loss": 0.263,
      "step": 46850
    },
    {
      "epoch": 0.8460085657125936,
      "grad_norm": 0.6211409568786621,
      "learning_rate": 1.1474277935471866e-06,
      "loss": 0.2004,
      "step": 46855
    },
    {
      "epoch": 0.8460988451454944,
      "grad_norm": 0.34853506088256836,
      "learning_rate": 1.1461090081100402e-06,
      "loss": 0.1077,
      "step": 46860
    },
    {
      "epoch": 0.846189124578395,
      "grad_norm": 0.41717976331710815,
      "learning_rate": 1.144790934904051e-06,
      "loss": 0.186,
      "step": 46865
    },
    {
      "epoch": 0.8462794040112958,
      "grad_norm": 0.4029599726200104,
      "learning_rate": 1.1434735740352465e-06,
      "loss": 0.2869,
      "step": 46870
    },
    {
      "epoch": 0.8463696834441965,
      "grad_norm": 0.47104987502098083,
      "learning_rate": 1.1421569256096032e-06,
      "loss": 0.2056,
      "step": 46875
    },
    {
      "epoch": 0.8464599628770972,
      "grad_norm": 0.627106249332428,
      "learning_rate": 1.140840989733032e-06,
      "loss": 0.2746,
      "step": 46880
    },
    {
      "epoch": 0.8465502423099979,
      "grad_norm": 0.5372933149337769,
      "learning_rate": 1.1395257665113878e-06,
      "loss": 0.2181,
      "step": 46885
    },
    {
      "epoch": 0.8466405217428986,
      "grad_norm": 0.482340544462204,
      "learning_rate": 1.1382112560504765e-06,
      "loss": 0.2103,
      "step": 46890
    },
    {
      "epoch": 0.8467308011757994,
      "grad_norm": 0.28699415922164917,
      "learning_rate": 1.1368974584560366e-06,
      "loss": 0.1307,
      "step": 46895
    },
    {
      "epoch": 0.8468210806087001,
      "grad_norm": 0.421001136302948,
      "learning_rate": 1.1355843738337546e-06,
      "loss": 0.2902,
      "step": 46900
    },
    {
      "epoch": 0.8469113600416007,
      "grad_norm": 0.503657341003418,
      "learning_rate": 1.1342720022892594e-06,
      "loss": 0.1634,
      "step": 46905
    },
    {
      "epoch": 0.8470016394745015,
      "grad_norm": 0.38650810718536377,
      "learning_rate": 1.1329603439281245e-06,
      "loss": 0.2721,
      "step": 46910
    },
    {
      "epoch": 0.8470919189074022,
      "grad_norm": 0.8284596800804138,
      "learning_rate": 1.1316493988558575e-06,
      "loss": 0.316,
      "step": 46915
    },
    {
      "epoch": 0.847182198340303,
      "grad_norm": 0.4707046449184418,
      "learning_rate": 1.1303391671779185e-06,
      "loss": 0.2384,
      "step": 46920
    },
    {
      "epoch": 0.8472724777732036,
      "grad_norm": 0.7109625935554504,
      "learning_rate": 1.1290296489997033e-06,
      "loss": 0.1903,
      "step": 46925
    },
    {
      "epoch": 0.8473627572061043,
      "grad_norm": 0.3998103439807892,
      "learning_rate": 1.1277208444265552e-06,
      "loss": 0.1523,
      "step": 46930
    },
    {
      "epoch": 0.8474530366390051,
      "grad_norm": 0.5555111765861511,
      "learning_rate": 1.126412753563757e-06,
      "loss": 0.2332,
      "step": 46935
    },
    {
      "epoch": 0.8475433160719058,
      "grad_norm": 0.7187251448631287,
      "learning_rate": 1.1251053765165377e-06,
      "loss": 0.3009,
      "step": 46940
    },
    {
      "epoch": 0.8476335955048064,
      "grad_norm": 0.275674045085907,
      "learning_rate": 1.1237987133900619e-06,
      "loss": 0.2232,
      "step": 46945
    },
    {
      "epoch": 0.8477238749377072,
      "grad_norm": 0.3866052031517029,
      "learning_rate": 1.1224927642894456e-06,
      "loss": 0.233,
      "step": 46950
    },
    {
      "epoch": 0.8478141543706079,
      "grad_norm": 0.6808643937110901,
      "learning_rate": 1.1211875293197394e-06,
      "loss": 0.2382,
      "step": 46955
    },
    {
      "epoch": 0.8479044338035087,
      "grad_norm": 0.3973676860332489,
      "learning_rate": 1.1198830085859425e-06,
      "loss": 0.1897,
      "step": 46960
    },
    {
      "epoch": 0.8479947132364093,
      "grad_norm": 0.4017499089241028,
      "learning_rate": 1.1185792021929909e-06,
      "loss": 0.1826,
      "step": 46965
    },
    {
      "epoch": 0.84808499266931,
      "grad_norm": 0.31418779492378235,
      "learning_rate": 1.1172761102457675e-06,
      "loss": 0.2113,
      "step": 46970
    },
    {
      "epoch": 0.8481752721022108,
      "grad_norm": 0.49413245916366577,
      "learning_rate": 1.1159737328491005e-06,
      "loss": 0.2781,
      "step": 46975
    },
    {
      "epoch": 0.8482655515351115,
      "grad_norm": 0.31502243876457214,
      "learning_rate": 1.1146720701077517e-06,
      "loss": 0.2216,
      "step": 46980
    },
    {
      "epoch": 0.8483558309680121,
      "grad_norm": 0.5428707003593445,
      "learning_rate": 1.113371122126433e-06,
      "loss": 0.1835,
      "step": 46985
    },
    {
      "epoch": 0.8484461104009129,
      "grad_norm": 0.9356114864349365,
      "learning_rate": 1.1120708890097954e-06,
      "loss": 0.1494,
      "step": 46990
    },
    {
      "epoch": 0.8485363898338136,
      "grad_norm": 0.696953296661377,
      "learning_rate": 1.1107713708624324e-06,
      "loss": 0.3059,
      "step": 46995
    },
    {
      "epoch": 0.8486266692667144,
      "grad_norm": 0.1654614359140396,
      "learning_rate": 1.1094725677888806e-06,
      "loss": 0.2274,
      "step": 47000
    },
    {
      "epoch": 0.848716948699615,
      "grad_norm": 0.2907618582248688,
      "learning_rate": 1.1081744798936235e-06,
      "loss": 0.1928,
      "step": 47005
    },
    {
      "epoch": 0.8488072281325157,
      "grad_norm": 0.27342113852500916,
      "learning_rate": 1.106877107281078e-06,
      "loss": 0.287,
      "step": 47010
    },
    {
      "epoch": 0.8488975075654165,
      "grad_norm": 0.21456903219223022,
      "learning_rate": 1.1055804500556112e-06,
      "loss": 0.2203,
      "step": 47015
    },
    {
      "epoch": 0.8489877869983172,
      "grad_norm": 0.44015249609947205,
      "learning_rate": 1.104284508321527e-06,
      "loss": 0.1979,
      "step": 47020
    },
    {
      "epoch": 0.8490780664312179,
      "grad_norm": 0.2882255017757416,
      "learning_rate": 1.1029892821830779e-06,
      "loss": 0.2393,
      "step": 47025
    },
    {
      "epoch": 0.8491683458641186,
      "grad_norm": 0.3677639663219452,
      "learning_rate": 1.1016947717444504e-06,
      "loss": 0.1739,
      "step": 47030
    },
    {
      "epoch": 0.8492586252970193,
      "grad_norm": 0.2677743136882782,
      "learning_rate": 1.1004009771097834e-06,
      "loss": 0.2938,
      "step": 47035
    },
    {
      "epoch": 0.8493489047299201,
      "grad_norm": 0.38641270995140076,
      "learning_rate": 1.0991078983831505e-06,
      "loss": 0.1543,
      "step": 47040
    },
    {
      "epoch": 0.8494391841628207,
      "grad_norm": 0.30329596996307373,
      "learning_rate": 1.0978155356685727e-06,
      "loss": 0.213,
      "step": 47045
    },
    {
      "epoch": 0.8495294635957215,
      "grad_norm": 0.31273511052131653,
      "learning_rate": 1.0965238890700091e-06,
      "loss": 0.2213,
      "step": 47050
    },
    {
      "epoch": 0.8496197430286222,
      "grad_norm": 0.6745008826255798,
      "learning_rate": 1.0952329586913657e-06,
      "loss": 0.1875,
      "step": 47055
    },
    {
      "epoch": 0.8497100224615229,
      "grad_norm": 0.5616105198860168,
      "learning_rate": 1.0939427446364848e-06,
      "loss": 0.2215,
      "step": 47060
    },
    {
      "epoch": 0.8498003018944236,
      "grad_norm": 0.16216596961021423,
      "learning_rate": 1.0926532470091567e-06,
      "loss": 0.163,
      "step": 47065
    },
    {
      "epoch": 0.8498905813273243,
      "grad_norm": 0.40241101384162903,
      "learning_rate": 1.091364465913115e-06,
      "loss": 0.2139,
      "step": 47070
    },
    {
      "epoch": 0.849980860760225,
      "grad_norm": 0.5022798180580139,
      "learning_rate": 1.0900764014520272e-06,
      "loss": 0.2423,
      "step": 47075
    },
    {
      "epoch": 0.8500711401931258,
      "grad_norm": 0.6277716755867004,
      "learning_rate": 1.0887890537295142e-06,
      "loss": 0.1428,
      "step": 47080
    },
    {
      "epoch": 0.8501614196260264,
      "grad_norm": 0.5615366697311401,
      "learning_rate": 1.0875024228491282e-06,
      "loss": 0.1823,
      "step": 47085
    },
    {
      "epoch": 0.8502516990589272,
      "grad_norm": 0.3080914318561554,
      "learning_rate": 1.086216508914374e-06,
      "loss": 0.1969,
      "step": 47090
    },
    {
      "epoch": 0.8503419784918279,
      "grad_norm": 0.38214749097824097,
      "learning_rate": 1.0849313120286908e-06,
      "loss": 0.1717,
      "step": 47095
    },
    {
      "epoch": 0.8504322579247287,
      "grad_norm": 0.5622016787528992,
      "learning_rate": 1.0836468322954651e-06,
      "loss": 0.2176,
      "step": 47100
    },
    {
      "epoch": 0.8505225373576293,
      "grad_norm": 0.316812664270401,
      "learning_rate": 1.082363069818023e-06,
      "loss": 0.1706,
      "step": 47105
    },
    {
      "epoch": 0.85061281679053,
      "grad_norm": 0.3119005560874939,
      "learning_rate": 1.0810800246996366e-06,
      "loss": 0.2445,
      "step": 47110
    },
    {
      "epoch": 0.8507030962234308,
      "grad_norm": 0.2569788694381714,
      "learning_rate": 1.0797976970435132e-06,
      "loss": 0.1961,
      "step": 47115
    },
    {
      "epoch": 0.8507933756563315,
      "grad_norm": 0.5018973350524902,
      "learning_rate": 1.078516086952811e-06,
      "loss": 0.1391,
      "step": 47120
    },
    {
      "epoch": 0.8508836550892321,
      "grad_norm": 0.4110466241836548,
      "learning_rate": 1.0772351945306225e-06,
      "loss": 0.2013,
      "step": 47125
    },
    {
      "epoch": 0.8509739345221329,
      "grad_norm": 0.37337103486061096,
      "learning_rate": 1.0759550198799873e-06,
      "loss": 0.2508,
      "step": 47130
    },
    {
      "epoch": 0.8510642139550336,
      "grad_norm": 0.581804633140564,
      "learning_rate": 1.0746755631038874e-06,
      "loss": 0.2345,
      "step": 47135
    },
    {
      "epoch": 0.8511544933879344,
      "grad_norm": 0.35834333300590515,
      "learning_rate": 1.0733968243052462e-06,
      "loss": 0.2535,
      "step": 47140
    },
    {
      "epoch": 0.851244772820835,
      "grad_norm": 0.5402718186378479,
      "learning_rate": 1.0721188035869267e-06,
      "loss": 0.2759,
      "step": 47145
    },
    {
      "epoch": 0.8513350522537357,
      "grad_norm": 0.4894947409629822,
      "learning_rate": 1.0708415010517392e-06,
      "loss": 0.2165,
      "step": 47150
    },
    {
      "epoch": 0.8514253316866365,
      "grad_norm": 0.44462037086486816,
      "learning_rate": 1.0695649168024313e-06,
      "loss": 0.1352,
      "step": 47155
    },
    {
      "epoch": 0.8515156111195372,
      "grad_norm": 0.419643759727478,
      "learning_rate": 1.0682890509416932e-06,
      "loss": 0.2462,
      "step": 47160
    },
    {
      "epoch": 0.851605890552438,
      "grad_norm": 0.964062511920929,
      "learning_rate": 1.0670139035721617e-06,
      "loss": 0.2357,
      "step": 47165
    },
    {
      "epoch": 0.8516961699853386,
      "grad_norm": 0.38403525948524475,
      "learning_rate": 1.0657394747964122e-06,
      "loss": 0.1935,
      "step": 47170
    },
    {
      "epoch": 0.8517864494182393,
      "grad_norm": 0.2649580240249634,
      "learning_rate": 1.0644657647169654e-06,
      "loss": 0.1805,
      "step": 47175
    },
    {
      "epoch": 0.8518767288511401,
      "grad_norm": 0.5285569429397583,
      "learning_rate": 1.0631927734362791e-06,
      "loss": 0.2362,
      "step": 47180
    },
    {
      "epoch": 0.8519670082840408,
      "grad_norm": 0.9157831072807312,
      "learning_rate": 1.0619205010567579e-06,
      "loss": 0.2776,
      "step": 47185
    },
    {
      "epoch": 0.8520572877169414,
      "grad_norm": 0.367585688829422,
      "learning_rate": 1.0606489476807458e-06,
      "loss": 0.2512,
      "step": 47190
    },
    {
      "epoch": 0.8521475671498422,
      "grad_norm": 0.30853065848350525,
      "learning_rate": 1.0593781134105296e-06,
      "loss": 0.236,
      "step": 47195
    },
    {
      "epoch": 0.8522378465827429,
      "grad_norm": 0.5924100279808044,
      "learning_rate": 1.05810799834834e-06,
      "loss": 0.1941,
      "step": 47200
    },
    {
      "epoch": 0.8523281260156437,
      "grad_norm": 0.4391089081764221,
      "learning_rate": 1.0568386025963496e-06,
      "loss": 0.2303,
      "step": 47205
    },
    {
      "epoch": 0.8524184054485443,
      "grad_norm": 0.44665244221687317,
      "learning_rate": 1.0555699262566688e-06,
      "loss": 0.1246,
      "step": 47210
    },
    {
      "epoch": 0.852508684881445,
      "grad_norm": 0.275681734085083,
      "learning_rate": 1.0543019694313573e-06,
      "loss": 0.1656,
      "step": 47215
    },
    {
      "epoch": 0.8525989643143458,
      "grad_norm": 0.34005236625671387,
      "learning_rate": 1.0530347322224087e-06,
      "loss": 0.1833,
      "step": 47220
    },
    {
      "epoch": 0.8526892437472465,
      "grad_norm": 0.5032612085342407,
      "learning_rate": 1.0517682147317665e-06,
      "loss": 0.1561,
      "step": 47225
    },
    {
      "epoch": 0.8527795231801472,
      "grad_norm": 0.3961392343044281,
      "learning_rate": 1.050502417061311e-06,
      "loss": 0.2852,
      "step": 47230
    },
    {
      "epoch": 0.8528698026130479,
      "grad_norm": 0.7540068030357361,
      "learning_rate": 1.0492373393128663e-06,
      "loss": 0.2878,
      "step": 47235
    },
    {
      "epoch": 0.8529600820459486,
      "grad_norm": 0.53920978307724,
      "learning_rate": 1.0479729815881988e-06,
      "loss": 0.2544,
      "step": 47240
    },
    {
      "epoch": 0.8530503614788494,
      "grad_norm": 0.26546889543533325,
      "learning_rate": 1.0467093439890198e-06,
      "loss": 0.1424,
      "step": 47245
    },
    {
      "epoch": 0.85314064091175,
      "grad_norm": 0.4271557033061981,
      "learning_rate": 1.0454464266169774e-06,
      "loss": 0.2831,
      "step": 47250
    },
    {
      "epoch": 0.8532309203446508,
      "grad_norm": 0.43881359696388245,
      "learning_rate": 1.0441842295736615e-06,
      "loss": 0.2006,
      "step": 47255
    },
    {
      "epoch": 0.8533211997775515,
      "grad_norm": 0.4485865831375122,
      "learning_rate": 1.04292275296061e-06,
      "loss": 0.1576,
      "step": 47260
    },
    {
      "epoch": 0.8534114792104522,
      "grad_norm": 0.39906686544418335,
      "learning_rate": 1.0416619968792985e-06,
      "loss": 0.2443,
      "step": 47265
    },
    {
      "epoch": 0.8535017586433529,
      "grad_norm": 0.6947904825210571,
      "learning_rate": 1.0404019614311479e-06,
      "loss": 0.2761,
      "step": 47270
    },
    {
      "epoch": 0.8535920380762536,
      "grad_norm": 0.7733624577522278,
      "learning_rate": 1.0391426467175147e-06,
      "loss": 0.2278,
      "step": 47275
    },
    {
      "epoch": 0.8536823175091544,
      "grad_norm": 1.3637889623641968,
      "learning_rate": 1.0378840528397061e-06,
      "loss": 0.21,
      "step": 47280
    },
    {
      "epoch": 0.8537725969420551,
      "grad_norm": 0.4050987958908081,
      "learning_rate": 1.036626179898963e-06,
      "loss": 0.2574,
      "step": 47285
    },
    {
      "epoch": 0.8538628763749557,
      "grad_norm": 0.29665687680244446,
      "learning_rate": 1.0353690279964746e-06,
      "loss": 0.2488,
      "step": 47290
    },
    {
      "epoch": 0.8539531558078565,
      "grad_norm": 0.3802777826786041,
      "learning_rate": 1.0341125972333676e-06,
      "loss": 0.1798,
      "step": 47295
    },
    {
      "epoch": 0.8540434352407572,
      "grad_norm": 0.2983674108982086,
      "learning_rate": 1.0328568877107137e-06,
      "loss": 0.1454,
      "step": 47300
    },
    {
      "epoch": 0.854133714673658,
      "grad_norm": 0.5490597486495972,
      "learning_rate": 1.0316018995295252e-06,
      "loss": 0.2243,
      "step": 47305
    },
    {
      "epoch": 0.8542239941065586,
      "grad_norm": 0.4573357105255127,
      "learning_rate": 1.0303476327907602e-06,
      "loss": 0.1936,
      "step": 47310
    },
    {
      "epoch": 0.8543142735394593,
      "grad_norm": 0.23389914631843567,
      "learning_rate": 1.0290940875953103e-06,
      "loss": 0.2392,
      "step": 47315
    },
    {
      "epoch": 0.8544045529723601,
      "grad_norm": 0.5954925417900085,
      "learning_rate": 1.0278412640440183e-06,
      "loss": 0.261,
      "step": 47320
    },
    {
      "epoch": 0.8544948324052608,
      "grad_norm": 0.30451729893684387,
      "learning_rate": 1.02658916223766e-06,
      "loss": 0.1524,
      "step": 47325
    },
    {
      "epoch": 0.8545851118381614,
      "grad_norm": 0.24599765241146088,
      "learning_rate": 1.0253377822769605e-06,
      "loss": 0.2205,
      "step": 47330
    },
    {
      "epoch": 0.8546753912710622,
      "grad_norm": 0.24942728877067566,
      "learning_rate": 1.0240871242625872e-06,
      "loss": 0.218,
      "step": 47335
    },
    {
      "epoch": 0.8547656707039629,
      "grad_norm": 0.17759248614311218,
      "learning_rate": 1.0228371882951416e-06,
      "loss": 0.2113,
      "step": 47340
    },
    {
      "epoch": 0.8548559501368637,
      "grad_norm": 0.4103125035762787,
      "learning_rate": 1.0215879744751756e-06,
      "loss": 0.1333,
      "step": 47345
    },
    {
      "epoch": 0.8549462295697643,
      "grad_norm": 0.31829023361206055,
      "learning_rate": 1.0203394829031754e-06,
      "loss": 0.2226,
      "step": 47350
    },
    {
      "epoch": 0.855036509002665,
      "grad_norm": 0.9971163868904114,
      "learning_rate": 1.0190917136795775e-06,
      "loss": 0.3087,
      "step": 47355
    },
    {
      "epoch": 0.8551267884355658,
      "grad_norm": 0.2358555644750595,
      "learning_rate": 1.0178446669047515e-06,
      "loss": 0.2073,
      "step": 47360
    },
    {
      "epoch": 0.8552170678684665,
      "grad_norm": 0.3916555643081665,
      "learning_rate": 1.016598342679016e-06,
      "loss": 0.2214,
      "step": 47365
    },
    {
      "epoch": 0.8553073473013671,
      "grad_norm": 0.5762766599655151,
      "learning_rate": 1.0153527411026287e-06,
      "loss": 0.2235,
      "step": 47370
    },
    {
      "epoch": 0.8553976267342679,
      "grad_norm": 0.6114965081214905,
      "learning_rate": 1.0141078622757894e-06,
      "loss": 0.1998,
      "step": 47375
    },
    {
      "epoch": 0.8554879061671686,
      "grad_norm": 0.711415708065033,
      "learning_rate": 1.012863706298638e-06,
      "loss": 0.2235,
      "step": 47380
    },
    {
      "epoch": 0.8555781856000694,
      "grad_norm": 0.4038284420967102,
      "learning_rate": 1.01162027327126e-06,
      "loss": 0.1959,
      "step": 47385
    },
    {
      "epoch": 0.85566846503297,
      "grad_norm": 0.2545495629310608,
      "learning_rate": 1.0103775632936764e-06,
      "loss": 0.1961,
      "step": 47390
    },
    {
      "epoch": 0.8557587444658707,
      "grad_norm": 0.429012656211853,
      "learning_rate": 1.0091355764658583e-06,
      "loss": 0.2066,
      "step": 47395
    },
    {
      "epoch": 0.8558490238987715,
      "grad_norm": 0.4435504674911499,
      "learning_rate": 1.0078943128877128e-06,
      "loss": 0.2632,
      "step": 47400
    },
    {
      "epoch": 0.8559393033316722,
      "grad_norm": 0.6207600235939026,
      "learning_rate": 1.0066537726590918e-06,
      "loss": 0.2314,
      "step": 47405
    },
    {
      "epoch": 0.8560295827645729,
      "grad_norm": 0.49283120036125183,
      "learning_rate": 1.0054139558797859e-06,
      "loss": 0.2113,
      "step": 47410
    },
    {
      "epoch": 0.8561198621974736,
      "grad_norm": 0.6071460247039795,
      "learning_rate": 1.004174862649533e-06,
      "loss": 0.2228,
      "step": 47415
    },
    {
      "epoch": 0.8562101416303743,
      "grad_norm": 0.3180447816848755,
      "learning_rate": 1.002936493068003e-06,
      "loss": 0.2133,
      "step": 47420
    },
    {
      "epoch": 0.8563004210632751,
      "grad_norm": 0.4284539520740509,
      "learning_rate": 1.00169884723482e-06,
      "loss": 0.2385,
      "step": 47425
    },
    {
      "epoch": 0.8563907004961757,
      "grad_norm": 0.45321059226989746,
      "learning_rate": 1.0004619252495385e-06,
      "loss": 0.2522,
      "step": 47430
    },
    {
      "epoch": 0.8564809799290765,
      "grad_norm": 0.5622091889381409,
      "learning_rate": 9.992257272116623e-07,
      "loss": 0.2924,
      "step": 47435
    },
    {
      "epoch": 0.8565712593619772,
      "grad_norm": 0.4963833689689636,
      "learning_rate": 9.979902532206365e-07,
      "loss": 0.1977,
      "step": 47440
    },
    {
      "epoch": 0.8566615387948779,
      "grad_norm": 0.2972092628479004,
      "learning_rate": 9.967555033758414e-07,
      "loss": 0.2081,
      "step": 47445
    },
    {
      "epoch": 0.8567518182277786,
      "grad_norm": 0.5037822127342224,
      "learning_rate": 9.955214777766076e-07,
      "loss": 0.2568,
      "step": 47450
    },
    {
      "epoch": 0.8568420976606793,
      "grad_norm": 0.5657833218574524,
      "learning_rate": 9.942881765222012e-07,
      "loss": 0.1971,
      "step": 47455
    },
    {
      "epoch": 0.85693237709358,
      "grad_norm": 0.6688538193702698,
      "learning_rate": 9.93055599711833e-07,
      "loss": 0.1707,
      "step": 47460
    },
    {
      "epoch": 0.8570226565264808,
      "grad_norm": 0.5159571766853333,
      "learning_rate": 9.918237474446546e-07,
      "loss": 0.2757,
      "step": 47465
    },
    {
      "epoch": 0.8571129359593814,
      "grad_norm": 0.4161568880081177,
      "learning_rate": 9.905926198197624e-07,
      "loss": 0.2097,
      "step": 47470
    },
    {
      "epoch": 0.8572032153922822,
      "grad_norm": 0.3604661524295807,
      "learning_rate": 9.893622169361871e-07,
      "loss": 0.1489,
      "step": 47475
    },
    {
      "epoch": 0.8572934948251829,
      "grad_norm": 0.4307427406311035,
      "learning_rate": 9.881325388929087e-07,
      "loss": 0.2146,
      "step": 47480
    },
    {
      "epoch": 0.8573837742580837,
      "grad_norm": 0.3729318082332611,
      "learning_rate": 9.869035857888443e-07,
      "loss": 0.1714,
      "step": 47485
    },
    {
      "epoch": 0.8574740536909843,
      "grad_norm": 0.4250018000602722,
      "learning_rate": 9.856753577228562e-07,
      "loss": 0.2264,
      "step": 47490
    },
    {
      "epoch": 0.857564333123885,
      "grad_norm": 0.5746373534202576,
      "learning_rate": 9.844478547937431e-07,
      "loss": 0.1523,
      "step": 47495
    },
    {
      "epoch": 0.8576546125567858,
      "grad_norm": 0.3057235777378082,
      "learning_rate": 9.832210771002515e-07,
      "loss": 0.1251,
      "step": 47500
    },
    {
      "epoch": 0.8577448919896865,
      "grad_norm": 0.3894713222980499,
      "learning_rate": 9.819950247410647e-07,
      "loss": 0.2702,
      "step": 47505
    },
    {
      "epoch": 0.8578351714225871,
      "grad_norm": 0.43752869963645935,
      "learning_rate": 9.80769697814814e-07,
      "loss": 0.2357,
      "step": 47510
    },
    {
      "epoch": 0.8579254508554879,
      "grad_norm": 0.34574222564697266,
      "learning_rate": 9.795450964200625e-07,
      "loss": 0.1833,
      "step": 47515
    },
    {
      "epoch": 0.8580157302883886,
      "grad_norm": 0.3923795521259308,
      "learning_rate": 9.78321220655325e-07,
      "loss": 0.2149,
      "step": 47520
    },
    {
      "epoch": 0.8581060097212894,
      "grad_norm": 0.3717537224292755,
      "learning_rate": 9.77098070619049e-07,
      "loss": 0.192,
      "step": 47525
    },
    {
      "epoch": 0.85819628915419,
      "grad_norm": 0.429400235414505,
      "learning_rate": 9.758756464096309e-07,
      "loss": 0.2578,
      "step": 47530
    },
    {
      "epoch": 0.8582865685870907,
      "grad_norm": 0.37884220480918884,
      "learning_rate": 9.74653948125407e-07,
      "loss": 0.2427,
      "step": 47535
    },
    {
      "epoch": 0.8583768480199915,
      "grad_norm": 0.5908001065254211,
      "learning_rate": 9.73432975864651e-07,
      "loss": 0.233,
      "step": 47540
    },
    {
      "epoch": 0.8584671274528922,
      "grad_norm": 0.3083002269268036,
      "learning_rate": 9.722127297255833e-07,
      "loss": 0.2563,
      "step": 47545
    },
    {
      "epoch": 0.8585574068857929,
      "grad_norm": 0.7070236802101135,
      "learning_rate": 9.709932098063624e-07,
      "loss": 0.2216,
      "step": 47550
    },
    {
      "epoch": 0.8586476863186936,
      "grad_norm": 0.7982986569404602,
      "learning_rate": 9.697744162050914e-07,
      "loss": 0.1687,
      "step": 47555
    },
    {
      "epoch": 0.8587379657515943,
      "grad_norm": 0.5032423138618469,
      "learning_rate": 9.685563490198113e-07,
      "loss": 0.259,
      "step": 47560
    },
    {
      "epoch": 0.8588282451844951,
      "grad_norm": 0.6070419549942017,
      "learning_rate": 9.67339008348508e-07,
      "loss": 0.2304,
      "step": 47565
    },
    {
      "epoch": 0.8589185246173958,
      "grad_norm": 0.4481718838214874,
      "learning_rate": 9.661223942891075e-07,
      "loss": 0.2453,
      "step": 47570
    },
    {
      "epoch": 0.8590088040502964,
      "grad_norm": 0.6688275337219238,
      "learning_rate": 9.649065069394802e-07,
      "loss": 0.268,
      "step": 47575
    },
    {
      "epoch": 0.8590990834831972,
      "grad_norm": 0.584083080291748,
      "learning_rate": 9.636913463974308e-07,
      "loss": 0.2024,
      "step": 47580
    },
    {
      "epoch": 0.8591893629160979,
      "grad_norm": 0.4952586591243744,
      "learning_rate": 9.624769127607137e-07,
      "loss": 0.2318,
      "step": 47585
    },
    {
      "epoch": 0.8592796423489987,
      "grad_norm": 1.0109870433807373,
      "learning_rate": 9.612632061270178e-07,
      "loss": 0.2133,
      "step": 47590
    },
    {
      "epoch": 0.8593699217818993,
      "grad_norm": 0.5469220876693726,
      "learning_rate": 9.600502265939793e-07,
      "loss": 0.1844,
      "step": 47595
    },
    {
      "epoch": 0.8594602012148,
      "grad_norm": 0.2093033343553543,
      "learning_rate": 9.588379742591736e-07,
      "loss": 0.1955,
      "step": 47600
    },
    {
      "epoch": 0.8595504806477008,
      "grad_norm": 0.34480932354927063,
      "learning_rate": 9.576264492201194e-07,
      "loss": 0.2472,
      "step": 47605
    },
    {
      "epoch": 0.8596407600806015,
      "grad_norm": 0.3347495198249817,
      "learning_rate": 9.564156515742705e-07,
      "loss": 0.3033,
      "step": 47610
    },
    {
      "epoch": 0.8597310395135022,
      "grad_norm": 0.41779398918151855,
      "learning_rate": 9.552055814190308e-07,
      "loss": 0.2663,
      "step": 47615
    },
    {
      "epoch": 0.8598213189464029,
      "grad_norm": 0.3597933351993561,
      "learning_rate": 9.53996238851741e-07,
      "loss": 0.2986,
      "step": 47620
    },
    {
      "epoch": 0.8599115983793036,
      "grad_norm": 0.8659226894378662,
      "learning_rate": 9.527876239696798e-07,
      "loss": 0.3131,
      "step": 47625
    },
    {
      "epoch": 0.8600018778122044,
      "grad_norm": 0.8438224792480469,
      "learning_rate": 9.515797368700764e-07,
      "loss": 0.2146,
      "step": 47630
    },
    {
      "epoch": 0.860092157245105,
      "grad_norm": 0.5149122476577759,
      "learning_rate": 9.503725776500927e-07,
      "loss": 0.313,
      "step": 47635
    },
    {
      "epoch": 0.8601824366780058,
      "grad_norm": 0.5562651753425598,
      "learning_rate": 9.491661464068413e-07,
      "loss": 0.1861,
      "step": 47640
    },
    {
      "epoch": 0.8602727161109065,
      "grad_norm": 0.39613303542137146,
      "learning_rate": 9.479604432373657e-07,
      "loss": 0.2227,
      "step": 47645
    },
    {
      "epoch": 0.8603629955438072,
      "grad_norm": 0.42106348276138306,
      "learning_rate": 9.467554682386593e-07,
      "loss": 0.1886,
      "step": 47650
    },
    {
      "epoch": 0.8604532749767079,
      "grad_norm": 0.620561420917511,
      "learning_rate": 9.455512215076501e-07,
      "loss": 0.378,
      "step": 47655
    },
    {
      "epoch": 0.8605435544096086,
      "grad_norm": 0.2884880304336548,
      "learning_rate": 9.443477031412129e-07,
      "loss": 0.2732,
      "step": 47660
    },
    {
      "epoch": 0.8606338338425094,
      "grad_norm": 0.2973548173904419,
      "learning_rate": 9.431449132361615e-07,
      "loss": 0.291,
      "step": 47665
    },
    {
      "epoch": 0.8607241132754101,
      "grad_norm": 0.65578693151474,
      "learning_rate": 9.419428518892548e-07,
      "loss": 0.295,
      "step": 47670
    },
    {
      "epoch": 0.8608143927083107,
      "grad_norm": 0.49643024802207947,
      "learning_rate": 9.407415191971847e-07,
      "loss": 0.1978,
      "step": 47675
    },
    {
      "epoch": 0.8609046721412115,
      "grad_norm": 0.29305776953697205,
      "learning_rate": 9.395409152565948e-07,
      "loss": 0.2002,
      "step": 47680
    },
    {
      "epoch": 0.8609949515741122,
      "grad_norm": 0.35219839215278625,
      "learning_rate": 9.383410401640602e-07,
      "loss": 0.1997,
      "step": 47685
    },
    {
      "epoch": 0.861085231007013,
      "grad_norm": 0.5114193558692932,
      "learning_rate": 9.371418940161059e-07,
      "loss": 0.1862,
      "step": 47690
    },
    {
      "epoch": 0.8611755104399136,
      "grad_norm": 0.7070506811141968,
      "learning_rate": 9.359434769091913e-07,
      "loss": 0.1949,
      "step": 47695
    },
    {
      "epoch": 0.8612657898728143,
      "grad_norm": 0.7334641814231873,
      "learning_rate": 9.347457889397215e-07,
      "loss": 0.2913,
      "step": 47700
    },
    {
      "epoch": 0.8613560693057151,
      "grad_norm": 0.42316699028015137,
      "learning_rate": 9.335488302040452e-07,
      "loss": 0.2535,
      "step": 47705
    },
    {
      "epoch": 0.8614463487386158,
      "grad_norm": 0.7839803099632263,
      "learning_rate": 9.32352600798443e-07,
      "loss": 0.1816,
      "step": 47710
    },
    {
      "epoch": 0.8615366281715164,
      "grad_norm": 0.41461873054504395,
      "learning_rate": 9.311571008191478e-07,
      "loss": 0.2394,
      "step": 47715
    },
    {
      "epoch": 0.8616269076044172,
      "grad_norm": 0.6121299862861633,
      "learning_rate": 9.299623303623262e-07,
      "loss": 0.2592,
      "step": 47720
    },
    {
      "epoch": 0.8617171870373179,
      "grad_norm": 0.4134901165962219,
      "learning_rate": 9.287682895240879e-07,
      "loss": 0.2361,
      "step": 47725
    },
    {
      "epoch": 0.8618074664702187,
      "grad_norm": 0.33365491032600403,
      "learning_rate": 9.275749784004884e-07,
      "loss": 0.2001,
      "step": 47730
    },
    {
      "epoch": 0.8618977459031193,
      "grad_norm": 0.5843504071235657,
      "learning_rate": 9.263823970875196e-07,
      "loss": 0.3159,
      "step": 47735
    },
    {
      "epoch": 0.86198802533602,
      "grad_norm": 0.4002785384654999,
      "learning_rate": 9.251905456811139e-07,
      "loss": 0.1652,
      "step": 47740
    },
    {
      "epoch": 0.8620783047689208,
      "grad_norm": 0.6355798840522766,
      "learning_rate": 9.239994242771499e-07,
      "loss": 0.205,
      "step": 47745
    },
    {
      "epoch": 0.8621685842018215,
      "grad_norm": 0.4560568034648895,
      "learning_rate": 9.228090329714423e-07,
      "loss": 0.2141,
      "step": 47750
    },
    {
      "epoch": 0.8622588636347222,
      "grad_norm": 0.5634968876838684,
      "learning_rate": 9.21619371859751e-07,
      "loss": 0.2183,
      "step": 47755
    },
    {
      "epoch": 0.8623491430676229,
      "grad_norm": 0.5417904853820801,
      "learning_rate": 9.20430441037774e-07,
      "loss": 0.1852,
      "step": 47760
    },
    {
      "epoch": 0.8624394225005236,
      "grad_norm": 0.18929065763950348,
      "learning_rate": 9.192422406011536e-07,
      "loss": 0.1941,
      "step": 47765
    },
    {
      "epoch": 0.8625297019334244,
      "grad_norm": 0.8262953758239746,
      "learning_rate": 9.180547706454712e-07,
      "loss": 0.1594,
      "step": 47770
    },
    {
      "epoch": 0.862619981366325,
      "grad_norm": 0.4193209707736969,
      "learning_rate": 9.168680312662515e-07,
      "loss": 0.2705,
      "step": 47775
    },
    {
      "epoch": 0.8627102607992257,
      "grad_norm": 0.45258623361587524,
      "learning_rate": 9.156820225589569e-07,
      "loss": 0.1757,
      "step": 47780
    },
    {
      "epoch": 0.8628005402321265,
      "grad_norm": 0.4728344678878784,
      "learning_rate": 9.144967446189956e-07,
      "loss": 0.2866,
      "step": 47785
    },
    {
      "epoch": 0.8628908196650272,
      "grad_norm": 0.5930164456367493,
      "learning_rate": 9.133121975417125e-07,
      "loss": 0.2212,
      "step": 47790
    },
    {
      "epoch": 0.8629810990979279,
      "grad_norm": 0.47831207513809204,
      "learning_rate": 9.121283814223946e-07,
      "loss": 0.2075,
      "step": 47795
    },
    {
      "epoch": 0.8630713785308286,
      "grad_norm": 0.6171165108680725,
      "learning_rate": 9.10945296356277e-07,
      "loss": 0.2753,
      "step": 47800
    },
    {
      "epoch": 0.8631616579637293,
      "grad_norm": 0.31840503215789795,
      "learning_rate": 9.097629424385235e-07,
      "loss": 0.1894,
      "step": 47805
    },
    {
      "epoch": 0.8632519373966301,
      "grad_norm": 0.717898428440094,
      "learning_rate": 9.085813197642502e-07,
      "loss": 0.1638,
      "step": 47810
    },
    {
      "epoch": 0.8633422168295307,
      "grad_norm": 0.4794361889362335,
      "learning_rate": 9.074004284285076e-07,
      "loss": 0.2876,
      "step": 47815
    },
    {
      "epoch": 0.8634324962624315,
      "grad_norm": 0.5447631478309631,
      "learning_rate": 9.062202685262933e-07,
      "loss": 0.2605,
      "step": 47820
    },
    {
      "epoch": 0.8635227756953322,
      "grad_norm": 0.32787272334098816,
      "learning_rate": 9.050408401525379e-07,
      "loss": 0.1382,
      "step": 47825
    },
    {
      "epoch": 0.8636130551282329,
      "grad_norm": 0.3508855700492859,
      "learning_rate": 9.038621434021199e-07,
      "loss": 0.2211,
      "step": 47830
    },
    {
      "epoch": 0.8637033345611336,
      "grad_norm": 0.3971794545650482,
      "learning_rate": 9.02684178369857e-07,
      "loss": 0.1168,
      "step": 47835
    },
    {
      "epoch": 0.8637936139940343,
      "grad_norm": 0.3002808690071106,
      "learning_rate": 9.015069451505109e-07,
      "loss": 0.2017,
      "step": 47840
    },
    {
      "epoch": 0.8638838934269351,
      "grad_norm": 0.3641931712627411,
      "learning_rate": 9.00330443838776e-07,
      "loss": 0.2921,
      "step": 47845
    },
    {
      "epoch": 0.8639741728598358,
      "grad_norm": 0.8863463401794434,
      "learning_rate": 8.991546745292979e-07,
      "loss": 0.2159,
      "step": 47850
    },
    {
      "epoch": 0.8640644522927364,
      "grad_norm": 0.6085312366485596,
      "learning_rate": 8.97979637316655e-07,
      "loss": 0.2143,
      "step": 47855
    },
    {
      "epoch": 0.8641547317256372,
      "grad_norm": 0.4391411542892456,
      "learning_rate": 8.968053322953718e-07,
      "loss": 0.2538,
      "step": 47860
    },
    {
      "epoch": 0.8642450111585379,
      "grad_norm": 0.4850060045719147,
      "learning_rate": 8.956317595599129e-07,
      "loss": 0.2463,
      "step": 47865
    },
    {
      "epoch": 0.8643352905914387,
      "grad_norm": 0.5466737151145935,
      "learning_rate": 8.94458919204686e-07,
      "loss": 0.1952,
      "step": 47870
    },
    {
      "epoch": 0.8644255700243393,
      "grad_norm": 0.29820647835731506,
      "learning_rate": 8.932868113240334e-07,
      "loss": 0.2013,
      "step": 47875
    },
    {
      "epoch": 0.86451584945724,
      "grad_norm": 0.585169792175293,
      "learning_rate": 8.921154360122464e-07,
      "loss": 0.1738,
      "step": 47880
    },
    {
      "epoch": 0.8646061288901408,
      "grad_norm": 0.8074939846992493,
      "learning_rate": 8.909447933635506e-07,
      "loss": 0.1304,
      "step": 47885
    },
    {
      "epoch": 0.8646964083230415,
      "grad_norm": 0.280657023191452,
      "learning_rate": 8.897748834721187e-07,
      "loss": 0.3266,
      "step": 47890
    },
    {
      "epoch": 0.8647866877559421,
      "grad_norm": 0.5078336000442505,
      "learning_rate": 8.886057064320574e-07,
      "loss": 0.1204,
      "step": 47895
    },
    {
      "epoch": 0.8648769671888429,
      "grad_norm": 0.4391881823539734,
      "learning_rate": 8.874372623374206e-07,
      "loss": 0.2035,
      "step": 47900
    },
    {
      "epoch": 0.8649672466217436,
      "grad_norm": 0.3612831234931946,
      "learning_rate": 8.862695512822039e-07,
      "loss": 0.2019,
      "step": 47905
    },
    {
      "epoch": 0.8650575260546444,
      "grad_norm": 0.2005240023136139,
      "learning_rate": 8.851025733603358e-07,
      "loss": 0.1793,
      "step": 47910
    },
    {
      "epoch": 0.865147805487545,
      "grad_norm": 0.3247433006763458,
      "learning_rate": 8.839363286656965e-07,
      "loss": 0.1645,
      "step": 47915
    },
    {
      "epoch": 0.8652380849204457,
      "grad_norm": 0.23706485331058502,
      "learning_rate": 8.827708172920968e-07,
      "loss": 0.2286,
      "step": 47920
    },
    {
      "epoch": 0.8653283643533465,
      "grad_norm": 0.42843496799468994,
      "learning_rate": 8.81606039333297e-07,
      "loss": 0.245,
      "step": 47925
    },
    {
      "epoch": 0.8654186437862472,
      "grad_norm": 0.6373015642166138,
      "learning_rate": 8.804419948829935e-07,
      "loss": 0.2401,
      "step": 47930
    },
    {
      "epoch": 0.8655089232191479,
      "grad_norm": 0.3249672055244446,
      "learning_rate": 8.792786840348289e-07,
      "loss": 0.1525,
      "step": 47935
    },
    {
      "epoch": 0.8655992026520486,
      "grad_norm": 0.25594303011894226,
      "learning_rate": 8.781161068823774e-07,
      "loss": 0.2066,
      "step": 47940
    },
    {
      "epoch": 0.8656894820849493,
      "grad_norm": 0.8266198635101318,
      "learning_rate": 8.769542635191641e-07,
      "loss": 0.2451,
      "step": 47945
    },
    {
      "epoch": 0.8657797615178501,
      "grad_norm": 0.3521641194820404,
      "learning_rate": 8.757931540386478e-07,
      "loss": 0.1713,
      "step": 47950
    },
    {
      "epoch": 0.8658700409507508,
      "grad_norm": 0.4017597436904907,
      "learning_rate": 8.746327785342357e-07,
      "loss": 0.2574,
      "step": 47955
    },
    {
      "epoch": 0.8659603203836514,
      "grad_norm": 0.4761274456977844,
      "learning_rate": 8.734731370992633e-07,
      "loss": 0.3343,
      "step": 47960
    },
    {
      "epoch": 0.8660505998165522,
      "grad_norm": 0.5980884432792664,
      "learning_rate": 8.72314229827026e-07,
      "loss": 0.2553,
      "step": 47965
    },
    {
      "epoch": 0.8661408792494529,
      "grad_norm": 0.6012890934944153,
      "learning_rate": 8.711560568107414e-07,
      "loss": 0.1958,
      "step": 47970
    },
    {
      "epoch": 0.8662311586823537,
      "grad_norm": 0.5191618204116821,
      "learning_rate": 8.699986181435815e-07,
      "loss": 0.2341,
      "step": 47975
    },
    {
      "epoch": 0.8663214381152543,
      "grad_norm": 0.4359510540962219,
      "learning_rate": 8.68841913918651e-07,
      "loss": 0.2036,
      "step": 47980
    },
    {
      "epoch": 0.866411717548155,
      "grad_norm": 0.32875490188598633,
      "learning_rate": 8.676859442289964e-07,
      "loss": 0.1804,
      "step": 47985
    },
    {
      "epoch": 0.8665019969810558,
      "grad_norm": 0.4370461106300354,
      "learning_rate": 8.66530709167609e-07,
      "loss": 0.1166,
      "step": 47990
    },
    {
      "epoch": 0.8665922764139565,
      "grad_norm": 0.15905004739761353,
      "learning_rate": 8.653762088274198e-07,
      "loss": 0.2479,
      "step": 47995
    },
    {
      "epoch": 0.8666825558468572,
      "grad_norm": 0.32449597120285034,
      "learning_rate": 8.642224433013014e-07,
      "loss": 0.2932,
      "step": 48000
    },
    {
      "epoch": 0.8667728352797579,
      "grad_norm": 0.26115724444389343,
      "learning_rate": 8.630694126820616e-07,
      "loss": 0.2006,
      "step": 48005
    },
    {
      "epoch": 0.8668631147126586,
      "grad_norm": 0.693935751914978,
      "learning_rate": 8.619171170624585e-07,
      "loss": 0.1629,
      "step": 48010
    },
    {
      "epoch": 0.8669533941455594,
      "grad_norm": 0.6370662450790405,
      "learning_rate": 8.607655565351803e-07,
      "loss": 0.2614,
      "step": 48015
    },
    {
      "epoch": 0.86704367357846,
      "grad_norm": 0.2275598645210266,
      "learning_rate": 8.59614731192866e-07,
      "loss": 0.1091,
      "step": 48020
    },
    {
      "epoch": 0.8671339530113608,
      "grad_norm": 0.2575340270996094,
      "learning_rate": 8.584646411280872e-07,
      "loss": 0.1735,
      "step": 48025
    },
    {
      "epoch": 0.8672242324442615,
      "grad_norm": 0.40870359539985657,
      "learning_rate": 8.573152864333645e-07,
      "loss": 0.2322,
      "step": 48030
    },
    {
      "epoch": 0.8673145118771622,
      "grad_norm": 0.33703476190567017,
      "learning_rate": 8.561666672011526e-07,
      "loss": 0.1138,
      "step": 48035
    },
    {
      "epoch": 0.8674047913100629,
      "grad_norm": 0.45342332124710083,
      "learning_rate": 8.550187835238521e-07,
      "loss": 0.2634,
      "step": 48040
    },
    {
      "epoch": 0.8674950707429636,
      "grad_norm": 0.39678117632865906,
      "learning_rate": 8.538716354937982e-07,
      "loss": 0.3507,
      "step": 48045
    },
    {
      "epoch": 0.8675853501758644,
      "grad_norm": 0.4366861581802368,
      "learning_rate": 8.527252232032734e-07,
      "loss": 0.2137,
      "step": 48050
    },
    {
      "epoch": 0.8676756296087651,
      "grad_norm": 0.3188689351081848,
      "learning_rate": 8.515795467444965e-07,
      "loss": 0.1916,
      "step": 48055
    },
    {
      "epoch": 0.8677659090416657,
      "grad_norm": 0.6067432165145874,
      "learning_rate": 8.504346062096291e-07,
      "loss": 0.2341,
      "step": 48060
    },
    {
      "epoch": 0.8678561884745665,
      "grad_norm": 0.47358977794647217,
      "learning_rate": 8.492904016907744e-07,
      "loss": 0.1646,
      "step": 48065
    },
    {
      "epoch": 0.8679464679074672,
      "grad_norm": 0.33584290742874146,
      "learning_rate": 8.481469332799763e-07,
      "loss": 0.2262,
      "step": 48070
    },
    {
      "epoch": 0.868036747340368,
      "grad_norm": 0.6119019389152527,
      "learning_rate": 8.470042010692159e-07,
      "loss": 0.1833,
      "step": 48075
    },
    {
      "epoch": 0.8681270267732686,
      "grad_norm": 0.4626547396183014,
      "learning_rate": 8.458622051504184e-07,
      "loss": 0.2844,
      "step": 48080
    },
    {
      "epoch": 0.8682173062061693,
      "grad_norm": 0.35874173045158386,
      "learning_rate": 8.447209456154503e-07,
      "loss": 0.2262,
      "step": 48085
    },
    {
      "epoch": 0.8683075856390701,
      "grad_norm": 0.36015933752059937,
      "learning_rate": 8.435804225561162e-07,
      "loss": 0.2485,
      "step": 48090
    },
    {
      "epoch": 0.8683978650719708,
      "grad_norm": 0.515663743019104,
      "learning_rate": 8.424406360641624e-07,
      "loss": 0.21,
      "step": 48095
    },
    {
      "epoch": 0.8684881445048714,
      "grad_norm": 0.47348344326019287,
      "learning_rate": 8.41301586231279e-07,
      "loss": 0.2061,
      "step": 48100
    },
    {
      "epoch": 0.8685784239377722,
      "grad_norm": 0.4087829887866974,
      "learning_rate": 8.401632731490927e-07,
      "loss": 0.2398,
      "step": 48105
    },
    {
      "epoch": 0.8686687033706729,
      "grad_norm": 0.7120229005813599,
      "learning_rate": 8.390256969091725e-07,
      "loss": 0.2717,
      "step": 48110
    },
    {
      "epoch": 0.8687589828035737,
      "grad_norm": 0.8429829478263855,
      "learning_rate": 8.378888576030287e-07,
      "loss": 0.1942,
      "step": 48115
    },
    {
      "epoch": 0.8688492622364743,
      "grad_norm": 0.3972378075122833,
      "learning_rate": 8.3675275532211e-07,
      "loss": 0.2245,
      "step": 48120
    },
    {
      "epoch": 0.868939541669375,
      "grad_norm": 0.515321671962738,
      "learning_rate": 8.356173901578091e-07,
      "loss": 0.2349,
      "step": 48125
    },
    {
      "epoch": 0.8690298211022758,
      "grad_norm": 1.08586847782135,
      "learning_rate": 8.344827622014584e-07,
      "loss": 0.2143,
      "step": 48130
    },
    {
      "epoch": 0.8691201005351765,
      "grad_norm": 0.41314950585365295,
      "learning_rate": 8.333488715443305e-07,
      "loss": 0.3164,
      "step": 48135
    },
    {
      "epoch": 0.8692103799680772,
      "grad_norm": 0.5639352798461914,
      "learning_rate": 8.322157182776358e-07,
      "loss": 0.1827,
      "step": 48140
    },
    {
      "epoch": 0.8693006594009779,
      "grad_norm": 0.6171694993972778,
      "learning_rate": 8.310833024925325e-07,
      "loss": 0.159,
      "step": 48145
    },
    {
      "epoch": 0.8693909388338786,
      "grad_norm": 0.5867541432380676,
      "learning_rate": 8.29951624280112e-07,
      "loss": 0.1949,
      "step": 48150
    },
    {
      "epoch": 0.8694812182667794,
      "grad_norm": 0.44068971276283264,
      "learning_rate": 8.288206837314116e-07,
      "loss": 0.2258,
      "step": 48155
    },
    {
      "epoch": 0.86957149769968,
      "grad_norm": 0.42447203397750854,
      "learning_rate": 8.276904809374031e-07,
      "loss": 0.2565,
      "step": 48160
    },
    {
      "epoch": 0.8696617771325807,
      "grad_norm": 0.4244731664657593,
      "learning_rate": 8.265610159890092e-07,
      "loss": 0.2274,
      "step": 48165
    },
    {
      "epoch": 0.8697520565654815,
      "grad_norm": 0.6324188113212585,
      "learning_rate": 8.254322889770849e-07,
      "loss": 0.2529,
      "step": 48170
    },
    {
      "epoch": 0.8698423359983822,
      "grad_norm": 0.4434747099876404,
      "learning_rate": 8.243042999924255e-07,
      "loss": 0.2762,
      "step": 48175
    },
    {
      "epoch": 0.8699326154312829,
      "grad_norm": 0.5678615570068359,
      "learning_rate": 8.231770491257729e-07,
      "loss": 0.1874,
      "step": 48180
    },
    {
      "epoch": 0.8700228948641836,
      "grad_norm": 0.44842442870140076,
      "learning_rate": 8.220505364678033e-07,
      "loss": 0.2059,
      "step": 48185
    },
    {
      "epoch": 0.8701131742970843,
      "grad_norm": 0.7900363206863403,
      "learning_rate": 8.209247621091376e-07,
      "loss": 0.2184,
      "step": 48190
    },
    {
      "epoch": 0.8702034537299851,
      "grad_norm": 0.3303584158420563,
      "learning_rate": 8.197997261403368e-07,
      "loss": 0.2753,
      "step": 48195
    },
    {
      "epoch": 0.8702937331628857,
      "grad_norm": 0.4439292550086975,
      "learning_rate": 8.186754286519027e-07,
      "loss": 0.2112,
      "step": 48200
    },
    {
      "epoch": 0.8703840125957865,
      "grad_norm": 0.3900735378265381,
      "learning_rate": 8.175518697342743e-07,
      "loss": 0.2861,
      "step": 48205
    },
    {
      "epoch": 0.8704742920286872,
      "grad_norm": 0.643632173538208,
      "learning_rate": 8.164290494778371e-07,
      "loss": 0.2954,
      "step": 48210
    },
    {
      "epoch": 0.8705645714615879,
      "grad_norm": 0.5918735861778259,
      "learning_rate": 8.153069679729098e-07,
      "loss": 0.3003,
      "step": 48215
    },
    {
      "epoch": 0.8706548508944886,
      "grad_norm": 0.40243905782699585,
      "learning_rate": 8.141856253097602e-07,
      "loss": 0.175,
      "step": 48220
    },
    {
      "epoch": 0.8707451303273893,
      "grad_norm": 2.0269851684570312,
      "learning_rate": 8.130650215785852e-07,
      "loss": 0.3366,
      "step": 48225
    },
    {
      "epoch": 0.8708354097602901,
      "grad_norm": 0.38487523794174194,
      "learning_rate": 8.11945156869538e-07,
      "loss": 0.2189,
      "step": 48230
    },
    {
      "epoch": 0.8709256891931908,
      "grad_norm": 0.34521669149398804,
      "learning_rate": 8.108260312726968e-07,
      "loss": 0.1645,
      "step": 48235
    },
    {
      "epoch": 0.8710159686260914,
      "grad_norm": 0.4212098717689514,
      "learning_rate": 8.097076448780916e-07,
      "loss": 0.1905,
      "step": 48240
    },
    {
      "epoch": 0.8711062480589922,
      "grad_norm": 0.5981873869895935,
      "learning_rate": 8.08589997775685e-07,
      "loss": 0.1878,
      "step": 48245
    },
    {
      "epoch": 0.8711965274918929,
      "grad_norm": 0.5543757081031799,
      "learning_rate": 8.074730900553873e-07,
      "loss": 0.2727,
      "step": 48250
    },
    {
      "epoch": 0.8712868069247937,
      "grad_norm": 0.4017278254032135,
      "learning_rate": 8.063569218070411e-07,
      "loss": 0.2687,
      "step": 48255
    },
    {
      "epoch": 0.8713770863576943,
      "grad_norm": 0.8131873607635498,
      "learning_rate": 8.052414931204366e-07,
      "loss": 0.2013,
      "step": 48260
    },
    {
      "epoch": 0.871467365790595,
      "grad_norm": 0.784367561340332,
      "learning_rate": 8.041268040853023e-07,
      "loss": 0.2989,
      "step": 48265
    },
    {
      "epoch": 0.8715576452234958,
      "grad_norm": 0.24246563017368317,
      "learning_rate": 8.030128547913052e-07,
      "loss": 0.3262,
      "step": 48270
    },
    {
      "epoch": 0.8716479246563965,
      "grad_norm": 0.36412227153778076,
      "learning_rate": 8.018996453280581e-07,
      "loss": 0.1953,
      "step": 48275
    },
    {
      "epoch": 0.8717382040892971,
      "grad_norm": 0.482963889837265,
      "learning_rate": 8.007871757851049e-07,
      "loss": 0.2645,
      "step": 48280
    },
    {
      "epoch": 0.8718284835221979,
      "grad_norm": 0.2874138057231903,
      "learning_rate": 7.996754462519408e-07,
      "loss": 0.1705,
      "step": 48285
    },
    {
      "epoch": 0.8719187629550986,
      "grad_norm": 0.36590391397476196,
      "learning_rate": 7.985644568179906e-07,
      "loss": 0.1881,
      "step": 48290
    },
    {
      "epoch": 0.8720090423879994,
      "grad_norm": 0.5138126015663147,
      "learning_rate": 7.974542075726322e-07,
      "loss": 0.1993,
      "step": 48295
    },
    {
      "epoch": 0.8720993218209,
      "grad_norm": 0.6720679402351379,
      "learning_rate": 7.963446986051715e-07,
      "loss": 0.2382,
      "step": 48300
    },
    {
      "epoch": 0.8721896012538007,
      "grad_norm": 0.4910641312599182,
      "learning_rate": 7.952359300048662e-07,
      "loss": 0.2453,
      "step": 48305
    },
    {
      "epoch": 0.8722798806867015,
      "grad_norm": 0.4402197003364563,
      "learning_rate": 7.941279018609027e-07,
      "loss": 0.2395,
      "step": 48310
    },
    {
      "epoch": 0.8723701601196022,
      "grad_norm": 0.41484811902046204,
      "learning_rate": 7.930206142624175e-07,
      "loss": 0.2043,
      "step": 48315
    },
    {
      "epoch": 0.8724604395525029,
      "grad_norm": 0.3428449034690857,
      "learning_rate": 7.919140672984815e-07,
      "loss": 0.2656,
      "step": 48320
    },
    {
      "epoch": 0.8725507189854036,
      "grad_norm": 0.5132912993431091,
      "learning_rate": 7.908082610581091e-07,
      "loss": 0.3005,
      "step": 48325
    },
    {
      "epoch": 0.8726409984183043,
      "grad_norm": 0.44725364446640015,
      "learning_rate": 7.897031956302547e-07,
      "loss": 0.1681,
      "step": 48330
    },
    {
      "epoch": 0.8727312778512051,
      "grad_norm": 0.46623513102531433,
      "learning_rate": 7.885988711038162e-07,
      "loss": 0.181,
      "step": 48335
    },
    {
      "epoch": 0.8728215572841057,
      "grad_norm": 0.30655795335769653,
      "learning_rate": 7.874952875676212e-07,
      "loss": 0.1926,
      "step": 48340
    },
    {
      "epoch": 0.8729118367170065,
      "grad_norm": 0.543652355670929,
      "learning_rate": 7.863924451104521e-07,
      "loss": 0.25,
      "step": 48345
    },
    {
      "epoch": 0.8730021161499072,
      "grad_norm": 0.5199833512306213,
      "learning_rate": 7.852903438210213e-07,
      "loss": 0.2849,
      "step": 48350
    },
    {
      "epoch": 0.8730923955828079,
      "grad_norm": 0.37067753076553345,
      "learning_rate": 7.841889837879812e-07,
      "loss": 0.1712,
      "step": 48355
    },
    {
      "epoch": 0.8731826750157087,
      "grad_norm": 0.3907431364059448,
      "learning_rate": 7.830883650999355e-07,
      "loss": 0.2518,
      "step": 48360
    },
    {
      "epoch": 0.8732729544486093,
      "grad_norm": 0.2768262028694153,
      "learning_rate": 7.819884878454154e-07,
      "loss": 0.2396,
      "step": 48365
    },
    {
      "epoch": 0.87336323388151,
      "grad_norm": 0.7335043549537659,
      "learning_rate": 7.808893521129013e-07,
      "loss": 0.1918,
      "step": 48370
    },
    {
      "epoch": 0.8734535133144108,
      "grad_norm": 0.5029274821281433,
      "learning_rate": 7.797909579908092e-07,
      "loss": 0.2167,
      "step": 48375
    },
    {
      "epoch": 0.8735437927473115,
      "grad_norm": 0.6301203966140747,
      "learning_rate": 7.786933055674973e-07,
      "loss": 0.2502,
      "step": 48380
    },
    {
      "epoch": 0.8736340721802122,
      "grad_norm": 0.35355889797210693,
      "learning_rate": 7.775963949312626e-07,
      "loss": 0.2063,
      "step": 48385
    },
    {
      "epoch": 0.8737243516131129,
      "grad_norm": 0.8525371551513672,
      "learning_rate": 7.765002261703448e-07,
      "loss": 0.21,
      "step": 48390
    },
    {
      "epoch": 0.8738146310460136,
      "grad_norm": 0.2836104929447174,
      "learning_rate": 7.754047993729219e-07,
      "loss": 0.1717,
      "step": 48395
    },
    {
      "epoch": 0.8739049104789144,
      "grad_norm": 0.3038562834262848,
      "learning_rate": 7.743101146271159e-07,
      "loss": 0.2272,
      "step": 48400
    },
    {
      "epoch": 0.873995189911815,
      "grad_norm": 0.9222632050514221,
      "learning_rate": 7.732161720209818e-07,
      "loss": 0.2373,
      "step": 48405
    },
    {
      "epoch": 0.8740854693447158,
      "grad_norm": 0.2853277623653412,
      "learning_rate": 7.721229716425227e-07,
      "loss": 0.2101,
      "step": 48410
    },
    {
      "epoch": 0.8741757487776165,
      "grad_norm": 0.6378380060195923,
      "learning_rate": 7.71030513579677e-07,
      "loss": 0.143,
      "step": 48415
    },
    {
      "epoch": 0.8742660282105172,
      "grad_norm": 0.589065670967102,
      "learning_rate": 7.699387979203243e-07,
      "loss": 0.2007,
      "step": 48420
    },
    {
      "epoch": 0.8743563076434179,
      "grad_norm": 0.4162118434906006,
      "learning_rate": 7.688478247522868e-07,
      "loss": 0.1579,
      "step": 48425
    },
    {
      "epoch": 0.8744465870763186,
      "grad_norm": 0.3636179268360138,
      "learning_rate": 7.677575941633264e-07,
      "loss": 0.2605,
      "step": 48430
    },
    {
      "epoch": 0.8745368665092194,
      "grad_norm": 0.2778651714324951,
      "learning_rate": 7.666681062411408e-07,
      "loss": 0.2255,
      "step": 48435
    },
    {
      "epoch": 0.8746271459421201,
      "grad_norm": 0.1925775110721588,
      "learning_rate": 7.655793610733753e-07,
      "loss": 0.2077,
      "step": 48440
    },
    {
      "epoch": 0.8747174253750207,
      "grad_norm": 0.21382668614387512,
      "learning_rate": 7.644913587476088e-07,
      "loss": 0.1934,
      "step": 48445
    },
    {
      "epoch": 0.8748077048079215,
      "grad_norm": 0.42845526337623596,
      "learning_rate": 7.634040993513625e-07,
      "loss": 0.2532,
      "step": 48450
    },
    {
      "epoch": 0.8748979842408222,
      "grad_norm": 0.5847483277320862,
      "learning_rate": 7.623175829720997e-07,
      "loss": 0.1947,
      "step": 48455
    },
    {
      "epoch": 0.874988263673723,
      "grad_norm": 0.5122021436691284,
      "learning_rate": 7.612318096972237e-07,
      "loss": 0.2392,
      "step": 48460
    },
    {
      "epoch": 0.8750785431066236,
      "grad_norm": 0.24058891832828522,
      "learning_rate": 7.601467796140771e-07,
      "loss": 0.2839,
      "step": 48465
    },
    {
      "epoch": 0.8751688225395243,
      "grad_norm": 0.38757041096687317,
      "learning_rate": 7.59062492809941e-07,
      "loss": 0.1499,
      "step": 48470
    },
    {
      "epoch": 0.8752591019724251,
      "grad_norm": 0.6410506963729858,
      "learning_rate": 7.579789493720402e-07,
      "loss": 0.2633,
      "step": 48475
    },
    {
      "epoch": 0.8753493814053258,
      "grad_norm": 0.38164740800857544,
      "learning_rate": 7.56896149387536e-07,
      "loss": 0.2471,
      "step": 48480
    },
    {
      "epoch": 0.8754396608382264,
      "grad_norm": 0.46518123149871826,
      "learning_rate": 7.55814092943532e-07,
      "loss": 0.1657,
      "step": 48485
    },
    {
      "epoch": 0.8755299402711272,
      "grad_norm": 0.18978314101696014,
      "learning_rate": 7.54732780127072e-07,
      "loss": 0.1636,
      "step": 48490
    },
    {
      "epoch": 0.8756202197040279,
      "grad_norm": 0.6388356685638428,
      "learning_rate": 7.536522110251432e-07,
      "loss": 0.2383,
      "step": 48495
    },
    {
      "epoch": 0.8757104991369287,
      "grad_norm": 0.3540574014186859,
      "learning_rate": 7.525723857246636e-07,
      "loss": 0.2615,
      "step": 48500
    },
    {
      "epoch": 0.8758007785698293,
      "grad_norm": 0.47638097405433655,
      "learning_rate": 7.514933043125027e-07,
      "loss": 0.1564,
      "step": 48505
    },
    {
      "epoch": 0.87589105800273,
      "grad_norm": 0.25722935795783997,
      "learning_rate": 7.504149668754612e-07,
      "loss": 0.2165,
      "step": 48510
    },
    {
      "epoch": 0.8759813374356308,
      "grad_norm": 0.1883164346218109,
      "learning_rate": 7.493373735002851e-07,
      "loss": 0.2019,
      "step": 48515
    },
    {
      "epoch": 0.8760716168685315,
      "grad_norm": 0.4336926341056824,
      "learning_rate": 7.482605242736573e-07,
      "loss": 0.1925,
      "step": 48520
    },
    {
      "epoch": 0.8761618963014322,
      "grad_norm": 0.24070729315280914,
      "learning_rate": 7.471844192822042e-07,
      "loss": 0.1779,
      "step": 48525
    },
    {
      "epoch": 0.8762521757343329,
      "grad_norm": 0.7458894848823547,
      "learning_rate": 7.461090586124897e-07,
      "loss": 0.1377,
      "step": 48530
    },
    {
      "epoch": 0.8763424551672336,
      "grad_norm": 0.2711096405982971,
      "learning_rate": 7.450344423510214e-07,
      "loss": 0.1506,
      "step": 48535
    },
    {
      "epoch": 0.8764327346001344,
      "grad_norm": 0.2870548665523529,
      "learning_rate": 7.439605705842412e-07,
      "loss": 0.2085,
      "step": 48540
    },
    {
      "epoch": 0.876523014033035,
      "grad_norm": 0.49196571111679077,
      "learning_rate": 7.428874433985333e-07,
      "loss": 0.2513,
      "step": 48545
    },
    {
      "epoch": 0.8766132934659358,
      "grad_norm": 0.4872879981994629,
      "learning_rate": 7.418150608802255e-07,
      "loss": 0.1888,
      "step": 48550
    },
    {
      "epoch": 0.8767035728988365,
      "grad_norm": 0.4504752457141876,
      "learning_rate": 7.40743423115583e-07,
      "loss": 0.2368,
      "step": 48555
    },
    {
      "epoch": 0.8767938523317372,
      "grad_norm": 0.6074302196502686,
      "learning_rate": 7.396725301908114e-07,
      "loss": 0.2713,
      "step": 48560
    },
    {
      "epoch": 0.8768841317646379,
      "grad_norm": 0.43019863963127136,
      "learning_rate": 7.386023821920552e-07,
      "loss": 0.27,
      "step": 48565
    },
    {
      "epoch": 0.8769744111975386,
      "grad_norm": 0.5306024551391602,
      "learning_rate": 7.375329792054009e-07,
      "loss": 0.1794,
      "step": 48570
    },
    {
      "epoch": 0.8770646906304393,
      "grad_norm": 0.5218394994735718,
      "learning_rate": 7.364643213168731e-07,
      "loss": 0.1527,
      "step": 48575
    },
    {
      "epoch": 0.8771549700633401,
      "grad_norm": 0.2932478189468384,
      "learning_rate": 7.353964086124388e-07,
      "loss": 0.3163,
      "step": 48580
    },
    {
      "epoch": 0.8772452494962407,
      "grad_norm": 0.4879951775074005,
      "learning_rate": 7.343292411780023e-07,
      "loss": 0.2522,
      "step": 48585
    },
    {
      "epoch": 0.8773355289291415,
      "grad_norm": 0.672153651714325,
      "learning_rate": 7.332628190994107e-07,
      "loss": 0.2887,
      "step": 48590
    },
    {
      "epoch": 0.8774258083620422,
      "grad_norm": 0.4504673480987549,
      "learning_rate": 7.321971424624497e-07,
      "loss": 0.2011,
      "step": 48595
    },
    {
      "epoch": 0.877516087794943,
      "grad_norm": 0.4887804090976715,
      "learning_rate": 7.311322113528463e-07,
      "loss": 0.2581,
      "step": 48600
    },
    {
      "epoch": 0.8776063672278436,
      "grad_norm": 0.2560054361820221,
      "learning_rate": 7.300680258562631e-07,
      "loss": 0.1815,
      "step": 48605
    },
    {
      "epoch": 0.8776966466607443,
      "grad_norm": 0.3547193109989166,
      "learning_rate": 7.290045860583106e-07,
      "loss": 0.2215,
      "step": 48610
    },
    {
      "epoch": 0.8777869260936451,
      "grad_norm": 0.524704098701477,
      "learning_rate": 7.279418920445314e-07,
      "loss": 0.2494,
      "step": 48615
    },
    {
      "epoch": 0.8778772055265458,
      "grad_norm": 0.3460308909416199,
      "learning_rate": 7.268799439004115e-07,
      "loss": 0.1966,
      "step": 48620
    },
    {
      "epoch": 0.8779674849594464,
      "grad_norm": 0.35949942469596863,
      "learning_rate": 7.258187417113804e-07,
      "loss": 0.2505,
      "step": 48625
    },
    {
      "epoch": 0.8780577643923472,
      "grad_norm": 0.3450348675251007,
      "learning_rate": 7.247582855627999e-07,
      "loss": 0.2108,
      "step": 48630
    },
    {
      "epoch": 0.8781480438252479,
      "grad_norm": 0.43432381749153137,
      "learning_rate": 7.236985755399794e-07,
      "loss": 0.2986,
      "step": 48635
    },
    {
      "epoch": 0.8782383232581487,
      "grad_norm": 0.45111775398254395,
      "learning_rate": 7.22639611728162e-07,
      "loss": 0.2615,
      "step": 48640
    },
    {
      "epoch": 0.8783286026910493,
      "grad_norm": 0.34187889099121094,
      "learning_rate": 7.215813942125361e-07,
      "loss": 0.2505,
      "step": 48645
    },
    {
      "epoch": 0.87841888212395,
      "grad_norm": 0.37597358226776123,
      "learning_rate": 7.205239230782246e-07,
      "loss": 0.2399,
      "step": 48650
    },
    {
      "epoch": 0.8785091615568508,
      "grad_norm": 0.5486959218978882,
      "learning_rate": 7.194671984102964e-07,
      "loss": 0.1959,
      "step": 48655
    },
    {
      "epoch": 0.8785994409897515,
      "grad_norm": 0.3595384657382965,
      "learning_rate": 7.184112202937543e-07,
      "loss": 0.1192,
      "step": 48660
    },
    {
      "epoch": 0.8786897204226521,
      "grad_norm": 0.3697947859764099,
      "learning_rate": 7.173559888135484e-07,
      "loss": 0.2608,
      "step": 48665
    },
    {
      "epoch": 0.8787799998555529,
      "grad_norm": 0.6005027294158936,
      "learning_rate": 7.163015040545607e-07,
      "loss": 0.2247,
      "step": 48670
    },
    {
      "epoch": 0.8788702792884536,
      "grad_norm": 0.472234308719635,
      "learning_rate": 7.152477661016189e-07,
      "loss": 0.2378,
      "step": 48675
    },
    {
      "epoch": 0.8789605587213544,
      "grad_norm": 0.9382433891296387,
      "learning_rate": 7.141947750394862e-07,
      "loss": 0.1628,
      "step": 48680
    },
    {
      "epoch": 0.879050838154255,
      "grad_norm": 0.37445417046546936,
      "learning_rate": 7.131425309528694e-07,
      "loss": 0.1783,
      "step": 48685
    },
    {
      "epoch": 0.8791411175871557,
      "grad_norm": 0.40765634179115295,
      "learning_rate": 7.120910339264142e-07,
      "loss": 0.1618,
      "step": 48690
    },
    {
      "epoch": 0.8792313970200565,
      "grad_norm": 0.5030569434165955,
      "learning_rate": 7.110402840447072e-07,
      "loss": 0.2147,
      "step": 48695
    },
    {
      "epoch": 0.8793216764529572,
      "grad_norm": 0.3929252326488495,
      "learning_rate": 7.09990281392271e-07,
      "loss": 0.232,
      "step": 48700
    },
    {
      "epoch": 0.8794119558858579,
      "grad_norm": 0.5757809281349182,
      "learning_rate": 7.089410260535723e-07,
      "loss": 0.264,
      "step": 48705
    },
    {
      "epoch": 0.8795022353187586,
      "grad_norm": 0.36466771364212036,
      "learning_rate": 7.078925181130147e-07,
      "loss": 0.1432,
      "step": 48710
    },
    {
      "epoch": 0.8795925147516593,
      "grad_norm": 0.6043946743011475,
      "learning_rate": 7.068447576549453e-07,
      "loss": 0.2092,
      "step": 48715
    },
    {
      "epoch": 0.8796827941845601,
      "grad_norm": 0.479689359664917,
      "learning_rate": 7.057977447636455e-07,
      "loss": 0.2223,
      "step": 48720
    },
    {
      "epoch": 0.8797730736174607,
      "grad_norm": 0.46360352635383606,
      "learning_rate": 7.047514795233423e-07,
      "loss": 0.2344,
      "step": 48725
    },
    {
      "epoch": 0.8798633530503615,
      "grad_norm": 0.7283722162246704,
      "learning_rate": 7.037059620182008e-07,
      "loss": 0.2391,
      "step": 48730
    },
    {
      "epoch": 0.8799536324832622,
      "grad_norm": 0.42849308252334595,
      "learning_rate": 7.026611923323224e-07,
      "loss": 0.3026,
      "step": 48735
    },
    {
      "epoch": 0.8800439119161629,
      "grad_norm": 0.5676857829093933,
      "learning_rate": 7.016171705497554e-07,
      "loss": 0.1848,
      "step": 48740
    },
    {
      "epoch": 0.8801341913490637,
      "grad_norm": 0.5181065201759338,
      "learning_rate": 7.005738967544784e-07,
      "loss": 0.1742,
      "step": 48745
    },
    {
      "epoch": 0.8802244707819643,
      "grad_norm": 0.5719236731529236,
      "learning_rate": 6.995313710304186e-07,
      "loss": 0.1925,
      "step": 48750
    },
    {
      "epoch": 0.880314750214865,
      "grad_norm": 0.5163266658782959,
      "learning_rate": 6.984895934614389e-07,
      "loss": 0.2706,
      "step": 48755
    },
    {
      "epoch": 0.8804050296477658,
      "grad_norm": 0.16679945588111877,
      "learning_rate": 6.974485641313444e-07,
      "loss": 0.1513,
      "step": 48760
    },
    {
      "epoch": 0.8804953090806665,
      "grad_norm": 0.4424380362033844,
      "learning_rate": 6.964082831238761e-07,
      "loss": 0.2493,
      "step": 48765
    },
    {
      "epoch": 0.8805855885135672,
      "grad_norm": 0.4516794681549072,
      "learning_rate": 6.95368750522718e-07,
      "loss": 0.2917,
      "step": 48770
    },
    {
      "epoch": 0.8806758679464679,
      "grad_norm": 0.3460313081741333,
      "learning_rate": 6.943299664114922e-07,
      "loss": 0.1721,
      "step": 48775
    },
    {
      "epoch": 0.8807661473793686,
      "grad_norm": 0.5408383011817932,
      "learning_rate": 6.932919308737629e-07,
      "loss": 0.2507,
      "step": 48780
    },
    {
      "epoch": 0.8808564268122694,
      "grad_norm": 0.32816874980926514,
      "learning_rate": 6.9225464399303e-07,
      "loss": 0.1181,
      "step": 48785
    },
    {
      "epoch": 0.88094670624517,
      "grad_norm": 0.24817796051502228,
      "learning_rate": 6.912181058527367e-07,
      "loss": 0.309,
      "step": 48790
    },
    {
      "epoch": 0.8810369856780708,
      "grad_norm": 0.5147075057029724,
      "learning_rate": 6.901823165362664e-07,
      "loss": 0.2264,
      "step": 48795
    },
    {
      "epoch": 0.8811272651109715,
      "grad_norm": 0.9248692989349365,
      "learning_rate": 6.891472761269413e-07,
      "loss": 0.244,
      "step": 48800
    },
    {
      "epoch": 0.8812175445438722,
      "grad_norm": 0.592526912689209,
      "learning_rate": 6.881129847080193e-07,
      "loss": 0.3732,
      "step": 48805
    },
    {
      "epoch": 0.8813078239767729,
      "grad_norm": 0.2904769778251648,
      "learning_rate": 6.870794423627047e-07,
      "loss": 0.248,
      "step": 48810
    },
    {
      "epoch": 0.8813981034096736,
      "grad_norm": 0.6007767915725708,
      "learning_rate": 6.860466491741368e-07,
      "loss": 0.2189,
      "step": 48815
    },
    {
      "epoch": 0.8814883828425744,
      "grad_norm": 0.7476814389228821,
      "learning_rate": 6.85014605225397e-07,
      "loss": 0.2788,
      "step": 48820
    },
    {
      "epoch": 0.8815786622754751,
      "grad_norm": 0.39375999569892883,
      "learning_rate": 6.839833105995064e-07,
      "loss": 0.1943,
      "step": 48825
    },
    {
      "epoch": 0.8816689417083757,
      "grad_norm": 0.6178081631660461,
      "learning_rate": 6.829527653794243e-07,
      "loss": 0.2376,
      "step": 48830
    },
    {
      "epoch": 0.8817592211412765,
      "grad_norm": 0.9571642279624939,
      "learning_rate": 6.819229696480511e-07,
      "loss": 0.2867,
      "step": 48835
    },
    {
      "epoch": 0.8818495005741772,
      "grad_norm": 0.4160026013851166,
      "learning_rate": 6.80893923488225e-07,
      "loss": 0.1502,
      "step": 48840
    },
    {
      "epoch": 0.881939780007078,
      "grad_norm": 0.3343513607978821,
      "learning_rate": 6.798656269827275e-07,
      "loss": 0.2608,
      "step": 48845
    },
    {
      "epoch": 0.8820300594399786,
      "grad_norm": 0.3635469973087311,
      "learning_rate": 6.788380802142758e-07,
      "loss": 0.1748,
      "step": 48850
    },
    {
      "epoch": 0.8821203388728793,
      "grad_norm": 0.1914890855550766,
      "learning_rate": 6.778112832655282e-07,
      "loss": 0.1421,
      "step": 48855
    },
    {
      "epoch": 0.8822106183057801,
      "grad_norm": 0.3439345061779022,
      "learning_rate": 6.767852362190841e-07,
      "loss": 0.1821,
      "step": 48860
    },
    {
      "epoch": 0.8823008977386808,
      "grad_norm": 0.5026911497116089,
      "learning_rate": 6.757599391574831e-07,
      "loss": 0.2356,
      "step": 48865
    },
    {
      "epoch": 0.8823911771715814,
      "grad_norm": 0.4740567207336426,
      "learning_rate": 6.747353921632005e-07,
      "loss": 0.2322,
      "step": 48870
    },
    {
      "epoch": 0.8824814566044822,
      "grad_norm": 0.4739016890525818,
      "learning_rate": 6.737115953186546e-07,
      "loss": 0.2387,
      "step": 48875
    },
    {
      "epoch": 0.8825717360373829,
      "grad_norm": 0.3513011038303375,
      "learning_rate": 6.726885487062007e-07,
      "loss": 0.1888,
      "step": 48880
    },
    {
      "epoch": 0.8826620154702837,
      "grad_norm": 0.5506095290184021,
      "learning_rate": 6.716662524081375e-07,
      "loss": 0.2461,
      "step": 48885
    },
    {
      "epoch": 0.8827522949031843,
      "grad_norm": 0.3321288228034973,
      "learning_rate": 6.706447065067012e-07,
      "loss": 0.1946,
      "step": 48890
    },
    {
      "epoch": 0.882842574336085,
      "grad_norm": 0.5968649387359619,
      "learning_rate": 6.696239110840686e-07,
      "loss": 0.1693,
      "step": 48895
    },
    {
      "epoch": 0.8829328537689858,
      "grad_norm": 0.3725471496582031,
      "learning_rate": 6.686038662223538e-07,
      "loss": 0.2356,
      "step": 48900
    },
    {
      "epoch": 0.8830231332018865,
      "grad_norm": 0.30136847496032715,
      "learning_rate": 6.675845720036101e-07,
      "loss": 0.2995,
      "step": 48905
    },
    {
      "epoch": 0.8831134126347872,
      "grad_norm": 0.4549270272254944,
      "learning_rate": 6.665660285098363e-07,
      "loss": 0.2596,
      "step": 48910
    },
    {
      "epoch": 0.8832036920676879,
      "grad_norm": 0.4814635217189789,
      "learning_rate": 6.655482358229636e-07,
      "loss": 0.1425,
      "step": 48915
    },
    {
      "epoch": 0.8832939715005886,
      "grad_norm": 0.31953999400138855,
      "learning_rate": 6.645311940248667e-07,
      "loss": 0.0754,
      "step": 48920
    },
    {
      "epoch": 0.8833842509334894,
      "grad_norm": 0.6461823582649231,
      "learning_rate": 6.635149031973598e-07,
      "loss": 0.1998,
      "step": 48925
    },
    {
      "epoch": 0.88347453036639,
      "grad_norm": 0.37079378962516785,
      "learning_rate": 6.624993634221987e-07,
      "loss": 0.181,
      "step": 48930
    },
    {
      "epoch": 0.8835648097992908,
      "grad_norm": 0.5783750414848328,
      "learning_rate": 6.614845747810706e-07,
      "loss": 0.2365,
      "step": 48935
    },
    {
      "epoch": 0.8836550892321915,
      "grad_norm": 0.3737196922302246,
      "learning_rate": 6.604705373556142e-07,
      "loss": 0.0863,
      "step": 48940
    },
    {
      "epoch": 0.8837453686650922,
      "grad_norm": 0.27370980381965637,
      "learning_rate": 6.594572512273956e-07,
      "loss": 0.2368,
      "step": 48945
    },
    {
      "epoch": 0.8838356480979929,
      "grad_norm": 0.42945805191993713,
      "learning_rate": 6.584447164779284e-07,
      "loss": 0.1998,
      "step": 48950
    },
    {
      "epoch": 0.8839259275308936,
      "grad_norm": 0.5229265689849854,
      "learning_rate": 6.574329331886642e-07,
      "loss": 0.2345,
      "step": 48955
    },
    {
      "epoch": 0.8840162069637943,
      "grad_norm": 0.43820658326148987,
      "learning_rate": 6.564219014409956e-07,
      "loss": 0.1946,
      "step": 48960
    },
    {
      "epoch": 0.8841064863966951,
      "grad_norm": 0.5245822072029114,
      "learning_rate": 6.554116213162487e-07,
      "loss": 0.2194,
      "step": 48965
    },
    {
      "epoch": 0.8841967658295957,
      "grad_norm": 0.3123713731765747,
      "learning_rate": 6.544020928956973e-07,
      "loss": 0.2327,
      "step": 48970
    },
    {
      "epoch": 0.8842870452624965,
      "grad_norm": 0.262265145778656,
      "learning_rate": 6.533933162605466e-07,
      "loss": 0.1715,
      "step": 48975
    },
    {
      "epoch": 0.8843773246953972,
      "grad_norm": 0.2101922333240509,
      "learning_rate": 6.523852914919482e-07,
      "loss": 0.2302,
      "step": 48980
    },
    {
      "epoch": 0.884467604128298,
      "grad_norm": 0.5431260466575623,
      "learning_rate": 6.513780186709884e-07,
      "loss": 0.1776,
      "step": 48985
    },
    {
      "epoch": 0.8845578835611986,
      "grad_norm": 0.4570768177509308,
      "learning_rate": 6.503714978786957e-07,
      "loss": 0.2269,
      "step": 48990
    },
    {
      "epoch": 0.8846481629940993,
      "grad_norm": 0.4027325510978699,
      "learning_rate": 6.493657291960398e-07,
      "loss": 0.2101,
      "step": 48995
    },
    {
      "epoch": 0.8847384424270001,
      "grad_norm": 0.18464119732379913,
      "learning_rate": 6.483607127039226e-07,
      "loss": 0.1624,
      "step": 49000
    },
    {
      "epoch": 0.8848287218599008,
      "grad_norm": 0.42464643716812134,
      "learning_rate": 6.473564484831962e-07,
      "loss": 0.2021,
      "step": 49005
    },
    {
      "epoch": 0.8849190012928014,
      "grad_norm": 0.5144774317741394,
      "learning_rate": 6.463529366146404e-07,
      "loss": 0.2546,
      "step": 49010
    },
    {
      "epoch": 0.8850092807257022,
      "grad_norm": 0.5907028913497925,
      "learning_rate": 6.453501771789839e-07,
      "loss": 0.2221,
      "step": 49015
    },
    {
      "epoch": 0.8850995601586029,
      "grad_norm": 0.44879671931266785,
      "learning_rate": 6.443481702568899e-07,
      "loss": 0.2271,
      "step": 49020
    },
    {
      "epoch": 0.8851898395915037,
      "grad_norm": 0.5192486643791199,
      "learning_rate": 6.433469159289662e-07,
      "loss": 0.1811,
      "step": 49025
    },
    {
      "epoch": 0.8852801190244043,
      "grad_norm": 0.5279654264450073,
      "learning_rate": 6.423464142757507e-07,
      "loss": 0.2762,
      "step": 49030
    },
    {
      "epoch": 0.885370398457305,
      "grad_norm": 0.41415759921073914,
      "learning_rate": 6.413466653777323e-07,
      "loss": 0.1521,
      "step": 49035
    },
    {
      "epoch": 0.8854606778902058,
      "grad_norm": 0.3615606725215912,
      "learning_rate": 6.403476693153287e-07,
      "loss": 0.1981,
      "step": 49040
    },
    {
      "epoch": 0.8855509573231065,
      "grad_norm": 0.5927579998970032,
      "learning_rate": 6.393494261689048e-07,
      "loss": 0.2668,
      "step": 49045
    },
    {
      "epoch": 0.8856412367560071,
      "grad_norm": 0.5542939305305481,
      "learning_rate": 6.383519360187606e-07,
      "loss": 0.2321,
      "step": 49050
    },
    {
      "epoch": 0.8857315161889079,
      "grad_norm": 0.48070257902145386,
      "learning_rate": 6.373551989451366e-07,
      "loss": 0.248,
      "step": 49055
    },
    {
      "epoch": 0.8858217956218086,
      "grad_norm": 0.5028548240661621,
      "learning_rate": 6.363592150282138e-07,
      "loss": 0.2422,
      "step": 49060
    },
    {
      "epoch": 0.8859120750547094,
      "grad_norm": 1.1969773769378662,
      "learning_rate": 6.353639843481141e-07,
      "loss": 0.2232,
      "step": 49065
    },
    {
      "epoch": 0.88600235448761,
      "grad_norm": 0.23896126449108124,
      "learning_rate": 6.34369506984892e-07,
      "loss": 0.2505,
      "step": 49070
    },
    {
      "epoch": 0.8860926339205107,
      "grad_norm": 0.4905920922756195,
      "learning_rate": 6.333757830185505e-07,
      "loss": 0.3466,
      "step": 49075
    },
    {
      "epoch": 0.8861829133534115,
      "grad_norm": 0.5092660784721375,
      "learning_rate": 6.323828125290244e-07,
      "loss": 0.196,
      "step": 49080
    },
    {
      "epoch": 0.8862731927863122,
      "grad_norm": 0.5876214504241943,
      "learning_rate": 6.313905955961908e-07,
      "loss": 0.2474,
      "step": 49085
    },
    {
      "epoch": 0.8863634722192129,
      "grad_norm": 0.4340628683567047,
      "learning_rate": 6.303991322998693e-07,
      "loss": 0.2652,
      "step": 49090
    },
    {
      "epoch": 0.8864537516521136,
      "grad_norm": 0.5022110939025879,
      "learning_rate": 6.294084227198127e-07,
      "loss": 0.3019,
      "step": 49095
    },
    {
      "epoch": 0.8865440310850143,
      "grad_norm": 0.32097309827804565,
      "learning_rate": 6.284184669357185e-07,
      "loss": 0.1647,
      "step": 49100
    },
    {
      "epoch": 0.8866343105179151,
      "grad_norm": 0.48288461565971375,
      "learning_rate": 6.274292650272196e-07,
      "loss": 0.2126,
      "step": 49105
    },
    {
      "epoch": 0.8867245899508157,
      "grad_norm": 0.33698907494544983,
      "learning_rate": 6.264408170738923e-07,
      "loss": 0.1959,
      "step": 49110
    },
    {
      "epoch": 0.8868148693837165,
      "grad_norm": 0.32553550601005554,
      "learning_rate": 6.254531231552485e-07,
      "loss": 0.2005,
      "step": 49115
    },
    {
      "epoch": 0.8869051488166172,
      "grad_norm": 0.3544366657733917,
      "learning_rate": 6.244661833507404e-07,
      "loss": 0.184,
      "step": 49120
    },
    {
      "epoch": 0.8869954282495179,
      "grad_norm": 0.1811123937368393,
      "learning_rate": 6.23479997739761e-07,
      "loss": 0.1693,
      "step": 49125
    },
    {
      "epoch": 0.8870857076824186,
      "grad_norm": 0.3863956034183502,
      "learning_rate": 6.224945664016435e-07,
      "loss": 0.1653,
      "step": 49130
    },
    {
      "epoch": 0.8871759871153193,
      "grad_norm": 0.2895120680332184,
      "learning_rate": 6.215098894156568e-07,
      "loss": 0.2131,
      "step": 49135
    },
    {
      "epoch": 0.88726626654822,
      "grad_norm": 0.3756144046783447,
      "learning_rate": 6.205259668610119e-07,
      "loss": 0.264,
      "step": 49140
    },
    {
      "epoch": 0.8873565459811208,
      "grad_norm": 0.8095720410346985,
      "learning_rate": 6.195427988168557e-07,
      "loss": 0.2005,
      "step": 49145
    },
    {
      "epoch": 0.8874468254140215,
      "grad_norm": 0.4172549247741699,
      "learning_rate": 6.18560385362279e-07,
      "loss": 0.192,
      "step": 49150
    },
    {
      "epoch": 0.8875371048469222,
      "grad_norm": 0.4497814476490021,
      "learning_rate": 6.1757872657631e-07,
      "loss": 0.153,
      "step": 49155
    },
    {
      "epoch": 0.8876273842798229,
      "grad_norm": 0.43509024381637573,
      "learning_rate": 6.165978225379165e-07,
      "loss": 0.302,
      "step": 49160
    },
    {
      "epoch": 0.8877176637127236,
      "grad_norm": 0.30748501420021057,
      "learning_rate": 6.156176733260033e-07,
      "loss": 0.1803,
      "step": 49165
    },
    {
      "epoch": 0.8878079431456244,
      "grad_norm": 0.4086261987686157,
      "learning_rate": 6.146382790194195e-07,
      "loss": 0.1683,
      "step": 49170
    },
    {
      "epoch": 0.887898222578525,
      "grad_norm": 0.47224515676498413,
      "learning_rate": 6.136596396969452e-07,
      "loss": 0.2976,
      "step": 49175
    },
    {
      "epoch": 0.8879885020114258,
      "grad_norm": 0.6250869631767273,
      "learning_rate": 6.126817554373088e-07,
      "loss": 0.222,
      "step": 49180
    },
    {
      "epoch": 0.8880787814443265,
      "grad_norm": 0.43033626675605774,
      "learning_rate": 6.117046263191729e-07,
      "loss": 0.2545,
      "step": 49185
    },
    {
      "epoch": 0.8881690608772272,
      "grad_norm": 0.3559838533401489,
      "learning_rate": 6.107282524211389e-07,
      "loss": 0.2645,
      "step": 49190
    },
    {
      "epoch": 0.8882593403101279,
      "grad_norm": 0.21485501527786255,
      "learning_rate": 6.09752633821753e-07,
      "loss": 0.3151,
      "step": 49195
    },
    {
      "epoch": 0.8883496197430286,
      "grad_norm": 0.34246349334716797,
      "learning_rate": 6.087777705994913e-07,
      "loss": 0.2091,
      "step": 49200
    },
    {
      "epoch": 0.8884398991759294,
      "grad_norm": 0.27413228154182434,
      "learning_rate": 6.078036628327788e-07,
      "loss": 0.2313,
      "step": 49205
    },
    {
      "epoch": 0.8885301786088301,
      "grad_norm": 0.4511106312274933,
      "learning_rate": 6.068303105999729e-07,
      "loss": 0.204,
      "step": 49210
    },
    {
      "epoch": 0.8886204580417307,
      "grad_norm": 0.37216225266456604,
      "learning_rate": 6.058577139793742e-07,
      "loss": 0.2052,
      "step": 49215
    },
    {
      "epoch": 0.8887107374746315,
      "grad_norm": 0.26895153522491455,
      "learning_rate": 6.048858730492191e-07,
      "loss": 0.1902,
      "step": 49220
    },
    {
      "epoch": 0.8888010169075322,
      "grad_norm": 0.28107526898384094,
      "learning_rate": 6.039147878876883e-07,
      "loss": 0.2358,
      "step": 49225
    },
    {
      "epoch": 0.888891296340433,
      "grad_norm": 0.4671335816383362,
      "learning_rate": 6.02944458572895e-07,
      "loss": 0.2322,
      "step": 49230
    },
    {
      "epoch": 0.8889815757733336,
      "grad_norm": 0.9049550890922546,
      "learning_rate": 6.019748851828988e-07,
      "loss": 0.2483,
      "step": 49235
    },
    {
      "epoch": 0.8890718552062343,
      "grad_norm": 0.9912333488464355,
      "learning_rate": 6.010060677956919e-07,
      "loss": 0.193,
      "step": 49240
    },
    {
      "epoch": 0.8891621346391351,
      "grad_norm": 0.38939374685287476,
      "learning_rate": 6.000380064892097e-07,
      "loss": 0.2529,
      "step": 49245
    },
    {
      "epoch": 0.8892524140720358,
      "grad_norm": 0.3491939306259155,
      "learning_rate": 5.990707013413243e-07,
      "loss": 0.2153,
      "step": 49250
    },
    {
      "epoch": 0.8893426935049364,
      "grad_norm": 0.7511391639709473,
      "learning_rate": 5.981041524298503e-07,
      "loss": 0.1904,
      "step": 49255
    },
    {
      "epoch": 0.8894329729378372,
      "grad_norm": 0.6481285095214844,
      "learning_rate": 5.971383598325386e-07,
      "loss": 0.2226,
      "step": 49260
    },
    {
      "epoch": 0.8895232523707379,
      "grad_norm": 0.412752628326416,
      "learning_rate": 5.961733236270817e-07,
      "loss": 0.1532,
      "step": 49265
    },
    {
      "epoch": 0.8896135318036387,
      "grad_norm": 0.9104444980621338,
      "learning_rate": 5.952090438911084e-07,
      "loss": 0.2151,
      "step": 49270
    },
    {
      "epoch": 0.8897038112365393,
      "grad_norm": 1.2514029741287231,
      "learning_rate": 5.942455207021868e-07,
      "loss": 0.227,
      "step": 49275
    },
    {
      "epoch": 0.88979409066944,
      "grad_norm": 0.4405597150325775,
      "learning_rate": 5.932827541378272e-07,
      "loss": 0.2554,
      "step": 49280
    },
    {
      "epoch": 0.8898843701023408,
      "grad_norm": 0.5971860289573669,
      "learning_rate": 5.923207442754764e-07,
      "loss": 0.2257,
      "step": 49285
    },
    {
      "epoch": 0.8899746495352415,
      "grad_norm": 0.5276842713356018,
      "learning_rate": 5.913594911925214e-07,
      "loss": 0.2626,
      "step": 49290
    },
    {
      "epoch": 0.8900649289681422,
      "grad_norm": 0.3187607228755951,
      "learning_rate": 5.90398994966288e-07,
      "loss": 0.2217,
      "step": 49295
    },
    {
      "epoch": 0.8901552084010429,
      "grad_norm": 0.3868699371814728,
      "learning_rate": 5.894392556740414e-07,
      "loss": 0.206,
      "step": 49300
    },
    {
      "epoch": 0.8902454878339436,
      "grad_norm": 0.4715498387813568,
      "learning_rate": 5.88480273392985e-07,
      "loss": 0.2874,
      "step": 49305
    },
    {
      "epoch": 0.8903357672668444,
      "grad_norm": 0.24980822205543518,
      "learning_rate": 5.875220482002631e-07,
      "loss": 0.1881,
      "step": 49310
    },
    {
      "epoch": 0.890426046699745,
      "grad_norm": 0.3385063111782074,
      "learning_rate": 5.865645801729558e-07,
      "loss": 0.1202,
      "step": 49315
    },
    {
      "epoch": 0.8905163261326458,
      "grad_norm": 0.7292124629020691,
      "learning_rate": 5.856078693880851e-07,
      "loss": 0.2565,
      "step": 49320
    },
    {
      "epoch": 0.8906066055655465,
      "grad_norm": 0.1540367305278778,
      "learning_rate": 5.846519159226127e-07,
      "loss": 0.2319,
      "step": 49325
    },
    {
      "epoch": 0.8906968849984472,
      "grad_norm": 0.4868296980857849,
      "learning_rate": 5.836967198534382e-07,
      "loss": 0.309,
      "step": 49330
    },
    {
      "epoch": 0.8907871644313479,
      "grad_norm": 0.4633415639400482,
      "learning_rate": 5.82742281257399e-07,
      "loss": 0.2276,
      "step": 49335
    },
    {
      "epoch": 0.8908774438642486,
      "grad_norm": 0.5018483400344849,
      "learning_rate": 5.817886002112738e-07,
      "loss": 0.1761,
      "step": 49340
    },
    {
      "epoch": 0.8909677232971493,
      "grad_norm": 0.3152850866317749,
      "learning_rate": 5.808356767917777e-07,
      "loss": 0.2455,
      "step": 49345
    },
    {
      "epoch": 0.8910580027300501,
      "grad_norm": 0.7689335942268372,
      "learning_rate": 5.798835110755674e-07,
      "loss": 0.1946,
      "step": 49350
    },
    {
      "epoch": 0.8911482821629507,
      "grad_norm": 0.3751264214515686,
      "learning_rate": 5.789321031392381e-07,
      "loss": 0.2964,
      "step": 49355
    },
    {
      "epoch": 0.8912385615958515,
      "grad_norm": 0.7395479083061218,
      "learning_rate": 5.779814530593241e-07,
      "loss": 0.2701,
      "step": 49360
    },
    {
      "epoch": 0.8913288410287522,
      "grad_norm": 0.23316659033298492,
      "learning_rate": 5.770315609122979e-07,
      "loss": 0.1797,
      "step": 49365
    },
    {
      "epoch": 0.891419120461653,
      "grad_norm": 0.2639722526073456,
      "learning_rate": 5.76082426774569e-07,
      "loss": 0.1985,
      "step": 49370
    },
    {
      "epoch": 0.8915093998945536,
      "grad_norm": 0.5189804434776306,
      "learning_rate": 5.751340507224923e-07,
      "loss": 0.3169,
      "step": 49375
    },
    {
      "epoch": 0.8915996793274543,
      "grad_norm": 0.8890037536621094,
      "learning_rate": 5.741864328323544e-07,
      "loss": 0.225,
      "step": 49380
    },
    {
      "epoch": 0.8916899587603551,
      "grad_norm": 0.4358655512332916,
      "learning_rate": 5.732395731803864e-07,
      "loss": 0.1619,
      "step": 49385
    },
    {
      "epoch": 0.8917802381932558,
      "grad_norm": 1.2555304765701294,
      "learning_rate": 5.722934718427542e-07,
      "loss": 0.2526,
      "step": 49390
    },
    {
      "epoch": 0.8918705176261564,
      "grad_norm": 0.43148449063301086,
      "learning_rate": 5.71348128895568e-07,
      "loss": 0.3112,
      "step": 49395
    },
    {
      "epoch": 0.8919607970590572,
      "grad_norm": 0.2432681769132614,
      "learning_rate": 5.704035444148715e-07,
      "loss": 0.2,
      "step": 49400
    },
    {
      "epoch": 0.8920510764919579,
      "grad_norm": 0.3374500572681427,
      "learning_rate": 5.694597184766504e-07,
      "loss": 0.242,
      "step": 49405
    },
    {
      "epoch": 0.8921413559248587,
      "grad_norm": 0.5422481298446655,
      "learning_rate": 5.685166511568263e-07,
      "loss": 0.2745,
      "step": 49410
    },
    {
      "epoch": 0.8922316353577593,
      "grad_norm": 0.910033643245697,
      "learning_rate": 5.675743425312652e-07,
      "loss": 0.1668,
      "step": 49415
    },
    {
      "epoch": 0.89232191479066,
      "grad_norm": 0.4588571786880493,
      "learning_rate": 5.666327926757675e-07,
      "loss": 0.2012,
      "step": 49420
    },
    {
      "epoch": 0.8924121942235608,
      "grad_norm": 0.5030219554901123,
      "learning_rate": 5.65692001666076e-07,
      "loss": 0.2414,
      "step": 49425
    },
    {
      "epoch": 0.8925024736564615,
      "grad_norm": 0.6208494305610657,
      "learning_rate": 5.64751969577868e-07,
      "loss": 0.2904,
      "step": 49430
    },
    {
      "epoch": 0.8925927530893621,
      "grad_norm": 0.4573724865913391,
      "learning_rate": 5.638126964867641e-07,
      "loss": 0.2345,
      "step": 49435
    },
    {
      "epoch": 0.8926830325222629,
      "grad_norm": 0.5691066384315491,
      "learning_rate": 5.628741824683193e-07,
      "loss": 0.2374,
      "step": 49440
    },
    {
      "epoch": 0.8927733119551636,
      "grad_norm": 0.732424795627594,
      "learning_rate": 5.619364275980333e-07,
      "loss": 0.1872,
      "step": 49445
    },
    {
      "epoch": 0.8928635913880644,
      "grad_norm": 0.564612865447998,
      "learning_rate": 5.609994319513401e-07,
      "loss": 0.2391,
      "step": 49450
    },
    {
      "epoch": 0.892953870820965,
      "grad_norm": 0.34582310914993286,
      "learning_rate": 5.600631956036129e-07,
      "loss": 0.307,
      "step": 49455
    },
    {
      "epoch": 0.8930441502538657,
      "grad_norm": 0.3481256067752838,
      "learning_rate": 5.591277186301692e-07,
      "loss": 0.2588,
      "step": 49460
    },
    {
      "epoch": 0.8931344296867665,
      "grad_norm": 0.46572649478912354,
      "learning_rate": 5.581930011062564e-07,
      "loss": 0.2656,
      "step": 49465
    },
    {
      "epoch": 0.8932247091196672,
      "grad_norm": 0.33579516410827637,
      "learning_rate": 5.572590431070702e-07,
      "loss": 0.125,
      "step": 49470
    },
    {
      "epoch": 0.8933149885525679,
      "grad_norm": 0.512161374092102,
      "learning_rate": 5.563258447077369e-07,
      "loss": 0.243,
      "step": 49475
    },
    {
      "epoch": 0.8934052679854686,
      "grad_norm": 0.33453431725502014,
      "learning_rate": 5.553934059833277e-07,
      "loss": 0.1899,
      "step": 49480
    },
    {
      "epoch": 0.8934955474183693,
      "grad_norm": 0.4576311409473419,
      "learning_rate": 5.544617270088493e-07,
      "loss": 0.2046,
      "step": 49485
    },
    {
      "epoch": 0.8935858268512701,
      "grad_norm": 0.3824627995491028,
      "learning_rate": 5.535308078592516e-07,
      "loss": 0.2438,
      "step": 49490
    },
    {
      "epoch": 0.8936761062841707,
      "grad_norm": 0.5188384056091309,
      "learning_rate": 5.52600648609416e-07,
      "loss": 0.3143,
      "step": 49495
    },
    {
      "epoch": 0.8937663857170715,
      "grad_norm": 0.6110566854476929,
      "learning_rate": 5.516712493341702e-07,
      "loss": 0.1958,
      "step": 49500
    },
    {
      "epoch": 0.8938566651499722,
      "grad_norm": 0.5411057472229004,
      "learning_rate": 5.507426101082758e-07,
      "loss": 0.3115,
      "step": 49505
    },
    {
      "epoch": 0.8939469445828729,
      "grad_norm": 0.3564608693122864,
      "learning_rate": 5.498147310064362e-07,
      "loss": 0.3119,
      "step": 49510
    },
    {
      "epoch": 0.8940372240157736,
      "grad_norm": 0.4241950809955597,
      "learning_rate": 5.488876121032915e-07,
      "loss": 0.2529,
      "step": 49515
    },
    {
      "epoch": 0.8941275034486743,
      "grad_norm": 0.5518417358398438,
      "learning_rate": 5.479612534734214e-07,
      "loss": 0.2723,
      "step": 49520
    },
    {
      "epoch": 0.894217782881575,
      "grad_norm": 0.6499502062797546,
      "learning_rate": 5.470356551913459e-07,
      "loss": 0.2468,
      "step": 49525
    },
    {
      "epoch": 0.8943080623144758,
      "grad_norm": 0.3124319911003113,
      "learning_rate": 5.461108173315233e-07,
      "loss": 0.2196,
      "step": 49530
    },
    {
      "epoch": 0.8943983417473765,
      "grad_norm": 0.6624701023101807,
      "learning_rate": 5.451867399683475e-07,
      "loss": 0.3122,
      "step": 49535
    },
    {
      "epoch": 0.8944886211802772,
      "grad_norm": 0.3819896876811981,
      "learning_rate": 5.442634231761568e-07,
      "loss": 0.1748,
      "step": 49540
    },
    {
      "epoch": 0.8945789006131779,
      "grad_norm": 0.4410114884376526,
      "learning_rate": 5.433408670292228e-07,
      "loss": 0.12,
      "step": 49545
    },
    {
      "epoch": 0.8946691800460786,
      "grad_norm": 0.3746069669723511,
      "learning_rate": 5.424190716017597e-07,
      "loss": 0.1849,
      "step": 49550
    },
    {
      "epoch": 0.8947594594789794,
      "grad_norm": 0.2565837502479553,
      "learning_rate": 5.414980369679201e-07,
      "loss": 0.1769,
      "step": 49555
    },
    {
      "epoch": 0.89484973891188,
      "grad_norm": 0.45501911640167236,
      "learning_rate": 5.405777632017917e-07,
      "loss": 0.1599,
      "step": 49560
    },
    {
      "epoch": 0.8949400183447808,
      "grad_norm": 0.5975123047828674,
      "learning_rate": 5.396582503774072e-07,
      "loss": 0.1931,
      "step": 49565
    },
    {
      "epoch": 0.8950302977776815,
      "grad_norm": 0.819442629814148,
      "learning_rate": 5.387394985687322e-07,
      "loss": 0.2122,
      "step": 49570
    },
    {
      "epoch": 0.8951205772105822,
      "grad_norm": 0.8892215490341187,
      "learning_rate": 5.378215078496751e-07,
      "loss": 0.3006,
      "step": 49575
    },
    {
      "epoch": 0.8952108566434829,
      "grad_norm": 0.41512638330459595,
      "learning_rate": 5.369042782940804e-07,
      "loss": 0.2028,
      "step": 49580
    },
    {
      "epoch": 0.8953011360763836,
      "grad_norm": 0.48824870586395264,
      "learning_rate": 5.359878099757332e-07,
      "loss": 0.2371,
      "step": 49585
    },
    {
      "epoch": 0.8953914155092844,
      "grad_norm": 0.27632448077201843,
      "learning_rate": 5.350721029683559e-07,
      "loss": 0.3257,
      "step": 49590
    },
    {
      "epoch": 0.8954816949421851,
      "grad_norm": 0.5071262121200562,
      "learning_rate": 5.341571573456128e-07,
      "loss": 0.2092,
      "step": 49595
    },
    {
      "epoch": 0.8955719743750857,
      "grad_norm": 0.3316676914691925,
      "learning_rate": 5.332429731811028e-07,
      "loss": 0.267,
      "step": 49600
    },
    {
      "epoch": 0.8956622538079865,
      "grad_norm": 0.5379388928413391,
      "learning_rate": 5.323295505483661e-07,
      "loss": 0.2527,
      "step": 49605
    },
    {
      "epoch": 0.8957525332408872,
      "grad_norm": 0.8905502557754517,
      "learning_rate": 5.314168895208782e-07,
      "loss": 0.3123,
      "step": 49610
    },
    {
      "epoch": 0.895842812673788,
      "grad_norm": 0.5016679763793945,
      "learning_rate": 5.305049901720593e-07,
      "loss": 0.1742,
      "step": 49615
    },
    {
      "epoch": 0.8959330921066886,
      "grad_norm": 0.6650755405426025,
      "learning_rate": 5.295938525752641e-07,
      "loss": 0.2197,
      "step": 49620
    },
    {
      "epoch": 0.8960233715395893,
      "grad_norm": 0.5120480060577393,
      "learning_rate": 5.286834768037885e-07,
      "loss": 0.1965,
      "step": 49625
    },
    {
      "epoch": 0.8961136509724901,
      "grad_norm": 0.4148460328578949,
      "learning_rate": 5.277738629308626e-07,
      "loss": 0.1623,
      "step": 49630
    },
    {
      "epoch": 0.8962039304053908,
      "grad_norm": 0.5100986361503601,
      "learning_rate": 5.268650110296602e-07,
      "loss": 0.2306,
      "step": 49635
    },
    {
      "epoch": 0.8962942098382914,
      "grad_norm": 0.400863915681839,
      "learning_rate": 5.259569211732929e-07,
      "loss": 0.2708,
      "step": 49640
    },
    {
      "epoch": 0.8963844892711922,
      "grad_norm": 0.28689926862716675,
      "learning_rate": 5.250495934348055e-07,
      "loss": 0.1492,
      "step": 49645
    },
    {
      "epoch": 0.8964747687040929,
      "grad_norm": 0.35647398233413696,
      "learning_rate": 5.241430278871895e-07,
      "loss": 0.2569,
      "step": 49650
    },
    {
      "epoch": 0.8965650481369937,
      "grad_norm": 0.4131743311882019,
      "learning_rate": 5.232372246033701e-07,
      "loss": 0.1356,
      "step": 49655
    },
    {
      "epoch": 0.8966553275698943,
      "grad_norm": 0.539134681224823,
      "learning_rate": 5.223321836562156e-07,
      "loss": 0.3402,
      "step": 49660
    },
    {
      "epoch": 0.896745607002795,
      "grad_norm": 0.35673341155052185,
      "learning_rate": 5.214279051185256e-07,
      "loss": 0.2032,
      "step": 49665
    },
    {
      "epoch": 0.8968358864356958,
      "grad_norm": 0.43705660104751587,
      "learning_rate": 5.205243890630462e-07,
      "loss": 0.1374,
      "step": 49670
    },
    {
      "epoch": 0.8969261658685965,
      "grad_norm": 0.47941845655441284,
      "learning_rate": 5.196216355624561e-07,
      "loss": 0.1932,
      "step": 49675
    },
    {
      "epoch": 0.8970164453014972,
      "grad_norm": 0.5962375998497009,
      "learning_rate": 5.187196446893761e-07,
      "loss": 0.3135,
      "step": 49680
    },
    {
      "epoch": 0.8971067247343979,
      "grad_norm": 0.47453346848487854,
      "learning_rate": 5.178184165163647e-07,
      "loss": 0.2893,
      "step": 49685
    },
    {
      "epoch": 0.8971970041672986,
      "grad_norm": 0.4467545747756958,
      "learning_rate": 5.169179511159217e-07,
      "loss": 0.2869,
      "step": 49690
    },
    {
      "epoch": 0.8972872836001994,
      "grad_norm": 0.24232818186283112,
      "learning_rate": 5.160182485604792e-07,
      "loss": 0.1522,
      "step": 49695
    },
    {
      "epoch": 0.8973775630331,
      "grad_norm": 0.26575037837028503,
      "learning_rate": 5.151193089224149e-07,
      "loss": 0.1616,
      "step": 49700
    },
    {
      "epoch": 0.8974678424660008,
      "grad_norm": 0.30552709102630615,
      "learning_rate": 5.142211322740398e-07,
      "loss": 0.2146,
      "step": 49705
    },
    {
      "epoch": 0.8975581218989015,
      "grad_norm": 0.37902095913887024,
      "learning_rate": 5.133237186876061e-07,
      "loss": 0.2451,
      "step": 49710
    },
    {
      "epoch": 0.8976484013318022,
      "grad_norm": 0.6796904802322388,
      "learning_rate": 5.124270682353028e-07,
      "loss": 0.271,
      "step": 49715
    },
    {
      "epoch": 0.8977386807647029,
      "grad_norm": 0.2675473392009735,
      "learning_rate": 5.115311809892631e-07,
      "loss": 0.1837,
      "step": 49720
    },
    {
      "epoch": 0.8978289601976036,
      "grad_norm": 0.4366501569747925,
      "learning_rate": 5.106360570215507e-07,
      "loss": 0.224,
      "step": 49725
    },
    {
      "epoch": 0.8979192396305044,
      "grad_norm": 0.36879366636276245,
      "learning_rate": 5.097416964041746e-07,
      "loss": 0.3035,
      "step": 49730
    },
    {
      "epoch": 0.8980095190634051,
      "grad_norm": 0.44806408882141113,
      "learning_rate": 5.088480992090783e-07,
      "loss": 0.1425,
      "step": 49735
    },
    {
      "epoch": 0.8980997984963057,
      "grad_norm": 0.7490373253822327,
      "learning_rate": 5.079552655081432e-07,
      "loss": 0.2077,
      "step": 49740
    },
    {
      "epoch": 0.8981900779292065,
      "grad_norm": 0.37198522686958313,
      "learning_rate": 5.070631953731931e-07,
      "loss": 0.0843,
      "step": 49745
    },
    {
      "epoch": 0.8982803573621072,
      "grad_norm": 0.3032682240009308,
      "learning_rate": 5.061718888759892e-07,
      "loss": 0.21,
      "step": 49750
    },
    {
      "epoch": 0.898370636795008,
      "grad_norm": 0.5252434015274048,
      "learning_rate": 5.052813460882311e-07,
      "loss": 0.1548,
      "step": 49755
    },
    {
      "epoch": 0.8984609162279086,
      "grad_norm": 0.265247642993927,
      "learning_rate": 5.043915670815536e-07,
      "loss": 0.1449,
      "step": 49760
    },
    {
      "epoch": 0.8985511956608093,
      "grad_norm": 0.3937067985534668,
      "learning_rate": 5.035025519275371e-07,
      "loss": 0.2638,
      "step": 49765
    },
    {
      "epoch": 0.8986414750937101,
      "grad_norm": 1.6205127239227295,
      "learning_rate": 5.026143006976914e-07,
      "loss": 0.2371,
      "step": 49770
    },
    {
      "epoch": 0.8987317545266108,
      "grad_norm": 0.48166754841804504,
      "learning_rate": 5.017268134634745e-07,
      "loss": 0.154,
      "step": 49775
    },
    {
      "epoch": 0.8988220339595114,
      "grad_norm": 0.4826033115386963,
      "learning_rate": 5.00840090296274e-07,
      "loss": 0.2296,
      "step": 49780
    },
    {
      "epoch": 0.8989123133924122,
      "grad_norm": 0.36453431844711304,
      "learning_rate": 4.999541312674261e-07,
      "loss": 0.1641,
      "step": 49785
    },
    {
      "epoch": 0.8990025928253129,
      "grad_norm": 0.2968432605266571,
      "learning_rate": 4.990689364481938e-07,
      "loss": 0.224,
      "step": 49790
    },
    {
      "epoch": 0.8990928722582137,
      "grad_norm": 0.3395954668521881,
      "learning_rate": 4.981845059097889e-07,
      "loss": 0.2494,
      "step": 49795
    },
    {
      "epoch": 0.8991831516911143,
      "grad_norm": 1.0246810913085938,
      "learning_rate": 4.973008397233536e-07,
      "loss": 0.2705,
      "step": 49800
    },
    {
      "epoch": 0.899273431124015,
      "grad_norm": 0.31981080770492554,
      "learning_rate": 4.964179379599765e-07,
      "loss": 0.2092,
      "step": 49805
    },
    {
      "epoch": 0.8993637105569158,
      "grad_norm": 0.8827393651008606,
      "learning_rate": 4.955358006906775e-07,
      "loss": 0.2263,
      "step": 49810
    },
    {
      "epoch": 0.8994539899898165,
      "grad_norm": 0.5777027010917664,
      "learning_rate": 4.946544279864185e-07,
      "loss": 0.2982,
      "step": 49815
    },
    {
      "epoch": 0.8995442694227171,
      "grad_norm": 0.31169047951698303,
      "learning_rate": 4.937738199181009e-07,
      "loss": 0.2608,
      "step": 49820
    },
    {
      "epoch": 0.8996345488556179,
      "grad_norm": 0.48567578196525574,
      "learning_rate": 4.928939765565633e-07,
      "loss": 0.2632,
      "step": 49825
    },
    {
      "epoch": 0.8997248282885186,
      "grad_norm": 0.9280194640159607,
      "learning_rate": 4.920148979725814e-07,
      "loss": 0.33,
      "step": 49830
    },
    {
      "epoch": 0.8998151077214194,
      "grad_norm": 0.18331050872802734,
      "learning_rate": 4.91136584236871e-07,
      "loss": 0.2385,
      "step": 49835
    },
    {
      "epoch": 0.89990538715432,
      "grad_norm": 0.3188425898551941,
      "learning_rate": 4.902590354200876e-07,
      "loss": 0.0719,
      "step": 49840
    },
    {
      "epoch": 0.8999956665872207,
      "grad_norm": 0.3170349597930908,
      "learning_rate": 4.893822515928193e-07,
      "loss": 0.2136,
      "step": 49845
    },
    {
      "epoch": 0.9000859460201215,
      "grad_norm": 0.49580907821655273,
      "learning_rate": 4.885062328256019e-07,
      "loss": 0.2086,
      "step": 49850
    },
    {
      "epoch": 0.9001762254530222,
      "grad_norm": 0.39910197257995605,
      "learning_rate": 4.876309791889023e-07,
      "loss": 0.2079,
      "step": 49855
    },
    {
      "epoch": 0.9002665048859229,
      "grad_norm": 0.7599525451660156,
      "learning_rate": 4.867564907531297e-07,
      "loss": 0.2537,
      "step": 49860
    },
    {
      "epoch": 0.9003567843188236,
      "grad_norm": 0.5778499245643616,
      "learning_rate": 4.858827675886291e-07,
      "loss": 0.1888,
      "step": 49865
    },
    {
      "epoch": 0.9004470637517243,
      "grad_norm": 0.5479825139045715,
      "learning_rate": 4.850098097656864e-07,
      "loss": 0.263,
      "step": 49870
    },
    {
      "epoch": 0.9005373431846251,
      "grad_norm": 0.5533686876296997,
      "learning_rate": 4.841376173545221e-07,
      "loss": 0.3033,
      "step": 49875
    },
    {
      "epoch": 0.9006276226175257,
      "grad_norm": 0.3194130063056946,
      "learning_rate": 4.832661904253e-07,
      "loss": 0.2485,
      "step": 49880
    },
    {
      "epoch": 0.9007179020504265,
      "grad_norm": 0.4575395882129669,
      "learning_rate": 4.823955290481186e-07,
      "loss": 0.1641,
      "step": 49885
    },
    {
      "epoch": 0.9008081814833272,
      "grad_norm": 0.40821582078933716,
      "learning_rate": 4.815256332930196e-07,
      "loss": 0.2354,
      "step": 49890
    },
    {
      "epoch": 0.9008984609162279,
      "grad_norm": 0.3677331507205963,
      "learning_rate": 4.806565032299759e-07,
      "loss": 0.1855,
      "step": 49895
    },
    {
      "epoch": 0.9009887403491286,
      "grad_norm": 0.4790988266468048,
      "learning_rate": 4.797881389289049e-07,
      "loss": 0.2774,
      "step": 49900
    },
    {
      "epoch": 0.9010790197820293,
      "grad_norm": 0.2946798801422119,
      "learning_rate": 4.789205404596587e-07,
      "loss": 0.2109,
      "step": 49905
    },
    {
      "epoch": 0.90116929921493,
      "grad_norm": 0.5090165734291077,
      "learning_rate": 4.780537078920311e-07,
      "loss": 0.1921,
      "step": 49910
    },
    {
      "epoch": 0.9012595786478308,
      "grad_norm": 0.5305168628692627,
      "learning_rate": 4.771876412957499e-07,
      "loss": 0.2644,
      "step": 49915
    },
    {
      "epoch": 0.9013498580807314,
      "grad_norm": 0.402736097574234,
      "learning_rate": 4.76322340740486e-07,
      "loss": 0.2345,
      "step": 49920
    },
    {
      "epoch": 0.9014401375136322,
      "grad_norm": 0.3757975697517395,
      "learning_rate": 4.7545780629584704e-07,
      "loss": 0.205,
      "step": 49925
    },
    {
      "epoch": 0.9015304169465329,
      "grad_norm": 0.5877687335014343,
      "learning_rate": 4.7459403803137627e-07,
      "loss": 0.3231,
      "step": 49930
    },
    {
      "epoch": 0.9016206963794337,
      "grad_norm": 0.38740038871765137,
      "learning_rate": 4.7373103601656033e-07,
      "loss": 0.1868,
      "step": 49935
    },
    {
      "epoch": 0.9017109758123344,
      "grad_norm": 0.4732867479324341,
      "learning_rate": 4.728688003208182e-07,
      "loss": 0.2084,
      "step": 49940
    },
    {
      "epoch": 0.901801255245235,
      "grad_norm": 0.45071929693222046,
      "learning_rate": 4.7200733101351204e-07,
      "loss": 0.2015,
      "step": 49945
    },
    {
      "epoch": 0.9018915346781358,
      "grad_norm": 0.36288556456565857,
      "learning_rate": 4.71146628163941e-07,
      "loss": 0.1347,
      "step": 49950
    },
    {
      "epoch": 0.9019818141110365,
      "grad_norm": 0.6755340695381165,
      "learning_rate": 4.7028669184134403e-07,
      "loss": 0.2061,
      "step": 49955
    },
    {
      "epoch": 0.9020720935439372,
      "grad_norm": 0.523026704788208,
      "learning_rate": 4.694275221148925e-07,
      "loss": 0.2287,
      "step": 49960
    },
    {
      "epoch": 0.9021623729768379,
      "grad_norm": 0.45862624049186707,
      "learning_rate": 4.685691190537056e-07,
      "loss": 0.2871,
      "step": 49965
    },
    {
      "epoch": 0.9022526524097386,
      "grad_norm": 0.6548789143562317,
      "learning_rate": 4.677114827268303e-07,
      "loss": 0.2517,
      "step": 49970
    },
    {
      "epoch": 0.9023429318426394,
      "grad_norm": 0.48521631956100464,
      "learning_rate": 4.668546132032614e-07,
      "loss": 0.2286,
      "step": 49975
    },
    {
      "epoch": 0.9024332112755401,
      "grad_norm": 0.17036576569080353,
      "learning_rate": 4.659985105519238e-07,
      "loss": 0.2146,
      "step": 49980
    },
    {
      "epoch": 0.9025234907084407,
      "grad_norm": 0.5720534324645996,
      "learning_rate": 4.6514317484169015e-07,
      "loss": 0.1847,
      "step": 49985
    },
    {
      "epoch": 0.9026137701413415,
      "grad_norm": 0.4207828938961029,
      "learning_rate": 4.642886061413621e-07,
      "loss": 0.2737,
      "step": 49990
    },
    {
      "epoch": 0.9027040495742422,
      "grad_norm": 0.32303088903427124,
      "learning_rate": 4.634348045196846e-07,
      "loss": 0.2367,
      "step": 49995
    },
    {
      "epoch": 0.902794329007143,
      "grad_norm": 0.604050874710083,
      "learning_rate": 4.625817700453383e-07,
      "loss": 0.2784,
      "step": 50000
    },
    {
      "epoch": 0.9028846084400436,
      "grad_norm": 0.5371136665344238,
      "learning_rate": 4.6172950278694616e-07,
      "loss": 0.1975,
      "step": 50005
    },
    {
      "epoch": 0.9029748878729443,
      "grad_norm": 0.3274342715740204,
      "learning_rate": 4.6087800281306437e-07,
      "loss": 0.1982,
      "step": 50010
    },
    {
      "epoch": 0.9030651673058451,
      "grad_norm": 0.6424617767333984,
      "learning_rate": 4.6002727019219153e-07,
      "loss": 0.2331,
      "step": 50015
    },
    {
      "epoch": 0.9031554467387458,
      "grad_norm": 0.5802335143089294,
      "learning_rate": 4.5917730499276393e-07,
      "loss": 0.2075,
      "step": 50020
    },
    {
      "epoch": 0.9032457261716464,
      "grad_norm": 0.4290149211883545,
      "learning_rate": 4.583281072831514e-07,
      "loss": 0.313,
      "step": 50025
    },
    {
      "epoch": 0.9033360056045472,
      "grad_norm": 0.6646602749824524,
      "learning_rate": 4.574796771316703e-07,
      "loss": 0.3307,
      "step": 50030
    },
    {
      "epoch": 0.9034262850374479,
      "grad_norm": 0.4131048321723938,
      "learning_rate": 4.5663201460656615e-07,
      "loss": 0.2217,
      "step": 50035
    },
    {
      "epoch": 0.9035165644703487,
      "grad_norm": 0.46411240100860596,
      "learning_rate": 4.557851197760321e-07,
      "loss": 0.1809,
      "step": 50040
    },
    {
      "epoch": 0.9036068439032493,
      "grad_norm": 0.4147179424762726,
      "learning_rate": 4.5493899270818817e-07,
      "loss": 0.2409,
      "step": 50045
    },
    {
      "epoch": 0.90369712333615,
      "grad_norm": 0.49063724279403687,
      "learning_rate": 4.5409363347110544e-07,
      "loss": 0.2961,
      "step": 50050
    },
    {
      "epoch": 0.9037874027690508,
      "grad_norm": 0.3859482705593109,
      "learning_rate": 4.53249042132784e-07,
      "loss": 0.178,
      "step": 50055
    },
    {
      "epoch": 0.9038776822019515,
      "grad_norm": 0.5122784376144409,
      "learning_rate": 4.5240521876116607e-07,
      "loss": 0.1957,
      "step": 50060
    },
    {
      "epoch": 0.9039679616348522,
      "grad_norm": 0.4450696110725403,
      "learning_rate": 4.5156216342412963e-07,
      "loss": 0.2349,
      "step": 50065
    },
    {
      "epoch": 0.9040582410677529,
      "grad_norm": 0.544794499874115,
      "learning_rate": 4.5071987618949485e-07,
      "loss": 0.2119,
      "step": 50070
    },
    {
      "epoch": 0.9041485205006536,
      "grad_norm": 0.3934589624404907,
      "learning_rate": 4.4987835712501425e-07,
      "loss": 0.125,
      "step": 50075
    },
    {
      "epoch": 0.9042387999335544,
      "grad_norm": 0.31174585223197937,
      "learning_rate": 4.4903760629838477e-07,
      "loss": 0.2428,
      "step": 50080
    },
    {
      "epoch": 0.904329079366455,
      "grad_norm": 0.5034710168838501,
      "learning_rate": 4.4819762377723675e-07,
      "loss": 0.2239,
      "step": 50085
    },
    {
      "epoch": 0.9044193587993558,
      "grad_norm": 0.31654495000839233,
      "learning_rate": 4.473584096291428e-07,
      "loss": 0.2392,
      "step": 50090
    },
    {
      "epoch": 0.9045096382322565,
      "grad_norm": 0.3663698136806488,
      "learning_rate": 4.465199639216089e-07,
      "loss": 0.2065,
      "step": 50095
    },
    {
      "epoch": 0.9045999176651572,
      "grad_norm": 0.5760450959205627,
      "learning_rate": 4.4568228672208557e-07,
      "loss": 0.2567,
      "step": 50100
    },
    {
      "epoch": 0.9046901970980579,
      "grad_norm": 0.4471501111984253,
      "learning_rate": 4.448453780979556e-07,
      "loss": 0.2105,
      "step": 50105
    },
    {
      "epoch": 0.9047804765309586,
      "grad_norm": 0.4944366216659546,
      "learning_rate": 4.440092381165384e-07,
      "loss": 0.2441,
      "step": 50110
    },
    {
      "epoch": 0.9048707559638594,
      "grad_norm": 0.7945095300674438,
      "learning_rate": 4.431738668451036e-07,
      "loss": 0.2615,
      "step": 50115
    },
    {
      "epoch": 0.9049610353967601,
      "grad_norm": 0.30083099007606506,
      "learning_rate": 4.4233926435084396e-07,
      "loss": 0.2695,
      "step": 50120
    },
    {
      "epoch": 0.9050513148296607,
      "grad_norm": 0.3817719519138336,
      "learning_rate": 4.4150543070090037e-07,
      "loss": 0.2053,
      "step": 50125
    },
    {
      "epoch": 0.9051415942625615,
      "grad_norm": 0.6499032378196716,
      "learning_rate": 4.4067236596234795e-07,
      "loss": 0.3102,
      "step": 50130
    },
    {
      "epoch": 0.9052318736954622,
      "grad_norm": 0.7429965138435364,
      "learning_rate": 4.39840070202201e-07,
      "loss": 0.1659,
      "step": 50135
    },
    {
      "epoch": 0.905322153128363,
      "grad_norm": 0.5219694375991821,
      "learning_rate": 4.390085434874103e-07,
      "loss": 0.185,
      "step": 50140
    },
    {
      "epoch": 0.9054124325612636,
      "grad_norm": 0.8456925749778748,
      "learning_rate": 4.381777858848668e-07,
      "loss": 0.175,
      "step": 50145
    },
    {
      "epoch": 0.9055027119941643,
      "grad_norm": 0.6609681844711304,
      "learning_rate": 4.3734779746140043e-07,
      "loss": 0.2725,
      "step": 50150
    },
    {
      "epoch": 0.9055929914270651,
      "grad_norm": 0.3738050162792206,
      "learning_rate": 4.3651857828377665e-07,
      "loss": 0.2405,
      "step": 50155
    },
    {
      "epoch": 0.9056832708599658,
      "grad_norm": 0.44865942001342773,
      "learning_rate": 4.3569012841869987e-07,
      "loss": 0.1327,
      "step": 50160
    },
    {
      "epoch": 0.9057735502928664,
      "grad_norm": 0.4371652901172638,
      "learning_rate": 4.348624479328134e-07,
      "loss": 0.3304,
      "step": 50165
    },
    {
      "epoch": 0.9058638297257672,
      "grad_norm": 0.22302810847759247,
      "learning_rate": 4.340355368926963e-07,
      "loss": 0.1925,
      "step": 50170
    },
    {
      "epoch": 0.9059541091586679,
      "grad_norm": 0.35875651240348816,
      "learning_rate": 4.3320939536486974e-07,
      "loss": 0.2129,
      "step": 50175
    },
    {
      "epoch": 0.9060443885915687,
      "grad_norm": 0.26657238602638245,
      "learning_rate": 4.3238402341578946e-07,
      "loss": 0.1229,
      "step": 50180
    },
    {
      "epoch": 0.9061346680244693,
      "grad_norm": 0.5273004770278931,
      "learning_rate": 4.3155942111185234e-07,
      "loss": 0.2458,
      "step": 50185
    },
    {
      "epoch": 0.90622494745737,
      "grad_norm": 0.5499250292778015,
      "learning_rate": 4.307355885193909e-07,
      "loss": 0.1427,
      "step": 50190
    },
    {
      "epoch": 0.9063152268902708,
      "grad_norm": 0.29581502079963684,
      "learning_rate": 4.2991252570467436e-07,
      "loss": 0.2336,
      "step": 50195
    },
    {
      "epoch": 0.9064055063231715,
      "grad_norm": 0.4888015687465668,
      "learning_rate": 4.290902327339153e-07,
      "loss": 0.2174,
      "step": 50200
    },
    {
      "epoch": 0.9064957857560721,
      "grad_norm": 0.33837902545928955,
      "learning_rate": 4.282687096732585e-07,
      "loss": 0.2551,
      "step": 50205
    },
    {
      "epoch": 0.9065860651889729,
      "grad_norm": 0.36375921964645386,
      "learning_rate": 4.274479565887912e-07,
      "loss": 0.1728,
      "step": 50210
    },
    {
      "epoch": 0.9066763446218736,
      "grad_norm": 0.30941715836524963,
      "learning_rate": 4.2662797354653616e-07,
      "loss": 0.1609,
      "step": 50215
    },
    {
      "epoch": 0.9067666240547744,
      "grad_norm": 0.5000976324081421,
      "learning_rate": 4.2580876061245726e-07,
      "loss": 0.2272,
      "step": 50220
    },
    {
      "epoch": 0.906856903487675,
      "grad_norm": 0.4815863072872162,
      "learning_rate": 4.249903178524517e-07,
      "loss": 0.2128,
      "step": 50225
    },
    {
      "epoch": 0.9069471829205757,
      "grad_norm": 0.3845180869102478,
      "learning_rate": 4.24172645332358e-07,
      "loss": 0.2306,
      "step": 50230
    },
    {
      "epoch": 0.9070374623534765,
      "grad_norm": 0.4279397130012512,
      "learning_rate": 4.2335574311795137e-07,
      "loss": 0.3058,
      "step": 50235
    },
    {
      "epoch": 0.9071277417863772,
      "grad_norm": 0.43979766964912415,
      "learning_rate": 4.225396112749458e-07,
      "loss": 0.2091,
      "step": 50240
    },
    {
      "epoch": 0.9072180212192779,
      "grad_norm": 0.6820843815803528,
      "learning_rate": 4.217242498689944e-07,
      "loss": 0.2498,
      "step": 50245
    },
    {
      "epoch": 0.9073083006521786,
      "grad_norm": 0.62982177734375,
      "learning_rate": 4.209096589656869e-07,
      "loss": 0.2269,
      "step": 50250
    },
    {
      "epoch": 0.9073985800850793,
      "grad_norm": 0.4188978672027588,
      "learning_rate": 4.200958386305498e-07,
      "loss": 0.2913,
      "step": 50255
    },
    {
      "epoch": 0.9074888595179801,
      "grad_norm": 0.40333935618400574,
      "learning_rate": 4.192827889290507e-07,
      "loss": 0.1762,
      "step": 50260
    },
    {
      "epoch": 0.9075791389508807,
      "grad_norm": 0.8141732811927795,
      "learning_rate": 4.184705099265918e-07,
      "loss": 0.2268,
      "step": 50265
    },
    {
      "epoch": 0.9076694183837815,
      "grad_norm": 0.6344169974327087,
      "learning_rate": 4.1765900168851737e-07,
      "loss": 0.2308,
      "step": 50270
    },
    {
      "epoch": 0.9077596978166822,
      "grad_norm": 0.4170890748500824,
      "learning_rate": 4.1684826428010414e-07,
      "loss": 0.2973,
      "step": 50275
    },
    {
      "epoch": 0.9078499772495829,
      "grad_norm": 0.522654116153717,
      "learning_rate": 4.160382977665722e-07,
      "loss": 0.254,
      "step": 50280
    },
    {
      "epoch": 0.9079402566824836,
      "grad_norm": 0.4280318319797516,
      "learning_rate": 4.1522910221307724e-07,
      "loss": 0.2998,
      "step": 50285
    },
    {
      "epoch": 0.9080305361153843,
      "grad_norm": 0.5495306849479675,
      "learning_rate": 4.1442067768471263e-07,
      "loss": 0.1829,
      "step": 50290
    },
    {
      "epoch": 0.908120815548285,
      "grad_norm": 0.6517475843429565,
      "learning_rate": 4.1361302424651197e-07,
      "loss": 0.3552,
      "step": 50295
    },
    {
      "epoch": 0.9082110949811858,
      "grad_norm": 0.5685386061668396,
      "learning_rate": 4.1280614196344215e-07,
      "loss": 0.2625,
      "step": 50300
    },
    {
      "epoch": 0.9083013744140864,
      "grad_norm": 0.7493937611579895,
      "learning_rate": 4.1200003090041353e-07,
      "loss": 0.2862,
      "step": 50305
    },
    {
      "epoch": 0.9083916538469872,
      "grad_norm": 0.4021660387516022,
      "learning_rate": 4.111946911222697e-07,
      "loss": 0.2414,
      "step": 50310
    },
    {
      "epoch": 0.9084819332798879,
      "grad_norm": 0.5879800915718079,
      "learning_rate": 4.1039012269379783e-07,
      "loss": 0.1154,
      "step": 50315
    },
    {
      "epoch": 0.9085722127127887,
      "grad_norm": 0.3413461148738861,
      "learning_rate": 4.09586325679715e-07,
      "loss": 0.2379,
      "step": 50320
    },
    {
      "epoch": 0.9086624921456893,
      "grad_norm": 0.5342974066734314,
      "learning_rate": 4.08783300144685e-07,
      "loss": 0.1862,
      "step": 50325
    },
    {
      "epoch": 0.90875277157859,
      "grad_norm": 0.4676631689071655,
      "learning_rate": 4.079810461533029e-07,
      "loss": 0.1875,
      "step": 50330
    },
    {
      "epoch": 0.9088430510114908,
      "grad_norm": 0.3508830666542053,
      "learning_rate": 4.071795637701059e-07,
      "loss": 0.1795,
      "step": 50335
    },
    {
      "epoch": 0.9089333304443915,
      "grad_norm": 0.7108884453773499,
      "learning_rate": 4.063788530595647e-07,
      "loss": 0.3063,
      "step": 50340
    },
    {
      "epoch": 0.9090236098772922,
      "grad_norm": 0.18200215697288513,
      "learning_rate": 4.055789140860922e-07,
      "loss": 0.2763,
      "step": 50345
    },
    {
      "epoch": 0.9091138893101929,
      "grad_norm": 0.4111839532852173,
      "learning_rate": 4.047797469140391e-07,
      "loss": 0.1744,
      "step": 50350
    },
    {
      "epoch": 0.9092041687430936,
      "grad_norm": 0.4100380837917328,
      "learning_rate": 4.039813516076918e-07,
      "loss": 0.1646,
      "step": 50355
    },
    {
      "epoch": 0.9092944481759944,
      "grad_norm": 0.5537596940994263,
      "learning_rate": 4.031837282312734e-07,
      "loss": 0.2745,
      "step": 50360
    },
    {
      "epoch": 0.9093847276088951,
      "grad_norm": 0.7820747494697571,
      "learning_rate": 4.023868768489503e-07,
      "loss": 0.2231,
      "step": 50365
    },
    {
      "epoch": 0.9094750070417957,
      "grad_norm": 0.40522339940071106,
      "learning_rate": 4.015907975248201e-07,
      "loss": 0.2613,
      "step": 50370
    },
    {
      "epoch": 0.9095652864746965,
      "grad_norm": 0.4584490954875946,
      "learning_rate": 4.007954903229239e-07,
      "loss": 0.1947,
      "step": 50375
    },
    {
      "epoch": 0.9096555659075972,
      "grad_norm": 0.4288844168186188,
      "learning_rate": 4.00000955307237e-07,
      "loss": 0.201,
      "step": 50380
    },
    {
      "epoch": 0.909745845340498,
      "grad_norm": 0.6215588450431824,
      "learning_rate": 3.9920719254167515e-07,
      "loss": 0.2047,
      "step": 50385
    },
    {
      "epoch": 0.9098361247733986,
      "grad_norm": 0.25456854701042175,
      "learning_rate": 3.984142020900905e-07,
      "loss": 0.1589,
      "step": 50390
    },
    {
      "epoch": 0.9099264042062993,
      "grad_norm": 0.33527636528015137,
      "learning_rate": 3.9762198401627095e-07,
      "loss": 0.2378,
      "step": 50395
    },
    {
      "epoch": 0.9100166836392001,
      "grad_norm": 1.9328597784042358,
      "learning_rate": 3.968305383839499e-07,
      "loss": 0.2021,
      "step": 50400
    },
    {
      "epoch": 0.9101069630721008,
      "grad_norm": 0.6006541848182678,
      "learning_rate": 3.960398652567876e-07,
      "loss": 0.2291,
      "step": 50405
    },
    {
      "epoch": 0.9101972425050014,
      "grad_norm": 0.296835333108902,
      "learning_rate": 3.952499646983909e-07,
      "loss": 0.2481,
      "step": 50410
    },
    {
      "epoch": 0.9102875219379022,
      "grad_norm": 0.25313642621040344,
      "learning_rate": 3.944608367723013e-07,
      "loss": 0.2859,
      "step": 50415
    },
    {
      "epoch": 0.9103778013708029,
      "grad_norm": 0.3476182818412781,
      "learning_rate": 3.93672481541999e-07,
      "loss": 0.2489,
      "step": 50420
    },
    {
      "epoch": 0.9104680808037037,
      "grad_norm": 0.3778388798236847,
      "learning_rate": 3.928848990709011e-07,
      "loss": 0.1478,
      "step": 50425
    },
    {
      "epoch": 0.9105583602366043,
      "grad_norm": 0.3250458836555481,
      "learning_rate": 3.9209808942236247e-07,
      "loss": 0.1719,
      "step": 50430
    },
    {
      "epoch": 0.910648639669505,
      "grad_norm": 0.48391348123550415,
      "learning_rate": 3.913120526596759e-07,
      "loss": 0.1562,
      "step": 50435
    },
    {
      "epoch": 0.9107389191024058,
      "grad_norm": 0.6892566680908203,
      "learning_rate": 3.9052678884607176e-07,
      "loss": 0.2654,
      "step": 50440
    },
    {
      "epoch": 0.9108291985353065,
      "grad_norm": 0.7143829464912415,
      "learning_rate": 3.8974229804472076e-07,
      "loss": 0.2494,
      "step": 50445
    },
    {
      "epoch": 0.9109194779682072,
      "grad_norm": 0.28805986046791077,
      "learning_rate": 3.8895858031873013e-07,
      "loss": 0.2096,
      "step": 50450
    },
    {
      "epoch": 0.9110097574011079,
      "grad_norm": 0.7758084535598755,
      "learning_rate": 3.881756357311406e-07,
      "loss": 0.2978,
      "step": 50455
    },
    {
      "epoch": 0.9111000368340086,
      "grad_norm": 0.6385540962219238,
      "learning_rate": 3.873934643449373e-07,
      "loss": 0.3422,
      "step": 50460
    },
    {
      "epoch": 0.9111903162669094,
      "grad_norm": 0.4996768534183502,
      "learning_rate": 3.8661206622303883e-07,
      "loss": 0.1942,
      "step": 50465
    },
    {
      "epoch": 0.91128059569981,
      "grad_norm": 0.6436729431152344,
      "learning_rate": 3.8583144142830485e-07,
      "loss": 0.2363,
      "step": 50470
    },
    {
      "epoch": 0.9113708751327108,
      "grad_norm": 0.29036223888397217,
      "learning_rate": 3.8505159002352853e-07,
      "loss": 0.1936,
      "step": 50475
    },
    {
      "epoch": 0.9114611545656115,
      "grad_norm": 0.3660075068473816,
      "learning_rate": 3.842725120714441e-07,
      "loss": 0.2528,
      "step": 50480
    },
    {
      "epoch": 0.9115514339985122,
      "grad_norm": 0.8161829113960266,
      "learning_rate": 3.834942076347248e-07,
      "loss": 0.3586,
      "step": 50485
    },
    {
      "epoch": 0.9116417134314129,
      "grad_norm": 0.37517833709716797,
      "learning_rate": 3.8271667677597603e-07,
      "loss": 0.2175,
      "step": 50490
    },
    {
      "epoch": 0.9117319928643136,
      "grad_norm": 0.5656657814979553,
      "learning_rate": 3.819399195577467e-07,
      "loss": 0.1661,
      "step": 50495
    },
    {
      "epoch": 0.9118222722972144,
      "grad_norm": 0.6404244899749756,
      "learning_rate": 3.811639360425201e-07,
      "loss": 0.213,
      "step": 50500
    },
    {
      "epoch": 0.9119125517301151,
      "grad_norm": 0.6275199055671692,
      "learning_rate": 3.8038872629271973e-07,
      "loss": 0.2142,
      "step": 50505
    },
    {
      "epoch": 0.9120028311630157,
      "grad_norm": 0.5586143732070923,
      "learning_rate": 3.7961429037070455e-07,
      "loss": 0.251,
      "step": 50510
    },
    {
      "epoch": 0.9120931105959165,
      "grad_norm": 0.27319806814193726,
      "learning_rate": 3.7884062833877355e-07,
      "loss": 0.2309,
      "step": 50515
    },
    {
      "epoch": 0.9121833900288172,
      "grad_norm": 0.5357292890548706,
      "learning_rate": 3.780677402591615e-07,
      "loss": 0.2423,
      "step": 50520
    },
    {
      "epoch": 0.912273669461718,
      "grad_norm": 0.4977418780326843,
      "learning_rate": 3.7729562619404083e-07,
      "loss": 0.2505,
      "step": 50525
    },
    {
      "epoch": 0.9123639488946186,
      "grad_norm": 0.5652428865432739,
      "learning_rate": 3.7652428620552293e-07,
      "loss": 0.2298,
      "step": 50530
    },
    {
      "epoch": 0.9124542283275193,
      "grad_norm": 0.2480408400297165,
      "learning_rate": 3.757537203556583e-07,
      "loss": 0.2606,
      "step": 50535
    },
    {
      "epoch": 0.9125445077604201,
      "grad_norm": 0.5349669456481934,
      "learning_rate": 3.749839287064305e-07,
      "loss": 0.2329,
      "step": 50540
    },
    {
      "epoch": 0.9126347871933208,
      "grad_norm": 0.6430692672729492,
      "learning_rate": 3.742149113197657e-07,
      "loss": 0.2907,
      "step": 50545
    },
    {
      "epoch": 0.9127250666262214,
      "grad_norm": 0.47174084186553955,
      "learning_rate": 3.734466682575233e-07,
      "loss": 0.3091,
      "step": 50550
    },
    {
      "epoch": 0.9128153460591222,
      "grad_norm": 0.44207125902175903,
      "learning_rate": 3.726791995815071e-07,
      "loss": 0.2782,
      "step": 50555
    },
    {
      "epoch": 0.9129056254920229,
      "grad_norm": 0.7439970970153809,
      "learning_rate": 3.7191250535345227e-07,
      "loss": 0.3023,
      "step": 50560
    },
    {
      "epoch": 0.9129959049249237,
      "grad_norm": 0.4343739151954651,
      "learning_rate": 3.711465856350316e-07,
      "loss": 0.2131,
      "step": 50565
    },
    {
      "epoch": 0.9130861843578243,
      "grad_norm": 0.08801253139972687,
      "learning_rate": 3.7038144048785916e-07,
      "loss": 0.1739,
      "step": 50570
    },
    {
      "epoch": 0.913176463790725,
      "grad_norm": 0.3697124421596527,
      "learning_rate": 3.696170699734858e-07,
      "loss": 0.1934,
      "step": 50575
    },
    {
      "epoch": 0.9132667432236258,
      "grad_norm": 0.8626468181610107,
      "learning_rate": 3.6885347415340113e-07,
      "loss": 0.3168,
      "step": 50580
    },
    {
      "epoch": 0.9133570226565265,
      "grad_norm": 0.5063561797142029,
      "learning_rate": 3.6809065308902824e-07,
      "loss": 0.1698,
      "step": 50585
    },
    {
      "epoch": 0.9134473020894271,
      "grad_norm": 0.5662740468978882,
      "learning_rate": 3.6732860684173255e-07,
      "loss": 0.1597,
      "step": 50590
    },
    {
      "epoch": 0.9135375815223279,
      "grad_norm": 0.4462043046951294,
      "learning_rate": 3.6656733547281273e-07,
      "loss": 0.292,
      "step": 50595
    },
    {
      "epoch": 0.9136278609552286,
      "grad_norm": 0.7503654360771179,
      "learning_rate": 3.658068390435099e-07,
      "loss": 0.2067,
      "step": 50600
    },
    {
      "epoch": 0.9137181403881294,
      "grad_norm": 0.3239396810531616,
      "learning_rate": 3.6504711761499947e-07,
      "loss": 0.2041,
      "step": 50605
    },
    {
      "epoch": 0.91380841982103,
      "grad_norm": 0.4279657006263733,
      "learning_rate": 3.6428817124839477e-07,
      "loss": 0.2585,
      "step": 50610
    },
    {
      "epoch": 0.9138986992539307,
      "grad_norm": 0.5558903813362122,
      "learning_rate": 3.635300000047481e-07,
      "loss": 0.2321,
      "step": 50615
    },
    {
      "epoch": 0.9139889786868315,
      "grad_norm": 0.5142278671264648,
      "learning_rate": 3.627726039450508e-07,
      "loss": 0.2074,
      "step": 50620
    },
    {
      "epoch": 0.9140792581197322,
      "grad_norm": 0.5785174369812012,
      "learning_rate": 3.620159831302261e-07,
      "loss": 0.2398,
      "step": 50625
    },
    {
      "epoch": 0.9141695375526329,
      "grad_norm": 0.4012088477611542,
      "learning_rate": 3.612601376211422e-07,
      "loss": 0.2587,
      "step": 50630
    },
    {
      "epoch": 0.9142598169855336,
      "grad_norm": 0.3788507282733917,
      "learning_rate": 3.605050674785993e-07,
      "loss": 0.2198,
      "step": 50635
    },
    {
      "epoch": 0.9143500964184343,
      "grad_norm": 0.6386476755142212,
      "learning_rate": 3.597507727633365e-07,
      "loss": 0.2315,
      "step": 50640
    },
    {
      "epoch": 0.9144403758513351,
      "grad_norm": 0.5969813466072083,
      "learning_rate": 3.589972535360342e-07,
      "loss": 0.1424,
      "step": 50645
    },
    {
      "epoch": 0.9145306552842357,
      "grad_norm": 0.535224437713623,
      "learning_rate": 3.582445098573062e-07,
      "loss": 0.2146,
      "step": 50650
    },
    {
      "epoch": 0.9146209347171365,
      "grad_norm": 0.46723034977912903,
      "learning_rate": 3.574925417877051e-07,
      "loss": 0.214,
      "step": 50655
    },
    {
      "epoch": 0.9147112141500372,
      "grad_norm": 0.2729646563529968,
      "learning_rate": 3.567413493877203e-07,
      "loss": 0.2081,
      "step": 50660
    },
    {
      "epoch": 0.9148014935829379,
      "grad_norm": 0.6797147989273071,
      "learning_rate": 3.5599093271778127e-07,
      "loss": 0.2154,
      "step": 50665
    },
    {
      "epoch": 0.9148917730158386,
      "grad_norm": 0.5753675699234009,
      "learning_rate": 3.5524129183825194e-07,
      "loss": 0.2217,
      "step": 50670
    },
    {
      "epoch": 0.9149820524487393,
      "grad_norm": 0.27585065364837646,
      "learning_rate": 3.544924268094374e-07,
      "loss": 0.1983,
      "step": 50675
    },
    {
      "epoch": 0.91507233188164,
      "grad_norm": 0.592746376991272,
      "learning_rate": 3.5374433769157614e-07,
      "loss": 0.236,
      "step": 50680
    },
    {
      "epoch": 0.9151626113145408,
      "grad_norm": 0.4515919089317322,
      "learning_rate": 3.5299702454485e-07,
      "loss": 0.2203,
      "step": 50685
    },
    {
      "epoch": 0.9152528907474414,
      "grad_norm": 0.396528422832489,
      "learning_rate": 3.5225048742937197e-07,
      "loss": 0.2219,
      "step": 50690
    },
    {
      "epoch": 0.9153431701803422,
      "grad_norm": 0.40932905673980713,
      "learning_rate": 3.515047264051963e-07,
      "loss": 0.0992,
      "step": 50695
    },
    {
      "epoch": 0.9154334496132429,
      "grad_norm": 0.7595083713531494,
      "learning_rate": 3.5075974153231384e-07,
      "loss": 0.168,
      "step": 50700
    },
    {
      "epoch": 0.9155237290461437,
      "grad_norm": 0.3031374216079712,
      "learning_rate": 3.5001553287065225e-07,
      "loss": 0.1691,
      "step": 50705
    },
    {
      "epoch": 0.9156140084790443,
      "grad_norm": 0.43266284465789795,
      "learning_rate": 3.492721004800803e-07,
      "loss": 0.1986,
      "step": 50710
    },
    {
      "epoch": 0.915704287911945,
      "grad_norm": 0.8472728133201599,
      "learning_rate": 3.4852944442040014e-07,
      "loss": 0.2709,
      "step": 50715
    },
    {
      "epoch": 0.9157945673448458,
      "grad_norm": 0.9106569886207581,
      "learning_rate": 3.4778756475135287e-07,
      "loss": 0.1658,
      "step": 50720
    },
    {
      "epoch": 0.9158848467777465,
      "grad_norm": 0.6724917888641357,
      "learning_rate": 3.4704646153261855e-07,
      "loss": 0.1571,
      "step": 50725
    },
    {
      "epoch": 0.9159751262106473,
      "grad_norm": 0.2121821939945221,
      "learning_rate": 3.463061348238117e-07,
      "loss": 0.224,
      "step": 50730
    },
    {
      "epoch": 0.9160654056435479,
      "grad_norm": 0.5543504357337952,
      "learning_rate": 3.45566584684488e-07,
      "loss": 0.2384,
      "step": 50735
    },
    {
      "epoch": 0.9161556850764486,
      "grad_norm": 0.5977898240089417,
      "learning_rate": 3.4482781117413764e-07,
      "loss": 0.2982,
      "step": 50740
    },
    {
      "epoch": 0.9162459645093494,
      "grad_norm": 0.618023157119751,
      "learning_rate": 3.440898143521898e-07,
      "loss": 0.1717,
      "step": 50745
    },
    {
      "epoch": 0.9163362439422501,
      "grad_norm": 0.4433342218399048,
      "learning_rate": 3.4335259427801135e-07,
      "loss": 0.1922,
      "step": 50750
    },
    {
      "epoch": 0.9164265233751507,
      "grad_norm": 0.38910043239593506,
      "learning_rate": 3.4261615101090607e-07,
      "loss": 0.2132,
      "step": 50755
    },
    {
      "epoch": 0.9165168028080515,
      "grad_norm": 1.1365541219711304,
      "learning_rate": 3.4188048461011534e-07,
      "loss": 0.3055,
      "step": 50760
    },
    {
      "epoch": 0.9166070822409522,
      "grad_norm": 0.19863960146903992,
      "learning_rate": 3.411455951348186e-07,
      "loss": 0.1518,
      "step": 50765
    },
    {
      "epoch": 0.916697361673853,
      "grad_norm": 0.3658634424209595,
      "learning_rate": 3.404114826441307e-07,
      "loss": 0.1368,
      "step": 50770
    },
    {
      "epoch": 0.9167876411067536,
      "grad_norm": 0.3282088041305542,
      "learning_rate": 3.396781471971078e-07,
      "loss": 0.3235,
      "step": 50775
    },
    {
      "epoch": 0.9168779205396543,
      "grad_norm": 0.3839718997478485,
      "learning_rate": 3.389455888527415e-07,
      "loss": 0.2708,
      "step": 50780
    },
    {
      "epoch": 0.9169681999725551,
      "grad_norm": 0.34831714630126953,
      "learning_rate": 3.382138076699593e-07,
      "loss": 0.1988,
      "step": 50785
    },
    {
      "epoch": 0.9170584794054558,
      "grad_norm": 0.26568031311035156,
      "learning_rate": 3.374828037076294e-07,
      "loss": 0.2896,
      "step": 50790
    },
    {
      "epoch": 0.9171487588383564,
      "grad_norm": 0.3606593906879425,
      "learning_rate": 3.3675257702455276e-07,
      "loss": 0.2307,
      "step": 50795
    },
    {
      "epoch": 0.9172390382712572,
      "grad_norm": 0.4034231901168823,
      "learning_rate": 3.3602312767947456e-07,
      "loss": 0.2867,
      "step": 50800
    },
    {
      "epoch": 0.9173293177041579,
      "grad_norm": 0.16336488723754883,
      "learning_rate": 3.3529445573107e-07,
      "loss": 0.1683,
      "step": 50805
    },
    {
      "epoch": 0.9174195971370587,
      "grad_norm": 0.357759028673172,
      "learning_rate": 3.3456656123795784e-07,
      "loss": 0.1406,
      "step": 50810
    },
    {
      "epoch": 0.9175098765699593,
      "grad_norm": 0.25141188502311707,
      "learning_rate": 3.338394442586912e-07,
      "loss": 0.1415,
      "step": 50815
    },
    {
      "epoch": 0.91760015600286,
      "grad_norm": 0.31993165612220764,
      "learning_rate": 3.331131048517622e-07,
      "loss": 0.2068,
      "step": 50820
    },
    {
      "epoch": 0.9176904354357608,
      "grad_norm": 0.7054659724235535,
      "learning_rate": 3.323875430755985e-07,
      "loss": 0.2548,
      "step": 50825
    },
    {
      "epoch": 0.9177807148686615,
      "grad_norm": 0.5586223602294922,
      "learning_rate": 3.316627589885668e-07,
      "loss": 0.2769,
      "step": 50830
    },
    {
      "epoch": 0.9178709943015622,
      "grad_norm": 0.4388918876647949,
      "learning_rate": 3.309387526489705e-07,
      "loss": 0.3059,
      "step": 50835
    },
    {
      "epoch": 0.9179612737344629,
      "grad_norm": 1.2424386739730835,
      "learning_rate": 3.3021552411504954e-07,
      "loss": 0.1947,
      "step": 50840
    },
    {
      "epoch": 0.9180515531673636,
      "grad_norm": 0.40078264474868774,
      "learning_rate": 3.2949307344498526e-07,
      "loss": 0.2116,
      "step": 50845
    },
    {
      "epoch": 0.9181418326002644,
      "grad_norm": 0.635747492313385,
      "learning_rate": 3.2877140069689116e-07,
      "loss": 0.1659,
      "step": 50850
    },
    {
      "epoch": 0.918232112033165,
      "grad_norm": 0.36611708998680115,
      "learning_rate": 3.2805050592882193e-07,
      "loss": 0.2203,
      "step": 50855
    },
    {
      "epoch": 0.9183223914660658,
      "grad_norm": 0.3921454846858978,
      "learning_rate": 3.273303891987667e-07,
      "loss": 0.1992,
      "step": 50860
    },
    {
      "epoch": 0.9184126708989665,
      "grad_norm": 0.4269610047340393,
      "learning_rate": 3.266110505646558e-07,
      "loss": 0.206,
      "step": 50865
    },
    {
      "epoch": 0.9185029503318672,
      "grad_norm": 0.34178775548934937,
      "learning_rate": 3.258924900843519e-07,
      "loss": 0.2422,
      "step": 50870
    },
    {
      "epoch": 0.9185932297647679,
      "grad_norm": 0.6991662979125977,
      "learning_rate": 3.251747078156597e-07,
      "loss": 0.2438,
      "step": 50875
    },
    {
      "epoch": 0.9186835091976686,
      "grad_norm": 0.4647800326347351,
      "learning_rate": 3.244577038163199e-07,
      "loss": 0.2393,
      "step": 50880
    },
    {
      "epoch": 0.9187737886305694,
      "grad_norm": 0.4164331555366516,
      "learning_rate": 3.237414781440107e-07,
      "loss": 0.1718,
      "step": 50885
    },
    {
      "epoch": 0.9188640680634701,
      "grad_norm": 0.4060598313808441,
      "learning_rate": 3.2302603085634486e-07,
      "loss": 0.2006,
      "step": 50890
    },
    {
      "epoch": 0.9189543474963707,
      "grad_norm": 0.41936469078063965,
      "learning_rate": 3.2231136201087754e-07,
      "loss": 0.1816,
      "step": 50895
    },
    {
      "epoch": 0.9190446269292715,
      "grad_norm": 0.6043384671211243,
      "learning_rate": 3.2159747166509716e-07,
      "loss": 0.3408,
      "step": 50900
    },
    {
      "epoch": 0.9191349063621722,
      "grad_norm": 0.4837489426136017,
      "learning_rate": 3.2088435987643106e-07,
      "loss": 0.2979,
      "step": 50905
    },
    {
      "epoch": 0.919225185795073,
      "grad_norm": 0.46801280975341797,
      "learning_rate": 3.2017202670224455e-07,
      "loss": 0.2624,
      "step": 50910
    },
    {
      "epoch": 0.9193154652279736,
      "grad_norm": 0.4052257537841797,
      "learning_rate": 3.194604721998396e-07,
      "loss": 0.309,
      "step": 50915
    },
    {
      "epoch": 0.9194057446608743,
      "grad_norm": 0.37288016080856323,
      "learning_rate": 3.187496964264547e-07,
      "loss": 0.2381,
      "step": 50920
    },
    {
      "epoch": 0.9194960240937751,
      "grad_norm": 0.45037224888801575,
      "learning_rate": 3.180396994392687e-07,
      "loss": 0.2332,
      "step": 50925
    },
    {
      "epoch": 0.9195863035266758,
      "grad_norm": 0.40705618262290955,
      "learning_rate": 3.173304812953937e-07,
      "loss": 0.185,
      "step": 50930
    },
    {
      "epoch": 0.9196765829595764,
      "grad_norm": 1.0847456455230713,
      "learning_rate": 3.1662204205188064e-07,
      "loss": 0.3054,
      "step": 50935
    },
    {
      "epoch": 0.9197668623924772,
      "grad_norm": 0.35280418395996094,
      "learning_rate": 3.159143817657195e-07,
      "loss": 0.1544,
      "step": 50940
    },
    {
      "epoch": 0.9198571418253779,
      "grad_norm": 0.5095433592796326,
      "learning_rate": 3.152075004938371e-07,
      "loss": 0.1899,
      "step": 50945
    },
    {
      "epoch": 0.9199474212582787,
      "grad_norm": 0.177003413438797,
      "learning_rate": 3.145013982930956e-07,
      "loss": 0.1124,
      "step": 50950
    },
    {
      "epoch": 0.9200377006911793,
      "grad_norm": 0.5592557787895203,
      "learning_rate": 3.137960752202962e-07,
      "loss": 0.1869,
      "step": 50955
    },
    {
      "epoch": 0.92012798012408,
      "grad_norm": 0.4516811966896057,
      "learning_rate": 3.1309153133217805e-07,
      "loss": 0.31,
      "step": 50960
    },
    {
      "epoch": 0.9202182595569808,
      "grad_norm": 0.35320770740509033,
      "learning_rate": 3.123877666854136e-07,
      "loss": 0.2033,
      "step": 50965
    },
    {
      "epoch": 0.9203085389898815,
      "grad_norm": 0.48333561420440674,
      "learning_rate": 3.116847813366186e-07,
      "loss": 0.2591,
      "step": 50970
    },
    {
      "epoch": 0.9203988184227822,
      "grad_norm": 0.4667608141899109,
      "learning_rate": 3.109825753423412e-07,
      "loss": 0.3112,
      "step": 50975
    },
    {
      "epoch": 0.9204890978556829,
      "grad_norm": 0.4377197325229645,
      "learning_rate": 3.1028114875907067e-07,
      "loss": 0.2024,
      "step": 50980
    },
    {
      "epoch": 0.9205793772885836,
      "grad_norm": 0.23441138863563538,
      "learning_rate": 3.0958050164323075e-07,
      "loss": 0.1782,
      "step": 50985
    },
    {
      "epoch": 0.9206696567214844,
      "grad_norm": 0.5151461362838745,
      "learning_rate": 3.0888063405118294e-07,
      "loss": 0.2156,
      "step": 50990
    },
    {
      "epoch": 0.920759936154385,
      "grad_norm": 0.7904641032218933,
      "learning_rate": 3.081815460392268e-07,
      "loss": 0.1485,
      "step": 50995
    },
    {
      "epoch": 0.9208502155872857,
      "grad_norm": 0.428873747587204,
      "learning_rate": 3.0748323766360054e-07,
      "loss": 0.2196,
      "step": 51000
    },
    {
      "epoch": 0.9209404950201865,
      "grad_norm": 0.29614976048469543,
      "learning_rate": 3.067857089804738e-07,
      "loss": 0.1931,
      "step": 51005
    },
    {
      "epoch": 0.9210307744530872,
      "grad_norm": 0.6210939288139343,
      "learning_rate": 3.0608896004596154e-07,
      "loss": 0.2243,
      "step": 51010
    },
    {
      "epoch": 0.9211210538859879,
      "grad_norm": 0.33381879329681396,
      "learning_rate": 3.053929909161102e-07,
      "loss": 0.2125,
      "step": 51015
    },
    {
      "epoch": 0.9212113333188886,
      "grad_norm": 0.44750410318374634,
      "learning_rate": 3.046978016469071e-07,
      "loss": 0.1754,
      "step": 51020
    },
    {
      "epoch": 0.9213016127517893,
      "grad_norm": 0.6163250207901001,
      "learning_rate": 3.0400339229427533e-07,
      "loss": 0.2049,
      "step": 51025
    },
    {
      "epoch": 0.9213918921846901,
      "grad_norm": 0.5448737740516663,
      "learning_rate": 3.0330976291407133e-07,
      "loss": 0.1916,
      "step": 51030
    },
    {
      "epoch": 0.9214821716175907,
      "grad_norm": 0.45817968249320984,
      "learning_rate": 3.02616913562096e-07,
      "loss": 0.1981,
      "step": 51035
    },
    {
      "epoch": 0.9215724510504915,
      "grad_norm": 0.3027898371219635,
      "learning_rate": 3.0192484429408363e-07,
      "loss": 0.2708,
      "step": 51040
    },
    {
      "epoch": 0.9216627304833922,
      "grad_norm": 0.43227240443229675,
      "learning_rate": 3.012335551657053e-07,
      "loss": 0.1804,
      "step": 51045
    },
    {
      "epoch": 0.9217530099162929,
      "grad_norm": 0.44979313015937805,
      "learning_rate": 3.005430462325709e-07,
      "loss": 0.2141,
      "step": 51050
    },
    {
      "epoch": 0.9218432893491936,
      "grad_norm": 0.19083057343959808,
      "learning_rate": 2.998533175502272e-07,
      "loss": 0.1663,
      "step": 51055
    },
    {
      "epoch": 0.9219335687820943,
      "grad_norm": 0.3405175805091858,
      "learning_rate": 2.991643691741564e-07,
      "loss": 0.2445,
      "step": 51060
    },
    {
      "epoch": 0.9220238482149951,
      "grad_norm": 0.4007622003555298,
      "learning_rate": 2.984762011597808e-07,
      "loss": 0.2264,
      "step": 51065
    },
    {
      "epoch": 0.9221141276478958,
      "grad_norm": 0.6208809614181519,
      "learning_rate": 2.9778881356245736e-07,
      "loss": 0.2062,
      "step": 51070
    },
    {
      "epoch": 0.9222044070807964,
      "grad_norm": 0.48946911096572876,
      "learning_rate": 2.9710220643748176e-07,
      "loss": 0.2162,
      "step": 51075
    },
    {
      "epoch": 0.9222946865136972,
      "grad_norm": 0.2105833739042282,
      "learning_rate": 2.9641637984008653e-07,
      "loss": 0.1951,
      "step": 51080
    },
    {
      "epoch": 0.9223849659465979,
      "grad_norm": 0.41803836822509766,
      "learning_rate": 2.9573133382544306e-07,
      "loss": 0.2241,
      "step": 51085
    },
    {
      "epoch": 0.9224752453794987,
      "grad_norm": 0.3910962641239166,
      "learning_rate": 2.95047068448655e-07,
      "loss": 0.2617,
      "step": 51090
    },
    {
      "epoch": 0.9225655248123993,
      "grad_norm": 0.1976976990699768,
      "learning_rate": 2.943635837647707e-07,
      "loss": 0.2,
      "step": 51095
    },
    {
      "epoch": 0.9226558042453,
      "grad_norm": 0.3609817922115326,
      "learning_rate": 2.9368087982876825e-07,
      "loss": 0.2004,
      "step": 51100
    },
    {
      "epoch": 0.9227460836782008,
      "grad_norm": 0.31081682443618774,
      "learning_rate": 2.92998956695566e-07,
      "loss": 0.2911,
      "step": 51105
    },
    {
      "epoch": 0.9228363631111015,
      "grad_norm": 0.32722151279449463,
      "learning_rate": 2.923178144200223e-07,
      "loss": 0.2769,
      "step": 51110
    },
    {
      "epoch": 0.9229266425440021,
      "grad_norm": 0.27847838401794434,
      "learning_rate": 2.9163745305692883e-07,
      "loss": 0.1281,
      "step": 51115
    },
    {
      "epoch": 0.9230169219769029,
      "grad_norm": 0.6383860111236572,
      "learning_rate": 2.9095787266101627e-07,
      "loss": 0.2101,
      "step": 51120
    },
    {
      "epoch": 0.9231072014098036,
      "grad_norm": 0.37489333748817444,
      "learning_rate": 2.9027907328694983e-07,
      "loss": 0.2812,
      "step": 51125
    },
    {
      "epoch": 0.9231974808427044,
      "grad_norm": 0.3819452226161957,
      "learning_rate": 2.896010549893369e-07,
      "loss": 0.2206,
      "step": 51130
    },
    {
      "epoch": 0.9232877602756051,
      "grad_norm": 0.42919087409973145,
      "learning_rate": 2.8892381782271606e-07,
      "loss": 0.2742,
      "step": 51135
    },
    {
      "epoch": 0.9233780397085057,
      "grad_norm": 0.32791903614997864,
      "learning_rate": 2.882473618415682e-07,
      "loss": 0.1806,
      "step": 51140
    },
    {
      "epoch": 0.9234683191414065,
      "grad_norm": 0.45618417859077454,
      "learning_rate": 2.875716871003087e-07,
      "loss": 0.274,
      "step": 51145
    },
    {
      "epoch": 0.9235585985743072,
      "grad_norm": 0.5508531928062439,
      "learning_rate": 2.868967936532918e-07,
      "loss": 0.1669,
      "step": 51150
    },
    {
      "epoch": 0.923648878007208,
      "grad_norm": 0.5229293704032898,
      "learning_rate": 2.862226815548064e-07,
      "loss": 0.304,
      "step": 51155
    },
    {
      "epoch": 0.9237391574401086,
      "grad_norm": 0.5950000286102295,
      "learning_rate": 2.8554935085908007e-07,
      "loss": 0.1958,
      "step": 51160
    },
    {
      "epoch": 0.9238294368730093,
      "grad_norm": 0.630179226398468,
      "learning_rate": 2.8487680162027745e-07,
      "loss": 0.2085,
      "step": 51165
    },
    {
      "epoch": 0.9239197163059101,
      "grad_norm": 0.5596440434455872,
      "learning_rate": 2.842050338924995e-07,
      "loss": 0.1888,
      "step": 51170
    },
    {
      "epoch": 0.9240099957388108,
      "grad_norm": 0.6010776162147522,
      "learning_rate": 2.8353404772978656e-07,
      "loss": 0.3384,
      "step": 51175
    },
    {
      "epoch": 0.9241002751717114,
      "grad_norm": 0.46091702580451965,
      "learning_rate": 2.8286384318611526e-07,
      "loss": 0.1176,
      "step": 51180
    },
    {
      "epoch": 0.9241905546046122,
      "grad_norm": 0.6372133493423462,
      "learning_rate": 2.821944203153959e-07,
      "loss": 0.1736,
      "step": 51185
    },
    {
      "epoch": 0.9242808340375129,
      "grad_norm": 0.3395882844924927,
      "learning_rate": 2.8152577917148203e-07,
      "loss": 0.287,
      "step": 51190
    },
    {
      "epoch": 0.9243711134704137,
      "grad_norm": 0.5429771542549133,
      "learning_rate": 2.8085791980815626e-07,
      "loss": 0.2081,
      "step": 51195
    },
    {
      "epoch": 0.9244613929033143,
      "grad_norm": 0.5102635025978088,
      "learning_rate": 2.801908422791488e-07,
      "loss": 0.1562,
      "step": 51200
    },
    {
      "epoch": 0.924551672336215,
      "grad_norm": 0.4636192321777344,
      "learning_rate": 2.795245466381158e-07,
      "loss": 0.2662,
      "step": 51205
    },
    {
      "epoch": 0.9246419517691158,
      "grad_norm": 0.21078205108642578,
      "learning_rate": 2.788590329386587e-07,
      "loss": 0.1973,
      "step": 51210
    },
    {
      "epoch": 0.9247322312020165,
      "grad_norm": 0.42441755533218384,
      "learning_rate": 2.781943012343147e-07,
      "loss": 0.2438,
      "step": 51215
    },
    {
      "epoch": 0.9248225106349172,
      "grad_norm": 0.16268227994441986,
      "learning_rate": 2.7753035157855323e-07,
      "loss": 0.15,
      "step": 51220
    },
    {
      "epoch": 0.9249127900678179,
      "grad_norm": 0.3093563914299011,
      "learning_rate": 2.7686718402478606e-07,
      "loss": 0.185,
      "step": 51225
    },
    {
      "epoch": 0.9250030695007186,
      "grad_norm": 0.4721256494522095,
      "learning_rate": 2.762047986263594e-07,
      "loss": 0.2026,
      "step": 51230
    },
    {
      "epoch": 0.9250933489336194,
      "grad_norm": 0.5277528762817383,
      "learning_rate": 2.755431954365584e-07,
      "loss": 0.3178,
      "step": 51235
    },
    {
      "epoch": 0.92518362836652,
      "grad_norm": 0.5060920119285583,
      "learning_rate": 2.7488237450860376e-07,
      "loss": 0.2432,
      "step": 51240
    },
    {
      "epoch": 0.9252739077994208,
      "grad_norm": 0.5695164799690247,
      "learning_rate": 2.7422233589565415e-07,
      "loss": 0.1769,
      "step": 51245
    },
    {
      "epoch": 0.9253641872323215,
      "grad_norm": 0.5778248906135559,
      "learning_rate": 2.7356307965080376e-07,
      "loss": 0.2684,
      "step": 51250
    },
    {
      "epoch": 0.9254544666652222,
      "grad_norm": 0.5844367146492004,
      "learning_rate": 2.729046058270857e-07,
      "loss": 0.2107,
      "step": 51255
    },
    {
      "epoch": 0.9255447460981229,
      "grad_norm": 0.7740316987037659,
      "learning_rate": 2.7224691447746866e-07,
      "loss": 0.235,
      "step": 51260
    },
    {
      "epoch": 0.9256350255310236,
      "grad_norm": 0.32647475600242615,
      "learning_rate": 2.715900056548615e-07,
      "loss": 0.2859,
      "step": 51265
    },
    {
      "epoch": 0.9257253049639244,
      "grad_norm": 0.5408141016960144,
      "learning_rate": 2.7093387941210413e-07,
      "loss": 0.3178,
      "step": 51270
    },
    {
      "epoch": 0.9258155843968251,
      "grad_norm": 0.42315220832824707,
      "learning_rate": 2.7027853580197994e-07,
      "loss": 0.1929,
      "step": 51275
    },
    {
      "epoch": 0.9259058638297257,
      "grad_norm": 0.6220344305038452,
      "learning_rate": 2.696239748772045e-07,
      "loss": 0.206,
      "step": 51280
    },
    {
      "epoch": 0.9259961432626265,
      "grad_norm": 0.408003032207489,
      "learning_rate": 2.689701966904357e-07,
      "loss": 0.1913,
      "step": 51285
    },
    {
      "epoch": 0.9260864226955272,
      "grad_norm": 0.5364543199539185,
      "learning_rate": 2.6831720129426144e-07,
      "loss": 0.2675,
      "step": 51290
    },
    {
      "epoch": 0.926176702128428,
      "grad_norm": 0.344568133354187,
      "learning_rate": 2.6766498874121306e-07,
      "loss": 0.2519,
      "step": 51295
    },
    {
      "epoch": 0.9262669815613286,
      "grad_norm": 0.5036171078681946,
      "learning_rate": 2.6701355908375525e-07,
      "loss": 0.1585,
      "step": 51300
    },
    {
      "epoch": 0.9263572609942293,
      "grad_norm": 0.6221877336502075,
      "learning_rate": 2.6636291237428945e-07,
      "loss": 0.2354,
      "step": 51305
    },
    {
      "epoch": 0.9264475404271301,
      "grad_norm": 0.48638632893562317,
      "learning_rate": 2.6571304866515936e-07,
      "loss": 0.2335,
      "step": 51310
    },
    {
      "epoch": 0.9265378198600308,
      "grad_norm": 0.5129445791244507,
      "learning_rate": 2.650639680086375e-07,
      "loss": 0.2657,
      "step": 51315
    },
    {
      "epoch": 0.9266280992929314,
      "grad_norm": 0.39693352580070496,
      "learning_rate": 2.644156704569412e-07,
      "loss": 0.176,
      "step": 51320
    },
    {
      "epoch": 0.9267183787258322,
      "grad_norm": 0.3678065836429596,
      "learning_rate": 2.637681560622185e-07,
      "loss": 0.344,
      "step": 51325
    },
    {
      "epoch": 0.9268086581587329,
      "grad_norm": 0.18631500005722046,
      "learning_rate": 2.631214248765601e-07,
      "loss": 0.243,
      "step": 51330
    },
    {
      "epoch": 0.9268989375916337,
      "grad_norm": 0.47880470752716064,
      "learning_rate": 2.624754769519866e-07,
      "loss": 0.1966,
      "step": 51335
    },
    {
      "epoch": 0.9269892170245343,
      "grad_norm": 0.2807424068450928,
      "learning_rate": 2.618303123404642e-07,
      "loss": 0.3082,
      "step": 51340
    },
    {
      "epoch": 0.927079496457435,
      "grad_norm": 0.8420343399047852,
      "learning_rate": 2.61185931093888e-07,
      "loss": 0.209,
      "step": 51345
    },
    {
      "epoch": 0.9271697758903358,
      "grad_norm": 0.34128251671791077,
      "learning_rate": 2.6054233326409775e-07,
      "loss": 0.2254,
      "step": 51350
    },
    {
      "epoch": 0.9272600553232365,
      "grad_norm": 0.35374507308006287,
      "learning_rate": 2.5989951890286305e-07,
      "loss": 0.2736,
      "step": 51355
    },
    {
      "epoch": 0.9273503347561372,
      "grad_norm": 0.5286627411842346,
      "learning_rate": 2.5925748806189476e-07,
      "loss": 0.1907,
      "step": 51360
    },
    {
      "epoch": 0.9274406141890379,
      "grad_norm": 0.37487369775772095,
      "learning_rate": 2.5861624079283945e-07,
      "loss": 0.1802,
      "step": 51365
    },
    {
      "epoch": 0.9275308936219386,
      "grad_norm": 0.3012831509113312,
      "learning_rate": 2.5797577714728015e-07,
      "loss": 0.1892,
      "step": 51370
    },
    {
      "epoch": 0.9276211730548394,
      "grad_norm": 0.38257500529289246,
      "learning_rate": 2.573360971767391e-07,
      "loss": 0.2031,
      "step": 51375
    },
    {
      "epoch": 0.92771145248774,
      "grad_norm": 0.2673783004283905,
      "learning_rate": 2.5669720093267404e-07,
      "loss": 0.1802,
      "step": 51380
    },
    {
      "epoch": 0.9278017319206407,
      "grad_norm": 0.3368590474128723,
      "learning_rate": 2.56059088466476e-07,
      "loss": 0.247,
      "step": 51385
    },
    {
      "epoch": 0.9278920113535415,
      "grad_norm": 0.2542653977870941,
      "learning_rate": 2.554217598294817e-07,
      "loss": 0.3199,
      "step": 51390
    },
    {
      "epoch": 0.9279822907864422,
      "grad_norm": 0.20993247628211975,
      "learning_rate": 2.547852150729557e-07,
      "loss": 0.2204,
      "step": 51395
    },
    {
      "epoch": 0.9280725702193429,
      "grad_norm": 0.5144657492637634,
      "learning_rate": 2.541494542481049e-07,
      "loss": 0.1955,
      "step": 51400
    },
    {
      "epoch": 0.9281628496522436,
      "grad_norm": 0.3926786780357361,
      "learning_rate": 2.535144774060705e-07,
      "loss": 0.1924,
      "step": 51405
    },
    {
      "epoch": 0.9282531290851443,
      "grad_norm": 0.47530850768089294,
      "learning_rate": 2.5288028459793167e-07,
      "loss": 0.2114,
      "step": 51410
    },
    {
      "epoch": 0.9283434085180451,
      "grad_norm": 0.6984689831733704,
      "learning_rate": 2.522468758747076e-07,
      "loss": 0.2636,
      "step": 51415
    },
    {
      "epoch": 0.9284336879509457,
      "grad_norm": 0.25490066409111023,
      "learning_rate": 2.5161425128734876e-07,
      "loss": 0.1891,
      "step": 51420
    },
    {
      "epoch": 0.9285239673838465,
      "grad_norm": 0.16208931803703308,
      "learning_rate": 2.5098241088674645e-07,
      "loss": 0.2138,
      "step": 51425
    },
    {
      "epoch": 0.9286142468167472,
      "grad_norm": 0.5864415168762207,
      "learning_rate": 2.5035135472372576e-07,
      "loss": 0.2718,
      "step": 51430
    },
    {
      "epoch": 0.9287045262496479,
      "grad_norm": 0.49469009041786194,
      "learning_rate": 2.497210828490515e-07,
      "loss": 0.27,
      "step": 51435
    },
    {
      "epoch": 0.9287948056825486,
      "grad_norm": 1.019500494003296,
      "learning_rate": 2.490915953134254e-07,
      "loss": 0.2049,
      "step": 51440
    },
    {
      "epoch": 0.9288850851154493,
      "grad_norm": 0.3572063446044922,
      "learning_rate": 2.484628921674848e-07,
      "loss": 0.2085,
      "step": 51445
    },
    {
      "epoch": 0.9289753645483501,
      "grad_norm": 0.36977294087409973,
      "learning_rate": 2.4783497346180353e-07,
      "loss": 0.2381,
      "step": 51450
    },
    {
      "epoch": 0.9290656439812508,
      "grad_norm": 0.4619942605495453,
      "learning_rate": 2.4720783924689464e-07,
      "loss": 0.1716,
      "step": 51455
    },
    {
      "epoch": 0.9291559234141514,
      "grad_norm": 0.338832288980484,
      "learning_rate": 2.4658148957320327e-07,
      "loss": 0.2055,
      "step": 51460
    },
    {
      "epoch": 0.9292462028470522,
      "grad_norm": 0.6222464442253113,
      "learning_rate": 2.459559244911192e-07,
      "loss": 0.1892,
      "step": 51465
    },
    {
      "epoch": 0.9293364822799529,
      "grad_norm": 0.6086744666099548,
      "learning_rate": 2.4533114405095984e-07,
      "loss": 0.2002,
      "step": 51470
    },
    {
      "epoch": 0.9294267617128537,
      "grad_norm": 0.729422390460968,
      "learning_rate": 2.4470714830298835e-07,
      "loss": 0.1867,
      "step": 51475
    },
    {
      "epoch": 0.9295170411457543,
      "grad_norm": 0.1963096708059311,
      "learning_rate": 2.440839372973991e-07,
      "loss": 0.2713,
      "step": 51480
    },
    {
      "epoch": 0.929607320578655,
      "grad_norm": 0.7741760015487671,
      "learning_rate": 2.4346151108432415e-07,
      "loss": 0.1345,
      "step": 51485
    },
    {
      "epoch": 0.9296976000115558,
      "grad_norm": 0.39452943205833435,
      "learning_rate": 2.4283986971383455e-07,
      "loss": 0.2173,
      "step": 51490
    },
    {
      "epoch": 0.9297878794444565,
      "grad_norm": 0.6194738745689392,
      "learning_rate": 2.4221901323593475e-07,
      "loss": 0.1802,
      "step": 51495
    },
    {
      "epoch": 0.9298781588773571,
      "grad_norm": 0.28373393416404724,
      "learning_rate": 2.415989417005693e-07,
      "loss": 0.2472,
      "step": 51500
    },
    {
      "epoch": 0.9299684383102579,
      "grad_norm": 0.29273226857185364,
      "learning_rate": 2.4097965515761933e-07,
      "loss": 0.2798,
      "step": 51505
    },
    {
      "epoch": 0.9300587177431586,
      "grad_norm": 0.33628299832344055,
      "learning_rate": 2.403611536569017e-07,
      "loss": 0.2241,
      "step": 51510
    },
    {
      "epoch": 0.9301489971760594,
      "grad_norm": 0.57901531457901,
      "learning_rate": 2.397434372481688e-07,
      "loss": 0.2853,
      "step": 51515
    },
    {
      "epoch": 0.9302392766089601,
      "grad_norm": 0.3291673958301544,
      "learning_rate": 2.391265059811132e-07,
      "loss": 0.1768,
      "step": 51520
    },
    {
      "epoch": 0.9303295560418607,
      "grad_norm": 0.48461225628852844,
      "learning_rate": 2.3851035990536066e-07,
      "loss": 0.2379,
      "step": 51525
    },
    {
      "epoch": 0.9304198354747615,
      "grad_norm": 0.4078727066516876,
      "learning_rate": 2.378949990704782e-07,
      "loss": 0.1943,
      "step": 51530
    },
    {
      "epoch": 0.9305101149076622,
      "grad_norm": 0.24602115154266357,
      "learning_rate": 2.3728042352596404e-07,
      "loss": 0.2172,
      "step": 51535
    },
    {
      "epoch": 0.930600394340563,
      "grad_norm": 0.34291091561317444,
      "learning_rate": 2.3666663332125972e-07,
      "loss": 0.1879,
      "step": 51540
    },
    {
      "epoch": 0.9306906737734636,
      "grad_norm": 0.5660107135772705,
      "learning_rate": 2.3605362850573688e-07,
      "loss": 0.1819,
      "step": 51545
    },
    {
      "epoch": 0.9307809532063643,
      "grad_norm": 0.5777392387390137,
      "learning_rate": 2.354414091287105e-07,
      "loss": 0.1868,
      "step": 51550
    },
    {
      "epoch": 0.9308712326392651,
      "grad_norm": 0.3865366578102112,
      "learning_rate": 2.3482997523942673e-07,
      "loss": 0.1685,
      "step": 51555
    },
    {
      "epoch": 0.9309615120721658,
      "grad_norm": 0.19842804968357086,
      "learning_rate": 2.3421932688707183e-07,
      "loss": 0.2207,
      "step": 51560
    },
    {
      "epoch": 0.9310517915050665,
      "grad_norm": 0.5594784021377563,
      "learning_rate": 2.3360946412076758e-07,
      "loss": 0.1626,
      "step": 51565
    },
    {
      "epoch": 0.9311420709379672,
      "grad_norm": 0.4227134883403778,
      "learning_rate": 2.3300038698957362e-07,
      "loss": 0.2835,
      "step": 51570
    },
    {
      "epoch": 0.9312323503708679,
      "grad_norm": 0.3799756169319153,
      "learning_rate": 2.323920955424863e-07,
      "loss": 0.1396,
      "step": 51575
    },
    {
      "epoch": 0.9313226298037687,
      "grad_norm": 0.3990386128425598,
      "learning_rate": 2.3178458982843655e-07,
      "loss": 0.1872,
      "step": 51580
    },
    {
      "epoch": 0.9314129092366693,
      "grad_norm": 0.24942395091056824,
      "learning_rate": 2.311778698962963e-07,
      "loss": 0.235,
      "step": 51585
    },
    {
      "epoch": 0.93150318866957,
      "grad_norm": 0.31736132502555847,
      "learning_rate": 2.305719357948688e-07,
      "loss": 0.1348,
      "step": 51590
    },
    {
      "epoch": 0.9315934681024708,
      "grad_norm": 0.41779983043670654,
      "learning_rate": 2.2996678757289836e-07,
      "loss": 0.2574,
      "step": 51595
    },
    {
      "epoch": 0.9316837475353715,
      "grad_norm": 0.4522228538990021,
      "learning_rate": 2.2936242527906495e-07,
      "loss": 0.3507,
      "step": 51600
    },
    {
      "epoch": 0.9317740269682722,
      "grad_norm": 0.52217698097229,
      "learning_rate": 2.2875884896198519e-07,
      "loss": 0.259,
      "step": 51605
    },
    {
      "epoch": 0.9318643064011729,
      "grad_norm": 0.3688105642795563,
      "learning_rate": 2.281560586702125e-07,
      "loss": 0.2284,
      "step": 51610
    },
    {
      "epoch": 0.9319545858340736,
      "grad_norm": 1.195549488067627,
      "learning_rate": 2.27554054452237e-07,
      "loss": 0.2717,
      "step": 51615
    },
    {
      "epoch": 0.9320448652669744,
      "grad_norm": 0.687444269657135,
      "learning_rate": 2.2695283635648435e-07,
      "loss": 0.2383,
      "step": 51620
    },
    {
      "epoch": 0.932135144699875,
      "grad_norm": 0.44127342104911804,
      "learning_rate": 2.2635240443132034e-07,
      "loss": 0.264,
      "step": 51625
    },
    {
      "epoch": 0.9322254241327758,
      "grad_norm": 0.4385707974433899,
      "learning_rate": 2.25752758725043e-07,
      "loss": 0.255,
      "step": 51630
    },
    {
      "epoch": 0.9323157035656765,
      "grad_norm": 0.9365828037261963,
      "learning_rate": 2.2515389928589038e-07,
      "loss": 0.2694,
      "step": 51635
    },
    {
      "epoch": 0.9324059829985772,
      "grad_norm": 0.40874359011650085,
      "learning_rate": 2.2455582616203732e-07,
      "loss": 0.2996,
      "step": 51640
    },
    {
      "epoch": 0.9324962624314779,
      "grad_norm": 0.44071194529533386,
      "learning_rate": 2.2395853940159417e-07,
      "loss": 0.2285,
      "step": 51645
    },
    {
      "epoch": 0.9325865418643786,
      "grad_norm": 0.3842937648296356,
      "learning_rate": 2.233620390526059e-07,
      "loss": 0.156,
      "step": 51650
    },
    {
      "epoch": 0.9326768212972794,
      "grad_norm": 0.5506059527397156,
      "learning_rate": 2.2276632516306074e-07,
      "loss": 0.3147,
      "step": 51655
    },
    {
      "epoch": 0.9327671007301801,
      "grad_norm": 0.6792159080505371,
      "learning_rate": 2.2217139778087594e-07,
      "loss": 0.2612,
      "step": 51660
    },
    {
      "epoch": 0.9328573801630807,
      "grad_norm": 0.4426134526729584,
      "learning_rate": 2.2157725695391207e-07,
      "loss": 0.1906,
      "step": 51665
    },
    {
      "epoch": 0.9329476595959815,
      "grad_norm": 0.6154745221138,
      "learning_rate": 2.2098390272996095e-07,
      "loss": 0.3561,
      "step": 51670
    },
    {
      "epoch": 0.9330379390288822,
      "grad_norm": 0.696847677230835,
      "learning_rate": 2.2039133515675327e-07,
      "loss": 0.1871,
      "step": 51675
    },
    {
      "epoch": 0.933128218461783,
      "grad_norm": 0.41318702697753906,
      "learning_rate": 2.1979955428195976e-07,
      "loss": 0.2813,
      "step": 51680
    },
    {
      "epoch": 0.9332184978946836,
      "grad_norm": 0.5265457034111023,
      "learning_rate": 2.1920856015318237e-07,
      "loss": 0.1673,
      "step": 51685
    },
    {
      "epoch": 0.9333087773275843,
      "grad_norm": 0.1904546171426773,
      "learning_rate": 2.1861835281796417e-07,
      "loss": 0.1848,
      "step": 51690
    },
    {
      "epoch": 0.9333990567604851,
      "grad_norm": 0.3450382351875305,
      "learning_rate": 2.1802893232378053e-07,
      "loss": 0.2669,
      "step": 51695
    },
    {
      "epoch": 0.9334893361933858,
      "grad_norm": 0.5738722681999207,
      "learning_rate": 2.1744029871804795e-07,
      "loss": 0.2131,
      "step": 51700
    },
    {
      "epoch": 0.9335796156262864,
      "grad_norm": 0.5521915555000305,
      "learning_rate": 2.1685245204811744e-07,
      "loss": 0.2703,
      "step": 51705
    },
    {
      "epoch": 0.9336698950591872,
      "grad_norm": 0.2938435971736908,
      "learning_rate": 2.1626539236127676e-07,
      "loss": 0.2245,
      "step": 51710
    },
    {
      "epoch": 0.9337601744920879,
      "grad_norm": 0.4499363601207733,
      "learning_rate": 2.156791197047503e-07,
      "loss": 0.2809,
      "step": 51715
    },
    {
      "epoch": 0.9338504539249887,
      "grad_norm": 2.3798131942749023,
      "learning_rate": 2.1509363412570038e-07,
      "loss": 0.3425,
      "step": 51720
    },
    {
      "epoch": 0.9339407333578893,
      "grad_norm": 0.4902951121330261,
      "learning_rate": 2.1450893567122378e-07,
      "loss": 0.2242,
      "step": 51725
    },
    {
      "epoch": 0.93403101279079,
      "grad_norm": 0.28112733364105225,
      "learning_rate": 2.1392502438835726e-07,
      "loss": 0.2116,
      "step": 51730
    },
    {
      "epoch": 0.9341212922236908,
      "grad_norm": 0.4254879057407379,
      "learning_rate": 2.1334190032406886e-07,
      "loss": 0.256,
      "step": 51735
    },
    {
      "epoch": 0.9342115716565915,
      "grad_norm": 0.41759738326072693,
      "learning_rate": 2.1275956352526994e-07,
      "loss": 0.1774,
      "step": 51740
    },
    {
      "epoch": 0.9343018510894922,
      "grad_norm": 0.2610630393028259,
      "learning_rate": 2.1217801403880412e-07,
      "loss": 0.1912,
      "step": 51745
    },
    {
      "epoch": 0.9343921305223929,
      "grad_norm": 0.3194853961467743,
      "learning_rate": 2.1159725191145287e-07,
      "loss": 0.2587,
      "step": 51750
    },
    {
      "epoch": 0.9344824099552936,
      "grad_norm": 0.5504977107048035,
      "learning_rate": 2.1101727718993437e-07,
      "loss": 0.2209,
      "step": 51755
    },
    {
      "epoch": 0.9345726893881944,
      "grad_norm": 1.0065027475357056,
      "learning_rate": 2.1043808992090243e-07,
      "loss": 0.1515,
      "step": 51760
    },
    {
      "epoch": 0.934662968821095,
      "grad_norm": 0.24719925224781036,
      "learning_rate": 2.0985969015094864e-07,
      "loss": 0.1959,
      "step": 51765
    },
    {
      "epoch": 0.9347532482539957,
      "grad_norm": 0.23635120689868927,
      "learning_rate": 2.092820779266014e-07,
      "loss": 0.1884,
      "step": 51770
    },
    {
      "epoch": 0.9348435276868965,
      "grad_norm": 0.17379799485206604,
      "learning_rate": 2.0870525329432678e-07,
      "loss": 0.1901,
      "step": 51775
    },
    {
      "epoch": 0.9349338071197972,
      "grad_norm": 0.17449960112571716,
      "learning_rate": 2.0812921630052442e-07,
      "loss": 0.1689,
      "step": 51780
    },
    {
      "epoch": 0.9350240865526979,
      "grad_norm": 0.24730928242206573,
      "learning_rate": 2.0755396699153385e-07,
      "loss": 0.192,
      "step": 51785
    },
    {
      "epoch": 0.9351143659855986,
      "grad_norm": 0.24990248680114746,
      "learning_rate": 2.06979505413627e-07,
      "loss": 0.1725,
      "step": 51790
    },
    {
      "epoch": 0.9352046454184993,
      "grad_norm": 0.37935662269592285,
      "learning_rate": 2.0640583161301796e-07,
      "loss": 0.2294,
      "step": 51795
    },
    {
      "epoch": 0.9352949248514001,
      "grad_norm": 0.5428740382194519,
      "learning_rate": 2.058329456358521e-07,
      "loss": 0.2519,
      "step": 51800
    },
    {
      "epoch": 0.9353852042843007,
      "grad_norm": 0.3672719895839691,
      "learning_rate": 2.0526084752821695e-07,
      "loss": 0.1027,
      "step": 51805
    },
    {
      "epoch": 0.9354754837172015,
      "grad_norm": 0.43215203285217285,
      "learning_rate": 2.046895373361313e-07,
      "loss": 0.2256,
      "step": 51810
    },
    {
      "epoch": 0.9355657631501022,
      "grad_norm": 0.5134435892105103,
      "learning_rate": 2.0411901510555386e-07,
      "loss": 0.3509,
      "step": 51815
    },
    {
      "epoch": 0.935656042583003,
      "grad_norm": 0.5209430456161499,
      "learning_rate": 2.0354928088237802e-07,
      "loss": 0.1723,
      "step": 51820
    },
    {
      "epoch": 0.9357463220159036,
      "grad_norm": 0.33558085560798645,
      "learning_rate": 2.0298033471243595e-07,
      "loss": 0.2411,
      "step": 51825
    },
    {
      "epoch": 0.9358366014488043,
      "grad_norm": 0.27937522530555725,
      "learning_rate": 2.0241217664149437e-07,
      "loss": 0.1683,
      "step": 51830
    },
    {
      "epoch": 0.9359268808817051,
      "grad_norm": 0.3353129029273987,
      "learning_rate": 2.0184480671525786e-07,
      "loss": 0.2219,
      "step": 51835
    },
    {
      "epoch": 0.9360171603146058,
      "grad_norm": 0.8231786489486694,
      "learning_rate": 2.0127822497936655e-07,
      "loss": 0.2422,
      "step": 51840
    },
    {
      "epoch": 0.9361074397475064,
      "grad_norm": 0.696200430393219,
      "learning_rate": 2.0071243147940067e-07,
      "loss": 0.3133,
      "step": 51845
    },
    {
      "epoch": 0.9361977191804072,
      "grad_norm": 0.3378933072090149,
      "learning_rate": 2.0014742626087046e-07,
      "loss": 0.1784,
      "step": 51850
    },
    {
      "epoch": 0.9362879986133079,
      "grad_norm": 0.2539663016796112,
      "learning_rate": 1.9958320936922848e-07,
      "loss": 0.1111,
      "step": 51855
    },
    {
      "epoch": 0.9363782780462087,
      "grad_norm": 0.19143852591514587,
      "learning_rate": 1.9901978084986062e-07,
      "loss": 0.1522,
      "step": 51860
    },
    {
      "epoch": 0.9364685574791093,
      "grad_norm": 0.4747665226459503,
      "learning_rate": 1.9845714074809062e-07,
      "loss": 0.1578,
      "step": 51865
    },
    {
      "epoch": 0.93655883691201,
      "grad_norm": 0.29267823696136475,
      "learning_rate": 1.978952891091812e-07,
      "loss": 0.1915,
      "step": 51870
    },
    {
      "epoch": 0.9366491163449108,
      "grad_norm": 0.4691658616065979,
      "learning_rate": 1.973342259783273e-07,
      "loss": 0.258,
      "step": 51875
    },
    {
      "epoch": 0.9367393957778115,
      "grad_norm": 1.3541239500045776,
      "learning_rate": 1.967739514006628e-07,
      "loss": 0.2346,
      "step": 51880
    },
    {
      "epoch": 0.9368296752107121,
      "grad_norm": 0.4629501700401306,
      "learning_rate": 1.962144654212572e-07,
      "loss": 0.2451,
      "step": 51885
    },
    {
      "epoch": 0.9369199546436129,
      "grad_norm": 0.5224490761756897,
      "learning_rate": 1.9565576808511788e-07,
      "loss": 0.2326,
      "step": 51890
    },
    {
      "epoch": 0.9370102340765136,
      "grad_norm": 0.40172770619392395,
      "learning_rate": 1.9509785943718662e-07,
      "loss": 0.2084,
      "step": 51895
    },
    {
      "epoch": 0.9371005135094144,
      "grad_norm": 0.22561655938625336,
      "learning_rate": 1.945407395223442e-07,
      "loss": 0.2111,
      "step": 51900
    },
    {
      "epoch": 0.937190792942315,
      "grad_norm": 0.5190891027450562,
      "learning_rate": 1.93984408385407e-07,
      "loss": 0.3468,
      "step": 51905
    },
    {
      "epoch": 0.9372810723752157,
      "grad_norm": 0.9866564869880676,
      "learning_rate": 1.934288660711281e-07,
      "loss": 0.2527,
      "step": 51910
    },
    {
      "epoch": 0.9373713518081165,
      "grad_norm": 0.8047907948493958,
      "learning_rate": 1.928741126241951e-07,
      "loss": 0.1663,
      "step": 51915
    },
    {
      "epoch": 0.9374616312410172,
      "grad_norm": 0.6488373875617981,
      "learning_rate": 1.9232014808923672e-07,
      "loss": 0.2403,
      "step": 51920
    },
    {
      "epoch": 0.937551910673918,
      "grad_norm": 0.36024126410484314,
      "learning_rate": 1.9176697251081177e-07,
      "loss": 0.1215,
      "step": 51925
    },
    {
      "epoch": 0.9376421901068186,
      "grad_norm": 0.5580694675445557,
      "learning_rate": 1.9121458593342246e-07,
      "loss": 0.2512,
      "step": 51930
    },
    {
      "epoch": 0.9377324695397193,
      "grad_norm": 0.3380827307701111,
      "learning_rate": 1.9066298840150211e-07,
      "loss": 0.2354,
      "step": 51935
    },
    {
      "epoch": 0.9378227489726201,
      "grad_norm": 0.4437563419342041,
      "learning_rate": 1.901121799594241e-07,
      "loss": 0.3053,
      "step": 51940
    },
    {
      "epoch": 0.9379130284055208,
      "grad_norm": 0.2973403334617615,
      "learning_rate": 1.8956216065149636e-07,
      "loss": 0.2504,
      "step": 51945
    },
    {
      "epoch": 0.9380033078384215,
      "grad_norm": 0.33137649297714233,
      "learning_rate": 1.8901293052196456e-07,
      "loss": 0.1983,
      "step": 51950
    },
    {
      "epoch": 0.9380935872713222,
      "grad_norm": 0.24417562782764435,
      "learning_rate": 1.8846448961500896e-07,
      "loss": 0.12,
      "step": 51955
    },
    {
      "epoch": 0.9381838667042229,
      "grad_norm": 0.5119628310203552,
      "learning_rate": 1.879168379747487e-07,
      "loss": 0.1649,
      "step": 51960
    },
    {
      "epoch": 0.9382741461371237,
      "grad_norm": 0.4930247366428375,
      "learning_rate": 1.873699756452374e-07,
      "loss": 0.2816,
      "step": 51965
    },
    {
      "epoch": 0.9383644255700243,
      "grad_norm": 0.19329699873924255,
      "learning_rate": 1.8682390267046768e-07,
      "loss": 0.2226,
      "step": 51970
    },
    {
      "epoch": 0.938454705002925,
      "grad_norm": 0.5004608035087585,
      "learning_rate": 1.862786190943666e-07,
      "loss": 0.2517,
      "step": 51975
    },
    {
      "epoch": 0.9385449844358258,
      "grad_norm": 0.401940256357193,
      "learning_rate": 1.857341249607969e-07,
      "loss": 0.1777,
      "step": 51980
    },
    {
      "epoch": 0.9386352638687265,
      "grad_norm": 0.3844544291496277,
      "learning_rate": 1.8519042031356127e-07,
      "loss": 0.2107,
      "step": 51985
    },
    {
      "epoch": 0.9387255433016272,
      "grad_norm": 0.48071059584617615,
      "learning_rate": 1.8464750519639473e-07,
      "loss": 0.2061,
      "step": 51990
    },
    {
      "epoch": 0.9388158227345279,
      "grad_norm": 0.395354300737381,
      "learning_rate": 1.8410537965297237e-07,
      "loss": 0.1423,
      "step": 51995
    },
    {
      "epoch": 0.9389061021674286,
      "grad_norm": 0.22170841693878174,
      "learning_rate": 1.8356404372690262e-07,
      "loss": 0.1431,
      "step": 52000
    },
    {
      "epoch": 0.9389963816003294,
      "grad_norm": 0.31132036447525024,
      "learning_rate": 1.830234974617351e-07,
      "loss": 0.1529,
      "step": 52005
    },
    {
      "epoch": 0.93908666103323,
      "grad_norm": 0.4811148941516876,
      "learning_rate": 1.8248374090094944e-07,
      "loss": 0.1695,
      "step": 52010
    },
    {
      "epoch": 0.9391769404661308,
      "grad_norm": 0.5122590661048889,
      "learning_rate": 1.8194477408796875e-07,
      "loss": 0.2564,
      "step": 52015
    },
    {
      "epoch": 0.9392672198990315,
      "grad_norm": 0.5909358859062195,
      "learning_rate": 1.8140659706614493e-07,
      "loss": 0.1825,
      "step": 52020
    },
    {
      "epoch": 0.9393574993319322,
      "grad_norm": 0.34511274099349976,
      "learning_rate": 1.8086920987877344e-07,
      "loss": 0.2724,
      "step": 52025
    },
    {
      "epoch": 0.9394477787648329,
      "grad_norm": 0.7616056799888611,
      "learning_rate": 1.8033261256908075e-07,
      "loss": 0.2216,
      "step": 52030
    },
    {
      "epoch": 0.9395380581977336,
      "grad_norm": 0.4209730625152588,
      "learning_rate": 1.7979680518023346e-07,
      "loss": 0.2234,
      "step": 52035
    },
    {
      "epoch": 0.9396283376306344,
      "grad_norm": 0.5408231019973755,
      "learning_rate": 1.7926178775533488e-07,
      "loss": 0.1385,
      "step": 52040
    },
    {
      "epoch": 0.9397186170635351,
      "grad_norm": 0.4944550693035126,
      "learning_rate": 1.7872756033742055e-07,
      "loss": 0.2979,
      "step": 52045
    },
    {
      "epoch": 0.9398088964964357,
      "grad_norm": 0.30177101492881775,
      "learning_rate": 1.7819412296946725e-07,
      "loss": 0.2299,
      "step": 52050
    },
    {
      "epoch": 0.9398991759293365,
      "grad_norm": 0.23153111338615417,
      "learning_rate": 1.7766147569438397e-07,
      "loss": 0.192,
      "step": 52055
    },
    {
      "epoch": 0.9399894553622372,
      "grad_norm": 0.4247056841850281,
      "learning_rate": 1.7712961855502087e-07,
      "loss": 0.3326,
      "step": 52060
    },
    {
      "epoch": 0.940079734795138,
      "grad_norm": 0.3396017849445343,
      "learning_rate": 1.7659855159416038e-07,
      "loss": 0.2892,
      "step": 52065
    },
    {
      "epoch": 0.9401700142280386,
      "grad_norm": 0.5537087321281433,
      "learning_rate": 1.7606827485452395e-07,
      "loss": 0.2346,
      "step": 52070
    },
    {
      "epoch": 0.9402602936609393,
      "grad_norm": 0.6220576167106628,
      "learning_rate": 1.7553878837876737e-07,
      "loss": 0.2804,
      "step": 52075
    },
    {
      "epoch": 0.9403505730938401,
      "grad_norm": 0.5737428069114685,
      "learning_rate": 1.750100922094844e-07,
      "loss": 0.1685,
      "step": 52080
    },
    {
      "epoch": 0.9404408525267408,
      "grad_norm": 0.507656455039978,
      "learning_rate": 1.7448218638920545e-07,
      "loss": 0.2028,
      "step": 52085
    },
    {
      "epoch": 0.9405311319596414,
      "grad_norm": 0.642570972442627,
      "learning_rate": 1.7395507096039654e-07,
      "loss": 0.1771,
      "step": 52090
    },
    {
      "epoch": 0.9406214113925422,
      "grad_norm": 0.56353759765625,
      "learning_rate": 1.7342874596545822e-07,
      "loss": 0.1809,
      "step": 52095
    },
    {
      "epoch": 0.9407116908254429,
      "grad_norm": 0.2150426059961319,
      "learning_rate": 1.7290321144673216e-07,
      "loss": 0.2133,
      "step": 52100
    },
    {
      "epoch": 0.9408019702583437,
      "grad_norm": 0.30578625202178955,
      "learning_rate": 1.7237846744649344e-07,
      "loss": 0.2366,
      "step": 52105
    },
    {
      "epoch": 0.9408922496912443,
      "grad_norm": 0.7995226979255676,
      "learning_rate": 1.7185451400695275e-07,
      "loss": 0.2499,
      "step": 52110
    },
    {
      "epoch": 0.940982529124145,
      "grad_norm": 0.4063202142715454,
      "learning_rate": 1.7133135117025856e-07,
      "loss": 0.2984,
      "step": 52115
    },
    {
      "epoch": 0.9410728085570458,
      "grad_norm": 0.5872689485549927,
      "learning_rate": 1.7080897897849723e-07,
      "loss": 0.2491,
      "step": 52120
    },
    {
      "epoch": 0.9411630879899465,
      "grad_norm": 1.0243349075317383,
      "learning_rate": 1.7028739747368738e-07,
      "loss": 0.1595,
      "step": 52125
    },
    {
      "epoch": 0.9412533674228472,
      "grad_norm": 0.46754711866378784,
      "learning_rate": 1.6976660669778655e-07,
      "loss": 0.243,
      "step": 52130
    },
    {
      "epoch": 0.9413436468557479,
      "grad_norm": 0.2676316797733307,
      "learning_rate": 1.6924660669269122e-07,
      "loss": 0.1927,
      "step": 52135
    },
    {
      "epoch": 0.9414339262886486,
      "grad_norm": 0.4048113226890564,
      "learning_rate": 1.6872739750022903e-07,
      "loss": 0.2179,
      "step": 52140
    },
    {
      "epoch": 0.9415242057215494,
      "grad_norm": 0.4328083395957947,
      "learning_rate": 1.682089791621677e-07,
      "loss": 0.2187,
      "step": 52145
    },
    {
      "epoch": 0.94161448515445,
      "grad_norm": 0.48077961802482605,
      "learning_rate": 1.6769135172020945e-07,
      "loss": 0.122,
      "step": 52150
    },
    {
      "epoch": 0.9417047645873508,
      "grad_norm": 0.5391274094581604,
      "learning_rate": 1.6717451521599536e-07,
      "loss": 0.1919,
      "step": 52155
    },
    {
      "epoch": 0.9417950440202515,
      "grad_norm": 0.4807625114917755,
      "learning_rate": 1.6665846969109888e-07,
      "loss": 0.2772,
      "step": 52160
    },
    {
      "epoch": 0.9418853234531522,
      "grad_norm": 1.064616322517395,
      "learning_rate": 1.6614321518703347e-07,
      "loss": 0.2024,
      "step": 52165
    },
    {
      "epoch": 0.9419756028860529,
      "grad_norm": 0.35680803656578064,
      "learning_rate": 1.6562875174524707e-07,
      "loss": 0.2104,
      "step": 52170
    },
    {
      "epoch": 0.9420658823189536,
      "grad_norm": 0.32618042826652527,
      "learning_rate": 1.6511507940712546e-07,
      "loss": 0.2215,
      "step": 52175
    },
    {
      "epoch": 0.9421561617518543,
      "grad_norm": 0.42294156551361084,
      "learning_rate": 1.6460219821398894e-07,
      "loss": 0.248,
      "step": 52180
    },
    {
      "epoch": 0.9422464411847551,
      "grad_norm": 0.3701134920120239,
      "learning_rate": 1.640901082070956e-07,
      "loss": 0.2675,
      "step": 52185
    },
    {
      "epoch": 0.9423367206176557,
      "grad_norm": 1.1385633945465088,
      "learning_rate": 1.6357880942763804e-07,
      "loss": 0.2164,
      "step": 52190
    },
    {
      "epoch": 0.9424270000505565,
      "grad_norm": 0.3978947103023529,
      "learning_rate": 1.6306830191674783e-07,
      "loss": 0.2493,
      "step": 52195
    },
    {
      "epoch": 0.9425172794834572,
      "grad_norm": 0.3072848320007324,
      "learning_rate": 1.62558585715491e-07,
      "loss": 0.1876,
      "step": 52200
    },
    {
      "epoch": 0.942607558916358,
      "grad_norm": 0.39237838983535767,
      "learning_rate": 1.620496608648725e-07,
      "loss": 0.2458,
      "step": 52205
    },
    {
      "epoch": 0.9426978383492586,
      "grad_norm": 0.49989035725593567,
      "learning_rate": 1.615415274058285e-07,
      "loss": 0.2544,
      "step": 52210
    },
    {
      "epoch": 0.9427881177821593,
      "grad_norm": 0.638972818851471,
      "learning_rate": 1.6103418537923742e-07,
      "loss": 0.1837,
      "step": 52215
    },
    {
      "epoch": 0.9428783972150601,
      "grad_norm": 0.32173827290534973,
      "learning_rate": 1.6052763482590995e-07,
      "loss": 0.1646,
      "step": 52220
    },
    {
      "epoch": 0.9429686766479608,
      "grad_norm": 0.14592932164669037,
      "learning_rate": 1.6002187578659344e-07,
      "loss": 0.1162,
      "step": 52225
    },
    {
      "epoch": 0.9430589560808614,
      "grad_norm": 0.47503066062927246,
      "learning_rate": 1.595169083019732e-07,
      "loss": 0.2758,
      "step": 52230
    },
    {
      "epoch": 0.9431492355137622,
      "grad_norm": 0.44942447543144226,
      "learning_rate": 1.5901273241267e-07,
      "loss": 0.349,
      "step": 52235
    },
    {
      "epoch": 0.9432395149466629,
      "grad_norm": 0.4225345551967621,
      "learning_rate": 1.5850934815924367e-07,
      "loss": 0.2135,
      "step": 52240
    },
    {
      "epoch": 0.9433297943795637,
      "grad_norm": 0.4064311981201172,
      "learning_rate": 1.5800675558218404e-07,
      "loss": 0.2737,
      "step": 52245
    },
    {
      "epoch": 0.9434200738124643,
      "grad_norm": 0.2946653366088867,
      "learning_rate": 1.575049547219232e-07,
      "loss": 0.1376,
      "step": 52250
    },
    {
      "epoch": 0.943510353245365,
      "grad_norm": 0.36680835485458374,
      "learning_rate": 1.5700394561882658e-07,
      "loss": 0.1992,
      "step": 52255
    },
    {
      "epoch": 0.9436006326782658,
      "grad_norm": 0.43676671385765076,
      "learning_rate": 1.565037283131976e-07,
      "loss": 0.1873,
      "step": 52260
    },
    {
      "epoch": 0.9436909121111665,
      "grad_norm": 0.4742371439933777,
      "learning_rate": 1.56004302845274e-07,
      "loss": 0.3535,
      "step": 52265
    },
    {
      "epoch": 0.9437811915440671,
      "grad_norm": 0.2743320167064667,
      "learning_rate": 1.5550566925523258e-07,
      "loss": 0.1865,
      "step": 52270
    },
    {
      "epoch": 0.9438714709769679,
      "grad_norm": 0.5959087014198303,
      "learning_rate": 1.5500782758318233e-07,
      "loss": 0.2989,
      "step": 52275
    },
    {
      "epoch": 0.9439617504098686,
      "grad_norm": 0.7350186705589294,
      "learning_rate": 1.5451077786917345e-07,
      "loss": 0.2391,
      "step": 52280
    },
    {
      "epoch": 0.9440520298427694,
      "grad_norm": 0.5229212045669556,
      "learning_rate": 1.5401452015318728e-07,
      "loss": 0.1791,
      "step": 52285
    },
    {
      "epoch": 0.94414230927567,
      "grad_norm": 0.39683276414871216,
      "learning_rate": 1.5351905447514747e-07,
      "loss": 0.323,
      "step": 52290
    },
    {
      "epoch": 0.9442325887085707,
      "grad_norm": 0.37934693694114685,
      "learning_rate": 1.5302438087490767e-07,
      "loss": 0.2551,
      "step": 52295
    },
    {
      "epoch": 0.9443228681414715,
      "grad_norm": 0.5581970810890198,
      "learning_rate": 1.5253049939226272e-07,
      "loss": 0.2526,
      "step": 52300
    },
    {
      "epoch": 0.9444131475743722,
      "grad_norm": 0.6557458639144897,
      "learning_rate": 1.5203741006694083e-07,
      "loss": 0.2037,
      "step": 52305
    },
    {
      "epoch": 0.944503427007273,
      "grad_norm": 0.5137622952461243,
      "learning_rate": 1.5154511293860807e-07,
      "loss": 0.2785,
      "step": 52310
    },
    {
      "epoch": 0.9445937064401736,
      "grad_norm": 0.47347983717918396,
      "learning_rate": 1.51053608046865e-07,
      "loss": 0.2556,
      "step": 52315
    },
    {
      "epoch": 0.9446839858730743,
      "grad_norm": 0.42270901799201965,
      "learning_rate": 1.50562895431251e-07,
      "loss": 0.2347,
      "step": 52320
    },
    {
      "epoch": 0.9447742653059751,
      "grad_norm": 0.44284293055534363,
      "learning_rate": 1.5007297513123908e-07,
      "loss": 0.314,
      "step": 52325
    },
    {
      "epoch": 0.9448645447388758,
      "grad_norm": 0.5739048719406128,
      "learning_rate": 1.4958384718624097e-07,
      "loss": 0.2303,
      "step": 52330
    },
    {
      "epoch": 0.9449548241717765,
      "grad_norm": 0.5385386943817139,
      "learning_rate": 1.49095511635603e-07,
      "loss": 0.2542,
      "step": 52335
    },
    {
      "epoch": 0.9450451036046772,
      "grad_norm": 0.6142498254776001,
      "learning_rate": 1.4860796851860814e-07,
      "loss": 0.3362,
      "step": 52340
    },
    {
      "epoch": 0.9451353830375779,
      "grad_norm": 0.28703126311302185,
      "learning_rate": 1.481212178744762e-07,
      "loss": 0.2327,
      "step": 52345
    },
    {
      "epoch": 0.9452256624704787,
      "grad_norm": 0.5515416264533997,
      "learning_rate": 1.4763525974236027e-07,
      "loss": 0.2608,
      "step": 52350
    },
    {
      "epoch": 0.9453159419033793,
      "grad_norm": 0.4463517963886261,
      "learning_rate": 1.4715009416135574e-07,
      "loss": 0.2052,
      "step": 52355
    },
    {
      "epoch": 0.94540622133628,
      "grad_norm": 0.36427173018455505,
      "learning_rate": 1.4666572117048917e-07,
      "loss": 0.3779,
      "step": 52360
    },
    {
      "epoch": 0.9454965007691808,
      "grad_norm": 0.3629922568798065,
      "learning_rate": 1.4618214080872272e-07,
      "loss": 0.2401,
      "step": 52365
    },
    {
      "epoch": 0.9455867802020815,
      "grad_norm": 0.3709052801132202,
      "learning_rate": 1.4569935311496087e-07,
      "loss": 0.1854,
      "step": 52370
    },
    {
      "epoch": 0.9456770596349822,
      "grad_norm": 0.698991596698761,
      "learning_rate": 1.4521735812803805e-07,
      "loss": 0.2725,
      "step": 52375
    },
    {
      "epoch": 0.9457673390678829,
      "grad_norm": 0.49911487102508545,
      "learning_rate": 1.4473615588672662e-07,
      "loss": 0.1953,
      "step": 52380
    },
    {
      "epoch": 0.9458576185007836,
      "grad_norm": 0.16943359375,
      "learning_rate": 1.4425574642973782e-07,
      "loss": 0.1517,
      "step": 52385
    },
    {
      "epoch": 0.9459478979336844,
      "grad_norm": 0.4964551329612732,
      "learning_rate": 1.437761297957152e-07,
      "loss": 0.2814,
      "step": 52390
    },
    {
      "epoch": 0.946038177366585,
      "grad_norm": 0.23414237797260284,
      "learning_rate": 1.4329730602324122e-07,
      "loss": 0.2067,
      "step": 52395
    },
    {
      "epoch": 0.9461284567994858,
      "grad_norm": 0.3901708722114563,
      "learning_rate": 1.4281927515083504e-07,
      "loss": 0.1551,
      "step": 52400
    },
    {
      "epoch": 0.9462187362323865,
      "grad_norm": 0.42421460151672363,
      "learning_rate": 1.4234203721694816e-07,
      "loss": 0.2296,
      "step": 52405
    },
    {
      "epoch": 0.9463090156652872,
      "grad_norm": 0.3116353750228882,
      "learning_rate": 1.4186559225997322e-07,
      "loss": 0.1827,
      "step": 52410
    },
    {
      "epoch": 0.9463992950981879,
      "grad_norm": 0.18112727999687195,
      "learning_rate": 1.4138994031823506e-07,
      "loss": 0.1922,
      "step": 52415
    },
    {
      "epoch": 0.9464895745310886,
      "grad_norm": 0.4041883945465088,
      "learning_rate": 1.4091508142999865e-07,
      "loss": 0.2435,
      "step": 52420
    },
    {
      "epoch": 0.9465798539639894,
      "grad_norm": 0.31055948138237,
      "learning_rate": 1.4044101563346014e-07,
      "loss": 0.2331,
      "step": 52425
    },
    {
      "epoch": 0.9466701333968901,
      "grad_norm": 0.2543698251247406,
      "learning_rate": 1.3996774296675564e-07,
      "loss": 0.2201,
      "step": 52430
    },
    {
      "epoch": 0.9467604128297907,
      "grad_norm": 0.37123754620552063,
      "learning_rate": 1.3949526346795695e-07,
      "loss": 0.1751,
      "step": 52435
    },
    {
      "epoch": 0.9468506922626915,
      "grad_norm": 0.38278284668922424,
      "learning_rate": 1.3902357717507254e-07,
      "loss": 0.3449,
      "step": 52440
    },
    {
      "epoch": 0.9469409716955922,
      "grad_norm": 0.6106435060501099,
      "learning_rate": 1.3855268412604427e-07,
      "loss": 0.2112,
      "step": 52445
    },
    {
      "epoch": 0.947031251128493,
      "grad_norm": 0.6555020213127136,
      "learning_rate": 1.3808258435875411e-07,
      "loss": 0.2682,
      "step": 52450
    },
    {
      "epoch": 0.9471215305613936,
      "grad_norm": 0.5614919066429138,
      "learning_rate": 1.376132779110151e-07,
      "loss": 0.2209,
      "step": 52455
    },
    {
      "epoch": 0.9472118099942943,
      "grad_norm": 0.22632837295532227,
      "learning_rate": 1.371447648205826e-07,
      "loss": 0.2477,
      "step": 52460
    },
    {
      "epoch": 0.9473020894271951,
      "grad_norm": 0.39801183342933655,
      "learning_rate": 1.3667704512514311e-07,
      "loss": 0.1773,
      "step": 52465
    },
    {
      "epoch": 0.9473923688600958,
      "grad_norm": 0.33802229166030884,
      "learning_rate": 1.3621011886232327e-07,
      "loss": 0.2323,
      "step": 52470
    },
    {
      "epoch": 0.9474826482929964,
      "grad_norm": 0.316730260848999,
      "learning_rate": 1.3574398606968186e-07,
      "loss": 0.2709,
      "step": 52475
    },
    {
      "epoch": 0.9475729277258972,
      "grad_norm": 0.6402896642684937,
      "learning_rate": 1.352786467847178e-07,
      "loss": 0.1968,
      "step": 52480
    },
    {
      "epoch": 0.9476632071587979,
      "grad_norm": 0.5211960673332214,
      "learning_rate": 1.3481410104486114e-07,
      "loss": 0.1975,
      "step": 52485
    },
    {
      "epoch": 0.9477534865916987,
      "grad_norm": 1.256142497062683,
      "learning_rate": 1.3435034888748534e-07,
      "loss": 0.2066,
      "step": 52490
    },
    {
      "epoch": 0.9478437660245993,
      "grad_norm": 0.3416600525379181,
      "learning_rate": 1.3388739034989274e-07,
      "loss": 0.1517,
      "step": 52495
    },
    {
      "epoch": 0.9479340454575,
      "grad_norm": 0.33880066871643066,
      "learning_rate": 1.334252254693258e-07,
      "loss": 0.2901,
      "step": 52500
    },
    {
      "epoch": 0.9480243248904008,
      "grad_norm": 0.46627411246299744,
      "learning_rate": 1.3296385428296255e-07,
      "loss": 0.3176,
      "step": 52505
    },
    {
      "epoch": 0.9481146043233015,
      "grad_norm": 0.42374828457832336,
      "learning_rate": 1.3250327682791663e-07,
      "loss": 0.2587,
      "step": 52510
    },
    {
      "epoch": 0.9482048837562022,
      "grad_norm": 0.4953526258468628,
      "learning_rate": 1.3204349314123954e-07,
      "loss": 0.2192,
      "step": 52515
    },
    {
      "epoch": 0.9482951631891029,
      "grad_norm": 0.2568039298057556,
      "learning_rate": 1.3158450325991613e-07,
      "loss": 0.1924,
      "step": 52520
    },
    {
      "epoch": 0.9483854426220036,
      "grad_norm": 0.22638952732086182,
      "learning_rate": 1.3112630722086795e-07,
      "loss": 0.1758,
      "step": 52525
    },
    {
      "epoch": 0.9484757220549044,
      "grad_norm": 0.37685781717300415,
      "learning_rate": 1.3066890506095552e-07,
      "loss": 0.1785,
      "step": 52530
    },
    {
      "epoch": 0.948566001487805,
      "grad_norm": 0.4343778192996979,
      "learning_rate": 1.3021229681697388e-07,
      "loss": 0.1827,
      "step": 52535
    },
    {
      "epoch": 0.9486562809207058,
      "grad_norm": 0.7477774024009705,
      "learning_rate": 1.2975648252565032e-07,
      "loss": 0.2522,
      "step": 52540
    },
    {
      "epoch": 0.9487465603536065,
      "grad_norm": 0.4742632806301117,
      "learning_rate": 1.2930146222365657e-07,
      "loss": 0.2197,
      "step": 52545
    },
    {
      "epoch": 0.9488368397865072,
      "grad_norm": 0.6767007112503052,
      "learning_rate": 1.2884723594759118e-07,
      "loss": 0.1603,
      "step": 52550
    },
    {
      "epoch": 0.9489271192194079,
      "grad_norm": 0.46887606382369995,
      "learning_rate": 1.2839380373399712e-07,
      "loss": 0.3118,
      "step": 52555
    },
    {
      "epoch": 0.9490173986523086,
      "grad_norm": 0.45550402998924255,
      "learning_rate": 1.2794116561934743e-07,
      "loss": 0.2259,
      "step": 52560
    },
    {
      "epoch": 0.9491076780852093,
      "grad_norm": 0.39493048191070557,
      "learning_rate": 1.2748932164005302e-07,
      "loss": 0.2386,
      "step": 52565
    },
    {
      "epoch": 0.9491979575181101,
      "grad_norm": 0.25530895590782166,
      "learning_rate": 1.2703827183246364e-07,
      "loss": 0.214,
      "step": 52570
    },
    {
      "epoch": 0.9492882369510107,
      "grad_norm": 0.2733272612094879,
      "learning_rate": 1.2658801623286254e-07,
      "loss": 0.1374,
      "step": 52575
    },
    {
      "epoch": 0.9493785163839115,
      "grad_norm": 0.43154987692832947,
      "learning_rate": 1.261385548774674e-07,
      "loss": 0.1989,
      "step": 52580
    },
    {
      "epoch": 0.9494687958168122,
      "grad_norm": 0.5476710796356201,
      "learning_rate": 1.2568988780243708e-07,
      "loss": 0.1594,
      "step": 52585
    },
    {
      "epoch": 0.949559075249713,
      "grad_norm": 0.7292976379394531,
      "learning_rate": 1.2524201504386157e-07,
      "loss": 0.2246,
      "step": 52590
    },
    {
      "epoch": 0.9496493546826136,
      "grad_norm": 0.30767586827278137,
      "learning_rate": 1.2479493663776875e-07,
      "loss": 0.2405,
      "step": 52595
    },
    {
      "epoch": 0.9497396341155143,
      "grad_norm": 0.5088165998458862,
      "learning_rate": 1.243486526201254e-07,
      "loss": 0.2151,
      "step": 52600
    },
    {
      "epoch": 0.9498299135484151,
      "grad_norm": 0.3825433552265167,
      "learning_rate": 1.2390316302682837e-07,
      "loss": 0.2879,
      "step": 52605
    },
    {
      "epoch": 0.9499201929813158,
      "grad_norm": 0.35412609577178955,
      "learning_rate": 1.2345846789371674e-07,
      "loss": 0.2937,
      "step": 52610
    },
    {
      "epoch": 0.9500104724142164,
      "grad_norm": 1.0277363061904907,
      "learning_rate": 1.230145672565608e-07,
      "loss": 0.2591,
      "step": 52615
    },
    {
      "epoch": 0.9501007518471172,
      "grad_norm": 1.2222204208374023,
      "learning_rate": 1.225714611510709e-07,
      "loss": 0.2046,
      "step": 52620
    },
    {
      "epoch": 0.9501910312800179,
      "grad_norm": 0.31255319714546204,
      "learning_rate": 1.221291496128918e-07,
      "loss": 0.2869,
      "step": 52625
    },
    {
      "epoch": 0.9502813107129187,
      "grad_norm": 0.5221478343009949,
      "learning_rate": 1.2168763267760175e-07,
      "loss": 0.2165,
      "step": 52630
    },
    {
      "epoch": 0.9503715901458193,
      "grad_norm": 0.2570783495903015,
      "learning_rate": 1.2124691038072124e-07,
      "loss": 0.2298,
      "step": 52635
    },
    {
      "epoch": 0.95046186957872,
      "grad_norm": 0.4935098886489868,
      "learning_rate": 1.2080698275770074e-07,
      "loss": 0.1582,
      "step": 52640
    },
    {
      "epoch": 0.9505521490116208,
      "grad_norm": 0.5720986723899841,
      "learning_rate": 1.203678498439287e-07,
      "loss": 0.1775,
      "step": 52645
    },
    {
      "epoch": 0.9506424284445215,
      "grad_norm": 0.4462292492389679,
      "learning_rate": 1.1992951167473344e-07,
      "loss": 0.231,
      "step": 52650
    },
    {
      "epoch": 0.9507327078774221,
      "grad_norm": 0.5290933847427368,
      "learning_rate": 1.1949196828537235e-07,
      "loss": 0.2116,
      "step": 52655
    },
    {
      "epoch": 0.9508229873103229,
      "grad_norm": 0.38296404480934143,
      "learning_rate": 1.1905521971104394e-07,
      "loss": 0.1249,
      "step": 52660
    },
    {
      "epoch": 0.9509132667432236,
      "grad_norm": 0.5403286814689636,
      "learning_rate": 1.1861926598688234e-07,
      "loss": 0.2489,
      "step": 52665
    },
    {
      "epoch": 0.9510035461761244,
      "grad_norm": 0.1880694478750229,
      "learning_rate": 1.1818410714795614e-07,
      "loss": 0.1901,
      "step": 52670
    },
    {
      "epoch": 0.951093825609025,
      "grad_norm": 0.4349796772003174,
      "learning_rate": 1.1774974322927068e-07,
      "loss": 0.2221,
      "step": 52675
    },
    {
      "epoch": 0.9511841050419257,
      "grad_norm": 0.3575138747692108,
      "learning_rate": 1.1731617426576802e-07,
      "loss": 0.1986,
      "step": 52680
    },
    {
      "epoch": 0.9512743844748265,
      "grad_norm": 0.39403751492500305,
      "learning_rate": 1.168834002923247e-07,
      "loss": 0.2165,
      "step": 52685
    },
    {
      "epoch": 0.9513646639077272,
      "grad_norm": 0.3441208302974701,
      "learning_rate": 1.164514213437551e-07,
      "loss": 0.1894,
      "step": 52690
    },
    {
      "epoch": 0.9514549433406279,
      "grad_norm": 0.3887169063091278,
      "learning_rate": 1.1602023745480694e-07,
      "loss": 0.1778,
      "step": 52695
    },
    {
      "epoch": 0.9515452227735286,
      "grad_norm": 0.6288618445396423,
      "learning_rate": 1.1558984866016809e-07,
      "loss": 0.25,
      "step": 52700
    },
    {
      "epoch": 0.9516355022064293,
      "grad_norm": 0.9938123226165771,
      "learning_rate": 1.1516025499445971e-07,
      "loss": 0.2619,
      "step": 52705
    },
    {
      "epoch": 0.9517257816393301,
      "grad_norm": 0.6329373121261597,
      "learning_rate": 1.147314564922386e-07,
      "loss": 0.1876,
      "step": 52710
    },
    {
      "epoch": 0.9518160610722308,
      "grad_norm": 0.38888344168663025,
      "learning_rate": 1.143034531879994e-07,
      "loss": 0.207,
      "step": 52715
    },
    {
      "epoch": 0.9519063405051315,
      "grad_norm": 0.44606176018714905,
      "learning_rate": 1.1387624511617013e-07,
      "loss": 0.3276,
      "step": 52720
    },
    {
      "epoch": 0.9519966199380322,
      "grad_norm": 0.4698202908039093,
      "learning_rate": 1.1344983231111884e-07,
      "loss": 0.2329,
      "step": 52725
    },
    {
      "epoch": 0.9520868993709329,
      "grad_norm": 0.34974223375320435,
      "learning_rate": 1.1302421480714588e-07,
      "loss": 0.2286,
      "step": 52730
    },
    {
      "epoch": 0.9521771788038337,
      "grad_norm": 0.48850712180137634,
      "learning_rate": 1.1259939263849052e-07,
      "loss": 0.225,
      "step": 52735
    },
    {
      "epoch": 0.9522674582367343,
      "grad_norm": 0.7180843353271484,
      "learning_rate": 1.1217536583932542e-07,
      "loss": 0.201,
      "step": 52740
    },
    {
      "epoch": 0.952357737669635,
      "grad_norm": 0.5990698337554932,
      "learning_rate": 1.117521344437611e-07,
      "loss": 0.2144,
      "step": 52745
    },
    {
      "epoch": 0.9524480171025358,
      "grad_norm": 0.41236624121665955,
      "learning_rate": 1.113296984858414e-07,
      "loss": 0.2034,
      "step": 52750
    },
    {
      "epoch": 0.9525382965354365,
      "grad_norm": 0.6211836338043213,
      "learning_rate": 1.109080579995514e-07,
      "loss": 0.2231,
      "step": 52755
    },
    {
      "epoch": 0.9526285759683372,
      "grad_norm": 0.6268085837364197,
      "learning_rate": 1.1048721301880728e-07,
      "loss": 0.2212,
      "step": 52760
    },
    {
      "epoch": 0.9527188554012379,
      "grad_norm": 0.3917376399040222,
      "learning_rate": 1.1006716357746194e-07,
      "loss": 0.2131,
      "step": 52765
    },
    {
      "epoch": 0.9528091348341386,
      "grad_norm": 0.8501490950584412,
      "learning_rate": 1.0964790970930839e-07,
      "loss": 0.1896,
      "step": 52770
    },
    {
      "epoch": 0.9528994142670394,
      "grad_norm": 0.34583914279937744,
      "learning_rate": 1.0922945144806963e-07,
      "loss": 0.254,
      "step": 52775
    },
    {
      "epoch": 0.95298969369994,
      "grad_norm": 0.22469790279865265,
      "learning_rate": 1.0881178882740872e-07,
      "loss": 0.1903,
      "step": 52780
    },
    {
      "epoch": 0.9530799731328408,
      "grad_norm": 0.44391101598739624,
      "learning_rate": 1.0839492188092327e-07,
      "loss": 0.2337,
      "step": 52785
    },
    {
      "epoch": 0.9531702525657415,
      "grad_norm": 0.4980946481227875,
      "learning_rate": 1.0797885064214752e-07,
      "loss": 0.2129,
      "step": 52790
    },
    {
      "epoch": 0.9532605319986422,
      "grad_norm": 0.35762354731559753,
      "learning_rate": 1.075635751445514e-07,
      "loss": 0.2233,
      "step": 52795
    },
    {
      "epoch": 0.9533508114315429,
      "grad_norm": 0.49305427074432373,
      "learning_rate": 1.071490954215415e-07,
      "loss": 0.2123,
      "step": 52800
    },
    {
      "epoch": 0.9534410908644436,
      "grad_norm": 0.41991478204727173,
      "learning_rate": 1.0673541150645783e-07,
      "loss": 0.1823,
      "step": 52805
    },
    {
      "epoch": 0.9535313702973444,
      "grad_norm": 0.3152017891407013,
      "learning_rate": 1.0632252343257931e-07,
      "loss": 0.1342,
      "step": 52810
    },
    {
      "epoch": 0.9536216497302451,
      "grad_norm": 0.2789387106895447,
      "learning_rate": 1.059104312331205e-07,
      "loss": 0.1647,
      "step": 52815
    },
    {
      "epoch": 0.9537119291631457,
      "grad_norm": 0.5095995664596558,
      "learning_rate": 1.0549913494123043e-07,
      "loss": 0.182,
      "step": 52820
    },
    {
      "epoch": 0.9538022085960465,
      "grad_norm": 0.6552473306655884,
      "learning_rate": 1.0508863458999485e-07,
      "loss": 0.232,
      "step": 52825
    },
    {
      "epoch": 0.9538924880289472,
      "grad_norm": 0.3909153938293457,
      "learning_rate": 1.0467893021243514e-07,
      "loss": 0.2768,
      "step": 52830
    },
    {
      "epoch": 0.953982767461848,
      "grad_norm": 0.37738755345344543,
      "learning_rate": 1.0427002184150936e-07,
      "loss": 0.2332,
      "step": 52835
    },
    {
      "epoch": 0.9540730468947486,
      "grad_norm": 0.2699531614780426,
      "learning_rate": 1.0386190951011233e-07,
      "loss": 0.2465,
      "step": 52840
    },
    {
      "epoch": 0.9541633263276493,
      "grad_norm": 0.3892914056777954,
      "learning_rate": 1.0345459325107221e-07,
      "loss": 0.2359,
      "step": 52845
    },
    {
      "epoch": 0.9542536057605501,
      "grad_norm": 0.12257574498653412,
      "learning_rate": 1.0304807309715615e-07,
      "loss": 0.1991,
      "step": 52850
    },
    {
      "epoch": 0.9543438851934508,
      "grad_norm": 0.4543693959712982,
      "learning_rate": 1.0264234908106352e-07,
      "loss": 0.1992,
      "step": 52855
    },
    {
      "epoch": 0.9544341646263514,
      "grad_norm": 0.5937663316726685,
      "learning_rate": 1.0223742123543378e-07,
      "loss": 0.2249,
      "step": 52860
    },
    {
      "epoch": 0.9545244440592522,
      "grad_norm": 0.32612088322639465,
      "learning_rate": 1.0183328959283977e-07,
      "loss": 0.2424,
      "step": 52865
    },
    {
      "epoch": 0.9546147234921529,
      "grad_norm": 0.5018953680992126,
      "learning_rate": 1.0142995418579105e-07,
      "loss": 0.1275,
      "step": 52870
    },
    {
      "epoch": 0.9547050029250537,
      "grad_norm": 0.5463442802429199,
      "learning_rate": 1.0102741504673386e-07,
      "loss": 0.2147,
      "step": 52875
    },
    {
      "epoch": 0.9547952823579543,
      "grad_norm": 0.32160958647727966,
      "learning_rate": 1.0062567220804676e-07,
      "loss": 0.1512,
      "step": 52880
    },
    {
      "epoch": 0.954885561790855,
      "grad_norm": 0.33108359575271606,
      "learning_rate": 1.0022472570205055e-07,
      "loss": 0.241,
      "step": 52885
    },
    {
      "epoch": 0.9549758412237558,
      "grad_norm": 0.2833552062511444,
      "learning_rate": 9.982457556099612e-08,
      "loss": 0.2393,
      "step": 52890
    },
    {
      "epoch": 0.9550661206566565,
      "grad_norm": 0.39436647295951843,
      "learning_rate": 9.942522181707326e-08,
      "loss": 0.3652,
      "step": 52895
    },
    {
      "epoch": 0.9551564000895572,
      "grad_norm": 0.7457765340805054,
      "learning_rate": 9.902666450240739e-08,
      "loss": 0.1964,
      "step": 52900
    },
    {
      "epoch": 0.9552466795224579,
      "grad_norm": 0.5097097754478455,
      "learning_rate": 9.862890364905953e-08,
      "loss": 0.1545,
      "step": 52905
    },
    {
      "epoch": 0.9553369589553586,
      "grad_norm": 0.3292642831802368,
      "learning_rate": 9.82319392890252e-08,
      "loss": 0.2628,
      "step": 52910
    },
    {
      "epoch": 0.9554272383882594,
      "grad_norm": 0.7061537504196167,
      "learning_rate": 9.783577145423995e-08,
      "loss": 0.1994,
      "step": 52915
    },
    {
      "epoch": 0.95551751782116,
      "grad_norm": 0.4803351163864136,
      "learning_rate": 9.744040017657053e-08,
      "loss": 0.237,
      "step": 52920
    },
    {
      "epoch": 0.9556077972540608,
      "grad_norm": 0.4237426519393921,
      "learning_rate": 9.70458254878226e-08,
      "loss": 0.2182,
      "step": 52925
    },
    {
      "epoch": 0.9556980766869615,
      "grad_norm": 0.7886397242546082,
      "learning_rate": 9.665204741973633e-08,
      "loss": 0.2221,
      "step": 52930
    },
    {
      "epoch": 0.9557883561198622,
      "grad_norm": 0.5053573846817017,
      "learning_rate": 9.62590660039886e-08,
      "loss": 0.1368,
      "step": 52935
    },
    {
      "epoch": 0.9558786355527629,
      "grad_norm": 0.5820841193199158,
      "learning_rate": 9.58668812721919e-08,
      "loss": 0.1859,
      "step": 52940
    },
    {
      "epoch": 0.9559689149856636,
      "grad_norm": 0.3846931457519531,
      "learning_rate": 9.547549325589544e-08,
      "loss": 0.214,
      "step": 52945
    },
    {
      "epoch": 0.9560591944185644,
      "grad_norm": 1.6069108247756958,
      "learning_rate": 9.508490198658183e-08,
      "loss": 0.3033,
      "step": 52950
    },
    {
      "epoch": 0.9561494738514651,
      "grad_norm": 0.403764009475708,
      "learning_rate": 9.469510749567257e-08,
      "loss": 0.2093,
      "step": 52955
    },
    {
      "epoch": 0.9562397532843657,
      "grad_norm": 0.22472268342971802,
      "learning_rate": 9.43061098145226e-08,
      "loss": 0.2622,
      "step": 52960
    },
    {
      "epoch": 0.9563300327172665,
      "grad_norm": 0.21604177355766296,
      "learning_rate": 9.391790897442577e-08,
      "loss": 0.2468,
      "step": 52965
    },
    {
      "epoch": 0.9564203121501672,
      "grad_norm": 0.3919534981250763,
      "learning_rate": 9.35305050066082e-08,
      "loss": 0.3053,
      "step": 52970
    },
    {
      "epoch": 0.956510591583068,
      "grad_norm": 0.6001983284950256,
      "learning_rate": 9.314389794223499e-08,
      "loss": 0.2177,
      "step": 52975
    },
    {
      "epoch": 0.9566008710159686,
      "grad_norm": 0.48056331276893616,
      "learning_rate": 9.27580878124057e-08,
      "loss": 0.2077,
      "step": 52980
    },
    {
      "epoch": 0.9566911504488693,
      "grad_norm": 0.31555619835853577,
      "learning_rate": 9.237307464815437e-08,
      "loss": 0.2056,
      "step": 52985
    },
    {
      "epoch": 0.9567814298817701,
      "grad_norm": 0.5026416182518005,
      "learning_rate": 9.198885848045402e-08,
      "loss": 0.2561,
      "step": 52990
    },
    {
      "epoch": 0.9568717093146708,
      "grad_norm": 1.641403079032898,
      "learning_rate": 9.160543934021216e-08,
      "loss": 0.1711,
      "step": 52995
    },
    {
      "epoch": 0.9569619887475714,
      "grad_norm": 0.5881418585777283,
      "learning_rate": 9.122281725827298e-08,
      "loss": 0.2398,
      "step": 53000
    },
    {
      "epoch": 0.9570522681804722,
      "grad_norm": 0.5745043158531189,
      "learning_rate": 9.084099226541298e-08,
      "loss": 0.1363,
      "step": 53005
    },
    {
      "epoch": 0.9571425476133729,
      "grad_norm": 0.3548424541950226,
      "learning_rate": 9.045996439234872e-08,
      "loss": 0.224,
      "step": 53010
    },
    {
      "epoch": 0.9572328270462737,
      "grad_norm": 0.8307169079780579,
      "learning_rate": 9.007973366973121e-08,
      "loss": 0.2867,
      "step": 53015
    },
    {
      "epoch": 0.9573231064791743,
      "grad_norm": 0.748156726360321,
      "learning_rate": 8.970030012814712e-08,
      "loss": 0.2731,
      "step": 53020
    },
    {
      "epoch": 0.957413385912075,
      "grad_norm": 0.29039686918258667,
      "learning_rate": 8.932166379811868e-08,
      "loss": 0.2667,
      "step": 53025
    },
    {
      "epoch": 0.9575036653449758,
      "grad_norm": 0.44262897968292236,
      "learning_rate": 8.894382471010487e-08,
      "loss": 0.1928,
      "step": 53030
    },
    {
      "epoch": 0.9575939447778765,
      "grad_norm": 0.35700151324272156,
      "learning_rate": 8.856678289450027e-08,
      "loss": 0.2164,
      "step": 53035
    },
    {
      "epoch": 0.9576842242107771,
      "grad_norm": 0.3340015411376953,
      "learning_rate": 8.819053838163505e-08,
      "loss": 0.1913,
      "step": 53040
    },
    {
      "epoch": 0.9577745036436779,
      "grad_norm": 0.4953787326812744,
      "learning_rate": 8.781509120177501e-08,
      "loss": 0.2271,
      "step": 53045
    },
    {
      "epoch": 0.9578647830765786,
      "grad_norm": 0.5200175642967224,
      "learning_rate": 8.744044138512265e-08,
      "loss": 0.2455,
      "step": 53050
    },
    {
      "epoch": 0.9579550625094794,
      "grad_norm": 0.46813035011291504,
      "learning_rate": 8.706658896181386e-08,
      "loss": 0.2572,
      "step": 53055
    },
    {
      "epoch": 0.95804534194238,
      "grad_norm": 0.31738176941871643,
      "learning_rate": 8.66935339619257e-08,
      "loss": 0.2258,
      "step": 53060
    },
    {
      "epoch": 0.9581356213752807,
      "grad_norm": 0.4683874249458313,
      "learning_rate": 8.632127641546528e-08,
      "loss": 0.2134,
      "step": 53065
    },
    {
      "epoch": 0.9582259008081815,
      "grad_norm": 0.40266793966293335,
      "learning_rate": 8.594981635237975e-08,
      "loss": 0.2604,
      "step": 53070
    },
    {
      "epoch": 0.9583161802410822,
      "grad_norm": 0.5166845321655273,
      "learning_rate": 8.557915380254855e-08,
      "loss": 0.2116,
      "step": 53075
    },
    {
      "epoch": 0.9584064596739829,
      "grad_norm": 0.31063374876976013,
      "learning_rate": 8.520928879579005e-08,
      "loss": 0.263,
      "step": 53080
    },
    {
      "epoch": 0.9584967391068836,
      "grad_norm": 0.31503283977508545,
      "learning_rate": 8.48402213618571e-08,
      "loss": 0.2032,
      "step": 53085
    },
    {
      "epoch": 0.9585870185397843,
      "grad_norm": 1.0543475151062012,
      "learning_rate": 8.447195153043708e-08,
      "loss": 0.1469,
      "step": 53090
    },
    {
      "epoch": 0.9586772979726851,
      "grad_norm": 0.45366400480270386,
      "learning_rate": 8.410447933115518e-08,
      "loss": 0.1851,
      "step": 53095
    },
    {
      "epoch": 0.9587675774055857,
      "grad_norm": 1.8845783472061157,
      "learning_rate": 8.373780479357329e-08,
      "loss": 0.231,
      "step": 53100
    },
    {
      "epoch": 0.9588578568384865,
      "grad_norm": 0.47017449140548706,
      "learning_rate": 8.337192794718784e-08,
      "loss": 0.2515,
      "step": 53105
    },
    {
      "epoch": 0.9589481362713872,
      "grad_norm": 0.439563512802124,
      "learning_rate": 8.300684882142862e-08,
      "loss": 0.1526,
      "step": 53110
    },
    {
      "epoch": 0.9590384157042879,
      "grad_norm": 0.7139844298362732,
      "learning_rate": 8.264256744566546e-08,
      "loss": 0.2517,
      "step": 53115
    },
    {
      "epoch": 0.9591286951371887,
      "grad_norm": 0.17191582918167114,
      "learning_rate": 8.227908384920158e-08,
      "loss": 0.297,
      "step": 53120
    },
    {
      "epoch": 0.9592189745700893,
      "grad_norm": 0.30800798535346985,
      "learning_rate": 8.191639806127582e-08,
      "loss": 0.1924,
      "step": 53125
    },
    {
      "epoch": 0.95930925400299,
      "grad_norm": 0.32242152094841003,
      "learning_rate": 8.155451011106485e-08,
      "loss": 0.1799,
      "step": 53130
    },
    {
      "epoch": 0.9593995334358908,
      "grad_norm": 0.4862722158432007,
      "learning_rate": 8.119342002767983e-08,
      "loss": 0.241,
      "step": 53135
    },
    {
      "epoch": 0.9594898128687915,
      "grad_norm": 0.5938839316368103,
      "learning_rate": 8.08331278401675e-08,
      "loss": 0.1877,
      "step": 53140
    },
    {
      "epoch": 0.9595800923016922,
      "grad_norm": 0.28677666187286377,
      "learning_rate": 8.047363357751137e-08,
      "loss": 0.1845,
      "step": 53145
    },
    {
      "epoch": 0.9596703717345929,
      "grad_norm": 0.2129906713962555,
      "learning_rate": 8.011493726862829e-08,
      "loss": 0.1233,
      "step": 53150
    },
    {
      "epoch": 0.9597606511674937,
      "grad_norm": 0.4362054169178009,
      "learning_rate": 7.975703894237407e-08,
      "loss": 0.2467,
      "step": 53155
    },
    {
      "epoch": 0.9598509306003944,
      "grad_norm": 0.6214504837989807,
      "learning_rate": 7.939993862753903e-08,
      "loss": 0.2055,
      "step": 53160
    },
    {
      "epoch": 0.959941210033295,
      "grad_norm": 0.4493671953678131,
      "learning_rate": 7.904363635284906e-08,
      "loss": 0.239,
      "step": 53165
    },
    {
      "epoch": 0.9600314894661958,
      "grad_norm": 0.4206450581550598,
      "learning_rate": 7.86881321469668e-08,
      "loss": 0.2206,
      "step": 53170
    },
    {
      "epoch": 0.9601217688990965,
      "grad_norm": 0.4071847200393677,
      "learning_rate": 7.833342603848936e-08,
      "loss": 0.2473,
      "step": 53175
    },
    {
      "epoch": 0.9602120483319972,
      "grad_norm": 1.0784540176391602,
      "learning_rate": 7.797951805595062e-08,
      "loss": 0.2786,
      "step": 53180
    },
    {
      "epoch": 0.9603023277648979,
      "grad_norm": 0.3914712071418762,
      "learning_rate": 7.762640822781886e-08,
      "loss": 0.1983,
      "step": 53185
    },
    {
      "epoch": 0.9603926071977986,
      "grad_norm": 0.4481930136680603,
      "learning_rate": 7.727409658249918e-08,
      "loss": 0.222,
      "step": 53190
    },
    {
      "epoch": 0.9604828866306994,
      "grad_norm": 0.3223609924316406,
      "learning_rate": 7.692258314833445e-08,
      "loss": 0.1913,
      "step": 53195
    },
    {
      "epoch": 0.9605731660636001,
      "grad_norm": 0.44718512892723083,
      "learning_rate": 7.657186795359983e-08,
      "loss": 0.245,
      "step": 53200
    },
    {
      "epoch": 0.9606634454965007,
      "grad_norm": 0.9813588261604309,
      "learning_rate": 7.62219510265083e-08,
      "loss": 0.3685,
      "step": 53205
    },
    {
      "epoch": 0.9607537249294015,
      "grad_norm": 0.4945777952671051,
      "learning_rate": 7.587283239520848e-08,
      "loss": 0.2221,
      "step": 53210
    },
    {
      "epoch": 0.9608440043623022,
      "grad_norm": 0.6341715455055237,
      "learning_rate": 7.552451208778233e-08,
      "loss": 0.1329,
      "step": 53215
    },
    {
      "epoch": 0.960934283795203,
      "grad_norm": 0.5160859227180481,
      "learning_rate": 7.517699013225299e-08,
      "loss": 0.2517,
      "step": 53220
    },
    {
      "epoch": 0.9610245632281036,
      "grad_norm": 0.495863676071167,
      "learning_rate": 7.483026655657255e-08,
      "loss": 0.2899,
      "step": 53225
    },
    {
      "epoch": 0.9611148426610043,
      "grad_norm": 0.5978807210922241,
      "learning_rate": 7.448434138863537e-08,
      "loss": 0.2264,
      "step": 53230
    },
    {
      "epoch": 0.9612051220939051,
      "grad_norm": 0.3395076394081116,
      "learning_rate": 7.413921465626806e-08,
      "loss": 0.2343,
      "step": 53235
    },
    {
      "epoch": 0.9612954015268058,
      "grad_norm": 0.3495229482650757,
      "learning_rate": 7.379488638723176e-08,
      "loss": 0.1622,
      "step": 53240
    },
    {
      "epoch": 0.9613856809597064,
      "grad_norm": 0.2799985706806183,
      "learning_rate": 7.345135660922765e-08,
      "loss": 0.2223,
      "step": 53245
    },
    {
      "epoch": 0.9614759603926072,
      "grad_norm": 0.6174278259277344,
      "learning_rate": 7.310862534988805e-08,
      "loss": 0.2237,
      "step": 53250
    },
    {
      "epoch": 0.9615662398255079,
      "grad_norm": 0.5111470222473145,
      "learning_rate": 7.276669263678316e-08,
      "loss": 0.2449,
      "step": 53255
    },
    {
      "epoch": 0.9616565192584087,
      "grad_norm": 0.20767705142498016,
      "learning_rate": 7.242555849742095e-08,
      "loss": 0.1808,
      "step": 53260
    },
    {
      "epoch": 0.9617467986913093,
      "grad_norm": 0.4902808964252472,
      "learning_rate": 7.20852229592417e-08,
      "loss": 0.1326,
      "step": 53265
    },
    {
      "epoch": 0.96183707812421,
      "grad_norm": 0.6132269501686096,
      "learning_rate": 7.17456860496235e-08,
      "loss": 0.1469,
      "step": 53270
    },
    {
      "epoch": 0.9619273575571108,
      "grad_norm": 0.3463001847267151,
      "learning_rate": 7.140694779588009e-08,
      "loss": 0.2011,
      "step": 53275
    },
    {
      "epoch": 0.9620176369900115,
      "grad_norm": 0.6098634004592896,
      "learning_rate": 7.106900822525964e-08,
      "loss": 0.1988,
      "step": 53280
    },
    {
      "epoch": 0.9621079164229122,
      "grad_norm": 0.2960973381996155,
      "learning_rate": 7.073186736494708e-08,
      "loss": 0.2279,
      "step": 53285
    },
    {
      "epoch": 0.9621981958558129,
      "grad_norm": 0.5420827269554138,
      "learning_rate": 7.039552524206183e-08,
      "loss": 0.2263,
      "step": 53290
    },
    {
      "epoch": 0.9622884752887136,
      "grad_norm": 0.5859010219573975,
      "learning_rate": 7.005998188366337e-08,
      "loss": 0.2199,
      "step": 53295
    },
    {
      "epoch": 0.9623787547216144,
      "grad_norm": 0.4214431941509247,
      "learning_rate": 6.972523731674008e-08,
      "loss": 0.2139,
      "step": 53300
    },
    {
      "epoch": 0.962469034154515,
      "grad_norm": 0.5375816226005554,
      "learning_rate": 6.939129156822267e-08,
      "loss": 0.2319,
      "step": 53305
    },
    {
      "epoch": 0.9625593135874158,
      "grad_norm": 0.3661133348941803,
      "learning_rate": 6.905814466497296e-08,
      "loss": 0.2255,
      "step": 53310
    },
    {
      "epoch": 0.9626495930203165,
      "grad_norm": 0.6833767890930176,
      "learning_rate": 6.872579663379065e-08,
      "loss": 0.2102,
      "step": 53315
    },
    {
      "epoch": 0.9627398724532172,
      "grad_norm": 0.34302306175231934,
      "learning_rate": 6.839424750140988e-08,
      "loss": 0.1961,
      "step": 53320
    },
    {
      "epoch": 0.9628301518861179,
      "grad_norm": 0.4643937051296234,
      "learning_rate": 6.806349729450268e-08,
      "loss": 0.1428,
      "step": 53325
    },
    {
      "epoch": 0.9629204313190186,
      "grad_norm": 0.890001654624939,
      "learning_rate": 6.773354603967552e-08,
      "loss": 0.2088,
      "step": 53330
    },
    {
      "epoch": 0.9630107107519194,
      "grad_norm": 0.37487348914146423,
      "learning_rate": 6.74043937634683e-08,
      "loss": 0.3113,
      "step": 53335
    },
    {
      "epoch": 0.9631009901848201,
      "grad_norm": 0.4764295220375061,
      "learning_rate": 6.707604049236316e-08,
      "loss": 0.1891,
      "step": 53340
    },
    {
      "epoch": 0.9631912696177207,
      "grad_norm": 0.16798852384090424,
      "learning_rate": 6.674848625276898e-08,
      "loss": 0.2206,
      "step": 53345
    },
    {
      "epoch": 0.9632815490506215,
      "grad_norm": 0.4167994260787964,
      "learning_rate": 6.64217310710391e-08,
      "loss": 0.1883,
      "step": 53350
    },
    {
      "epoch": 0.9633718284835222,
      "grad_norm": 0.5369573831558228,
      "learning_rate": 6.609577497345476e-08,
      "loss": 0.2549,
      "step": 53355
    },
    {
      "epoch": 0.963462107916423,
      "grad_norm": 0.4055411219596863,
      "learning_rate": 6.577061798624051e-08,
      "loss": 0.1853,
      "step": 53360
    },
    {
      "epoch": 0.9635523873493236,
      "grad_norm": 0.3651867210865021,
      "learning_rate": 6.544626013555099e-08,
      "loss": 0.1254,
      "step": 53365
    },
    {
      "epoch": 0.9636426667822243,
      "grad_norm": 0.2772580683231354,
      "learning_rate": 6.512270144747867e-08,
      "loss": 0.1921,
      "step": 53370
    },
    {
      "epoch": 0.9637329462151251,
      "grad_norm": 0.3850838840007782,
      "learning_rate": 6.47999419480505e-08,
      "loss": 0.2453,
      "step": 53375
    },
    {
      "epoch": 0.9638232256480258,
      "grad_norm": 0.4329755902290344,
      "learning_rate": 6.447798166323238e-08,
      "loss": 0.2601,
      "step": 53380
    },
    {
      "epoch": 0.9639135050809264,
      "grad_norm": 0.2568431496620178,
      "learning_rate": 6.415682061892248e-08,
      "loss": 0.1775,
      "step": 53385
    },
    {
      "epoch": 0.9640037845138272,
      "grad_norm": 0.7551835179328918,
      "learning_rate": 6.383645884095457e-08,
      "loss": 0.2182,
      "step": 53390
    },
    {
      "epoch": 0.9640940639467279,
      "grad_norm": 0.3586956560611725,
      "learning_rate": 6.351689635510139e-08,
      "loss": 0.2476,
      "step": 53395
    },
    {
      "epoch": 0.9641843433796287,
      "grad_norm": 0.48093101382255554,
      "learning_rate": 6.319813318706903e-08,
      "loss": 0.2034,
      "step": 53400
    },
    {
      "epoch": 0.9642746228125293,
      "grad_norm": 0.6688908338546753,
      "learning_rate": 6.288016936249808e-08,
      "loss": 0.256,
      "step": 53405
    },
    {
      "epoch": 0.96436490224543,
      "grad_norm": 0.26767441630363464,
      "learning_rate": 6.25630049069692e-08,
      "loss": 0.1174,
      "step": 53410
    },
    {
      "epoch": 0.9644551816783308,
      "grad_norm": 0.5681080222129822,
      "learning_rate": 6.224663984599422e-08,
      "loss": 0.2648,
      "step": 53415
    },
    {
      "epoch": 0.9645454611112315,
      "grad_norm": 0.19477705657482147,
      "learning_rate": 6.193107420502164e-08,
      "loss": 0.2603,
      "step": 53420
    },
    {
      "epoch": 0.9646357405441321,
      "grad_norm": 0.4034990668296814,
      "learning_rate": 6.161630800943786e-08,
      "loss": 0.2282,
      "step": 53425
    },
    {
      "epoch": 0.9647260199770329,
      "grad_norm": 0.49816352128982544,
      "learning_rate": 6.130234128456258e-08,
      "loss": 0.2564,
      "step": 53430
    },
    {
      "epoch": 0.9648162994099336,
      "grad_norm": 0.3927150070667267,
      "learning_rate": 6.098917405565341e-08,
      "loss": 0.1817,
      "step": 53435
    },
    {
      "epoch": 0.9649065788428344,
      "grad_norm": 0.512175977230072,
      "learning_rate": 6.067680634790129e-08,
      "loss": 0.1719,
      "step": 53440
    },
    {
      "epoch": 0.964996858275735,
      "grad_norm": 0.30420270562171936,
      "learning_rate": 6.036523818643502e-08,
      "loss": 0.2328,
      "step": 53445
    },
    {
      "epoch": 0.9650871377086357,
      "grad_norm": 0.8397376537322998,
      "learning_rate": 6.005446959631567e-08,
      "loss": 0.1231,
      "step": 53450
    },
    {
      "epoch": 0.9651774171415365,
      "grad_norm": 0.18549306690692902,
      "learning_rate": 5.974450060254433e-08,
      "loss": 0.1716,
      "step": 53455
    },
    {
      "epoch": 0.9652676965744372,
      "grad_norm": 0.35602420568466187,
      "learning_rate": 5.943533123005552e-08,
      "loss": 0.1806,
      "step": 53460
    },
    {
      "epoch": 0.9653579760073379,
      "grad_norm": 0.5836049318313599,
      "learning_rate": 5.9126961503720437e-08,
      "loss": 0.2974,
      "step": 53465
    },
    {
      "epoch": 0.9654482554402386,
      "grad_norm": 0.5807859897613525,
      "learning_rate": 5.881939144834259e-08,
      "loss": 0.2645,
      "step": 53470
    },
    {
      "epoch": 0.9655385348731393,
      "grad_norm": 0.4430500268936157,
      "learning_rate": 5.851262108866662e-08,
      "loss": 0.2112,
      "step": 53475
    },
    {
      "epoch": 0.9656288143060401,
      "grad_norm": 0.39009523391723633,
      "learning_rate": 5.820665044936724e-08,
      "loss": 0.1803,
      "step": 53480
    },
    {
      "epoch": 0.9657190937389407,
      "grad_norm": 0.44665607810020447,
      "learning_rate": 5.7901479555060314e-08,
      "loss": 0.2182,
      "step": 53485
    },
    {
      "epoch": 0.9658093731718415,
      "grad_norm": 0.3570907711982727,
      "learning_rate": 5.759710843029287e-08,
      "loss": 0.2087,
      "step": 53490
    },
    {
      "epoch": 0.9658996526047422,
      "grad_norm": 0.49240604043006897,
      "learning_rate": 5.729353709954977e-08,
      "loss": 0.2469,
      "step": 53495
    },
    {
      "epoch": 0.9659899320376429,
      "grad_norm": 0.744106113910675,
      "learning_rate": 5.699076558725036e-08,
      "loss": 0.3114,
      "step": 53500
    },
    {
      "epoch": 0.9660802114705437,
      "grad_norm": 0.5872280597686768,
      "learning_rate": 5.668879391775295e-08,
      "loss": 0.2079,
      "step": 53505
    },
    {
      "epoch": 0.9661704909034443,
      "grad_norm": 0.641364574432373,
      "learning_rate": 5.6387622115345877e-08,
      "loss": 0.2465,
      "step": 53510
    },
    {
      "epoch": 0.966260770336345,
      "grad_norm": 0.38412678241729736,
      "learning_rate": 5.608725020425754e-08,
      "loss": 0.2284,
      "step": 53515
    },
    {
      "epoch": 0.9663510497692458,
      "grad_norm": 0.3273337483406067,
      "learning_rate": 5.578767820865083e-08,
      "loss": 0.2677,
      "step": 53520
    },
    {
      "epoch": 0.9664413292021465,
      "grad_norm": 0.3032999634742737,
      "learning_rate": 5.548890615262426e-08,
      "loss": 0.2144,
      "step": 53525
    },
    {
      "epoch": 0.9665316086350472,
      "grad_norm": 0.40187644958496094,
      "learning_rate": 5.519093406021192e-08,
      "loss": 0.2244,
      "step": 53530
    },
    {
      "epoch": 0.9666218880679479,
      "grad_norm": 0.2785539925098419,
      "learning_rate": 5.489376195538243e-08,
      "loss": 0.2216,
      "step": 53535
    },
    {
      "epoch": 0.9667121675008487,
      "grad_norm": 0.38451331853866577,
      "learning_rate": 5.4597389862042216e-08,
      "loss": 0.2483,
      "step": 53540
    },
    {
      "epoch": 0.9668024469337494,
      "grad_norm": 0.41696181893348694,
      "learning_rate": 5.430181780403221e-08,
      "loss": 0.2812,
      "step": 53545
    },
    {
      "epoch": 0.96689272636665,
      "grad_norm": 0.3749529719352722,
      "learning_rate": 5.400704580512894e-08,
      "loss": 0.1698,
      "step": 53550
    },
    {
      "epoch": 0.9669830057995508,
      "grad_norm": 0.5485641956329346,
      "learning_rate": 5.371307388904345e-08,
      "loss": 0.299,
      "step": 53555
    },
    {
      "epoch": 0.9670732852324515,
      "grad_norm": 0.6841853857040405,
      "learning_rate": 5.341990207942571e-08,
      "loss": 0.2233,
      "step": 53560
    },
    {
      "epoch": 0.9671635646653522,
      "grad_norm": 0.6185765862464905,
      "learning_rate": 5.312753039985685e-08,
      "loss": 0.1981,
      "step": 53565
    },
    {
      "epoch": 0.9672538440982529,
      "grad_norm": 0.25791651010513306,
      "learning_rate": 5.283595887385917e-08,
      "loss": 0.1868,
      "step": 53570
    },
    {
      "epoch": 0.9673441235311536,
      "grad_norm": 0.34983909130096436,
      "learning_rate": 5.2545187524885024e-08,
      "loss": 0.25,
      "step": 53575
    },
    {
      "epoch": 0.9674344029640544,
      "grad_norm": 0.5966994166374207,
      "learning_rate": 5.225521637632569e-08,
      "loss": 0.3569,
      "step": 53580
    },
    {
      "epoch": 0.9675246823969551,
      "grad_norm": 0.2861376404762268,
      "learning_rate": 5.196604545150585e-08,
      "loss": 0.1452,
      "step": 53585
    },
    {
      "epoch": 0.9676149618298557,
      "grad_norm": 0.4372887909412384,
      "learning_rate": 5.1677674773690234e-08,
      "loss": 0.1444,
      "step": 53590
    },
    {
      "epoch": 0.9677052412627565,
      "grad_norm": 0.18914885818958282,
      "learning_rate": 5.1390104366072505e-08,
      "loss": 0.2534,
      "step": 53595
    },
    {
      "epoch": 0.9677955206956572,
      "grad_norm": 0.45735007524490356,
      "learning_rate": 5.110333425178971e-08,
      "loss": 0.2267,
      "step": 53600
    },
    {
      "epoch": 0.967885800128558,
      "grad_norm": 0.2701546251773834,
      "learning_rate": 5.0817364453906727e-08,
      "loss": 0.2206,
      "step": 53605
    },
    {
      "epoch": 0.9679760795614586,
      "grad_norm": 0.41877853870391846,
      "learning_rate": 5.053219499542961e-08,
      "loss": 0.183,
      "step": 53610
    },
    {
      "epoch": 0.9680663589943593,
      "grad_norm": 0.5866236686706543,
      "learning_rate": 5.0247825899297773e-08,
      "loss": 0.2052,
      "step": 53615
    },
    {
      "epoch": 0.9681566384272601,
      "grad_norm": 0.5392400622367859,
      "learning_rate": 4.9964257188386264e-08,
      "loss": 0.1449,
      "step": 53620
    },
    {
      "epoch": 0.9682469178601608,
      "grad_norm": 0.5776108503341675,
      "learning_rate": 4.9681488885506835e-08,
      "loss": 0.2987,
      "step": 53625
    },
    {
      "epoch": 0.9683371972930614,
      "grad_norm": 0.5677412748336792,
      "learning_rate": 4.939952101340573e-08,
      "loss": 0.2401,
      "step": 53630
    },
    {
      "epoch": 0.9684274767259622,
      "grad_norm": 0.5053970813751221,
      "learning_rate": 4.911835359476591e-08,
      "loss": 0.2752,
      "step": 53635
    },
    {
      "epoch": 0.9685177561588629,
      "grad_norm": 0.5509980916976929,
      "learning_rate": 4.883798665220374e-08,
      "loss": 0.2565,
      "step": 53640
    },
    {
      "epoch": 0.9686080355917637,
      "grad_norm": 0.30780690908432007,
      "learning_rate": 4.85584202082745e-08,
      "loss": 0.3142,
      "step": 53645
    },
    {
      "epoch": 0.9686983150246643,
      "grad_norm": 0.3872641623020172,
      "learning_rate": 4.827965428546577e-08,
      "loss": 0.1649,
      "step": 53650
    },
    {
      "epoch": 0.968788594457565,
      "grad_norm": 0.3725968301296234,
      "learning_rate": 4.800168890620294e-08,
      "loss": 0.2596,
      "step": 53655
    },
    {
      "epoch": 0.9688788738904658,
      "grad_norm": 0.6582515835762024,
      "learning_rate": 4.7724524092845895e-08,
      "loss": 0.2926,
      "step": 53660
    },
    {
      "epoch": 0.9689691533233665,
      "grad_norm": 0.48744910955429077,
      "learning_rate": 4.7448159867691244e-08,
      "loss": 0.1601,
      "step": 53665
    },
    {
      "epoch": 0.9690594327562672,
      "grad_norm": 0.3776699900627136,
      "learning_rate": 4.717259625297011e-08,
      "loss": 0.2596,
      "step": 53670
    },
    {
      "epoch": 0.9691497121891679,
      "grad_norm": 0.5383025407791138,
      "learning_rate": 4.689783327085029e-08,
      "loss": 0.2028,
      "step": 53675
    },
    {
      "epoch": 0.9692399916220686,
      "grad_norm": 0.48457786440849304,
      "learning_rate": 4.662387094343301e-08,
      "loss": 0.2108,
      "step": 53680
    },
    {
      "epoch": 0.9693302710549694,
      "grad_norm": 0.9042235016822815,
      "learning_rate": 4.6350709292756205e-08,
      "loss": 0.2879,
      "step": 53685
    },
    {
      "epoch": 0.96942055048787,
      "grad_norm": 0.5203196406364441,
      "learning_rate": 4.607834834079672e-08,
      "loss": 0.2075,
      "step": 53690
    },
    {
      "epoch": 0.9695108299207708,
      "grad_norm": 0.42074522376060486,
      "learning_rate": 4.5806788109460373e-08,
      "loss": 0.1693,
      "step": 53695
    },
    {
      "epoch": 0.9696011093536715,
      "grad_norm": 0.8473619818687439,
      "learning_rate": 4.553602862059525e-08,
      "loss": 0.2932,
      "step": 53700
    },
    {
      "epoch": 0.9696913887865722,
      "grad_norm": 0.6807908415794373,
      "learning_rate": 4.5266069895980594e-08,
      "loss": 0.2788,
      "step": 53705
    },
    {
      "epoch": 0.9697816682194729,
      "grad_norm": 0.30918651819229126,
      "learning_rate": 4.499691195733236e-08,
      "loss": 0.2647,
      "step": 53710
    },
    {
      "epoch": 0.9698719476523736,
      "grad_norm": 0.4880746603012085,
      "learning_rate": 4.472855482630212e-08,
      "loss": 0.2247,
      "step": 53715
    },
    {
      "epoch": 0.9699622270852744,
      "grad_norm": 0.30650821328163147,
      "learning_rate": 4.446099852447816e-08,
      "loss": 0.2894,
      "step": 53720
    },
    {
      "epoch": 0.9700525065181751,
      "grad_norm": 0.35868415236473083,
      "learning_rate": 4.419424307338327e-08,
      "loss": 0.2829,
      "step": 53725
    },
    {
      "epoch": 0.9701427859510757,
      "grad_norm": 0.6965757608413696,
      "learning_rate": 4.392828849447583e-08,
      "loss": 0.2363,
      "step": 53730
    },
    {
      "epoch": 0.9702330653839765,
      "grad_norm": 0.5765778422355652,
      "learning_rate": 4.366313480914986e-08,
      "loss": 0.2097,
      "step": 53735
    },
    {
      "epoch": 0.9703233448168772,
      "grad_norm": 0.3773828446865082,
      "learning_rate": 4.339878203873604e-08,
      "loss": 0.2333,
      "step": 53740
    },
    {
      "epoch": 0.970413624249778,
      "grad_norm": 0.457940012216568,
      "learning_rate": 4.3135230204497394e-08,
      "loss": 0.2028,
      "step": 53745
    },
    {
      "epoch": 0.9705039036826786,
      "grad_norm": 0.471177339553833,
      "learning_rate": 4.287247932763694e-08,
      "loss": 0.1695,
      "step": 53750
    },
    {
      "epoch": 0.9705941831155793,
      "grad_norm": 0.5329426527023315,
      "learning_rate": 4.2610529429290005e-08,
      "loss": 0.2217,
      "step": 53755
    },
    {
      "epoch": 0.9706844625484801,
      "grad_norm": 0.45549896359443665,
      "learning_rate": 4.234938053052862e-08,
      "loss": 0.2349,
      "step": 53760
    },
    {
      "epoch": 0.9707747419813808,
      "grad_norm": 0.532945990562439,
      "learning_rate": 4.2089032652360415e-08,
      "loss": 0.2345,
      "step": 53765
    },
    {
      "epoch": 0.9708650214142814,
      "grad_norm": 0.41036728024482727,
      "learning_rate": 4.182948581572866e-08,
      "loss": 0.2363,
      "step": 53770
    },
    {
      "epoch": 0.9709553008471822,
      "grad_norm": 0.4287985861301422,
      "learning_rate": 4.157074004151107e-08,
      "loss": 0.1736,
      "step": 53775
    },
    {
      "epoch": 0.9710455802800829,
      "grad_norm": 0.4294351637363434,
      "learning_rate": 4.131279535052435e-08,
      "loss": 0.1674,
      "step": 53780
    },
    {
      "epoch": 0.9711358597129837,
      "grad_norm": 0.3271455764770508,
      "learning_rate": 4.1055651763515226e-08,
      "loss": 0.2346,
      "step": 53785
    },
    {
      "epoch": 0.9712261391458843,
      "grad_norm": 0.44987601041793823,
      "learning_rate": 4.079930930117049e-08,
      "loss": 0.2933,
      "step": 53790
    },
    {
      "epoch": 0.971316418578785,
      "grad_norm": 0.5628604888916016,
      "learning_rate": 4.0543767984111417e-08,
      "loss": 0.1721,
      "step": 53795
    },
    {
      "epoch": 0.9714066980116858,
      "grad_norm": 0.46557164192199707,
      "learning_rate": 4.028902783289379e-08,
      "loss": 0.237,
      "step": 53800
    },
    {
      "epoch": 0.9714969774445865,
      "grad_norm": 0.4409957230091095,
      "learning_rate": 4.003508886801011e-08,
      "loss": 0.2053,
      "step": 53805
    },
    {
      "epoch": 0.9715872568774871,
      "grad_norm": 0.5898836851119995,
      "learning_rate": 3.9781951109887364e-08,
      "loss": 0.1727,
      "step": 53810
    },
    {
      "epoch": 0.9716775363103879,
      "grad_norm": 0.47835275530815125,
      "learning_rate": 3.952961457888815e-08,
      "loss": 0.1725,
      "step": 53815
    },
    {
      "epoch": 0.9717678157432886,
      "grad_norm": 0.46880072355270386,
      "learning_rate": 3.92780792953118e-08,
      "loss": 0.2866,
      "step": 53820
    },
    {
      "epoch": 0.9718580951761894,
      "grad_norm": 0.14378659427165985,
      "learning_rate": 3.902734527939434e-08,
      "loss": 0.1799,
      "step": 53825
    },
    {
      "epoch": 0.97194837460909,
      "grad_norm": 0.4346258342266083,
      "learning_rate": 3.877741255130185e-08,
      "loss": 0.2799,
      "step": 53830
    },
    {
      "epoch": 0.9720386540419907,
      "grad_norm": 0.2553173303604126,
      "learning_rate": 3.852828113114271e-08,
      "loss": 0.1726,
      "step": 53835
    },
    {
      "epoch": 0.9721289334748915,
      "grad_norm": 0.33480632305145264,
      "learning_rate": 3.827995103895532e-08,
      "loss": 0.2129,
      "step": 53840
    },
    {
      "epoch": 0.9722192129077922,
      "grad_norm": 0.5246515870094299,
      "learning_rate": 3.8032422294718154e-08,
      "loss": 0.2672,
      "step": 53845
    },
    {
      "epoch": 0.9723094923406929,
      "grad_norm": 0.12980012595653534,
      "learning_rate": 3.778569491834083e-08,
      "loss": 0.1648,
      "step": 53850
    },
    {
      "epoch": 0.9723997717735936,
      "grad_norm": 0.38328951597213745,
      "learning_rate": 3.753976892967304e-08,
      "loss": 0.3238,
      "step": 53855
    },
    {
      "epoch": 0.9724900512064943,
      "grad_norm": 0.5784655809402466,
      "learning_rate": 3.729464434849672e-08,
      "loss": 0.1323,
      "step": 53860
    },
    {
      "epoch": 0.9725803306393951,
      "grad_norm": 0.38006168603897095,
      "learning_rate": 3.705032119453056e-08,
      "loss": 0.2283,
      "step": 53865
    },
    {
      "epoch": 0.9726706100722957,
      "grad_norm": 0.6768224239349365,
      "learning_rate": 3.6806799487427716e-08,
      "loss": 0.1879,
      "step": 53870
    },
    {
      "epoch": 0.9727608895051965,
      "grad_norm": 0.42994022369384766,
      "learning_rate": 3.656407924677918e-08,
      "loss": 0.1891,
      "step": 53875
    },
    {
      "epoch": 0.9728511689380972,
      "grad_norm": 0.824230968952179,
      "learning_rate": 3.632216049210824e-08,
      "loss": 0.2687,
      "step": 53880
    },
    {
      "epoch": 0.9729414483709979,
      "grad_norm": 0.1999301314353943,
      "learning_rate": 3.608104324287709e-08,
      "loss": 0.2878,
      "step": 53885
    },
    {
      "epoch": 0.9730317278038986,
      "grad_norm": 0.14496801793575287,
      "learning_rate": 3.584072751848133e-08,
      "loss": 0.2171,
      "step": 53890
    },
    {
      "epoch": 0.9731220072367993,
      "grad_norm": 0.4517415761947632,
      "learning_rate": 3.5601213338253284e-08,
      "loss": 0.2225,
      "step": 53895
    },
    {
      "epoch": 0.9732122866697,
      "grad_norm": 0.32180681824684143,
      "learning_rate": 3.5362500721458635e-08,
      "loss": 0.294,
      "step": 53900
    },
    {
      "epoch": 0.9733025661026008,
      "grad_norm": 0.3482528328895569,
      "learning_rate": 3.5124589687300926e-08,
      "loss": 0.2145,
      "step": 53905
    },
    {
      "epoch": 0.9733928455355015,
      "grad_norm": 0.2886830270290375,
      "learning_rate": 3.4887480254919285e-08,
      "loss": 0.1946,
      "step": 53910
    },
    {
      "epoch": 0.9734831249684022,
      "grad_norm": 0.6015817523002625,
      "learning_rate": 3.465117244338512e-08,
      "loss": 0.1654,
      "step": 53915
    },
    {
      "epoch": 0.9735734044013029,
      "grad_norm": 0.5138819813728333,
      "learning_rate": 3.4415666271709893e-08,
      "loss": 0.2839,
      "step": 53920
    },
    {
      "epoch": 0.9736636838342037,
      "grad_norm": 0.39462965726852417,
      "learning_rate": 3.418096175883734e-08,
      "loss": 0.165,
      "step": 53925
    },
    {
      "epoch": 0.9737539632671044,
      "grad_norm": 0.2814905643463135,
      "learning_rate": 3.3947058923647916e-08,
      "loss": 0.1518,
      "step": 53930
    },
    {
      "epoch": 0.973844242700005,
      "grad_norm": 0.556002140045166,
      "learning_rate": 3.3713957784956566e-08,
      "loss": 0.238,
      "step": 53935
    },
    {
      "epoch": 0.9739345221329058,
      "grad_norm": 0.4952412247657776,
      "learning_rate": 3.3481658361517174e-08,
      "loss": 0.1439,
      "step": 53940
    },
    {
      "epoch": 0.9740248015658065,
      "grad_norm": 0.3634929656982422,
      "learning_rate": 3.32501606720137e-08,
      "loss": 0.1923,
      "step": 53945
    },
    {
      "epoch": 0.9741150809987073,
      "grad_norm": 0.5335572361946106,
      "learning_rate": 3.3019464735069004e-08,
      "loss": 0.2511,
      "step": 53950
    },
    {
      "epoch": 0.9742053604316079,
      "grad_norm": 0.732477605342865,
      "learning_rate": 3.278957056924159e-08,
      "loss": 0.1523,
      "step": 53955
    },
    {
      "epoch": 0.9742956398645086,
      "grad_norm": 0.823394238948822,
      "learning_rate": 3.256047819302555e-08,
      "loss": 0.1512,
      "step": 53960
    },
    {
      "epoch": 0.9743859192974094,
      "grad_norm": 0.2745742201805115,
      "learning_rate": 3.2332187624848355e-08,
      "loss": 0.2163,
      "step": 53965
    },
    {
      "epoch": 0.9744761987303101,
      "grad_norm": 0.20988920331001282,
      "learning_rate": 3.210469888307422e-08,
      "loss": 0.2594,
      "step": 53970
    },
    {
      "epoch": 0.9745664781632107,
      "grad_norm": 0.43002596497535706,
      "learning_rate": 3.1878011986004044e-08,
      "loss": 0.1908,
      "step": 53975
    },
    {
      "epoch": 0.9746567575961115,
      "grad_norm": 0.2920987904071808,
      "learning_rate": 3.165212695187214e-08,
      "loss": 0.2134,
      "step": 53980
    },
    {
      "epoch": 0.9747470370290122,
      "grad_norm": 0.4996824860572815,
      "learning_rate": 3.1427043798849536e-08,
      "loss": 0.2194,
      "step": 53985
    },
    {
      "epoch": 0.974837316461913,
      "grad_norm": 0.35674232244491577,
      "learning_rate": 3.120276254504395e-08,
      "loss": 0.2013,
      "step": 53990
    },
    {
      "epoch": 0.9749275958948136,
      "grad_norm": 0.9280580878257751,
      "learning_rate": 3.0979283208495416e-08,
      "loss": 0.2114,
      "step": 53995
    },
    {
      "epoch": 0.9750178753277143,
      "grad_norm": 0.6595668792724609,
      "learning_rate": 3.075660580718065e-08,
      "loss": 0.2647,
      "step": 54000
    },
    {
      "epoch": 0.9751081547606151,
      "grad_norm": 0.45414796471595764,
      "learning_rate": 3.053473035901422e-08,
      "loss": 0.2197,
      "step": 54005
    },
    {
      "epoch": 0.9751984341935158,
      "grad_norm": 0.702100932598114,
      "learning_rate": 3.031365688184407e-08,
      "loss": 0.3289,
      "step": 54010
    },
    {
      "epoch": 0.9752887136264164,
      "grad_norm": 0.2315141260623932,
      "learning_rate": 3.009338539345374e-08,
      "loss": 0.1687,
      "step": 54015
    },
    {
      "epoch": 0.9753789930593172,
      "grad_norm": 0.4079658091068268,
      "learning_rate": 2.98739159115613e-08,
      "loss": 0.1738,
      "step": 54020
    },
    {
      "epoch": 0.9754692724922179,
      "grad_norm": 0.35546764731407166,
      "learning_rate": 2.9655248453823726e-08,
      "loss": 0.1492,
      "step": 54025
    },
    {
      "epoch": 0.9755595519251187,
      "grad_norm": 0.3246750831604004,
      "learning_rate": 2.943738303782917e-08,
      "loss": 0.1771,
      "step": 54030
    },
    {
      "epoch": 0.9756498313580193,
      "grad_norm": 0.5031123161315918,
      "learning_rate": 2.9220319681104724e-08,
      "loss": 0.1374,
      "step": 54035
    },
    {
      "epoch": 0.97574011079092,
      "grad_norm": 0.34943029284477234,
      "learning_rate": 2.900405840111087e-08,
      "loss": 0.2455,
      "step": 54040
    },
    {
      "epoch": 0.9758303902238208,
      "grad_norm": 0.2979680895805359,
      "learning_rate": 2.8788599215243683e-08,
      "loss": 0.1988,
      "step": 54045
    },
    {
      "epoch": 0.9759206696567215,
      "grad_norm": 0.838577151298523,
      "learning_rate": 2.8573942140837084e-08,
      "loss": 0.25,
      "step": 54050
    },
    {
      "epoch": 0.9760109490896222,
      "grad_norm": 0.37349367141723633,
      "learning_rate": 2.8360087195156148e-08,
      "loss": 0.1397,
      "step": 54055
    },
    {
      "epoch": 0.9761012285225229,
      "grad_norm": 0.2899876534938812,
      "learning_rate": 2.8147034395406003e-08,
      "loss": 0.2271,
      "step": 54060
    },
    {
      "epoch": 0.9761915079554236,
      "grad_norm": 0.7934709787368774,
      "learning_rate": 2.793478375872516e-08,
      "loss": 0.2183,
      "step": 54065
    },
    {
      "epoch": 0.9762817873883244,
      "grad_norm": 0.4726581871509552,
      "learning_rate": 2.7723335302186628e-08,
      "loss": 0.1587,
      "step": 54070
    },
    {
      "epoch": 0.976372066821225,
      "grad_norm": 0.39323365688323975,
      "learning_rate": 2.7512689042800134e-08,
      "loss": 0.3155,
      "step": 54075
    },
    {
      "epoch": 0.9764623462541258,
      "grad_norm": 0.4772615134716034,
      "learning_rate": 2.7302844997511013e-08,
      "loss": 0.2656,
      "step": 54080
    },
    {
      "epoch": 0.9765526256870265,
      "grad_norm": 0.29938217997550964,
      "learning_rate": 2.7093803183199096e-08,
      "loss": 0.2542,
      "step": 54085
    },
    {
      "epoch": 0.9766429051199272,
      "grad_norm": 0.5492769479751587,
      "learning_rate": 2.6885563616682042e-08,
      "loss": 0.2459,
      "step": 54090
    },
    {
      "epoch": 0.9767331845528279,
      "grad_norm": 0.7767098546028137,
      "learning_rate": 2.6678126314708675e-08,
      "loss": 0.2211,
      "step": 54095
    },
    {
      "epoch": 0.9768234639857286,
      "grad_norm": 0.5957434177398682,
      "learning_rate": 2.6471491293966755e-08,
      "loss": 0.2162,
      "step": 54100
    },
    {
      "epoch": 0.9769137434186294,
      "grad_norm": 0.4204747676849365,
      "learning_rate": 2.6265658571079654e-08,
      "loss": 0.2596,
      "step": 54105
    },
    {
      "epoch": 0.9770040228515301,
      "grad_norm": 0.5886346101760864,
      "learning_rate": 2.606062816260413e-08,
      "loss": 0.3572,
      "step": 54110
    },
    {
      "epoch": 0.9770943022844307,
      "grad_norm": 0.45081228017807007,
      "learning_rate": 2.585640008503254e-08,
      "loss": 0.2742,
      "step": 54115
    },
    {
      "epoch": 0.9771845817173315,
      "grad_norm": 0.6001715064048767,
      "learning_rate": 2.565297435479508e-08,
      "loss": 0.1913,
      "step": 54120
    },
    {
      "epoch": 0.9772748611502322,
      "grad_norm": 0.479575514793396,
      "learning_rate": 2.5450350988255325e-08,
      "loss": 0.2184,
      "step": 54125
    },
    {
      "epoch": 0.977365140583133,
      "grad_norm": 1.1889973878860474,
      "learning_rate": 2.5248530001713567e-08,
      "loss": 0.2626,
      "step": 54130
    },
    {
      "epoch": 0.9774554200160336,
      "grad_norm": 0.4684038758277893,
      "learning_rate": 2.5047511411403492e-08,
      "loss": 0.3636,
      "step": 54135
    },
    {
      "epoch": 0.9775456994489343,
      "grad_norm": 0.6237125992774963,
      "learning_rate": 2.4847295233495493e-08,
      "loss": 0.2402,
      "step": 54140
    },
    {
      "epoch": 0.9776359788818351,
      "grad_norm": 0.4272877871990204,
      "learning_rate": 2.46478814840978e-08,
      "loss": 0.2284,
      "step": 54145
    },
    {
      "epoch": 0.9777262583147358,
      "grad_norm": 0.5578311681747437,
      "learning_rate": 2.4449270179248695e-08,
      "loss": 0.2594,
      "step": 54150
    },
    {
      "epoch": 0.9778165377476364,
      "grad_norm": 0.9206238389015198,
      "learning_rate": 2.425146133492762e-08,
      "loss": 0.2056,
      "step": 54155
    },
    {
      "epoch": 0.9779068171805372,
      "grad_norm": 0.6731491088867188,
      "learning_rate": 2.4054454967045172e-08,
      "loss": 0.3118,
      "step": 54160
    },
    {
      "epoch": 0.9779970966134379,
      "grad_norm": 0.5745200514793396,
      "learning_rate": 2.385825109144979e-08,
      "loss": 0.1717,
      "step": 54165
    },
    {
      "epoch": 0.9780873760463387,
      "grad_norm": 0.25734376907348633,
      "learning_rate": 2.3662849723924407e-08,
      "loss": 0.2262,
      "step": 54170
    },
    {
      "epoch": 0.9781776554792393,
      "grad_norm": 0.39289623498916626,
      "learning_rate": 2.346825088018867e-08,
      "loss": 0.2184,
      "step": 54175
    },
    {
      "epoch": 0.97826793491214,
      "grad_norm": 0.43563422560691833,
      "learning_rate": 2.32744545758945e-08,
      "loss": 0.2393,
      "step": 54180
    },
    {
      "epoch": 0.9783582143450408,
      "grad_norm": 0.34247323870658875,
      "learning_rate": 2.3081460826633874e-08,
      "loss": 0.2528,
      "step": 54185
    },
    {
      "epoch": 0.9784484937779415,
      "grad_norm": 0.25040972232818604,
      "learning_rate": 2.2889269647929922e-08,
      "loss": 0.2516,
      "step": 54190
    },
    {
      "epoch": 0.9785387732108421,
      "grad_norm": 0.3073778450489044,
      "learning_rate": 2.269788105524362e-08,
      "loss": 0.2036,
      "step": 54195
    },
    {
      "epoch": 0.9786290526437429,
      "grad_norm": 0.29352474212646484,
      "learning_rate": 2.2507295063970425e-08,
      "loss": 0.1861,
      "step": 54200
    },
    {
      "epoch": 0.9787193320766436,
      "grad_norm": 0.4617655575275421,
      "learning_rate": 2.231751168944252e-08,
      "loss": 0.2416,
      "step": 54205
    },
    {
      "epoch": 0.9788096115095444,
      "grad_norm": 0.32537949085235596,
      "learning_rate": 2.2128530946925463e-08,
      "loss": 0.2118,
      "step": 54210
    },
    {
      "epoch": 0.978899890942445,
      "grad_norm": 0.3557182252407074,
      "learning_rate": 2.1940352851622658e-08,
      "loss": 0.2294,
      "step": 54215
    },
    {
      "epoch": 0.9789901703753457,
      "grad_norm": 0.7561355829238892,
      "learning_rate": 2.1752977418670885e-08,
      "loss": 0.2576,
      "step": 54220
    },
    {
      "epoch": 0.9790804498082465,
      "grad_norm": 0.23972244560718536,
      "learning_rate": 2.1566404663142527e-08,
      "loss": 0.2196,
      "step": 54225
    },
    {
      "epoch": 0.9791707292411472,
      "grad_norm": 0.32039886713027954,
      "learning_rate": 2.138063460004669e-08,
      "loss": 0.2188,
      "step": 54230
    },
    {
      "epoch": 0.9792610086740479,
      "grad_norm": 0.5600818395614624,
      "learning_rate": 2.119566724432809e-08,
      "loss": 0.2569,
      "step": 54235
    },
    {
      "epoch": 0.9793512881069486,
      "grad_norm": 0.6730336546897888,
      "learning_rate": 2.101150261086482e-08,
      "loss": 0.2534,
      "step": 54240
    },
    {
      "epoch": 0.9794415675398493,
      "grad_norm": 0.3314119577407837,
      "learning_rate": 2.08281407144717e-08,
      "loss": 0.2092,
      "step": 54245
    },
    {
      "epoch": 0.9795318469727501,
      "grad_norm": 0.36345618963241577,
      "learning_rate": 2.0645581569899153e-08,
      "loss": 0.1609,
      "step": 54250
    },
    {
      "epoch": 0.9796221264056507,
      "grad_norm": 0.5022575259208679,
      "learning_rate": 2.0463825191833207e-08,
      "loss": 0.2596,
      "step": 54255
    },
    {
      "epoch": 0.9797124058385515,
      "grad_norm": 0.45627039670944214,
      "learning_rate": 2.0282871594893284e-08,
      "loss": 0.2048,
      "step": 54260
    },
    {
      "epoch": 0.9798026852714522,
      "grad_norm": 0.37281811237335205,
      "learning_rate": 2.0102720793637732e-08,
      "loss": 0.1054,
      "step": 54265
    },
    {
      "epoch": 0.9798929647043529,
      "grad_norm": 0.4168417751789093,
      "learning_rate": 1.9923372802556073e-08,
      "loss": 0.2717,
      "step": 54270
    },
    {
      "epoch": 0.9799832441372536,
      "grad_norm": 0.5583621263504028,
      "learning_rate": 1.974482763607788e-08,
      "loss": 0.176,
      "step": 54275
    },
    {
      "epoch": 0.9800735235701543,
      "grad_norm": 0.6229156255722046,
      "learning_rate": 1.9567085308564994e-08,
      "loss": 0.2551,
      "step": 54280
    },
    {
      "epoch": 0.9801638030030551,
      "grad_norm": 0.5340616106987,
      "learning_rate": 1.9390145834314867e-08,
      "loss": 0.2462,
      "step": 54285
    },
    {
      "epoch": 0.9802540824359558,
      "grad_norm": 0.5986381769180298,
      "learning_rate": 1.921400922756167e-08,
      "loss": 0.2074,
      "step": 54290
    },
    {
      "epoch": 0.9803443618688565,
      "grad_norm": 0.6956779360771179,
      "learning_rate": 1.9038675502474068e-08,
      "loss": 0.2119,
      "step": 54295
    },
    {
      "epoch": 0.9804346413017572,
      "grad_norm": 0.5290546417236328,
      "learning_rate": 1.8864144673156336e-08,
      "loss": 0.2377,
      "step": 54300
    },
    {
      "epoch": 0.9805249207346579,
      "grad_norm": 0.6634647846221924,
      "learning_rate": 1.8690416753648356e-08,
      "loss": 0.1975,
      "step": 54305
    },
    {
      "epoch": 0.9806152001675587,
      "grad_norm": 0.5973348617553711,
      "learning_rate": 1.851749175792561e-08,
      "loss": 0.28,
      "step": 54310
    },
    {
      "epoch": 0.9807054796004594,
      "grad_norm": 0.3736438453197479,
      "learning_rate": 1.8345369699898085e-08,
      "loss": 0.2068,
      "step": 54315
    },
    {
      "epoch": 0.98079575903336,
      "grad_norm": 0.5507653951644897,
      "learning_rate": 1.8174050593411375e-08,
      "loss": 0.1457,
      "step": 54320
    },
    {
      "epoch": 0.9808860384662608,
      "grad_norm": 0.3628789484500885,
      "learning_rate": 1.8003534452247786e-08,
      "loss": 0.2868,
      "step": 54325
    },
    {
      "epoch": 0.9809763178991615,
      "grad_norm": 0.39264750480651855,
      "learning_rate": 1.783382129012412e-08,
      "loss": 0.1769,
      "step": 54330
    },
    {
      "epoch": 0.9810665973320623,
      "grad_norm": 0.4039103090763092,
      "learning_rate": 1.7664911120691684e-08,
      "loss": 0.2148,
      "step": 54335
    },
    {
      "epoch": 0.9811568767649629,
      "grad_norm": 0.19867925345897675,
      "learning_rate": 1.74968039575385e-08,
      "loss": 0.2144,
      "step": 54340
    },
    {
      "epoch": 0.9812471561978636,
      "grad_norm": 0.3533666431903839,
      "learning_rate": 1.7329499814187077e-08,
      "loss": 0.2703,
      "step": 54345
    },
    {
      "epoch": 0.9813374356307644,
      "grad_norm": 0.5271742939949036,
      "learning_rate": 1.7162998704097768e-08,
      "loss": 0.2183,
      "step": 54350
    },
    {
      "epoch": 0.9814277150636651,
      "grad_norm": 0.2934906780719757,
      "learning_rate": 1.6997300640662075e-08,
      "loss": 0.1751,
      "step": 54355
    },
    {
      "epoch": 0.9815179944965657,
      "grad_norm": 0.3343862295150757,
      "learning_rate": 1.6832405637210447e-08,
      "loss": 0.262,
      "step": 54360
    },
    {
      "epoch": 0.9816082739294665,
      "grad_norm": 0.1872626692056656,
      "learning_rate": 1.6668313707006723e-08,
      "loss": 0.2004,
      "step": 54365
    },
    {
      "epoch": 0.9816985533623672,
      "grad_norm": 0.4585855305194855,
      "learning_rate": 1.650502486325145e-08,
      "loss": 0.2252,
      "step": 54370
    },
    {
      "epoch": 0.981788832795268,
      "grad_norm": 0.6149754524230957,
      "learning_rate": 1.634253911908079e-08,
      "loss": 0.2997,
      "step": 54375
    },
    {
      "epoch": 0.9818791122281686,
      "grad_norm": 0.3708350956439972,
      "learning_rate": 1.618085648756318e-08,
      "loss": 0.2633,
      "step": 54380
    },
    {
      "epoch": 0.9819693916610693,
      "grad_norm": 0.43999844789505005,
      "learning_rate": 1.6019976981707096e-08,
      "loss": 0.3044,
      "step": 54385
    },
    {
      "epoch": 0.9820596710939701,
      "grad_norm": 0.4915013909339905,
      "learning_rate": 1.5859900614453306e-08,
      "loss": 0.1663,
      "step": 54390
    },
    {
      "epoch": 0.9821499505268708,
      "grad_norm": 0.6268812417984009,
      "learning_rate": 1.570062739867817e-08,
      "loss": 0.357,
      "step": 54395
    },
    {
      "epoch": 0.9822402299597714,
      "grad_norm": 0.6899821162223816,
      "learning_rate": 1.5542157347195885e-08,
      "loss": 0.2998,
      "step": 54400
    },
    {
      "epoch": 0.9823305093926722,
      "grad_norm": 0.2525346577167511,
      "learning_rate": 1.5384490472751813e-08,
      "loss": 0.2131,
      "step": 54405
    },
    {
      "epoch": 0.9824207888255729,
      "grad_norm": 0.601879358291626,
      "learning_rate": 1.5227626788031358e-08,
      "loss": 0.1923,
      "step": 54410
    },
    {
      "epoch": 0.9825110682584737,
      "grad_norm": 0.6370570659637451,
      "learning_rate": 1.5071566305651096e-08,
      "loss": 0.1979,
      "step": 54415
    },
    {
      "epoch": 0.9826013476913743,
      "grad_norm": 0.44733312726020813,
      "learning_rate": 1.4916309038166542e-08,
      "loss": 0.2806,
      "step": 54420
    },
    {
      "epoch": 0.982691627124275,
      "grad_norm": 0.4717170000076294,
      "learning_rate": 1.4761854998065483e-08,
      "loss": 0.1165,
      "step": 54425
    },
    {
      "epoch": 0.9827819065571758,
      "grad_norm": 0.39846286177635193,
      "learning_rate": 1.4608204197774644e-08,
      "loss": 0.2344,
      "step": 54430
    },
    {
      "epoch": 0.9828721859900765,
      "grad_norm": 0.760108232498169,
      "learning_rate": 1.4455356649651919e-08,
      "loss": 0.2154,
      "step": 54435
    },
    {
      "epoch": 0.9829624654229772,
      "grad_norm": 0.3395760655403137,
      "learning_rate": 1.4303312365995248e-08,
      "loss": 0.1984,
      "step": 54440
    },
    {
      "epoch": 0.9830527448558779,
      "grad_norm": 0.5740180611610413,
      "learning_rate": 1.4152071359032626e-08,
      "loss": 0.0967,
      "step": 54445
    },
    {
      "epoch": 0.9831430242887786,
      "grad_norm": 0.21744437515735626,
      "learning_rate": 1.40016336409321e-08,
      "loss": 0.2766,
      "step": 54450
    },
    {
      "epoch": 0.9832333037216794,
      "grad_norm": 0.4723064601421356,
      "learning_rate": 1.3851999223795098e-08,
      "loss": 0.253,
      "step": 54455
    },
    {
      "epoch": 0.98332358315458,
      "grad_norm": 0.6886090636253357,
      "learning_rate": 1.3703168119658661e-08,
      "loss": 0.1891,
      "step": 54460
    },
    {
      "epoch": 0.9834138625874808,
      "grad_norm": 0.6787614226341248,
      "learning_rate": 1.3555140340494322e-08,
      "loss": 0.1867,
      "step": 54465
    },
    {
      "epoch": 0.9835041420203815,
      "grad_norm": 0.42908579111099243,
      "learning_rate": 1.3407915898211443e-08,
      "loss": 0.1815,
      "step": 54470
    },
    {
      "epoch": 0.9835944214532822,
      "grad_norm": 0.4250125288963318,
      "learning_rate": 1.3261494804651664e-08,
      "loss": 0.2078,
      "step": 54475
    },
    {
      "epoch": 0.9836847008861829,
      "grad_norm": 0.5903279185295105,
      "learning_rate": 1.3115877071593341e-08,
      "loss": 0.2514,
      "step": 54480
    },
    {
      "epoch": 0.9837749803190836,
      "grad_norm": 0.3856804072856903,
      "learning_rate": 1.2971062710751547e-08,
      "loss": 0.2237,
      "step": 54485
    },
    {
      "epoch": 0.9838652597519844,
      "grad_norm": 0.3865649104118347,
      "learning_rate": 1.2827051733775852e-08,
      "loss": 0.2022,
      "step": 54490
    },
    {
      "epoch": 0.9839555391848851,
      "grad_norm": 0.4912815988063812,
      "learning_rate": 1.2683844152249214e-08,
      "loss": 0.1831,
      "step": 54495
    },
    {
      "epoch": 0.9840458186177857,
      "grad_norm": 0.3872782289981842,
      "learning_rate": 1.2541439977692415e-08,
      "loss": 0.2566,
      "step": 54500
    },
    {
      "epoch": 0.9841360980506865,
      "grad_norm": 0.6470474600791931,
      "learning_rate": 1.239983922156185e-08,
      "loss": 0.2493,
      "step": 54505
    },
    {
      "epoch": 0.9842263774835872,
      "grad_norm": 0.3223435580730438,
      "learning_rate": 1.2259041895246182e-08,
      "loss": 0.1579,
      "step": 54510
    },
    {
      "epoch": 0.984316656916488,
      "grad_norm": 0.07871793955564499,
      "learning_rate": 1.211904801007302e-08,
      "loss": 0.182,
      "step": 54515
    },
    {
      "epoch": 0.9844069363493886,
      "grad_norm": 0.6256330013275146,
      "learning_rate": 1.1979857577304465e-08,
      "loss": 0.2576,
      "step": 54520
    },
    {
      "epoch": 0.9844972157822893,
      "grad_norm": 0.44302675127983093,
      "learning_rate": 1.1841470608134897e-08,
      "loss": 0.1846,
      "step": 54525
    },
    {
      "epoch": 0.9845874952151901,
      "grad_norm": 0.5845201015472412,
      "learning_rate": 1.1703887113698742e-08,
      "loss": 0.2834,
      "step": 54530
    },
    {
      "epoch": 0.9846777746480908,
      "grad_norm": 0.3106643557548523,
      "learning_rate": 1.1567107105063812e-08,
      "loss": 0.1771,
      "step": 54535
    },
    {
      "epoch": 0.9847680540809914,
      "grad_norm": 0.656623899936676,
      "learning_rate": 1.143113059323131e-08,
      "loss": 0.1233,
      "step": 54540
    },
    {
      "epoch": 0.9848583335138922,
      "grad_norm": 0.6997153162956238,
      "learning_rate": 1.129595758914026e-08,
      "loss": 0.346,
      "step": 54545
    },
    {
      "epoch": 0.9849486129467929,
      "grad_norm": 0.3024485409259796,
      "learning_rate": 1.1161588103665299e-08,
      "loss": 0.2499,
      "step": 54550
    },
    {
      "epoch": 0.9850388923796937,
      "grad_norm": 0.49105653166770935,
      "learning_rate": 1.1028022147614448e-08,
      "loss": 0.2373,
      "step": 54555
    },
    {
      "epoch": 0.9851291718125943,
      "grad_norm": 0.3709699213504791,
      "learning_rate": 1.0895259731731334e-08,
      "loss": 0.1944,
      "step": 54560
    },
    {
      "epoch": 0.985219451245495,
      "grad_norm": 0.4455859661102295,
      "learning_rate": 1.0763300866698523e-08,
      "loss": 0.2043,
      "step": 54565
    },
    {
      "epoch": 0.9853097306783958,
      "grad_norm": 0.5328179597854614,
      "learning_rate": 1.0632145563128638e-08,
      "loss": 0.2008,
      "step": 54570
    },
    {
      "epoch": 0.9854000101112965,
      "grad_norm": 0.5087750554084778,
      "learning_rate": 1.0501793831572127e-08,
      "loss": 0.2165,
      "step": 54575
    },
    {
      "epoch": 0.9854902895441972,
      "grad_norm": 0.4027363657951355,
      "learning_rate": 1.0372245682516157e-08,
      "loss": 0.1802,
      "step": 54580
    },
    {
      "epoch": 0.9855805689770979,
      "grad_norm": 0.47149139642715454,
      "learning_rate": 1.0243501126381283e-08,
      "loss": 0.2732,
      "step": 54585
    },
    {
      "epoch": 0.9856708484099986,
      "grad_norm": 0.4707593619823456,
      "learning_rate": 1.0115560173524774e-08,
      "loss": 0.1656,
      "step": 54590
    },
    {
      "epoch": 0.9857611278428994,
      "grad_norm": 1.3330811262130737,
      "learning_rate": 9.988422834237287e-09,
      "loss": 0.2148,
      "step": 54595
    },
    {
      "epoch": 0.9858514072758,
      "grad_norm": 0.5004041194915771,
      "learning_rate": 9.862089118747309e-09,
      "loss": 0.2962,
      "step": 54600
    },
    {
      "epoch": 0.9859416867087007,
      "grad_norm": 0.3979488015174866,
      "learning_rate": 9.736559037216708e-09,
      "loss": 0.2138,
      "step": 54605
    },
    {
      "epoch": 0.9860319661416015,
      "grad_norm": 0.41310417652130127,
      "learning_rate": 9.611832599744075e-09,
      "loss": 0.1875,
      "step": 54610
    },
    {
      "epoch": 0.9861222455745022,
      "grad_norm": 0.39090973138809204,
      "learning_rate": 9.487909816361384e-09,
      "loss": 0.181,
      "step": 54615
    },
    {
      "epoch": 0.9862125250074029,
      "grad_norm": 0.6974276900291443,
      "learning_rate": 9.36479069703955e-09,
      "loss": 0.173,
      "step": 54620
    },
    {
      "epoch": 0.9863028044403036,
      "grad_norm": 0.3585265874862671,
      "learning_rate": 9.24247525168065e-09,
      "loss": 0.2758,
      "step": 54625
    },
    {
      "epoch": 0.9863930838732043,
      "grad_norm": 0.6234022974967957,
      "learning_rate": 9.12096349012459e-09,
      "loss": 0.3286,
      "step": 54630
    },
    {
      "epoch": 0.9864833633061051,
      "grad_norm": 0.3395076394081116,
      "learning_rate": 9.000255422146886e-09,
      "loss": 0.201,
      "step": 54635
    },
    {
      "epoch": 0.9865736427390057,
      "grad_norm": 0.24648065865039825,
      "learning_rate": 8.880351057456437e-09,
      "loss": 0.1741,
      "step": 54640
    },
    {
      "epoch": 0.9866639221719065,
      "grad_norm": 0.3816898763179779,
      "learning_rate": 8.761250405699973e-09,
      "loss": 0.1558,
      "step": 54645
    },
    {
      "epoch": 0.9867542016048072,
      "grad_norm": 0.6216617226600647,
      "learning_rate": 8.642953476457604e-09,
      "loss": 0.228,
      "step": 54650
    },
    {
      "epoch": 0.9868444810377079,
      "grad_norm": 0.4171766936779022,
      "learning_rate": 8.525460279245057e-09,
      "loss": 0.2411,
      "step": 54655
    },
    {
      "epoch": 0.9869347604706086,
      "grad_norm": 0.17013856768608093,
      "learning_rate": 8.408770823513657e-09,
      "loss": 0.1842,
      "step": 54660
    },
    {
      "epoch": 0.9870250399035093,
      "grad_norm": 0.7724552154541016,
      "learning_rate": 8.29288511865145e-09,
      "loss": 0.1417,
      "step": 54665
    },
    {
      "epoch": 0.9871153193364101,
      "grad_norm": 0.2612602412700653,
      "learning_rate": 8.17780317397987e-09,
      "loss": 0.2186,
      "step": 54670
    },
    {
      "epoch": 0.9872055987693108,
      "grad_norm": 0.43372562527656555,
      "learning_rate": 8.063524998755956e-09,
      "loss": 0.2414,
      "step": 54675
    },
    {
      "epoch": 0.9872958782022114,
      "grad_norm": 0.6091335415840149,
      "learning_rate": 7.950050602173465e-09,
      "loss": 0.2543,
      "step": 54680
    },
    {
      "epoch": 0.9873861576351122,
      "grad_norm": 0.4017021059989929,
      "learning_rate": 7.83737999336065e-09,
      "loss": 0.2791,
      "step": 54685
    },
    {
      "epoch": 0.9874764370680129,
      "grad_norm": 0.584742546081543,
      "learning_rate": 7.725513181380262e-09,
      "loss": 0.274,
      "step": 54690
    },
    {
      "epoch": 0.9875667165009137,
      "grad_norm": 0.38122114539146423,
      "learning_rate": 7.614450175230658e-09,
      "loss": 0.1558,
      "step": 54695
    },
    {
      "epoch": 0.9876569959338144,
      "grad_norm": 0.4123266637325287,
      "learning_rate": 7.504190983848025e-09,
      "loss": 0.2442,
      "step": 54700
    },
    {
      "epoch": 0.987747275366715,
      "grad_norm": 0.29697564244270325,
      "learning_rate": 7.394735616099713e-09,
      "loss": 0.2092,
      "step": 54705
    },
    {
      "epoch": 0.9878375547996158,
      "grad_norm": 0.3298875391483307,
      "learning_rate": 7.286084080793121e-09,
      "loss": 0.2783,
      "step": 54710
    },
    {
      "epoch": 0.9879278342325165,
      "grad_norm": 0.6056180596351624,
      "learning_rate": 7.1782363866657045e-09,
      "loss": 0.201,
      "step": 54715
    },
    {
      "epoch": 0.9880181136654173,
      "grad_norm": 0.8074268102645874,
      "learning_rate": 7.071192542394967e-09,
      "loss": 0.2275,
      "step": 54720
    },
    {
      "epoch": 0.9881083930983179,
      "grad_norm": 0.6272867918014526,
      "learning_rate": 6.964952556590687e-09,
      "loss": 0.2298,
      "step": 54725
    },
    {
      "epoch": 0.9881986725312186,
      "grad_norm": 0.3642462491989136,
      "learning_rate": 6.859516437799363e-09,
      "loss": 0.2761,
      "step": 54730
    },
    {
      "epoch": 0.9882889519641194,
      "grad_norm": 0.52139812707901,
      "learning_rate": 6.754884194503097e-09,
      "loss": 0.2101,
      "step": 54735
    },
    {
      "epoch": 0.9883792313970201,
      "grad_norm": 0.5058034062385559,
      "learning_rate": 6.651055835118492e-09,
      "loss": 0.1981,
      "step": 54740
    },
    {
      "epoch": 0.9884695108299207,
      "grad_norm": 0.38746005296707153,
      "learning_rate": 6.548031367998864e-09,
      "loss": 0.188,
      "step": 54745
    },
    {
      "epoch": 0.9885597902628215,
      "grad_norm": 0.2977273762226105,
      "learning_rate": 6.445810801428698e-09,
      "loss": 0.1936,
      "step": 54750
    },
    {
      "epoch": 0.9886500696957222,
      "grad_norm": 0.47060200572013855,
      "learning_rate": 6.344394143634747e-09,
      "loss": 0.1698,
      "step": 54755
    },
    {
      "epoch": 0.988740349128623,
      "grad_norm": 0.5865183472633362,
      "learning_rate": 6.243781402772709e-09,
      "loss": 0.2928,
      "step": 54760
    },
    {
      "epoch": 0.9888306285615236,
      "grad_norm": 0.414611279964447,
      "learning_rate": 6.143972586936109e-09,
      "loss": 0.2012,
      "step": 54765
    },
    {
      "epoch": 0.9889209079944243,
      "grad_norm": 0.49423834681510925,
      "learning_rate": 6.0449677041551914e-09,
      "loss": 0.2292,
      "step": 54770
    },
    {
      "epoch": 0.9890111874273251,
      "grad_norm": 0.4946470260620117,
      "learning_rate": 5.9467667623935854e-09,
      "loss": 0.2609,
      "step": 54775
    },
    {
      "epoch": 0.9891014668602258,
      "grad_norm": 0.4100695848464966,
      "learning_rate": 5.849369769550528e-09,
      "loss": 0.1228,
      "step": 54780
    },
    {
      "epoch": 0.9891917462931265,
      "grad_norm": 0.45060092210769653,
      "learning_rate": 5.752776733461973e-09,
      "loss": 0.2696,
      "step": 54785
    },
    {
      "epoch": 0.9892820257260272,
      "grad_norm": 0.4754217565059662,
      "learning_rate": 5.656987661896152e-09,
      "loss": 0.201,
      "step": 54790
    },
    {
      "epoch": 0.9893723051589279,
      "grad_norm": 0.3832783102989197,
      "learning_rate": 5.562002562560232e-09,
      "loss": 0.2685,
      "step": 54795
    },
    {
      "epoch": 0.9894625845918287,
      "grad_norm": 0.3240828216075897,
      "learning_rate": 5.467821443095877e-09,
      "loss": 0.2837,
      "step": 54800
    },
    {
      "epoch": 0.9895528640247293,
      "grad_norm": 0.5363389849662781,
      "learning_rate": 5.374444311075921e-09,
      "loss": 0.2026,
      "step": 54805
    },
    {
      "epoch": 0.98964314345763,
      "grad_norm": 0.5036905407905579,
      "learning_rate": 5.281871174015463e-09,
      "loss": 0.2955,
      "step": 54810
    },
    {
      "epoch": 0.9897334228905308,
      "grad_norm": 0.39396196603775024,
      "learning_rate": 5.190102039358546e-09,
      "loss": 0.2446,
      "step": 54815
    },
    {
      "epoch": 0.9898237023234315,
      "grad_norm": 0.5660658478736877,
      "learning_rate": 5.099136914490377e-09,
      "loss": 0.1853,
      "step": 54820
    },
    {
      "epoch": 0.9899139817563322,
      "grad_norm": 0.4224463105201721,
      "learning_rate": 5.0089758067251026e-09,
      "loss": 0.1955,
      "step": 54825
    },
    {
      "epoch": 0.9900042611892329,
      "grad_norm": 0.8481043577194214,
      "learning_rate": 4.919618723318032e-09,
      "loss": 0.1669,
      "step": 54830
    },
    {
      "epoch": 0.9900945406221336,
      "grad_norm": 0.3382450342178345,
      "learning_rate": 4.831065671455637e-09,
      "loss": 0.2177,
      "step": 54835
    },
    {
      "epoch": 0.9901848200550344,
      "grad_norm": 0.38149264454841614,
      "learning_rate": 4.743316658263331e-09,
      "loss": 0.24,
      "step": 54840
    },
    {
      "epoch": 0.990275099487935,
      "grad_norm": 0.36693939566612244,
      "learning_rate": 4.65637169079769e-09,
      "loss": 0.1988,
      "step": 54845
    },
    {
      "epoch": 0.9903653789208358,
      "grad_norm": 1.147173523902893,
      "learning_rate": 4.570230776053119e-09,
      "loss": 0.2925,
      "step": 54850
    },
    {
      "epoch": 0.9904556583537365,
      "grad_norm": 0.6585434079170227,
      "learning_rate": 4.484893920960742e-09,
      "loss": 0.219,
      "step": 54855
    },
    {
      "epoch": 0.9905459377866372,
      "grad_norm": 0.4392758905887604,
      "learning_rate": 4.400361132383957e-09,
      "loss": 0.2885,
      "step": 54860
    },
    {
      "epoch": 0.9906362172195379,
      "grad_norm": 0.37770336866378784,
      "learning_rate": 4.316632417122879e-09,
      "loss": 0.2936,
      "step": 54865
    },
    {
      "epoch": 0.9907264966524386,
      "grad_norm": 0.5151588916778564,
      "learning_rate": 4.233707781914342e-09,
      "loss": 0.1824,
      "step": 54870
    },
    {
      "epoch": 0.9908167760853394,
      "grad_norm": 0.45082736015319824,
      "learning_rate": 4.151587233426346e-09,
      "loss": 0.2648,
      "step": 54875
    },
    {
      "epoch": 0.9909070555182401,
      "grad_norm": 0.6206878423690796,
      "learning_rate": 4.070270778266938e-09,
      "loss": 0.2074,
      "step": 54880
    },
    {
      "epoch": 0.9909973349511407,
      "grad_norm": 0.7082839608192444,
      "learning_rate": 3.989758422976442e-09,
      "loss": 0.205,
      "step": 54885
    },
    {
      "epoch": 0.9910876143840415,
      "grad_norm": 0.5329179763793945,
      "learning_rate": 3.910050174031898e-09,
      "loss": 0.1578,
      "step": 54890
    },
    {
      "epoch": 0.9911778938169422,
      "grad_norm": 0.5654190182685852,
      "learning_rate": 3.8311460378459564e-09,
      "loss": 0.1589,
      "step": 54895
    },
    {
      "epoch": 0.991268173249843,
      "grad_norm": 0.5214343070983887,
      "learning_rate": 3.75304602076354e-09,
      "loss": 0.1954,
      "step": 54900
    },
    {
      "epoch": 0.9913584526827436,
      "grad_norm": 0.5556568503379822,
      "learning_rate": 3.675750129070732e-09,
      "loss": 0.2024,
      "step": 54905
    },
    {
      "epoch": 0.9914487321156443,
      "grad_norm": 0.4347390830516815,
      "learning_rate": 3.599258368982561e-09,
      "loss": 0.2587,
      "step": 54910
    },
    {
      "epoch": 0.9915390115485451,
      "grad_norm": 0.7272732853889465,
      "learning_rate": 3.5235707466529935e-09,
      "loss": 0.1688,
      "step": 54915
    },
    {
      "epoch": 0.9916292909814458,
      "grad_norm": 0.3668297231197357,
      "learning_rate": 3.448687268170492e-09,
      "loss": 0.2342,
      "step": 54920
    },
    {
      "epoch": 0.9917195704143464,
      "grad_norm": 0.26720502972602844,
      "learning_rate": 3.374607939560237e-09,
      "loss": 0.1538,
      "step": 54925
    },
    {
      "epoch": 0.9918098498472472,
      "grad_norm": 0.5334031581878662,
      "learning_rate": 3.3013327667796857e-09,
      "loss": 0.1935,
      "step": 54930
    },
    {
      "epoch": 0.9919001292801479,
      "grad_norm": 1.0811445713043213,
      "learning_rate": 3.2288617557241218e-09,
      "loss": 0.333,
      "step": 54935
    },
    {
      "epoch": 0.9919904087130487,
      "grad_norm": 0.4220045804977417,
      "learning_rate": 3.1571949122222167e-09,
      "loss": 0.1772,
      "step": 54940
    },
    {
      "epoch": 0.9920806881459493,
      "grad_norm": 0.41664108633995056,
      "learning_rate": 3.0863322420404685e-09,
      "loss": 0.2029,
      "step": 54945
    },
    {
      "epoch": 0.99217096757885,
      "grad_norm": 0.353105366230011,
      "learning_rate": 3.016273750878762e-09,
      "loss": 0.1825,
      "step": 54950
    },
    {
      "epoch": 0.9922612470117508,
      "grad_norm": 0.35080665349960327,
      "learning_rate": 2.9470194443736997e-09,
      "loss": 0.1995,
      "step": 54955
    },
    {
      "epoch": 0.9923515264446515,
      "grad_norm": 0.31684285402297974,
      "learning_rate": 2.8785693280941606e-09,
      "loss": 0.1536,
      "step": 54960
    },
    {
      "epoch": 0.9924418058775522,
      "grad_norm": 0.5460395216941833,
      "learning_rate": 2.810923407547961e-09,
      "loss": 0.2148,
      "step": 54965
    },
    {
      "epoch": 0.9925320853104529,
      "grad_norm": 0.5535667538642883,
      "learning_rate": 2.744081688176303e-09,
      "loss": 0.2236,
      "step": 54970
    },
    {
      "epoch": 0.9926223647433536,
      "grad_norm": 0.39629554748535156,
      "learning_rate": 2.6780441753559983e-09,
      "loss": 0.1486,
      "step": 54975
    },
    {
      "epoch": 0.9927126441762544,
      "grad_norm": 0.9714711308479309,
      "learning_rate": 2.612810874399463e-09,
      "loss": 0.2227,
      "step": 54980
    },
    {
      "epoch": 0.992802923609155,
      "grad_norm": 0.7687849402427673,
      "learning_rate": 2.5483817905547215e-09,
      "loss": 0.2099,
      "step": 54985
    },
    {
      "epoch": 0.9928932030420557,
      "grad_norm": 0.4090625047683716,
      "learning_rate": 2.4847569290031847e-09,
      "loss": 0.2424,
      "step": 54990
    },
    {
      "epoch": 0.9929834824749565,
      "grad_norm": 0.2441577911376953,
      "learning_rate": 2.4219362948652016e-09,
      "loss": 0.1599,
      "step": 54995
    },
    {
      "epoch": 0.9930737619078572,
      "grad_norm": 0.3369239866733551,
      "learning_rate": 2.359919893192286e-09,
      "loss": 0.1663,
      "step": 55000
    },
    {
      "epoch": 0.9931640413407579,
      "grad_norm": 0.5537863373756409,
      "learning_rate": 2.2987077289748914e-09,
      "loss": 0.1604,
      "step": 55005
    },
    {
      "epoch": 0.9932543207736586,
      "grad_norm": 0.4537431597709656,
      "learning_rate": 2.2382998071346362e-09,
      "loss": 0.1988,
      "step": 55010
    },
    {
      "epoch": 0.9933446002065593,
      "grad_norm": 0.3401337265968323,
      "learning_rate": 2.1786961325331867e-09,
      "loss": 0.14,
      "step": 55015
    },
    {
      "epoch": 0.9934348796394601,
      "grad_norm": 0.38408929109573364,
      "learning_rate": 2.1198967099644864e-09,
      "loss": 0.2203,
      "step": 55020
    },
    {
      "epoch": 0.9935251590723607,
      "grad_norm": 0.17645908892154694,
      "learning_rate": 2.061901544159195e-09,
      "loss": 0.2402,
      "step": 55025
    },
    {
      "epoch": 0.9936154385052615,
      "grad_norm": 0.8458005785942078,
      "learning_rate": 2.0047106397813598e-09,
      "loss": 0.2522,
      "step": 55030
    },
    {
      "epoch": 0.9937057179381622,
      "grad_norm": 0.4414557218551636,
      "learning_rate": 1.948324001431745e-09,
      "loss": 0.1801,
      "step": 55035
    },
    {
      "epoch": 0.993795997371063,
      "grad_norm": 0.29727786779403687,
      "learning_rate": 1.892741633646722e-09,
      "loss": 0.1798,
      "step": 55040
    },
    {
      "epoch": 0.9938862768039636,
      "grad_norm": 0.2330741137266159,
      "learning_rate": 1.8379635408971586e-09,
      "loss": 0.1889,
      "step": 55045
    },
    {
      "epoch": 0.9939765562368643,
      "grad_norm": 0.7773925065994263,
      "learning_rate": 1.7839897275895301e-09,
      "loss": 0.2301,
      "step": 55050
    },
    {
      "epoch": 0.9940668356697651,
      "grad_norm": 0.48192688822746277,
      "learning_rate": 1.7308201980670291e-09,
      "loss": 0.1987,
      "step": 55055
    },
    {
      "epoch": 0.9941571151026658,
      "grad_norm": 0.23969459533691406,
      "learning_rate": 1.6784549566051245e-09,
      "loss": 0.2059,
      "step": 55060
    },
    {
      "epoch": 0.9942473945355664,
      "grad_norm": 1.0368404388427734,
      "learning_rate": 1.6268940074160022e-09,
      "loss": 0.2315,
      "step": 55065
    },
    {
      "epoch": 0.9943376739684672,
      "grad_norm": 0.4173521101474762,
      "learning_rate": 1.576137354648566e-09,
      "loss": 0.2434,
      "step": 55070
    },
    {
      "epoch": 0.9944279534013679,
      "grad_norm": 0.5417553186416626,
      "learning_rate": 1.5261850023851055e-09,
      "loss": 0.1798,
      "step": 55075
    },
    {
      "epoch": 0.9945182328342687,
      "grad_norm": 0.7902389168739319,
      "learning_rate": 1.4770369546446284e-09,
      "loss": 0.2487,
      "step": 55080
    },
    {
      "epoch": 0.9946085122671694,
      "grad_norm": 0.6232701539993286,
      "learning_rate": 1.4286932153806388e-09,
      "loss": 0.2372,
      "step": 55085
    },
    {
      "epoch": 0.99469879170007,
      "grad_norm": 0.7035704851150513,
      "learning_rate": 1.3811537884800274e-09,
      "loss": 0.2957,
      "step": 55090
    },
    {
      "epoch": 0.9947890711329708,
      "grad_norm": 0.37746915221214294,
      "learning_rate": 1.3344186777697332e-09,
      "loss": 0.1739,
      "step": 55095
    },
    {
      "epoch": 0.9948793505658715,
      "grad_norm": 0.566070556640625,
      "learning_rate": 1.288487887007861e-09,
      "loss": 0.2797,
      "step": 55100
    },
    {
      "epoch": 0.9949696299987723,
      "grad_norm": 0.4011896252632141,
      "learning_rate": 1.2433614198903432e-09,
      "loss": 0.1882,
      "step": 55105
    },
    {
      "epoch": 0.9950599094316729,
      "grad_norm": 0.5778722763061523,
      "learning_rate": 1.1990392800453888e-09,
      "loss": 0.2647,
      "step": 55110
    },
    {
      "epoch": 0.9951501888645736,
      "grad_norm": 0.42765867710113525,
      "learning_rate": 1.1555214710390339e-09,
      "loss": 0.238,
      "step": 55115
    },
    {
      "epoch": 0.9952404682974744,
      "grad_norm": 0.4244074523448944,
      "learning_rate": 1.1128079963729222e-09,
      "loss": 0.0957,
      "step": 55120
    },
    {
      "epoch": 0.9953307477303751,
      "grad_norm": 0.36028406023979187,
      "learning_rate": 1.0708988594831937e-09,
      "loss": 0.2058,
      "step": 55125
    },
    {
      "epoch": 0.9954210271632757,
      "grad_norm": 0.30583226680755615,
      "learning_rate": 1.0297940637393755e-09,
      "loss": 0.2262,
      "step": 55130
    },
    {
      "epoch": 0.9955113065961765,
      "grad_norm": 0.6024458408355713,
      "learning_rate": 9.894936124499322e-10,
      "loss": 0.2973,
      "step": 55135
    },
    {
      "epoch": 0.9956015860290772,
      "grad_norm": 0.4497954845428467,
      "learning_rate": 9.499975088556046e-10,
      "loss": 0.2808,
      "step": 55140
    },
    {
      "epoch": 0.995691865461978,
      "grad_norm": 0.5909044146537781,
      "learning_rate": 9.11305756132741e-10,
      "loss": 0.2521,
      "step": 55145
    },
    {
      "epoch": 0.9957821448948786,
      "grad_norm": 0.40942302346229553,
      "learning_rate": 8.734183573966271e-10,
      "loss": 0.2495,
      "step": 55150
    },
    {
      "epoch": 0.9958724243277793,
      "grad_norm": 0.18043410778045654,
      "learning_rate": 8.363353156926046e-10,
      "loss": 0.2246,
      "step": 55155
    },
    {
      "epoch": 0.9959627037606801,
      "grad_norm": 0.6366596817970276,
      "learning_rate": 8.00056634003843e-10,
      "loss": 0.2331,
      "step": 55160
    },
    {
      "epoch": 0.9960529831935808,
      "grad_norm": 0.4459744095802307,
      "learning_rate": 7.645823152502285e-10,
      "loss": 0.2308,
      "step": 55165
    },
    {
      "epoch": 0.9961432626264815,
      "grad_norm": 0.64663165807724,
      "learning_rate": 7.299123622839243e-10,
      "loss": 0.2971,
      "step": 55170
    },
    {
      "epoch": 0.9962335420593822,
      "grad_norm": 0.4321961998939514,
      "learning_rate": 6.960467778949209e-10,
      "loss": 0.262,
      "step": 55175
    },
    {
      "epoch": 0.9963238214922829,
      "grad_norm": 0.5614007115364075,
      "learning_rate": 6.629855648065952e-10,
      "loss": 0.2206,
      "step": 55180
    },
    {
      "epoch": 0.9964141009251837,
      "grad_norm": 0.3029267489910126,
      "learning_rate": 6.307287256790417e-10,
      "loss": 0.3297,
      "step": 55185
    },
    {
      "epoch": 0.9965043803580843,
      "grad_norm": 0.42381399869918823,
      "learning_rate": 5.992762631068516e-10,
      "loss": 0.2677,
      "step": 55190
    },
    {
      "epoch": 0.996594659790985,
      "grad_norm": 0.864939272403717,
      "learning_rate": 5.686281796202231e-10,
      "loss": 0.225,
      "step": 55195
    },
    {
      "epoch": 0.9966849392238858,
      "grad_norm": 0.5262264013290405,
      "learning_rate": 5.387844776849615e-10,
      "loss": 0.3182,
      "step": 55200
    },
    {
      "epoch": 0.9967752186567865,
      "grad_norm": 0.1654919683933258,
      "learning_rate": 5.09745159700259e-10,
      "loss": 0.1415,
      "step": 55205
    },
    {
      "epoch": 0.9968654980896872,
      "grad_norm": 0.39461642503738403,
      "learning_rate": 4.815102280042449e-10,
      "loss": 0.116,
      "step": 55210
    },
    {
      "epoch": 0.9969557775225879,
      "grad_norm": 0.37544846534729004,
      "learning_rate": 4.540796848673257e-10,
      "loss": 0.1863,
      "step": 55215
    },
    {
      "epoch": 0.9970460569554886,
      "grad_norm": 0.46736714243888855,
      "learning_rate": 4.2745353249551423e-10,
      "loss": 0.2557,
      "step": 55220
    },
    {
      "epoch": 0.9971363363883894,
      "grad_norm": 0.49816399812698364,
      "learning_rate": 4.016317730315411e-10,
      "loss": 0.3284,
      "step": 55225
    },
    {
      "epoch": 0.99722661582129,
      "grad_norm": 0.658227264881134,
      "learning_rate": 3.766144085515233e-10,
      "loss": 0.2312,
      "step": 55230
    },
    {
      "epoch": 0.9973168952541908,
      "grad_norm": 0.62874436378479,
      "learning_rate": 3.5240144106940546e-10,
      "loss": 0.2392,
      "step": 55235
    },
    {
      "epoch": 0.9974071746870915,
      "grad_norm": 0.5400856137275696,
      "learning_rate": 3.2899287253140846e-10,
      "loss": 0.2038,
      "step": 55240
    },
    {
      "epoch": 0.9974974541199922,
      "grad_norm": 0.5790485739707947,
      "learning_rate": 3.063887048215808e-10,
      "loss": 0.2685,
      "step": 55245
    },
    {
      "epoch": 0.9975877335528929,
      "grad_norm": 0.4721094071865082,
      "learning_rate": 2.8458893975735756e-10,
      "loss": 0.1797,
      "step": 55250
    },
    {
      "epoch": 0.9976780129857936,
      "grad_norm": 0.47890564799308777,
      "learning_rate": 2.635935790940014e-10,
      "loss": 0.2093,
      "step": 55255
    },
    {
      "epoch": 0.9977682924186944,
      "grad_norm": 0.7733480334281921,
      "learning_rate": 2.4340262451905126e-10,
      "loss": 0.2791,
      "step": 55260
    },
    {
      "epoch": 0.9978585718515951,
      "grad_norm": 0.4151161015033722,
      "learning_rate": 2.2401607765676346e-10,
      "loss": 0.2732,
      "step": 55265
    },
    {
      "epoch": 0.9979488512844957,
      "grad_norm": 1.0067189931869507,
      "learning_rate": 2.0543394006811156e-10,
      "loss": 0.2369,
      "step": 55270
    },
    {
      "epoch": 0.9980391307173965,
      "grad_norm": 0.37664952874183655,
      "learning_rate": 1.876562132452353e-10,
      "loss": 0.2181,
      "step": 55275
    },
    {
      "epoch": 0.9981294101502972,
      "grad_norm": 0.2857028841972351,
      "learning_rate": 1.706828986203224e-10,
      "loss": 0.1915,
      "step": 55280
    },
    {
      "epoch": 0.998219689583198,
      "grad_norm": 0.4453427493572235,
      "learning_rate": 1.5451399755894713e-10,
      "loss": 0.2189,
      "step": 55285
    },
    {
      "epoch": 0.9983099690160986,
      "grad_norm": 0.3707076609134674,
      "learning_rate": 1.391495113600705e-10,
      "loss": 0.2552,
      "step": 55290
    },
    {
      "epoch": 0.9984002484489993,
      "grad_norm": 0.45158451795578003,
      "learning_rate": 1.2458944126159113e-10,
      "loss": 0.2085,
      "step": 55295
    },
    {
      "epoch": 0.9984905278819001,
      "grad_norm": 0.49200868606567383,
      "learning_rate": 1.1083378843257386e-10,
      "loss": 0.2165,
      "step": 55300
    },
    {
      "epoch": 0.9985808073148008,
      "grad_norm": 0.5954471230506897,
      "learning_rate": 9.788255398102131e-11,
      "loss": 0.2107,
      "step": 55305
    },
    {
      "epoch": 0.9986710867477014,
      "grad_norm": 0.6071249842643738,
      "learning_rate": 8.573573894943288e-11,
      "loss": 0.2585,
      "step": 55310
    },
    {
      "epoch": 0.9987613661806022,
      "grad_norm": 0.17548207938671112,
      "learning_rate": 7.43933443136946e-11,
      "loss": 0.1225,
      "step": 55315
    },
    {
      "epoch": 0.9988516456135029,
      "grad_norm": 0.5218843817710876,
      "learning_rate": 6.385537098529959e-11,
      "loss": 0.1846,
      "step": 55320
    },
    {
      "epoch": 0.9989419250464037,
      "grad_norm": 0.5301579236984253,
      "learning_rate": 5.4121819814678675e-11,
      "loss": 0.2038,
      "step": 55325
    },
    {
      "epoch": 0.9990322044793043,
      "grad_norm": 0.38880106806755066,
      "learning_rate": 4.519269158231865e-11,
      "loss": 0.3044,
      "step": 55330
    },
    {
      "epoch": 0.999122483912205,
      "grad_norm": 0.41869696974754333,
      "learning_rate": 3.7067987007644023e-11,
      "loss": 0.3242,
      "step": 55335
    },
    {
      "epoch": 0.9992127633451058,
      "grad_norm": 0.19082844257354736,
      "learning_rate": 2.974770674457617e-11,
      "loss": 0.1967,
      "step": 55340
    },
    {
      "epoch": 0.9993030427780065,
      "grad_norm": 0.6308541297912598,
      "learning_rate": 2.3231851381533276e-11,
      "loss": 0.2056,
      "step": 55345
    },
    {
      "epoch": 0.9993933222109072,
      "grad_norm": 0.7585052251815796,
      "learning_rate": 1.7520421442540625e-11,
      "loss": 0.2667,
      "step": 55350
    },
    {
      "epoch": 0.9994836016438079,
      "grad_norm": 0.32018110156059265,
      "learning_rate": 1.2613417387230542e-11,
      "loss": 0.1958,
      "step": 55355
    },
    {
      "epoch": 0.9995738810767086,
      "grad_norm": 0.4806329309940338,
      "learning_rate": 8.510839609732201e-12,
      "loss": 0.2796,
      "step": 55360
    },
    {
      "epoch": 0.9996641605096094,
      "grad_norm": 0.6360764503479004,
      "learning_rate": 5.212688440892066e-12,
      "loss": 0.2305,
      "step": 55365
    },
    {
      "epoch": 0.99975443994251,
      "grad_norm": 0.5003007054328918,
      "learning_rate": 2.7189641449432146e-12,
      "loss": 0.2016,
      "step": 55370
    },
    {
      "epoch": 0.9998447193754108,
      "grad_norm": 0.30162450671195984,
      "learning_rate": 1.0296669239462376e-12,
      "loss": 0.2986,
      "step": 55375
    },
    {
      "epoch": 0.9999349988083115,
      "grad_norm": 0.2668619751930237,
      "learning_rate": 1.44796913348344e-13,
      "loss": 0.2414,
      "step": 55380
    },
    {
      "epoch": 0.9999891664680519,
      "step": 55383,
      "total_flos": 6.312971135569623e+19,
      "train_loss": 0.2639624638028189,
      "train_runtime": 248662.3087,
      "train_samples_per_second": 4.455,
      "train_steps_per_second": 0.223
    }
  ],
  "logging_steps": 5,
  "max_steps": 55383,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 15000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 6.312971135569623e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}