{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.0631463871386596,
  "eval_steps": 500,
  "global_step": 8000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 27.876121520996094,
      "learning_rate": 5e-06,
      "loss": 12.5932,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 33.38088607788086,
      "learning_rate": 1e-05,
      "loss": 12.4804,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 47.88199996948242,
      "learning_rate": 1.5e-05,
      "loss": 12.0102,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.78128433227539,
      "learning_rate": 2e-05,
      "loss": 11.3536,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 17.0174617767334,
      "learning_rate": 2.5e-05,
      "loss": 10.612,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.87566375732422,
      "learning_rate": 3e-05,
      "loss": 9.9503,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.851903915405273,
      "learning_rate": 3.5e-05,
      "loss": 9.297,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.8008832931518555,
      "learning_rate": 4e-05,
      "loss": 8.8373,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8600993156433105,
      "learning_rate": 4.5e-05,
      "loss": 8.4592,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8724172115325928,
      "learning_rate": 5e-05,
      "loss": 8.2595,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.987183094024658,
      "learning_rate": 5.500000000000001e-05,
      "loss": 8.147,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7816776037216187,
      "learning_rate": 6e-05,
      "loss": 8.0736,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7206008434295654,
      "learning_rate": 6.500000000000001e-05,
      "loss": 8.0315,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5158601999282837,
      "learning_rate": 7e-05,
      "loss": 8.0424,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.456038475036621,
      "learning_rate": 7.500000000000001e-05,
      "loss": 8.0199,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3331818580627441,
      "learning_rate": 8e-05,
      "loss": 8.0158,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3727846145629883,
      "learning_rate": 8.5e-05,
      "loss": 7.9636,
      "step": 85
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7198413610458374,
      "learning_rate": 9e-05,
      "loss": 7.9121,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0740188360214233,
      "learning_rate": 9.5e-05,
      "loss": 7.8248,
      "step": 95
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.686797857284546,
      "learning_rate": 0.0001,
      "loss": 7.7606,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.506991386413574,
      "learning_rate": 9.999999961506464e-05,
      "loss": 7.7291,
      "step": 105
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3725061416625977,
      "learning_rate": 9.999999846025851e-05,
      "loss": 7.714,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6619765758514404,
      "learning_rate": 9.999999653558166e-05,
      "loss": 7.7032,
      "step": 115
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6985408067703247,
      "learning_rate": 9.999999384103412e-05,
      "loss": 7.6134,
      "step": 120
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5308122634887695,
      "learning_rate": 9.999999037661592e-05,
      "loss": 7.5795,
      "step": 125
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.9915060997009277,
      "learning_rate": 9.999998614232713e-05,
      "loss": 7.5811,
      "step": 130
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5330960750579834,
      "learning_rate": 9.99999811381678e-05,
      "loss": 7.4158,
      "step": 135
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.707305908203125,
      "learning_rate": 9.9999975364138e-05,
      "loss": 7.4482,
      "step": 140
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9988445043563843,
      "learning_rate": 9.999996882023784e-05,
      "loss": 7.3237,
      "step": 145
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7773005962371826,
      "learning_rate": 9.99999615064674e-05,
      "loss": 7.2467,
      "step": 150
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.413421869277954,
      "learning_rate": 9.99999534228268e-05,
      "loss": 7.1401,
      "step": 155
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.0084333419799805,
      "learning_rate": 9.999994456931619e-05,
      "loss": 7.0291,
      "step": 160
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2569665908813477,
      "learning_rate": 9.999993494593566e-05,
      "loss": 6.9673,
      "step": 165
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.333170413970947,
      "learning_rate": 9.999992455268539e-05,
      "loss": 6.9405,
      "step": 170
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.4605367183685303,
      "learning_rate": 9.999991338956554e-05,
      "loss": 6.911,
      "step": 175
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.577528953552246,
      "learning_rate": 9.999990145657626e-05,
      "loss": 6.8194,
      "step": 180
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.965916156768799,
      "learning_rate": 9.999988875371776e-05,
      "loss": 6.7825,
      "step": 185
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.154845714569092,
      "learning_rate": 9.999987528099021e-05,
      "loss": 6.7229,
      "step": 190
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.153703689575195,
      "learning_rate": 9.999986103839384e-05,
      "loss": 6.769,
      "step": 195
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.805217266082764,
      "learning_rate": 9.999984602592884e-05,
      "loss": 6.6112,
      "step": 200
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.591383934020996,
      "learning_rate": 9.999983024359549e-05,
      "loss": 6.6432,
      "step": 205
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.83391809463501,
      "learning_rate": 9.9999813691394e-05,
      "loss": 6.5506,
      "step": 210
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1975998878479004,
      "learning_rate": 9.999979636932461e-05,
      "loss": 6.5025,
      "step": 215
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.858977794647217,
      "learning_rate": 9.999977827738762e-05,
      "loss": 6.5038,
      "step": 220
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.761500835418701,
      "learning_rate": 9.999975941558328e-05,
      "loss": 6.416,
      "step": 225
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8495373725891113,
      "learning_rate": 9.999973978391191e-05,
      "loss": 6.3792,
      "step": 230
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.4267616271972656,
      "learning_rate": 9.99997193823738e-05,
      "loss": 6.303,
      "step": 235
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.342633247375488,
      "learning_rate": 9.999969821096924e-05,
      "loss": 6.2643,
      "step": 240
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0076143741607666,
      "learning_rate": 9.99996762696986e-05,
      "loss": 6.1959,
      "step": 245
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.826104164123535,
      "learning_rate": 9.999965355856218e-05,
      "loss": 6.2774,
      "step": 250
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.76304030418396,
      "learning_rate": 9.999963007756035e-05,
      "loss": 6.2056,
      "step": 255
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2733840942382812,
      "learning_rate": 9.999960582669346e-05,
      "loss": 6.0891,
      "step": 260
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.243439197540283,
      "learning_rate": 9.99995808059619e-05,
      "loss": 6.1236,
      "step": 265
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.158951759338379,
      "learning_rate": 9.999955501536603e-05,
      "loss": 6.1872,
      "step": 270
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.149489879608154,
      "learning_rate": 9.999952845490626e-05,
      "loss": 6.1483,
      "step": 275
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.7918243408203125,
      "learning_rate": 9.999950112458302e-05,
      "loss": 6.0476,
      "step": 280
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.137300968170166,
      "learning_rate": 9.99994730243967e-05,
      "loss": 6.1032,
      "step": 285
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.770310163497925,
      "learning_rate": 9.999944415434774e-05,
      "loss": 5.9817,
      "step": 290
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.315665245056152,
      "learning_rate": 9.99994145144366e-05,
      "loss": 6.0705,
      "step": 295
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.784292697906494,
      "learning_rate": 9.999938410466371e-05,
      "loss": 5.9705,
      "step": 300
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.829956531524658,
      "learning_rate": 9.999935292502958e-05,
      "loss": 5.9342,
      "step": 305
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.353908061981201,
      "learning_rate": 9.999932097553463e-05,
      "loss": 5.8559,
      "step": 310
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0690224170684814,
      "learning_rate": 9.999928825617941e-05,
      "loss": 5.9629,
      "step": 315
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.486865520477295,
      "learning_rate": 9.999925476696439e-05,
      "loss": 5.789,
      "step": 320
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3331587314605713,
      "learning_rate": 9.99992205078901e-05,
      "loss": 5.8865,
      "step": 325
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6461687088012695,
      "learning_rate": 9.999918547895705e-05,
      "loss": 5.9514,
      "step": 330
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.570477485656738,
      "learning_rate": 9.999914968016579e-05,
      "loss": 5.829,
      "step": 335
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.204243183135986,
      "learning_rate": 9.999911311151689e-05,
      "loss": 5.8427,
      "step": 340
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.031857967376709,
      "learning_rate": 9.999907577301089e-05,
      "loss": 5.7698,
      "step": 345
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9448566436767578,
      "learning_rate": 9.999903766464837e-05,
      "loss": 5.788,
      "step": 350
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.893794059753418,
      "learning_rate": 9.99989987864299e-05,
      "loss": 5.7842,
      "step": 355
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.1854119300842285,
      "learning_rate": 9.999895913835613e-05,
      "loss": 5.7082,
      "step": 360
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1127471923828125,
      "learning_rate": 9.999891872042762e-05,
      "loss": 5.7396,
      "step": 365
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5149290561676025,
      "learning_rate": 9.9998877532645e-05,
      "loss": 5.6465,
      "step": 370
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.902657985687256,
      "learning_rate": 9.999883557500892e-05,
      "loss": 5.6004,
      "step": 375
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.77569317817688,
      "learning_rate": 9.999879284752003e-05,
      "loss": 5.6005,
      "step": 380
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.9350929260253906,
      "learning_rate": 9.999874935017897e-05,
      "loss": 5.6191,
      "step": 385
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.020569324493408,
      "learning_rate": 9.999870508298641e-05,
      "loss": 5.6004,
      "step": 390
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.165491580963135,
      "learning_rate": 9.999866004594303e-05,
      "loss": 5.5383,
      "step": 395
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.954424858093262,
      "learning_rate": 9.999861423904954e-05,
      "loss": 5.5615,
      "step": 400
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2333877086639404,
      "learning_rate": 9.999856766230665e-05,
      "loss": 5.6033,
      "step": 405
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6083648204803467,
      "learning_rate": 9.999852031571506e-05,
      "loss": 5.643,
      "step": 410
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6841142177581787,
      "learning_rate": 9.999847219927549e-05,
      "loss": 5.451,
      "step": 415
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.3548808097839355,
      "learning_rate": 9.999842331298872e-05,
      "loss": 5.4857,
      "step": 420
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7186849117279053,
      "learning_rate": 9.999837365685544e-05,
      "loss": 5.374,
      "step": 425
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6831889152526855,
      "learning_rate": 9.999832323087649e-05,
      "loss": 5.4313,
      "step": 430
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3699824810028076,
      "learning_rate": 9.999827203505259e-05,
      "loss": 5.3524,
      "step": 435
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.926292061805725,
      "learning_rate": 9.999822006938455e-05,
      "loss": 5.3183,
      "step": 440
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6304750442504883,
      "learning_rate": 9.999816733387316e-05,
      "loss": 5.5003,
      "step": 445
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.1798224449157715,
      "learning_rate": 9.999811382851924e-05,
      "loss": 5.2356,
      "step": 450
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4884047508239746,
      "learning_rate": 9.999805955332362e-05,
      "loss": 5.3895,
      "step": 455
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.8663625717163086,
      "learning_rate": 9.999800450828711e-05,
      "loss": 5.2157,
      "step": 460
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.258092880249023,
      "learning_rate": 9.99979486934106e-05,
      "loss": 5.2185,
      "step": 465
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.349142551422119,
      "learning_rate": 9.99978921086949e-05,
      "loss": 5.1709,
      "step": 470
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.071136951446533,
      "learning_rate": 9.999783475414093e-05,
      "loss": 5.2357,
      "step": 475
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.015169143676758,
      "learning_rate": 9.999777662974952e-05,
      "loss": 5.2275,
      "step": 480
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8933837413787842,
      "learning_rate": 9.999771773552162e-05,
      "loss": 5.0911,
      "step": 485
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.3265419006347656,
      "learning_rate": 9.999765807145811e-05,
      "loss": 5.0683,
      "step": 490
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.7656900882720947,
      "learning_rate": 9.99975976375599e-05,
      "loss": 5.1039,
      "step": 495
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.8807759284973145,
      "learning_rate": 9.999753643382793e-05,
      "loss": 5.0649,
      "step": 500
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.449495553970337,
      "learning_rate": 9.999747446026315e-05,
      "loss": 5.1116,
      "step": 505
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.55258846282959,
      "learning_rate": 9.999741171686649e-05,
      "loss": 5.0347,
      "step": 510
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9589213132858276,
      "learning_rate": 9.999734820363895e-05,
      "loss": 5.0986,
      "step": 515
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.284320592880249,
      "learning_rate": 9.99972839205815e-05,
      "loss": 5.1101,
      "step": 520
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.100949764251709,
      "learning_rate": 9.999721886769511e-05,
      "loss": 4.9835,
      "step": 525
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.074458122253418,
      "learning_rate": 9.999715304498078e-05,
      "loss": 4.9139,
      "step": 530
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7766720056533813,
      "learning_rate": 9.999708645243955e-05,
      "loss": 4.9851,
      "step": 535
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.9218993186950684,
      "learning_rate": 9.999701909007242e-05,
      "loss": 4.9359,
      "step": 540
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.10783052444458,
      "learning_rate": 9.999695095788044e-05,
      "loss": 4.8765,
      "step": 545
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8996570110321045,
      "learning_rate": 9.999688205586468e-05,
      "loss": 4.9302,
      "step": 550
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2307090759277344,
      "learning_rate": 9.999681238402615e-05,
      "loss": 4.8296,
      "step": 555
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.489236354827881,
      "learning_rate": 9.999674194236597e-05,
      "loss": 4.8288,
      "step": 560
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.654552459716797,
      "learning_rate": 9.999667073088519e-05,
      "loss": 4.8523,
      "step": 565
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.793175220489502,
      "learning_rate": 9.999659874958493e-05,
      "loss": 4.8327,
      "step": 570
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8369104862213135,
      "learning_rate": 9.99965259984663e-05,
      "loss": 4.8769,
      "step": 575
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4289119243621826,
      "learning_rate": 9.999645247753039e-05,
      "loss": 4.7697,
      "step": 580
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.508596897125244,
      "learning_rate": 9.999637818677836e-05,
      "loss": 4.7625,
      "step": 585
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.263563394546509,
      "learning_rate": 9.999630312621136e-05,
      "loss": 4.7774,
      "step": 590
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.725827932357788,
      "learning_rate": 9.999622729583051e-05,
      "loss": 4.4813,
      "step": 595
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0028059482574463,
      "learning_rate": 9.999615069563701e-05,
      "loss": 4.8321,
      "step": 600
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9981409311294556,
      "learning_rate": 9.999607332563205e-05,
      "loss": 4.8613,
      "step": 605
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.346536159515381,
      "learning_rate": 9.999599518581676e-05,
      "loss": 4.7415,
      "step": 610
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6441526412963867,
      "learning_rate": 9.999591627619242e-05,
      "loss": 4.7101,
      "step": 615
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.681407928466797,
      "learning_rate": 9.999583659676019e-05,
      "loss": 4.6543,
      "step": 620
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.81906795501709,
      "learning_rate": 9.999575614752133e-05,
      "loss": 4.788,
      "step": 625
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1597816944122314,
      "learning_rate": 9.999567492847706e-05,
      "loss": 4.837,
      "step": 630
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.409820795059204,
      "learning_rate": 9.999559293962864e-05,
      "loss": 4.6613,
      "step": 635
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2003366947174072,
      "learning_rate": 9.99955101809773e-05,
      "loss": 4.6055,
      "step": 640
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.659848690032959,
      "learning_rate": 9.999542665252437e-05,
      "loss": 4.503,
      "step": 645
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.420382022857666,
      "learning_rate": 9.999534235427109e-05,
      "loss": 4.6862,
      "step": 650
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.433791399002075,
      "learning_rate": 9.999525728621879e-05,
      "loss": 4.6671,
      "step": 655
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3587777614593506,
      "learning_rate": 9.999517144836876e-05,
      "loss": 4.6807,
      "step": 660
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8520917892456055,
      "learning_rate": 9.999508484072233e-05,
      "loss": 4.6877,
      "step": 665
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2715208530426025,
      "learning_rate": 9.999499746328083e-05,
      "loss": 4.5028,
      "step": 670
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8923941850662231,
      "learning_rate": 9.99949093160456e-05,
      "loss": 4.4864,
      "step": 675
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.125587224960327,
      "learning_rate": 9.9994820399018e-05,
      "loss": 4.5253,
      "step": 680
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.377746105194092,
      "learning_rate": 9.999473071219941e-05,
      "loss": 4.6415,
      "step": 685
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8352751731872559,
      "learning_rate": 9.999464025559121e-05,
      "loss": 4.513,
      "step": 690
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.669769287109375,
      "learning_rate": 9.999454902919478e-05,
      "loss": 4.5463,
      "step": 695
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6786859035491943,
      "learning_rate": 9.999445703301154e-05,
      "loss": 4.5925,
      "step": 700
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7276148796081543,
      "learning_rate": 9.999436426704287e-05,
      "loss": 4.6006,
      "step": 705
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2088770866394043,
      "learning_rate": 9.999427073129026e-05,
      "loss": 4.5148,
      "step": 710
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.261117696762085,
      "learning_rate": 9.999417642575509e-05,
      "loss": 4.6388,
      "step": 715
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8530731201171875,
      "learning_rate": 9.999408135043886e-05,
      "loss": 4.6318,
      "step": 720
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7427523136138916,
      "learning_rate": 9.999398550534298e-05,
      "loss": 4.3433,
      "step": 725
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0432887077331543,
      "learning_rate": 9.999388889046899e-05,
      "loss": 4.4283,
      "step": 730
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.773913621902466,
      "learning_rate": 9.999379150581833e-05,
      "loss": 4.5184,
      "step": 735
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.154879331588745,
      "learning_rate": 9.999369335139251e-05,
      "loss": 4.5572,
      "step": 740
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.524599075317383,
      "learning_rate": 9.999359442719306e-05,
      "loss": 4.4268,
      "step": 745
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4532240629196167,
      "learning_rate": 9.999349473322149e-05,
      "loss": 4.4529,
      "step": 750
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3587846755981445,
      "learning_rate": 9.999339426947932e-05,
      "loss": 4.5208,
      "step": 755
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4776296615600586,
      "learning_rate": 9.999329303596812e-05,
      "loss": 4.4477,
      "step": 760
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9740087985992432,
      "learning_rate": 9.999319103268945e-05,
      "loss": 4.5592,
      "step": 765
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2760045528411865,
      "learning_rate": 9.999308825964485e-05,
      "loss": 4.3836,
      "step": 770
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4244019985198975,
      "learning_rate": 9.999298471683593e-05,
      "loss": 4.4567,
      "step": 775
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0621178150177,
      "learning_rate": 9.999288040426429e-05,
      "loss": 4.3633,
      "step": 780
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8395153284072876,
      "learning_rate": 9.999277532193151e-05,
      "loss": 4.3881,
      "step": 785
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9404289722442627,
      "learning_rate": 9.999266946983923e-05,
      "loss": 4.247,
      "step": 790
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8988182544708252,
      "learning_rate": 9.999256284798906e-05,
      "loss": 4.4471,
      "step": 795
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1905274391174316,
      "learning_rate": 9.999245545638267e-05,
      "loss": 4.2996,
      "step": 800
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1765809059143066,
      "learning_rate": 9.999234729502167e-05,
      "loss": 4.2839,
      "step": 805
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.41542649269104,
      "learning_rate": 9.999223836390777e-05,
      "loss": 4.2675,
      "step": 810
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6735470294952393,
      "learning_rate": 9.999212866304262e-05,
      "loss": 4.4014,
      "step": 815
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3618645668029785,
      "learning_rate": 9.999201819242794e-05,
      "loss": 4.4287,
      "step": 820
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.147860527038574,
      "learning_rate": 9.99919069520654e-05,
      "loss": 4.4596,
      "step": 825
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.711953043937683,
      "learning_rate": 9.99917949419567e-05,
      "loss": 4.2474,
      "step": 830
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.41106915473938,
      "learning_rate": 9.999168216210362e-05,
      "loss": 4.2805,
      "step": 835
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6366605758666992,
      "learning_rate": 9.999156861250784e-05,
      "loss": 4.3756,
      "step": 840
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6821074485778809,
      "learning_rate": 9.999145429317114e-05,
      "loss": 4.3938,
      "step": 845
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.797267198562622,
      "learning_rate": 9.999133920409527e-05,
      "loss": 4.3444,
      "step": 850
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.291905164718628,
      "learning_rate": 9.999122334528199e-05,
      "loss": 4.3399,
      "step": 855
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9095951318740845,
      "learning_rate": 9.999110671673313e-05,
      "loss": 4.3625,
      "step": 860
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9491018056869507,
      "learning_rate": 9.999098931845044e-05,
      "loss": 4.3306,
      "step": 865
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9286075830459595,
      "learning_rate": 9.999087115043573e-05,
      "loss": 4.2583,
      "step": 870
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.225546360015869,
      "learning_rate": 9.999075221269083e-05,
      "loss": 4.2437,
      "step": 875
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3416944742202759,
      "learning_rate": 9.999063250521759e-05,
      "loss": 4.215,
      "step": 880
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9249601364135742,
      "learning_rate": 9.999051202801781e-05,
      "loss": 4.3215,
      "step": 885
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9533398151397705,
      "learning_rate": 9.999039078109341e-05,
      "loss": 4.3574,
      "step": 890
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7104941606521606,
      "learning_rate": 9.999026876444617e-05,
      "loss": 4.2446,
      "step": 895
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7254068851470947,
      "learning_rate": 9.999014597807805e-05,
      "loss": 4.1337,
      "step": 900
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9523411989212036,
      "learning_rate": 9.99900224219909e-05,
      "loss": 4.1699,
      "step": 905
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1077957153320312,
      "learning_rate": 9.998989809618662e-05,
      "loss": 4.1827,
      "step": 910
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8629860877990723,
      "learning_rate": 9.998977300066715e-05,
      "loss": 4.2235,
      "step": 915
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0885443687438965,
      "learning_rate": 9.998964713543437e-05,
      "loss": 4.0693,
      "step": 920
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.773579478263855,
      "learning_rate": 9.998952050049026e-05,
      "loss": 4.1769,
      "step": 925
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0322937965393066,
      "learning_rate": 9.998939309583677e-05,
      "loss": 4.3255,
      "step": 930
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.670674204826355,
      "learning_rate": 9.998926492147585e-05,
      "loss": 4.3377,
      "step": 935
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9845823049545288,
      "learning_rate": 9.998913597740945e-05,
      "loss": 4.1305,
      "step": 940
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7269636392593384,
      "learning_rate": 9.99890062636396e-05,
      "loss": 4.2384,
      "step": 945
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.715166449546814,
      "learning_rate": 9.998887578016827e-05,
      "loss": 3.9799,
      "step": 950
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1728355884552,
      "learning_rate": 9.998874452699747e-05,
      "loss": 4.1521,
      "step": 955
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6808676719665527,
      "learning_rate": 9.998861250412922e-05,
      "loss": 4.1888,
      "step": 960
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6475008726119995,
      "learning_rate": 9.998847971156556e-05,
      "loss": 4.1859,
      "step": 965
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5650328397750854,
      "learning_rate": 9.998834614930855e-05,
      "loss": 4.177,
      "step": 970
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.528485894203186,
      "learning_rate": 9.99882118173602e-05,
      "loss": 4.2595,
      "step": 975
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.415407657623291,
      "learning_rate": 9.998807671572263e-05,
      "loss": 4.2375,
      "step": 980
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4295876026153564,
      "learning_rate": 9.998794084439789e-05,
      "loss": 4.1299,
      "step": 985
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.374717116355896,
      "learning_rate": 9.998780420338809e-05,
      "loss": 4.1049,
      "step": 990
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5122811794281006,
      "learning_rate": 9.998766679269531e-05,
      "loss": 4.1367,
      "step": 995
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.448702096939087,
      "learning_rate": 9.998752861232168e-05,
      "loss": 4.1369,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.92279052734375,
      "learning_rate": 9.998738966226933e-05,
      "loss": 4.2653,
      "step": 1005
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5852283239364624,
      "learning_rate": 9.998724994254041e-05,
      "loss": 4.1013,
      "step": 1010
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.022915840148926,
      "learning_rate": 9.998710945313705e-05,
      "loss": 4.199,
      "step": 1015
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.648411273956299,
      "learning_rate": 9.998696819406141e-05,
      "loss": 4.0925,
      "step": 1020
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6134710311889648,
      "learning_rate": 9.99868261653157e-05,
      "loss": 4.1539,
      "step": 1025
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8423235416412354,
      "learning_rate": 9.998668336690206e-05,
      "loss": 4.0834,
      "step": 1030
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5607178211212158,
      "learning_rate": 9.998653979882271e-05,
      "loss": 4.1667,
      "step": 1035
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.507724404335022,
      "learning_rate": 9.998639546107989e-05,
      "loss": 4.1183,
      "step": 1040
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6256999969482422,
      "learning_rate": 9.998625035367577e-05,
      "loss": 4.0105,
      "step": 1045
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4240740537643433,
      "learning_rate": 9.998610447661261e-05,
      "loss": 4.0721,
      "step": 1050
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3977088928222656,
      "learning_rate": 9.998595782989266e-05,
      "loss": 4.1215,
      "step": 1055
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6170176267623901,
      "learning_rate": 9.998581041351818e-05,
      "loss": 4.0369,
      "step": 1060
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.095721483230591,
      "learning_rate": 9.998566222749142e-05,
      "loss": 4.0252,
      "step": 1065
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8155313730239868,
      "learning_rate": 9.998551327181468e-05,
      "loss": 4.0868,
      "step": 1070
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6068108081817627,
      "learning_rate": 9.998536354649026e-05,
      "loss": 4.1387,
      "step": 1075
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5400094985961914,
      "learning_rate": 9.998521305152045e-05,
      "loss": 4.0084,
      "step": 1080
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4838355779647827,
      "learning_rate": 9.998506178690755e-05,
      "loss": 4.0414,
      "step": 1085
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2195415496826172,
      "learning_rate": 9.998490975265393e-05,
      "loss": 4.084,
      "step": 1090
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6410263776779175,
      "learning_rate": 9.99847569487619e-05,
      "loss": 3.9677,
      "step": 1095
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6714084148406982,
      "learning_rate": 9.998460337523383e-05,
      "loss": 4.1894,
      "step": 1100
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.739774227142334,
      "learning_rate": 9.998444903207207e-05,
      "loss": 4.1052,
      "step": 1105
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.569242238998413,
      "learning_rate": 9.998429391927902e-05,
      "loss": 4.0954,
      "step": 1110
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4891986846923828,
      "learning_rate": 9.998413803685705e-05,
      "loss": 4.0774,
      "step": 1115
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4582473039627075,
      "learning_rate": 9.998398138480855e-05,
      "loss": 4.0041,
      "step": 1120
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3005480766296387,
      "learning_rate": 9.998382396313597e-05,
      "loss": 4.0193,
      "step": 1125
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.008749008178711,
      "learning_rate": 9.998366577184168e-05,
      "loss": 3.9152,
      "step": 1130
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.259828805923462,
      "learning_rate": 9.998350681092817e-05,
      "loss": 3.8865,
      "step": 1135
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4959806203842163,
      "learning_rate": 9.998334708039785e-05,
      "loss": 3.9924,
      "step": 1140
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6365115642547607,
      "learning_rate": 9.998318658025318e-05,
      "loss": 4.0255,
      "step": 1145
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6874918937683105,
      "learning_rate": 9.998302531049667e-05,
      "loss": 4.0276,
      "step": 1150
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.459473729133606,
      "learning_rate": 9.998286327113076e-05,
      "loss": 4.0493,
      "step": 1155
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.478223204612732,
      "learning_rate": 9.998270046215797e-05,
      "loss": 3.9731,
      "step": 1160
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4293180704116821,
      "learning_rate": 9.998253688358078e-05,
      "loss": 3.8683,
      "step": 1165
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.69686758518219,
      "learning_rate": 9.998237253540174e-05,
      "loss": 4.087,
      "step": 1170
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5213356018066406,
      "learning_rate": 9.998220741762338e-05,
      "loss": 4.0013,
      "step": 1175
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7112723588943481,
      "learning_rate": 9.998204153024821e-05,
      "loss": 3.9447,
      "step": 1180
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2025816440582275,
      "learning_rate": 9.998187487327882e-05,
      "loss": 3.7893,
      "step": 1185
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3971811532974243,
      "learning_rate": 9.998170744671775e-05,
      "loss": 3.9351,
      "step": 1190
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2900769710540771,
      "learning_rate": 9.998153925056758e-05,
      "loss": 3.9523,
      "step": 1195
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1375789642333984,
      "learning_rate": 9.998137028483092e-05,
      "loss": 4.0112,
      "step": 1200
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.337507963180542,
      "learning_rate": 9.998120054951035e-05,
      "loss": 3.9898,
      "step": 1205
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6715366840362549,
      "learning_rate": 9.99810300446085e-05,
      "loss": 4.0032,
      "step": 1210
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8074537515640259,
      "learning_rate": 9.9980858770128e-05,
      "loss": 3.855,
      "step": 1215
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.393588662147522,
      "learning_rate": 9.998068672607146e-05,
      "loss": 3.9624,
      "step": 1220
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4749821424484253,
      "learning_rate": 9.998051391244155e-05,
      "loss": 4.03,
      "step": 1225
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6638877391815186,
      "learning_rate": 9.998034032924092e-05,
      "loss": 4.035,
      "step": 1230
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3646574020385742,
      "learning_rate": 9.998016597647225e-05,
      "loss": 3.9641,
      "step": 1235
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6863949298858643,
      "learning_rate": 9.997999085413823e-05,
      "loss": 3.9216,
      "step": 1240
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.396281361579895,
      "learning_rate": 9.997981496224153e-05,
      "loss": 3.9631,
      "step": 1245
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2718080282211304,
      "learning_rate": 9.99796383007849e-05,
      "loss": 3.8685,
      "step": 1250
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4302754402160645,
      "learning_rate": 9.997946086977103e-05,
      "loss": 4.0005,
      "step": 1255
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.361426591873169,
      "learning_rate": 9.997928266920266e-05,
      "loss": 3.8474,
      "step": 1260
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2314462661743164,
      "learning_rate": 9.997910369908254e-05,
      "loss": 3.9611,
      "step": 1265
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.48407781124115,
      "learning_rate": 9.99789239594134e-05,
      "loss": 3.8192,
      "step": 1270
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4937589168548584,
      "learning_rate": 9.997874345019805e-05,
      "loss": 3.9464,
      "step": 1275
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2191518545150757,
      "learning_rate": 9.997856217143921e-05,
      "loss": 3.8344,
      "step": 1280
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.455945372581482,
      "learning_rate": 9.997838012313974e-05,
      "loss": 3.8479,
      "step": 1285
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4489750862121582,
      "learning_rate": 9.99781973053024e-05,
      "loss": 3.8951,
      "step": 1290
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6571656465530396,
      "learning_rate": 9.997801371793001e-05,
      "loss": 3.8528,
      "step": 1295
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.286756992340088,
      "learning_rate": 9.997782936102542e-05,
      "loss": 3.9123,
      "step": 1300
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3976352214813232,
      "learning_rate": 9.997764423459144e-05,
      "loss": 3.8062,
      "step": 1305
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2056770324707031,
      "learning_rate": 9.997745833863092e-05,
      "loss": 4.0636,
      "step": 1310
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3087226152420044,
      "learning_rate": 9.997727167314676e-05,
      "loss": 3.9636,
      "step": 1315
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3142720460891724,
      "learning_rate": 9.997708423814178e-05,
      "loss": 3.8453,
      "step": 1320
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5223431587219238,
      "learning_rate": 9.99768960336189e-05,
      "loss": 3.8582,
      "step": 1325
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4130308628082275,
      "learning_rate": 9.997670705958102e-05,
      "loss": 3.8497,
      "step": 1330
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3284647464752197,
      "learning_rate": 9.997651731603104e-05,
      "loss": 3.8006,
      "step": 1335
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2974432706832886,
      "learning_rate": 9.997632680297188e-05,
      "loss": 3.7314,
      "step": 1340
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2121127843856812,
      "learning_rate": 9.997613552040648e-05,
      "loss": 3.8743,
      "step": 1345
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.342211365699768,
      "learning_rate": 9.997594346833779e-05,
      "loss": 3.8356,
      "step": 1350
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2535418272018433,
      "learning_rate": 9.997575064676875e-05,
      "loss": 3.8479,
      "step": 1355
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2163220643997192,
      "learning_rate": 9.997555705570232e-05,
      "loss": 3.7908,
      "step": 1360
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2057090997695923,
      "learning_rate": 9.997536269514152e-05,
      "loss": 3.7489,
      "step": 1365
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1692609786987305,
      "learning_rate": 9.99751675650893e-05,
      "loss": 3.8809,
      "step": 1370
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2686443328857422,
      "learning_rate": 9.99749716655487e-05,
      "loss": 3.8923,
      "step": 1375
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5202538967132568,
      "learning_rate": 9.997477499652273e-05,
      "loss": 3.895,
      "step": 1380
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1383185386657715,
      "learning_rate": 9.99745775580144e-05,
      "loss": 4.0239,
      "step": 1385
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3303546905517578,
      "learning_rate": 9.997437935002676e-05,
      "loss": 3.8153,
      "step": 1390
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4921525716781616,
      "learning_rate": 9.997418037256285e-05,
      "loss": 3.9478,
      "step": 1395
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.183221459388733,
      "learning_rate": 9.997398062562575e-05,
      "loss": 3.7559,
      "step": 1400
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3215347528457642,
      "learning_rate": 9.997378010921856e-05,
      "loss": 3.8197,
      "step": 1405
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3516408205032349,
      "learning_rate": 9.99735788233443e-05,
      "loss": 3.8299,
      "step": 1410
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3312804698944092,
      "learning_rate": 9.997337676800613e-05,
      "loss": 3.9281,
      "step": 1415
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1919550895690918,
      "learning_rate": 9.997317394320713e-05,
      "loss": 3.8334,
      "step": 1420
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.542568564414978,
      "learning_rate": 9.997297034895044e-05,
      "loss": 3.8606,
      "step": 1425
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3972443342208862,
      "learning_rate": 9.99727659852392e-05,
      "loss": 3.8723,
      "step": 1430
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4193700551986694,
      "learning_rate": 9.997256085207652e-05,
      "loss": 3.7605,
      "step": 1435
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1687391996383667,
      "learning_rate": 9.997235494946557e-05,
      "loss": 3.8882,
      "step": 1440
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0899901390075684,
      "learning_rate": 9.997214827740955e-05,
      "loss": 3.8342,
      "step": 1445
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3528944253921509,
      "learning_rate": 9.997194083591164e-05,
      "loss": 3.8773,
      "step": 1450
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3315227031707764,
      "learning_rate": 9.9971732624975e-05,
      "loss": 3.7766,
      "step": 1455
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0130810737609863,
      "learning_rate": 9.997152364460287e-05,
      "loss": 3.664,
      "step": 1460
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5584934949874878,
      "learning_rate": 9.997131389479844e-05,
      "loss": 3.8546,
      "step": 1465
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2053862810134888,
      "learning_rate": 9.997110337556494e-05,
      "loss": 3.7302,
      "step": 1470
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1595443487167358,
      "learning_rate": 9.997089208690564e-05,
      "loss": 3.8455,
      "step": 1475
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3403232097625732,
      "learning_rate": 9.997068002882377e-05,
      "loss": 3.8297,
      "step": 1480
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3555865287780762,
      "learning_rate": 9.997046720132261e-05,
      "loss": 3.86,
      "step": 1485
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1468091011047363,
      "learning_rate": 9.997025360440541e-05,
      "loss": 3.7792,
      "step": 1490
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3253270387649536,
      "learning_rate": 9.997003923807551e-05,
      "loss": 3.7879,
      "step": 1495
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2670639753341675,
      "learning_rate": 9.996982410233615e-05,
      "loss": 3.7822,
      "step": 1500
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9859651327133179,
      "learning_rate": 9.996960819719068e-05,
      "loss": 3.7415,
      "step": 1505
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0997586250305176,
      "learning_rate": 9.996939152264242e-05,
      "loss": 3.8024,
      "step": 1510
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1661698818206787,
      "learning_rate": 9.99691740786947e-05,
      "loss": 3.6827,
      "step": 1515
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2775877714157104,
      "learning_rate": 9.996895586535086e-05,
      "loss": 3.7629,
      "step": 1520
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0463752746582031,
      "learning_rate": 9.996873688261428e-05,
      "loss": 3.7837,
      "step": 1525
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2637611627578735,
      "learning_rate": 9.996851713048831e-05,
      "loss": 3.8752,
      "step": 1530
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2414253950119019,
      "learning_rate": 9.996829660897637e-05,
      "loss": 3.7332,
      "step": 1535
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3822253942489624,
      "learning_rate": 9.996807531808181e-05,
      "loss": 3.7639,
      "step": 1540
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0833503007888794,
      "learning_rate": 9.996785325780806e-05,
      "loss": 3.8395,
      "step": 1545
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.148206353187561,
      "learning_rate": 9.996763042815853e-05,
      "loss": 3.7885,
      "step": 1550
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1983190774917603,
      "learning_rate": 9.996740682913668e-05,
      "loss": 3.7172,
      "step": 1555
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1160719394683838,
      "learning_rate": 9.996718246074591e-05,
      "loss": 3.796,
      "step": 1560
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1344895362854004,
      "learning_rate": 9.99669573229897e-05,
      "loss": 3.6759,
      "step": 1565
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1262212991714478,
      "learning_rate": 9.996673141587151e-05,
      "loss": 3.7567,
      "step": 1570
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5000883340835571,
      "learning_rate": 9.996650473939482e-05,
      "loss": 3.7238,
      "step": 1575
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.368445873260498,
      "learning_rate": 9.996627729356314e-05,
      "loss": 3.6976,
      "step": 1580
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2350101470947266,
      "learning_rate": 9.996604907837993e-05,
      "loss": 3.7799,
      "step": 1585
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5796904563903809,
      "learning_rate": 9.996582009384874e-05,
      "loss": 3.6844,
      "step": 1590
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5272090435028076,
      "learning_rate": 9.996559033997307e-05,
      "loss": 3.6553,
      "step": 1595
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1256613731384277,
      "learning_rate": 9.996535981675648e-05,
      "loss": 3.5805,
      "step": 1600
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1103079319000244,
      "learning_rate": 9.996512852420251e-05,
      "loss": 3.6393,
      "step": 1605
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0625386238098145,
      "learning_rate": 9.996489646231471e-05,
      "loss": 3.6839,
      "step": 1610
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0707393884658813,
      "learning_rate": 9.996466363109667e-05,
      "loss": 3.5516,
      "step": 1615
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0898019075393677,
      "learning_rate": 9.996443003055196e-05,
      "loss": 3.8269,
      "step": 1620
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2631019353866577,
      "learning_rate": 9.99641956606842e-05,
      "loss": 3.7921,
      "step": 1625
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.237564206123352,
      "learning_rate": 9.996396052149698e-05,
      "loss": 3.763,
      "step": 1630
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9420194029808044,
      "learning_rate": 9.996372461299392e-05,
      "loss": 3.733,
      "step": 1635
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0645323991775513,
      "learning_rate": 9.996348793517866e-05,
      "loss": 3.8324,
      "step": 1640
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.016371726989746,
      "learning_rate": 9.996325048805484e-05,
      "loss": 3.652,
      "step": 1645
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0370053052902222,
      "learning_rate": 9.996301227162612e-05,
      "loss": 3.6233,
      "step": 1650
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0382436513900757,
      "learning_rate": 9.996277328589618e-05,
      "loss": 3.725,
      "step": 1655
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6462538242340088,
      "learning_rate": 9.996253353086866e-05,
      "loss": 3.6682,
      "step": 1660
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2156929969787598,
      "learning_rate": 9.996229300654729e-05,
      "loss": 3.7289,
      "step": 1665
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0971386432647705,
      "learning_rate": 9.996205171293576e-05,
      "loss": 3.6208,
      "step": 1670
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1641781330108643,
      "learning_rate": 9.996180965003778e-05,
      "loss": 3.6167,
      "step": 1675
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.241957426071167,
      "learning_rate": 9.996156681785708e-05,
      "loss": 3.6958,
      "step": 1680
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1770355701446533,
      "learning_rate": 9.996132321639741e-05,
      "loss": 3.6898,
      "step": 1685
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.122515082359314,
      "learning_rate": 9.996107884566252e-05,
      "loss": 3.7914,
      "step": 1690
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.08939528465271,
      "learning_rate": 9.996083370565616e-05,
      "loss": 3.7322,
      "step": 1695
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9389724731445312,
      "learning_rate": 9.99605877963821e-05,
      "loss": 3.5773,
      "step": 1700
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9614108204841614,
      "learning_rate": 9.996034111784413e-05,
      "loss": 3.7456,
      "step": 1705
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0695230960845947,
      "learning_rate": 9.996009367004607e-05,
      "loss": 3.7395,
      "step": 1710
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5205737352371216,
      "learning_rate": 9.995984545299171e-05,
      "loss": 3.5413,
      "step": 1715
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9975587725639343,
      "learning_rate": 9.995959646668488e-05,
      "loss": 3.6818,
      "step": 1720
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1887078285217285,
      "learning_rate": 9.995934671112941e-05,
      "loss": 3.6174,
      "step": 1725
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0418992042541504,
      "learning_rate": 9.995909618632915e-05,
      "loss": 3.5976,
      "step": 1730
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2081197500228882,
      "learning_rate": 9.995884489228794e-05,
      "loss": 3.6395,
      "step": 1735
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.166056752204895,
      "learning_rate": 9.995859282900967e-05,
      "loss": 3.6516,
      "step": 1740
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1711153984069824,
      "learning_rate": 9.995833999649821e-05,
      "loss": 3.6587,
      "step": 1745
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0777111053466797,
      "learning_rate": 9.995808639475746e-05,
      "loss": 3.5286,
      "step": 1750
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2093099355697632,
      "learning_rate": 9.995783202379131e-05,
      "loss": 3.7071,
      "step": 1755
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8598945140838623,
      "learning_rate": 9.99575768836037e-05,
      "loss": 3.619,
      "step": 1760
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1166030168533325,
      "learning_rate": 9.995732097419855e-05,
      "loss": 3.6801,
      "step": 1765
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9624553322792053,
      "learning_rate": 9.995706429557978e-05,
      "loss": 3.6588,
      "step": 1770
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9502954483032227,
      "learning_rate": 9.995680684775137e-05,
      "loss": 3.544,
      "step": 1775
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9735146164894104,
      "learning_rate": 9.995654863071729e-05,
      "loss": 3.7236,
      "step": 1780
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0256701707839966,
      "learning_rate": 9.995628964448147e-05,
      "loss": 3.6176,
      "step": 1785
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9160292744636536,
      "learning_rate": 9.995602988904793e-05,
      "loss": 3.6033,
      "step": 1790
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8587386608123779,
      "learning_rate": 9.995576936442068e-05,
      "loss": 3.6353,
      "step": 1795
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0511562824249268,
      "learning_rate": 9.995550807060371e-05,
      "loss": 3.6715,
      "step": 1800
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.273176908493042,
      "learning_rate": 9.995524600760105e-05,
      "loss": 3.5689,
      "step": 1805
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1116704940795898,
      "learning_rate": 9.995498317541674e-05,
      "loss": 3.6907,
      "step": 1810
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9578545689582825,
      "learning_rate": 9.995471957405483e-05,
      "loss": 3.5862,
      "step": 1815
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.901204526424408,
      "learning_rate": 9.995445520351935e-05,
      "loss": 3.571,
      "step": 1820
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.986279308795929,
      "learning_rate": 9.995419006381442e-05,
      "loss": 3.5482,
      "step": 1825
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3870453834533691,
      "learning_rate": 9.995392415494407e-05,
      "loss": 3.6398,
      "step": 1830
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1633667945861816,
      "learning_rate": 9.995365747691243e-05,
      "loss": 3.4995,
      "step": 1835
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.051268458366394,
      "learning_rate": 9.995339002972359e-05,
      "loss": 3.5606,
      "step": 1840
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.099851369857788,
      "learning_rate": 9.995312181338168e-05,
      "loss": 3.6801,
      "step": 1845
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1902729272842407,
      "learning_rate": 9.995285282789082e-05,
      "loss": 3.6235,
      "step": 1850
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9518100023269653,
      "learning_rate": 9.995258307325515e-05,
      "loss": 3.6267,
      "step": 1855
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0606943368911743,
      "learning_rate": 9.995231254947883e-05,
      "loss": 3.5607,
      "step": 1860
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0986220836639404,
      "learning_rate": 9.995204125656603e-05,
      "loss": 3.6654,
      "step": 1865
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.053001880645752,
      "learning_rate": 9.995176919452092e-05,
      "loss": 3.5674,
      "step": 1870
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8718825578689575,
      "learning_rate": 9.995149636334769e-05,
      "loss": 3.4152,
      "step": 1875
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1395750045776367,
      "learning_rate": 9.995122276305054e-05,
      "loss": 3.5912,
      "step": 1880
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8094372749328613,
      "learning_rate": 9.995094839363368e-05,
      "loss": 3.485,
      "step": 1885
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9941425919532776,
      "learning_rate": 9.995067325510132e-05,
      "loss": 3.5732,
      "step": 1890
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9548867344856262,
      "learning_rate": 9.995039734745774e-05,
      "loss": 3.5996,
      "step": 1895
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9619913697242737,
      "learning_rate": 9.995012067070715e-05,
      "loss": 3.607,
      "step": 1900
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9063735604286194,
      "learning_rate": 9.994984322485383e-05,
      "loss": 3.5471,
      "step": 1905
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0066523551940918,
      "learning_rate": 9.994956500990205e-05,
      "loss": 3.6157,
      "step": 1910
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.185215950012207,
      "learning_rate": 9.994928602585608e-05,
      "loss": 3.4414,
      "step": 1915
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0985796451568604,
      "learning_rate": 9.994900627272023e-05,
      "loss": 3.4922,
      "step": 1920
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9418334364891052,
      "learning_rate": 9.99487257504988e-05,
      "loss": 3.5603,
      "step": 1925
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1406140327453613,
      "learning_rate": 9.994844445919612e-05,
      "loss": 3.5077,
      "step": 1930
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2536481618881226,
      "learning_rate": 9.994816239881648e-05,
      "loss": 3.6416,
      "step": 1935
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0355080366134644,
      "learning_rate": 9.99478795693643e-05,
      "loss": 3.5253,
      "step": 1940
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8323686718940735,
      "learning_rate": 9.994759597084384e-05,
      "loss": 3.7135,
      "step": 1945
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9537489414215088,
      "learning_rate": 9.994731160325955e-05,
      "loss": 3.5499,
      "step": 1950
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0943478345870972,
      "learning_rate": 9.994702646661577e-05,
      "loss": 3.5695,
      "step": 1955
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1690897941589355,
      "learning_rate": 9.994674056091689e-05,
      "loss": 3.5245,
      "step": 1960
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9295859932899475,
      "learning_rate": 9.994645388616732e-05,
      "loss": 3.5914,
      "step": 1965
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1724251508712769,
      "learning_rate": 9.994616644237148e-05,
      "loss": 3.4758,
      "step": 1970
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1768447160720825,
      "learning_rate": 9.994587822953378e-05,
      "loss": 3.6727,
      "step": 1975
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1149611473083496,
      "learning_rate": 9.994558924765866e-05,
      "loss": 3.5212,
      "step": 1980
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.206658124923706,
      "learning_rate": 9.99452994967506e-05,
      "loss": 3.6364,
      "step": 1985
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0348880290985107,
      "learning_rate": 9.994500897681401e-05,
      "loss": 3.6467,
      "step": 1990
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.886435866355896,
      "learning_rate": 9.994471768785339e-05,
      "loss": 3.5752,
      "step": 1995
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8619951605796814,
      "learning_rate": 9.994442562987324e-05,
      "loss": 3.6079,
      "step": 2000
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.129574179649353,
      "learning_rate": 9.994413280287803e-05,
      "loss": 3.5576,
      "step": 2005
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9984617233276367,
      "learning_rate": 9.994383920687229e-05,
      "loss": 3.664,
      "step": 2010
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0818992853164673,
      "learning_rate": 9.994354484186053e-05,
      "loss": 3.5483,
      "step": 2015
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9386975765228271,
      "learning_rate": 9.994324970784727e-05,
      "loss": 3.3879,
      "step": 2020
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.881239116191864,
      "learning_rate": 9.994295380483709e-05,
      "loss": 3.5852,
      "step": 2025
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8454223275184631,
      "learning_rate": 9.99426571328345e-05,
      "loss": 3.4855,
      "step": 2030
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.757887840270996,
      "learning_rate": 9.994235969184411e-05,
      "loss": 3.4476,
      "step": 2035
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9315691590309143,
      "learning_rate": 9.994206148187046e-05,
      "loss": 3.458,
      "step": 2040
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8333036303520203,
      "learning_rate": 9.994176250291818e-05,
      "loss": 3.526,
      "step": 2045
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.398749589920044,
      "learning_rate": 9.994146275499186e-05,
      "loss": 3.4919,
      "step": 2050
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8594757914543152,
      "learning_rate": 9.994116223809611e-05,
      "loss": 3.5424,
      "step": 2055
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8984076976776123,
      "learning_rate": 9.994086095223557e-05,
      "loss": 3.34,
      "step": 2060
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9739062786102295,
      "learning_rate": 9.994055889741485e-05,
      "loss": 3.5549,
      "step": 2065
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8851149678230286,
      "learning_rate": 9.994025607363862e-05,
      "loss": 3.5826,
      "step": 2070
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1962389945983887,
      "learning_rate": 9.993995248091155e-05,
      "loss": 3.4462,
      "step": 2075
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8680959939956665,
      "learning_rate": 9.993964811923831e-05,
      "loss": 3.4285,
      "step": 2080
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8808024525642395,
      "learning_rate": 9.993934298862358e-05,
      "loss": 3.6156,
      "step": 2085
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.936768651008606,
      "learning_rate": 9.993903708907205e-05,
      "loss": 3.5516,
      "step": 2090
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8678238391876221,
      "learning_rate": 9.993873042058845e-05,
      "loss": 3.5196,
      "step": 2095
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.256488561630249,
      "learning_rate": 9.993842298317749e-05,
      "loss": 3.5247,
      "step": 2100
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1896398067474365,
      "learning_rate": 9.993811477684392e-05,
      "loss": 3.3924,
      "step": 2105
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9546697735786438,
      "learning_rate": 9.993780580159246e-05,
      "loss": 3.516,
      "step": 2110
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9228548407554626,
      "learning_rate": 9.993749605742788e-05,
      "loss": 3.5816,
      "step": 2115
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9277712106704712,
      "learning_rate": 9.993718554435495e-05,
      "loss": 3.5773,
      "step": 2120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.911045253276825,
      "learning_rate": 9.993687426237846e-05,
      "loss": 3.4222,
      "step": 2125
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6116458177566528,
      "learning_rate": 9.993656221150318e-05,
      "loss": 3.4411,
      "step": 2130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8525874614715576,
      "learning_rate": 9.993624939173394e-05,
      "loss": 3.3517,
      "step": 2135
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9243764281272888,
      "learning_rate": 9.993593580307553e-05,
      "loss": 3.5251,
      "step": 2140
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8601652383804321,
      "learning_rate": 9.99356214455328e-05,
      "loss": 3.6116,
      "step": 2145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8833007216453552,
      "learning_rate": 9.993530631911059e-05,
      "loss": 3.445,
      "step": 2150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9162232875823975,
      "learning_rate": 9.993499042381373e-05,
      "loss": 3.4663,
      "step": 2155
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9348765015602112,
      "learning_rate": 9.993467375964711e-05,
      "loss": 3.4573,
      "step": 2160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.891520619392395,
      "learning_rate": 9.99343563266156e-05,
      "loss": 3.3256,
      "step": 2165
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.000100016593933,
      "learning_rate": 9.993403812472408e-05,
      "loss": 3.5815,
      "step": 2170
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8600257039070129,
      "learning_rate": 9.993371915397745e-05,
      "loss": 3.4509,
      "step": 2175
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8629732131958008,
      "learning_rate": 9.993339941438062e-05,
      "loss": 3.4743,
      "step": 2180
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8462866544723511,
      "learning_rate": 9.993307890593853e-05,
      "loss": 3.4767,
      "step": 2185
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8340736627578735,
      "learning_rate": 9.993275762865609e-05,
      "loss": 3.4448,
      "step": 2190
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8478630185127258,
      "learning_rate": 9.993243558253826e-05,
      "loss": 3.4542,
      "step": 2195
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8294188976287842,
      "learning_rate": 9.993211276759e-05,
      "loss": 3.4679,
      "step": 2200
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8729796409606934,
      "learning_rate": 9.993178918381628e-05,
      "loss": 3.3859,
      "step": 2205
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8798531293869019,
      "learning_rate": 9.993146483122207e-05,
      "loss": 3.4452,
      "step": 2210
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.874874472618103,
      "learning_rate": 9.993113970981238e-05,
      "loss": 3.5393,
      "step": 2215
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.844140350818634,
      "learning_rate": 9.993081381959221e-05,
      "loss": 3.3382,
      "step": 2220
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8777813911437988,
      "learning_rate": 9.99304871605666e-05,
      "loss": 3.4406,
      "step": 2225
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8876941800117493,
      "learning_rate": 9.993015973274053e-05,
      "loss": 3.3691,
      "step": 2230
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9664689898490906,
      "learning_rate": 9.992983153611908e-05,
      "loss": 3.4944,
      "step": 2235
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9407105445861816,
      "learning_rate": 9.99295025707073e-05,
      "loss": 3.4977,
      "step": 2240
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8512258529663086,
      "learning_rate": 9.992917283651024e-05,
      "loss": 3.473,
      "step": 2245
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8954763412475586,
      "learning_rate": 9.992884233353298e-05,
      "loss": 3.5134,
      "step": 2250
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8768230676651001,
      "learning_rate": 9.992851106178063e-05,
      "loss": 3.3514,
      "step": 2255
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8083854913711548,
      "learning_rate": 9.992817902125826e-05,
      "loss": 3.3488,
      "step": 2260
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8098512887954712,
      "learning_rate": 9.9927846211971e-05,
      "loss": 3.434,
      "step": 2265
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8368965983390808,
      "learning_rate": 9.992751263392397e-05,
      "loss": 3.3709,
      "step": 2270
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.780312716960907,
      "learning_rate": 9.992717828712232e-05,
      "loss": 3.406,
      "step": 2275
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8234829902648926,
      "learning_rate": 9.992684317157117e-05,
      "loss": 3.3671,
      "step": 2280
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7574958801269531,
      "learning_rate": 9.992650728727571e-05,
      "loss": 3.5098,
      "step": 2285
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9158136248588562,
      "learning_rate": 9.99261706342411e-05,
      "loss": 3.3116,
      "step": 2290
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7725308537483215,
      "learning_rate": 9.99258332124725e-05,
      "loss": 3.4853,
      "step": 2295
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8513344526290894,
      "learning_rate": 9.992549502197514e-05,
      "loss": 3.4288,
      "step": 2300
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8616483807563782,
      "learning_rate": 9.992515606275423e-05,
      "loss": 3.3989,
      "step": 2305
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8529488444328308,
      "learning_rate": 9.992481633481495e-05,
      "loss": 3.4369,
      "step": 2310
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4446547031402588,
      "learning_rate": 9.992447583816257e-05,
      "loss": 3.4726,
      "step": 2315
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8019496202468872,
      "learning_rate": 9.992413457280231e-05,
      "loss": 3.3732,
      "step": 2320
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4860488176345825,
      "learning_rate": 9.992379253873943e-05,
      "loss": 3.4108,
      "step": 2325
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8782339096069336,
      "learning_rate": 9.99234497359792e-05,
      "loss": 3.377,
      "step": 2330
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7930213809013367,
      "learning_rate": 9.992310616452689e-05,
      "loss": 3.3955,
      "step": 2335
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9351562857627869,
      "learning_rate": 9.99227618243878e-05,
      "loss": 3.3463,
      "step": 2340
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7620325684547424,
      "learning_rate": 9.992241671556724e-05,
      "loss": 3.3527,
      "step": 2345
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8109075427055359,
      "learning_rate": 9.99220708380705e-05,
      "loss": 3.4071,
      "step": 2350
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.175560712814331,
      "learning_rate": 9.99217241919029e-05,
      "loss": 3.2791,
      "step": 2355
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8343840837478638,
      "learning_rate": 9.992137677706982e-05,
      "loss": 3.3348,
      "step": 2360
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8932955265045166,
      "learning_rate": 9.99210285935766e-05,
      "loss": 3.394,
      "step": 2365
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8180989027023315,
      "learning_rate": 9.992067964142856e-05,
      "loss": 3.3787,
      "step": 2370
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.821746826171875,
      "learning_rate": 9.992032992063109e-05,
      "loss": 3.3326,
      "step": 2375
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.780275821685791,
      "learning_rate": 9.99199794311896e-05,
      "loss": 3.4446,
      "step": 2380
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.150073528289795,
      "learning_rate": 9.991962817310947e-05,
      "loss": 3.3297,
      "step": 2385
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9100016355514526,
      "learning_rate": 9.991927614639612e-05,
      "loss": 3.3296,
      "step": 2390
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8502523899078369,
      "learning_rate": 9.991892335105493e-05,
      "loss": 3.3356,
      "step": 2395
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9188058972358704,
      "learning_rate": 9.991856978709137e-05,
      "loss": 3.4162,
      "step": 2400
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8709559440612793,
      "learning_rate": 9.99182154545109e-05,
      "loss": 3.3197,
      "step": 2405
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.820497989654541,
      "learning_rate": 9.991786035331893e-05,
      "loss": 3.4706,
      "step": 2410
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7680765986442566,
      "learning_rate": 9.991750448352096e-05,
      "loss": 3.3475,
      "step": 2415
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8154876232147217,
      "learning_rate": 9.991714784512243e-05,
      "loss": 3.32,
      "step": 2420
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8859189748764038,
      "learning_rate": 9.991679043812887e-05,
      "loss": 3.345,
      "step": 2425
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.73581463098526,
      "learning_rate": 9.99164322625458e-05,
      "loss": 3.3046,
      "step": 2430
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8547416925430298,
      "learning_rate": 9.991607331837867e-05,
      "loss": 3.2779,
      "step": 2435
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8052057027816772,
      "learning_rate": 9.991571360563306e-05,
      "loss": 3.3573,
      "step": 2440
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8639572858810425,
      "learning_rate": 9.99153531243145e-05,
      "loss": 3.343,
      "step": 2445
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8732950687408447,
      "learning_rate": 9.991499187442852e-05,
      "loss": 3.2825,
      "step": 2450
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7967109680175781,
      "learning_rate": 9.99146298559807e-05,
      "loss": 3.4219,
      "step": 2455
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.811153769493103,
      "learning_rate": 9.991426706897661e-05,
      "loss": 3.3002,
      "step": 2460
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.716399610042572,
      "learning_rate": 9.991390351342185e-05,
      "loss": 3.2856,
      "step": 2465
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8309920430183411,
      "learning_rate": 9.991353918932198e-05,
      "loss": 3.3337,
      "step": 2470
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8790179491043091,
      "learning_rate": 9.991317409668264e-05,
      "loss": 3.3531,
      "step": 2475
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8215545415878296,
      "learning_rate": 9.991280823550945e-05,
      "loss": 3.2456,
      "step": 2480
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8807469606399536,
      "learning_rate": 9.991244160580803e-05,
      "loss": 3.3456,
      "step": 2485
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7865097522735596,
      "learning_rate": 9.991207420758404e-05,
      "loss": 3.2503,
      "step": 2490
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6726062297821045,
      "learning_rate": 9.991170604084314e-05,
      "loss": 3.3593,
      "step": 2495
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8313099145889282,
      "learning_rate": 9.991133710559097e-05,
      "loss": 3.3203,
      "step": 2500
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7395262718200684,
      "learning_rate": 9.991096740183324e-05,
      "loss": 3.3469,
      "step": 2505
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1887556314468384,
      "learning_rate": 9.991059692957563e-05,
      "loss": 3.2972,
      "step": 2510
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8035768866539001,
      "learning_rate": 9.991022568882384e-05,
      "loss": 3.224,
      "step": 2515
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8527531623840332,
      "learning_rate": 9.99098536795836e-05,
      "loss": 3.2748,
      "step": 2520
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8359070420265198,
      "learning_rate": 9.990948090186063e-05,
      "loss": 3.2032,
      "step": 2525
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7952890992164612,
      "learning_rate": 9.990910735566065e-05,
      "loss": 3.2557,
      "step": 2530
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8101961612701416,
      "learning_rate": 9.990873304098946e-05,
      "loss": 3.4309,
      "step": 2535
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9587768316268921,
      "learning_rate": 9.990835795785278e-05,
      "loss": 3.3557,
      "step": 2540
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7527632713317871,
      "learning_rate": 9.990798210625642e-05,
      "loss": 3.412,
      "step": 2545
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8132131695747375,
      "learning_rate": 9.990760548620613e-05,
      "loss": 3.1682,
      "step": 2550
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7347726225852966,
      "learning_rate": 9.990722809770773e-05,
      "loss": 3.2339,
      "step": 2555
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7867166996002197,
      "learning_rate": 9.990684994076702e-05,
      "loss": 3.2579,
      "step": 2560
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7665725946426392,
      "learning_rate": 9.990647101538983e-05,
      "loss": 3.4403,
      "step": 2565
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7059378623962402,
      "learning_rate": 9.990609132158201e-05,
      "loss": 3.2649,
      "step": 2570
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8164868354797363,
      "learning_rate": 9.990571085934938e-05,
      "loss": 3.281,
      "step": 2575
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8156791925430298,
      "learning_rate": 9.990532962869783e-05,
      "loss": 3.2932,
      "step": 2580
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8037603497505188,
      "learning_rate": 9.990494762963318e-05,
      "loss": 3.2641,
      "step": 2585
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8995884656906128,
      "learning_rate": 9.990456486216136e-05,
      "loss": 3.2658,
      "step": 2590
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8237126469612122,
      "learning_rate": 9.990418132628824e-05,
      "loss": 3.3598,
      "step": 2595
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7586016058921814,
      "learning_rate": 9.990379702201974e-05,
      "loss": 3.2328,
      "step": 2600
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7745185494422913,
      "learning_rate": 9.990341194936173e-05,
      "loss": 3.2217,
      "step": 2605
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.737471342086792,
      "learning_rate": 9.990302610832021e-05,
      "loss": 3.301,
      "step": 2610
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.761242687702179,
      "learning_rate": 9.990263949890108e-05,
      "loss": 3.2692,
      "step": 2615
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9848249554634094,
      "learning_rate": 9.990225212111029e-05,
      "loss": 3.3069,
      "step": 2620
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7448675632476807,
      "learning_rate": 9.990186397495383e-05,
      "loss": 3.2917,
      "step": 2625
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6836724877357483,
      "learning_rate": 9.990147506043766e-05,
      "loss": 3.2129,
      "step": 2630
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8708077669143677,
      "learning_rate": 9.990108537756775e-05,
      "loss": 3.1994,
      "step": 2635
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7983249425888062,
      "learning_rate": 9.990069492635015e-05,
      "loss": 3.296,
      "step": 2640
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7326323986053467,
      "learning_rate": 9.990030370679082e-05,
      "loss": 3.3292,
      "step": 2645
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6904963254928589,
      "learning_rate": 9.989991171889581e-05,
      "loss": 3.2884,
      "step": 2650
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7780689597129822,
      "learning_rate": 9.989951896267115e-05,
      "loss": 3.2472,
      "step": 2655
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8310834765434265,
      "learning_rate": 9.989912543812289e-05,
      "loss": 3.1513,
      "step": 2660
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7430053353309631,
      "learning_rate": 9.989873114525709e-05,
      "loss": 3.1933,
      "step": 2665
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7677866816520691,
      "learning_rate": 9.989833608407981e-05,
      "loss": 3.2804,
      "step": 2670
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8345364332199097,
      "learning_rate": 9.989794025459715e-05,
      "loss": 3.2285,
      "step": 2675
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8181478977203369,
      "learning_rate": 9.989754365681519e-05,
      "loss": 3.2897,
      "step": 2680
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7227663397789001,
      "learning_rate": 9.989714629074006e-05,
      "loss": 3.1931,
      "step": 2685
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7723978757858276,
      "learning_rate": 9.989674815637784e-05,
      "loss": 3.3631,
      "step": 2690
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8158310055732727,
      "learning_rate": 9.989634925373468e-05,
      "loss": 3.1184,
      "step": 2695
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8147916197776794,
      "learning_rate": 9.989594958281673e-05,
      "loss": 3.2507,
      "step": 2700
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7860249280929565,
      "learning_rate": 9.989554914363015e-05,
      "loss": 3.2479,
      "step": 2705
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8432801961898804,
      "learning_rate": 9.989514793618109e-05,
      "loss": 3.2959,
      "step": 2710
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0103390216827393,
      "learning_rate": 9.989474596047572e-05,
      "loss": 3.2864,
      "step": 2715
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.820296049118042,
      "learning_rate": 9.989434321652024e-05,
      "loss": 3.2273,
      "step": 2720
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7908734083175659,
      "learning_rate": 9.989393970432085e-05,
      "loss": 3.2216,
      "step": 2725
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8387293815612793,
      "learning_rate": 9.989353542388378e-05,
      "loss": 3.199,
      "step": 2730
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8564770221710205,
      "learning_rate": 9.989313037521524e-05,
      "loss": 3.1539,
      "step": 2735
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2320681810379028,
      "learning_rate": 9.989272455832145e-05,
      "loss": 3.167,
      "step": 2740
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7062599062919617,
      "learning_rate": 9.989231797320869e-05,
      "loss": 3.1822,
      "step": 2745
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8994221687316895,
      "learning_rate": 9.989191061988319e-05,
      "loss": 3.2404,
      "step": 2750
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7819957137107849,
      "learning_rate": 9.989150249835125e-05,
      "loss": 3.2352,
      "step": 2755
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8571491837501526,
      "learning_rate": 9.989109360861913e-05,
      "loss": 3.2871,
      "step": 2760
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7846237421035767,
      "learning_rate": 9.989068395069313e-05,
      "loss": 3.2532,
      "step": 2765
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7159004211425781,
      "learning_rate": 9.989027352457958e-05,
      "loss": 3.1969,
      "step": 2770
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7262901067733765,
      "learning_rate": 9.988986233028478e-05,
      "loss": 3.1929,
      "step": 2775
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.716117262840271,
      "learning_rate": 9.988945036781507e-05,
      "loss": 3.1906,
      "step": 2780
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6612688302993774,
      "learning_rate": 9.988903763717679e-05,
      "loss": 3.2109,
      "step": 2785
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8182170987129211,
      "learning_rate": 9.988862413837628e-05,
      "loss": 3.1201,
      "step": 2790
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6866785883903503,
      "learning_rate": 9.988820987141993e-05,
      "loss": 3.2207,
      "step": 2795
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7542786002159119,
      "learning_rate": 9.988779483631409e-05,
      "loss": 3.2495,
      "step": 2800
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7545336484909058,
      "learning_rate": 9.98873790330652e-05,
      "loss": 3.219,
      "step": 2805
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8212763071060181,
      "learning_rate": 9.988696246167961e-05,
      "loss": 3.2126,
      "step": 2810
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7151721119880676,
      "learning_rate": 9.988654512216378e-05,
      "loss": 3.2187,
      "step": 2815
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6824684143066406,
      "learning_rate": 9.988612701452409e-05,
      "loss": 3.2369,
      "step": 2820
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6970787048339844,
      "learning_rate": 9.988570813876701e-05,
      "loss": 3.2458,
      "step": 2825
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7109242677688599,
      "learning_rate": 9.988528849489897e-05,
      "loss": 3.2407,
      "step": 2830
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6991509199142456,
      "learning_rate": 9.988486808292646e-05,
      "loss": 3.2476,
      "step": 2835
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8480271100997925,
      "learning_rate": 9.988444690285592e-05,
      "loss": 3.1222,
      "step": 2840
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6860089898109436,
      "learning_rate": 9.988402495469385e-05,
      "loss": 3.1667,
      "step": 2845
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.756851315498352,
      "learning_rate": 9.988360223844673e-05,
      "loss": 3.124,
      "step": 2850
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6788793802261353,
      "learning_rate": 9.98831787541211e-05,
      "loss": 3.2506,
      "step": 2855
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7682697176933289,
      "learning_rate": 9.988275450172346e-05,
      "loss": 3.1588,
      "step": 2860
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6875393390655518,
      "learning_rate": 9.988232948126037e-05,
      "loss": 3.2061,
      "step": 2865
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6801357269287109,
      "learning_rate": 9.988190369273833e-05,
      "loss": 3.2573,
      "step": 2870
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.800896942615509,
      "learning_rate": 9.988147713616391e-05,
      "loss": 3.217,
      "step": 2875
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6984912753105164,
      "learning_rate": 9.988104981154371e-05,
      "loss": 3.2432,
      "step": 2880
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0920158624649048,
      "learning_rate": 9.988062171888429e-05,
      "loss": 3.2609,
      "step": 2885
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7151643633842468,
      "learning_rate": 9.988019285819221e-05,
      "loss": 3.1871,
      "step": 2890
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7269686460494995,
      "learning_rate": 9.987976322947411e-05,
      "loss": 3.0879,
      "step": 2895
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6963534355163574,
      "learning_rate": 9.98793328327366e-05,
      "loss": 3.2251,
      "step": 2900
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8605373501777649,
      "learning_rate": 9.98789016679863e-05,
      "loss": 3.1601,
      "step": 2905
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6729230284690857,
      "learning_rate": 9.987846973522985e-05,
      "loss": 3.1066,
      "step": 2910
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7581908702850342,
      "learning_rate": 9.98780370344739e-05,
      "loss": 3.1711,
      "step": 2915
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7415596842765808,
      "learning_rate": 9.987760356572512e-05,
      "loss": 3.2066,
      "step": 2920
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7093629240989685,
      "learning_rate": 9.987716932899018e-05,
      "loss": 3.1667,
      "step": 2925
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7570719122886658,
      "learning_rate": 9.987673432427576e-05,
      "loss": 3.1826,
      "step": 2930
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7846543192863464,
      "learning_rate": 9.987629855158855e-05,
      "loss": 3.0908,
      "step": 2935
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7132800817489624,
      "learning_rate": 9.98758620109353e-05,
      "loss": 3.2501,
      "step": 2940
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7441908121109009,
      "learning_rate": 9.98754247023227e-05,
      "loss": 3.128,
      "step": 2945
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7207724452018738,
      "learning_rate": 9.987498662575747e-05,
      "loss": 3.1059,
      "step": 2950
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6907110214233398,
      "learning_rate": 9.987454778124639e-05,
      "loss": 3.1416,
      "step": 2955
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6930436491966248,
      "learning_rate": 9.987410816879619e-05,
      "loss": 3.225,
      "step": 2960
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.647905707359314,
      "learning_rate": 9.987366778841365e-05,
      "loss": 3.228,
      "step": 2965
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7706412672996521,
      "learning_rate": 9.987322664010554e-05,
      "loss": 2.9979,
      "step": 2970
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6616899967193604,
      "learning_rate": 9.987278472387867e-05,
      "loss": 3.12,
      "step": 2975
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.012879729270935,
      "learning_rate": 9.987234203973986e-05,
      "loss": 3.1513,
      "step": 2980
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7267111539840698,
      "learning_rate": 9.987189858769586e-05,
      "loss": 3.2152,
      "step": 2985
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8209304809570312,
      "learning_rate": 9.987145436775357e-05,
      "loss": 3.1048,
      "step": 2990
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7600386142730713,
      "learning_rate": 9.98710093799198e-05,
      "loss": 3.0365,
      "step": 2995
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7507590651512146,
      "learning_rate": 9.987056362420138e-05,
      "loss": 3.1073,
      "step": 3000
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6948139071464539,
      "learning_rate": 9.98701171006052e-05,
      "loss": 3.0935,
      "step": 3005
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7858610153198242,
      "learning_rate": 9.986966980913817e-05,
      "loss": 3.0849,
      "step": 3010
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6845154762268066,
      "learning_rate": 9.986922174980709e-05,
      "loss": 3.0629,
      "step": 3015
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6574291586875916,
      "learning_rate": 9.986877292261893e-05,
      "loss": 3.1177,
      "step": 3020
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6983383893966675,
      "learning_rate": 9.986832332758058e-05,
      "loss": 3.2202,
      "step": 3025
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6973050236701965,
      "learning_rate": 9.986787296469895e-05,
      "loss": 3.1592,
      "step": 3030
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9584047198295593,
      "learning_rate": 9.986742183398099e-05,
      "loss": 3.0089,
      "step": 3035
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7567125558853149,
      "learning_rate": 9.986696993543366e-05,
      "loss": 3.1741,
      "step": 3040
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7055335640907288,
      "learning_rate": 9.986651726906387e-05,
      "loss": 3.1582,
      "step": 3045
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7041676044464111,
      "learning_rate": 9.986606383487862e-05,
      "loss": 3.0788,
      "step": 3050
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7270480990409851,
      "learning_rate": 9.98656096328849e-05,
      "loss": 3.0995,
      "step": 3055
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7157003879547119,
      "learning_rate": 9.98651546630897e-05,
      "loss": 3.1154,
      "step": 3060
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6704452037811279,
      "learning_rate": 9.986469892550001e-05,
      "loss": 3.0541,
      "step": 3065
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6464080810546875,
      "learning_rate": 9.986424242012287e-05,
      "loss": 3.153,
      "step": 3070
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6497782468795776,
      "learning_rate": 9.986378514696528e-05,
      "loss": 3.0797,
      "step": 3075
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7691593766212463,
      "learning_rate": 9.98633271060343e-05,
      "loss": 3.0781,
      "step": 3080
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7210003137588501,
      "learning_rate": 9.9862868297337e-05,
      "loss": 3.1053,
      "step": 3085
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7439016699790955,
      "learning_rate": 9.98624087208804e-05,
      "loss": 3.1576,
      "step": 3090
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7134301066398621,
      "learning_rate": 9.98619483766716e-05,
      "loss": 3.2192,
      "step": 3095
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.688418984413147,
      "learning_rate": 9.98614872647177e-05,
      "loss": 3.0201,
      "step": 3100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7210434675216675,
      "learning_rate": 9.986102538502579e-05,
      "loss": 3.041,
      "step": 3105
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.691558837890625,
      "learning_rate": 9.986056273760297e-05,
      "loss": 2.9863,
      "step": 3110
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7917752861976624,
      "learning_rate": 9.986009932245637e-05,
      "loss": 2.9741,
      "step": 3115
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6922580003738403,
      "learning_rate": 9.985963513959313e-05,
      "loss": 3.087,
      "step": 3120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7060474157333374,
      "learning_rate": 9.985917018902041e-05,
      "loss": 3.1235,
      "step": 3125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7813776135444641,
      "learning_rate": 9.985870447074535e-05,
      "loss": 3.1106,
      "step": 3130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.727481484413147,
      "learning_rate": 9.985823798477512e-05,
      "loss": 3.0785,
      "step": 3135
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7690613269805908,
      "learning_rate": 9.985777073111691e-05,
      "loss": 3.0952,
      "step": 3140
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6825624108314514,
      "learning_rate": 9.985730270977791e-05,
      "loss": 3.0269,
      "step": 3145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6597278714179993,
      "learning_rate": 9.985683392076534e-05,
      "loss": 3.1021,
      "step": 3150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6765249967575073,
      "learning_rate": 9.98563643640864e-05,
      "loss": 3.185,
      "step": 3155
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6701353192329407,
      "learning_rate": 9.985589403974833e-05,
      "loss": 3.1478,
      "step": 3160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6609770059585571,
      "learning_rate": 9.985542294775838e-05,
      "loss": 3.0515,
      "step": 3165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6828314065933228,
      "learning_rate": 9.985495108812377e-05,
      "loss": 3.099,
      "step": 3170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6115306615829468,
      "learning_rate": 9.985447846085181e-05,
      "loss": 3.0284,
      "step": 3175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6610054969787598,
      "learning_rate": 9.985400506594974e-05,
      "loss": 3.065,
      "step": 3180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7117329239845276,
      "learning_rate": 9.985353090342487e-05,
      "loss": 3.1471,
      "step": 3185
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.66950923204422,
      "learning_rate": 9.985305597328449e-05,
      "loss": 3.1503,
      "step": 3190
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7331110239028931,
      "learning_rate": 9.985258027553593e-05,
      "loss": 3.2387,
      "step": 3195
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6562609672546387,
      "learning_rate": 9.98521038101865e-05,
      "loss": 3.2091,
      "step": 3200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7370874881744385,
      "learning_rate": 9.985162657724354e-05,
      "loss": 3.0692,
      "step": 3205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6741855144500732,
      "learning_rate": 9.98511485767144e-05,
      "loss": 3.083,
      "step": 3210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6670575141906738,
      "learning_rate": 9.985066980860642e-05,
      "loss": 3.06,
      "step": 3215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7293885350227356,
      "learning_rate": 9.9850190272927e-05,
      "loss": 3.1338,
      "step": 3220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6651206612586975,
      "learning_rate": 9.984970996968352e-05,
      "loss": 3.0709,
      "step": 3225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6478111147880554,
      "learning_rate": 9.984922889888335e-05,
      "loss": 3.0733,
      "step": 3230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7272320985794067,
      "learning_rate": 9.984874706053393e-05,
      "loss": 3.0975,
      "step": 3235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6405625343322754,
      "learning_rate": 9.984826445464266e-05,
      "loss": 3.0098,
      "step": 3240
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6816385388374329,
      "learning_rate": 9.984778108121698e-05,
      "loss": 2.8973,
      "step": 3245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6794739961624146,
      "learning_rate": 9.984729694026431e-05,
      "loss": 2.9682,
      "step": 3250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7222070097923279,
      "learning_rate": 9.984681203179214e-05,
      "loss": 3.1077,
      "step": 3255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6771973371505737,
      "learning_rate": 9.984632635580792e-05,
      "loss": 3.1705,
      "step": 3260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7048766613006592,
      "learning_rate": 9.984583991231913e-05,
      "loss": 2.9253,
      "step": 3265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6086748242378235,
      "learning_rate": 9.984535270133324e-05,
      "loss": 3.0543,
      "step": 3270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7050692439079285,
      "learning_rate": 9.984486472285777e-05,
      "loss": 2.9189,
      "step": 3275
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.680128812789917,
      "learning_rate": 9.984437597690024e-05,
      "loss": 3.1075,
      "step": 3280
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6455293297767639,
      "learning_rate": 9.984388646346817e-05,
      "loss": 3.0048,
      "step": 3285
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6734024286270142,
      "learning_rate": 9.984339618256909e-05,
      "loss": 3.0151,
      "step": 3290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.662703812122345,
      "learning_rate": 9.984290513421057e-05,
      "loss": 3.0054,
      "step": 3295
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.668911337852478,
      "learning_rate": 9.984241331840015e-05,
      "loss": 3.1149,
      "step": 3300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6501004099845886,
      "learning_rate": 9.984192073514538e-05,
      "loss": 3.0476,
      "step": 3305
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6546483635902405,
      "learning_rate": 9.98414273844539e-05,
      "loss": 3.0044,
      "step": 3310
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8378316760063171,
      "learning_rate": 9.984093326633325e-05,
      "loss": 2.9227,
      "step": 3315
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7579377293586731,
      "learning_rate": 9.98404383807911e-05,
      "loss": 2.9744,
      "step": 3320
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8332241773605347,
      "learning_rate": 9.983994272783502e-05,
      "loss": 3.0533,
      "step": 3325
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6539364457130432,
      "learning_rate": 9.983944630747266e-05,
      "loss": 3.0767,
      "step": 3330
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6874090433120728,
      "learning_rate": 9.983894911971167e-05,
      "loss": 2.9512,
      "step": 3335
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6601637005805969,
      "learning_rate": 9.983845116455969e-05,
      "loss": 3.0084,
      "step": 3340
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6404942274093628,
      "learning_rate": 9.983795244202439e-05,
      "loss": 2.9873,
      "step": 3345
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6778912544250488,
      "learning_rate": 9.983745295211347e-05,
      "loss": 3.0494,
      "step": 3350
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6898072361946106,
      "learning_rate": 9.983695269483459e-05,
      "loss": 3.0353,
      "step": 3355
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.622125506401062,
      "learning_rate": 9.983645167019546e-05,
      "loss": 3.0142,
      "step": 3360
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7027627229690552,
      "learning_rate": 9.983594987820381e-05,
      "loss": 2.9498,
      "step": 3365
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1987366676330566,
      "learning_rate": 9.983544731886735e-05,
      "loss": 3.0551,
      "step": 3370
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6601062417030334,
      "learning_rate": 9.983494399219386e-05,
      "loss": 3.0239,
      "step": 3375
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7387123107910156,
      "learning_rate": 9.983443989819103e-05,
      "loss": 3.0071,
      "step": 3380
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6793323159217834,
      "learning_rate": 9.983393503686665e-05,
      "loss": 2.9856,
      "step": 3385
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6871797442436218,
      "learning_rate": 9.983342940822849e-05,
      "loss": 2.9365,
      "step": 3390
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.661694347858429,
      "learning_rate": 9.983292301228432e-05,
      "loss": 3.0598,
      "step": 3395
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6496928930282593,
      "learning_rate": 9.983241584904197e-05,
      "loss": 3.0601,
      "step": 3400
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6569991707801819,
      "learning_rate": 9.983190791850925e-05,
      "loss": 2.873,
      "step": 3405
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6024008393287659,
      "learning_rate": 9.983139922069394e-05,
      "loss": 2.9371,
      "step": 3410
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6510295867919922,
      "learning_rate": 9.983088975560391e-05,
      "loss": 3.102,
      "step": 3415
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6605210304260254,
      "learning_rate": 9.983037952324699e-05,
      "loss": 3.0029,
      "step": 3420
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6675822138786316,
      "learning_rate": 9.982986852363102e-05,
      "loss": 2.9457,
      "step": 3425
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6606597304344177,
      "learning_rate": 9.98293567567639e-05,
      "loss": 3.0917,
      "step": 3430
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6386340260505676,
      "learning_rate": 9.982884422265349e-05,
      "loss": 3.0624,
      "step": 3435
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4312800168991089,
      "learning_rate": 9.982833092130768e-05,
      "loss": 2.994,
      "step": 3440
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6451196074485779,
      "learning_rate": 9.982781685273438e-05,
      "loss": 2.9892,
      "step": 3445
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6639716625213623,
      "learning_rate": 9.982730201694153e-05,
      "loss": 2.9304,
      "step": 3450
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6283800005912781,
      "learning_rate": 9.982678641393701e-05,
      "loss": 3.0952,
      "step": 3455
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6407597064971924,
      "learning_rate": 9.98262700437288e-05,
      "loss": 2.9362,
      "step": 3460
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7734404802322388,
      "learning_rate": 9.982575290632481e-05,
      "loss": 2.9518,
      "step": 3465
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6312546730041504,
      "learning_rate": 9.982523500173304e-05,
      "loss": 2.9926,
      "step": 3470
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6450769901275635,
      "learning_rate": 9.982471632996144e-05,
      "loss": 3.0648,
      "step": 3475
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6390717029571533,
      "learning_rate": 9.9824196891018e-05,
      "loss": 3.0899,
      "step": 3480
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6482605934143066,
      "learning_rate": 9.982367668491075e-05,
      "loss": 2.969,
      "step": 3485
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6661344766616821,
      "learning_rate": 9.982315571164765e-05,
      "loss": 2.9593,
      "step": 3490
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6754226684570312,
      "learning_rate": 9.982263397123675e-05,
      "loss": 3.0315,
      "step": 3495
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6187918186187744,
      "learning_rate": 9.982211146368608e-05,
      "loss": 2.9546,
      "step": 3500
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7692302465438843,
      "learning_rate": 9.982158818900369e-05,
      "loss": 2.9602,
      "step": 3505
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7444223761558533,
      "learning_rate": 9.982106414719762e-05,
      "loss": 2.9803,
      "step": 3510
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7428085803985596,
      "learning_rate": 9.982053933827595e-05,
      "loss": 3.1014,
      "step": 3515
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6346836686134338,
      "learning_rate": 9.982001376224677e-05,
      "loss": 3.0221,
      "step": 3520
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6290015578269958,
      "learning_rate": 9.981948741911816e-05,
      "loss": 2.9929,
      "step": 3525
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7137219905853271,
      "learning_rate": 9.981896030889821e-05,
      "loss": 2.9826,
      "step": 3530
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8584983944892883,
      "learning_rate": 9.981843243159507e-05,
      "loss": 2.9097,
      "step": 3535
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6207117438316345,
      "learning_rate": 9.981790378721684e-05,
      "loss": 2.9104,
      "step": 3540
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6150648593902588,
      "learning_rate": 9.98173743757717e-05,
      "loss": 3.0122,
      "step": 3545
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6964576840400696,
      "learning_rate": 9.981684419726775e-05,
      "loss": 3.0513,
      "step": 3550
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.626272439956665,
      "learning_rate": 9.981631325171318e-05,
      "loss": 3.0584,
      "step": 3555
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8664392828941345,
      "learning_rate": 9.981578153911616e-05,
      "loss": 2.917,
      "step": 3560
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6531164646148682,
      "learning_rate": 9.981524905948487e-05,
      "loss": 2.8918,
      "step": 3565
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6600880026817322,
      "learning_rate": 9.981471581282753e-05,
      "loss": 2.974,
      "step": 3570
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.805077075958252,
      "learning_rate": 9.981418179915234e-05,
      "loss": 2.9687,
      "step": 3575
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5846509337425232,
      "learning_rate": 9.981364701846752e-05,
      "loss": 2.9011,
      "step": 3580
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6690158843994141,
      "learning_rate": 9.981311147078128e-05,
      "loss": 2.9889,
      "step": 3585
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6842199563980103,
      "learning_rate": 9.981257515610192e-05,
      "loss": 2.9221,
      "step": 3590
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6372672915458679,
      "learning_rate": 9.981203807443764e-05,
      "loss": 3.0029,
      "step": 3595
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6136438250541687,
      "learning_rate": 9.981150022579676e-05,
      "loss": 2.904,
      "step": 3600
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6045554876327515,
      "learning_rate": 9.981096161018754e-05,
      "loss": 2.8719,
      "step": 3605
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7316902279853821,
      "learning_rate": 9.981042222761826e-05,
      "loss": 2.8933,
      "step": 3610
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7150322198867798,
      "learning_rate": 9.980988207809726e-05,
      "loss": 3.0391,
      "step": 3615
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5937066674232483,
      "learning_rate": 9.98093411616328e-05,
      "loss": 2.9283,
      "step": 3620
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6155052781105042,
      "learning_rate": 9.980879947823327e-05,
      "loss": 2.9352,
      "step": 3625
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6316401958465576,
      "learning_rate": 9.980825702790696e-05,
      "loss": 2.9079,
      "step": 3630
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6248288750648499,
      "learning_rate": 9.980771381066227e-05,
      "loss": 2.9013,
      "step": 3635
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6647264361381531,
      "learning_rate": 9.980716982650753e-05,
      "loss": 2.9101,
      "step": 3640
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6168761849403381,
      "learning_rate": 9.980662507545113e-05,
      "loss": 2.8653,
      "step": 3645
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6285346150398254,
      "learning_rate": 9.980607955750146e-05,
      "loss": 2.9562,
      "step": 3650
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6122355461120605,
      "learning_rate": 9.98055332726669e-05,
      "loss": 2.8897,
      "step": 3655
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5891215801239014,
      "learning_rate": 9.980498622095588e-05,
      "loss": 2.9276,
      "step": 3660
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6383939385414124,
      "learning_rate": 9.980443840237681e-05,
      "loss": 2.9117,
      "step": 3665
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6450744867324829,
      "learning_rate": 9.980388981693816e-05,
      "loss": 2.8493,
      "step": 3670
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7293107509613037,
      "learning_rate": 9.980334046464832e-05,
      "loss": 2.9854,
      "step": 3675
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6698068976402283,
      "learning_rate": 9.980279034551579e-05,
      "loss": 2.9535,
      "step": 3680
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6676675081253052,
      "learning_rate": 9.980223945954902e-05,
      "loss": 2.8875,
      "step": 3685
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6608633995056152,
      "learning_rate": 9.980168780675651e-05,
      "loss": 3.0654,
      "step": 3690
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.649146318435669,
      "learning_rate": 9.980113538714675e-05,
      "loss": 2.9135,
      "step": 3695
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5925349593162537,
      "learning_rate": 9.980058220072825e-05,
      "loss": 2.9705,
      "step": 3700
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8126382827758789,
      "learning_rate": 9.98000282475095e-05,
      "loss": 2.9398,
      "step": 3705
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5908036828041077,
      "learning_rate": 9.979947352749906e-05,
      "loss": 2.9587,
      "step": 3710
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6106910705566406,
      "learning_rate": 9.979891804070545e-05,
      "loss": 2.9693,
      "step": 3715
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7378579378128052,
      "learning_rate": 9.979836178713724e-05,
      "loss": 2.9456,
      "step": 3720
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.81907057762146,
      "learning_rate": 9.979780476680298e-05,
      "loss": 2.834,
      "step": 3725
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.636466383934021,
      "learning_rate": 9.979724697971125e-05,
      "loss": 2.8825,
      "step": 3730
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7709857225418091,
      "learning_rate": 9.979668842587065e-05,
      "loss": 2.8309,
      "step": 3735
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5964935421943665,
      "learning_rate": 9.979612910528976e-05,
      "loss": 2.8922,
      "step": 3740
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6499475240707397,
      "learning_rate": 9.979556901797723e-05,
      "loss": 2.8839,
      "step": 3745
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7891850471496582,
      "learning_rate": 9.979500816394165e-05,
      "loss": 2.8116,
      "step": 3750
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.628230094909668,
      "learning_rate": 9.979444654319165e-05,
      "loss": 2.9266,
      "step": 3755
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7809752821922302,
      "learning_rate": 9.979388415573591e-05,
      "loss": 2.8993,
      "step": 3760
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6779226660728455,
      "learning_rate": 9.979332100158307e-05,
      "loss": 2.9298,
      "step": 3765
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7264147400856018,
      "learning_rate": 9.979275708074179e-05,
      "loss": 2.8301,
      "step": 3770
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6389162540435791,
      "learning_rate": 9.979219239322078e-05,
      "loss": 2.8921,
      "step": 3775
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.740480899810791,
      "learning_rate": 9.979162693902871e-05,
      "loss": 2.9493,
      "step": 3780
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7303812503814697,
      "learning_rate": 9.97910607181743e-05,
      "loss": 2.9908,
      "step": 3785
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6659772396087646,
      "learning_rate": 9.979049373066626e-05,
      "loss": 2.9293,
      "step": 3790
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6809698939323425,
      "learning_rate": 9.978992597651334e-05,
      "loss": 2.9301,
      "step": 3795
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.600370466709137,
      "learning_rate": 9.978935745572424e-05,
      "loss": 2.8581,
      "step": 3800
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5917452573776245,
      "learning_rate": 9.978878816830776e-05,
      "loss": 2.937,
      "step": 3805
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5923593044281006,
      "learning_rate": 9.978821811427263e-05,
      "loss": 2.9362,
      "step": 3810
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6336028575897217,
      "learning_rate": 9.978764729362766e-05,
      "loss": 2.8484,
      "step": 3815
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6873530745506287,
      "learning_rate": 9.978707570638162e-05,
      "loss": 2.9292,
      "step": 3820
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6014951467514038,
      "learning_rate": 9.978650335254332e-05,
      "loss": 2.8998,
      "step": 3825
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6296197772026062,
      "learning_rate": 9.978593023212155e-05,
      "loss": 2.9462,
      "step": 3830
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5723553895950317,
      "learning_rate": 9.978535634512515e-05,
      "loss": 2.8896,
      "step": 3835
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6474579572677612,
      "learning_rate": 9.978478169156298e-05,
      "loss": 2.8457,
      "step": 3840
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5892506241798401,
      "learning_rate": 9.978420627144385e-05,
      "loss": 2.9283,
      "step": 3845
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.693993866443634,
      "learning_rate": 9.978363008477665e-05,
      "loss": 2.9549,
      "step": 3850
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.567287266254425,
      "learning_rate": 9.978305313157022e-05,
      "loss": 3.0145,
      "step": 3855
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5880354046821594,
      "learning_rate": 9.978247541183348e-05,
      "loss": 2.9666,
      "step": 3860
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5902069211006165,
      "learning_rate": 9.978189692557529e-05,
      "loss": 2.881,
      "step": 3865
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5918734073638916,
      "learning_rate": 9.978131767280456e-05,
      "loss": 2.9989,
      "step": 3870
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5947348475456238,
      "learning_rate": 9.978073765353026e-05,
      "loss": 2.9248,
      "step": 3875
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6042597889900208,
      "learning_rate": 9.978015686776125e-05,
      "loss": 2.7824,
      "step": 3880
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7034199833869934,
      "learning_rate": 9.977957531550652e-05,
      "loss": 3.024,
      "step": 3885
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6814021468162537,
      "learning_rate": 9.977899299677502e-05,
      "loss": 2.88,
      "step": 3890
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6425178050994873,
      "learning_rate": 9.97784099115757e-05,
      "loss": 2.9488,
      "step": 3895
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5803279876708984,
      "learning_rate": 9.977782605991754e-05,
      "loss": 2.9132,
      "step": 3900
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6155019402503967,
      "learning_rate": 9.977724144180953e-05,
      "loss": 2.9123,
      "step": 3905
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5978789329528809,
      "learning_rate": 9.977665605726067e-05,
      "loss": 2.966,
      "step": 3910
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6574641466140747,
      "learning_rate": 9.977606990628001e-05,
      "loss": 2.9176,
      "step": 3915
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6395916938781738,
      "learning_rate": 9.977548298887651e-05,
      "loss": 2.7803,
      "step": 3920
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5976806879043579,
      "learning_rate": 9.977489530505927e-05,
      "loss": 2.8235,
      "step": 3925
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6236398220062256,
      "learning_rate": 9.97743068548373e-05,
      "loss": 2.9431,
      "step": 3930
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6292185187339783,
      "learning_rate": 9.977371763821966e-05,
      "loss": 2.8828,
      "step": 3935
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.711495578289032,
      "learning_rate": 9.977312765521545e-05,
      "loss": 2.9136,
      "step": 3940
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6037873029708862,
      "learning_rate": 9.977253690583372e-05,
      "loss": 2.9594,
      "step": 3945
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7125998139381409,
      "learning_rate": 9.97719453900836e-05,
      "loss": 2.897,
      "step": 3950
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5490726828575134,
      "learning_rate": 9.977135310797418e-05,
      "loss": 2.9697,
      "step": 3955
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5779038071632385,
      "learning_rate": 9.977076005951458e-05,
      "loss": 2.8346,
      "step": 3960
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5951065421104431,
      "learning_rate": 9.977016624471392e-05,
      "loss": 2.9629,
      "step": 3965
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5694071054458618,
      "learning_rate": 9.976957166358137e-05,
      "loss": 2.8439,
      "step": 3970
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5733669996261597,
      "learning_rate": 9.976897631612607e-05,
      "loss": 2.9122,
      "step": 3975
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6106358170509338,
      "learning_rate": 9.97683802023572e-05,
      "loss": 2.8858,
      "step": 3980
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.550674557685852,
      "learning_rate": 9.976778332228389e-05,
      "loss": 2.9319,
      "step": 3985
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5632244944572449,
      "learning_rate": 9.97671856759154e-05,
      "loss": 2.8907,
      "step": 3990
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5563514828681946,
      "learning_rate": 9.976658726326088e-05,
      "loss": 2.9004,
      "step": 3995
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5770056247711182,
      "learning_rate": 9.976598808432957e-05,
      "loss": 2.9113,
      "step": 4000
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5799848437309265,
      "learning_rate": 9.976538813913068e-05,
      "loss": 2.8489,
      "step": 4005
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5374472141265869,
      "learning_rate": 9.976478742767347e-05,
      "loss": 2.7478,
      "step": 4010
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7343278527259827,
      "learning_rate": 9.976418594996716e-05,
      "loss": 2.9457,
      "step": 4015
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6015961170196533,
      "learning_rate": 9.976358370602103e-05,
      "loss": 2.9049,
      "step": 4020
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5591393709182739,
      "learning_rate": 9.976298069584438e-05,
      "loss": 2.8732,
      "step": 4025
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6846256256103516,
      "learning_rate": 9.976237691944643e-05,
      "loss": 2.8509,
      "step": 4030
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5750730037689209,
      "learning_rate": 9.976177237683653e-05,
      "loss": 2.8814,
      "step": 4035
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5473635196685791,
      "learning_rate": 9.976116706802395e-05,
      "loss": 2.9909,
      "step": 4040
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6361514329910278,
      "learning_rate": 9.976056099301805e-05,
      "loss": 2.9245,
      "step": 4045
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6505343317985535,
      "learning_rate": 9.975995415182812e-05,
      "loss": 2.8543,
      "step": 4050
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6956175565719604,
      "learning_rate": 9.975934654446355e-05,
      "loss": 2.9286,
      "step": 4055
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7333725690841675,
      "learning_rate": 9.975873817093367e-05,
      "loss": 2.911,
      "step": 4060
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6539108157157898,
      "learning_rate": 9.975812903124782e-05,
      "loss": 2.902,
      "step": 4065
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5854659080505371,
      "learning_rate": 9.975751912541544e-05,
      "loss": 2.8263,
      "step": 4070
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.602601945400238,
      "learning_rate": 9.975690845344588e-05,
      "loss": 2.8705,
      "step": 4075
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5977951884269714,
      "learning_rate": 9.975629701534854e-05,
      "loss": 2.8376,
      "step": 4080
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7529337406158447,
      "learning_rate": 9.975568481113285e-05,
      "loss": 2.7147,
      "step": 4085
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6307843327522278,
      "learning_rate": 9.975507184080823e-05,
      "loss": 2.8778,
      "step": 4090
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5689871907234192,
      "learning_rate": 9.975445810438413e-05,
      "loss": 2.8998,
      "step": 4095
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6293722987174988,
      "learning_rate": 9.975384360186999e-05,
      "loss": 2.8479,
      "step": 4100
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7138576507568359,
      "learning_rate": 9.975322833327527e-05,
      "loss": 2.8331,
      "step": 4105
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6521368026733398,
      "learning_rate": 9.975261229860944e-05,
      "loss": 2.9586,
      "step": 4110
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5539219379425049,
      "learning_rate": 9.975199549788199e-05,
      "loss": 2.825,
      "step": 4115
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6253762245178223,
      "learning_rate": 9.975137793110243e-05,
      "loss": 2.7672,
      "step": 4120
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6437821388244629,
      "learning_rate": 9.975075959828026e-05,
      "loss": 2.8927,
      "step": 4125
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6088517904281616,
      "learning_rate": 9.975014049942499e-05,
      "loss": 2.787,
      "step": 4130
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6032080054283142,
      "learning_rate": 9.974952063454615e-05,
      "loss": 2.854,
      "step": 4135
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5884603261947632,
      "learning_rate": 9.974890000365332e-05,
      "loss": 2.7727,
      "step": 4140
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5526155233383179,
      "learning_rate": 9.9748278606756e-05,
      "loss": 2.8008,
      "step": 4145
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5841876864433289,
      "learning_rate": 9.974765644386382e-05,
      "loss": 2.8637,
      "step": 4150
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6179796457290649,
      "learning_rate": 9.974703351498631e-05,
      "loss": 2.8822,
      "step": 4155
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5693957209587097,
      "learning_rate": 9.974640982013307e-05,
      "loss": 2.8763,
      "step": 4160
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5932303667068481,
      "learning_rate": 9.974578535931374e-05,
      "loss": 2.8605,
      "step": 4165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5779699087142944,
      "learning_rate": 9.974516013253789e-05,
      "loss": 2.8226,
      "step": 4170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6314557790756226,
      "learning_rate": 9.974453413981517e-05,
      "loss": 2.8811,
      "step": 4175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.618886411190033,
      "learning_rate": 9.974390738115521e-05,
      "loss": 2.8288,
      "step": 4180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5721433162689209,
      "learning_rate": 9.974327985656767e-05,
      "loss": 2.7986,
      "step": 4185
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7093735337257385,
      "learning_rate": 9.974265156606221e-05,
      "loss": 2.802,
      "step": 4190
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5887966752052307,
      "learning_rate": 9.974202250964849e-05,
      "loss": 2.7613,
      "step": 4195
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6274653077125549,
      "learning_rate": 9.974139268733623e-05,
      "loss": 2.9704,
      "step": 4200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.617312490940094,
      "learning_rate": 9.974076209913508e-05,
      "loss": 2.8121,
      "step": 4205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5675191283226013,
      "learning_rate": 9.974013074505477e-05,
      "loss": 2.8464,
      "step": 4210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5687738060951233,
      "learning_rate": 9.973949862510503e-05,
      "loss": 2.8923,
      "step": 4215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5681830644607544,
      "learning_rate": 9.97388657392956e-05,
      "loss": 2.8646,
      "step": 4220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5673846006393433,
      "learning_rate": 9.97382320876362e-05,
      "loss": 2.7667,
      "step": 4225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5855274200439453,
      "learning_rate": 9.973759767013662e-05,
      "loss": 2.826,
      "step": 4230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5810365080833435,
      "learning_rate": 9.973696248680659e-05,
      "loss": 2.848,
      "step": 4235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5380073189735413,
      "learning_rate": 9.97363265376559e-05,
      "loss": 2.8072,
      "step": 4240
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5675109028816223,
      "learning_rate": 9.973568982269436e-05,
      "loss": 2.9097,
      "step": 4245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5617326498031616,
      "learning_rate": 9.973505234193177e-05,
      "loss": 2.8116,
      "step": 4250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.629545271396637,
      "learning_rate": 9.973441409537793e-05,
      "loss": 2.8504,
      "step": 4255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5640160441398621,
      "learning_rate": 9.973377508304269e-05,
      "loss": 2.8169,
      "step": 4260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5504751801490784,
      "learning_rate": 9.973313530493586e-05,
      "loss": 2.9612,
      "step": 4265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5442126989364624,
      "learning_rate": 9.973249476106732e-05,
      "loss": 2.8335,
      "step": 4270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5623832941055298,
      "learning_rate": 9.97318534514469e-05,
      "loss": 2.8181,
      "step": 4275
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5777313709259033,
      "learning_rate": 9.97312113760845e-05,
      "loss": 2.7531,
      "step": 4280
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6078979969024658,
      "learning_rate": 9.973056853499001e-05,
      "loss": 2.8534,
      "step": 4285
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6778735518455505,
      "learning_rate": 9.972992492817331e-05,
      "loss": 2.6995,
      "step": 4290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6337717771530151,
      "learning_rate": 9.972928055564432e-05,
      "loss": 2.7766,
      "step": 4295
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7907583117485046,
      "learning_rate": 9.972863541741297e-05,
      "loss": 2.883,
      "step": 4300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.626194953918457,
      "learning_rate": 9.972798951348918e-05,
      "loss": 2.92,
      "step": 4305
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5689208507537842,
      "learning_rate": 9.972734284388288e-05,
      "loss": 2.8163,
      "step": 4310
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6129360198974609,
      "learning_rate": 9.972669540860406e-05,
      "loss": 2.8739,
      "step": 4315
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5575409531593323,
      "learning_rate": 9.972604720766268e-05,
      "loss": 2.8616,
      "step": 4320
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6557300686836243,
      "learning_rate": 9.97253982410687e-05,
      "loss": 2.7806,
      "step": 4325
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6057844161987305,
      "learning_rate": 9.972474850883213e-05,
      "loss": 2.8129,
      "step": 4330
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5926105976104736,
      "learning_rate": 9.972409801096299e-05,
      "loss": 2.8954,
      "step": 4335
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5883350372314453,
      "learning_rate": 9.972344674747125e-05,
      "loss": 2.911,
      "step": 4340
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5962880849838257,
      "learning_rate": 9.972279471836698e-05,
      "loss": 2.8364,
      "step": 4345
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5524152517318726,
      "learning_rate": 9.972214192366021e-05,
      "loss": 2.7979,
      "step": 4350
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5887129902839661,
      "learning_rate": 9.972148836336098e-05,
      "loss": 2.8602,
      "step": 4355
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.571533203125,
      "learning_rate": 9.972083403747937e-05,
      "loss": 2.7212,
      "step": 4360
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5729742050170898,
      "learning_rate": 9.972017894602542e-05,
      "loss": 2.7095,
      "step": 4365
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5727760195732117,
      "learning_rate": 9.971952308900925e-05,
      "loss": 2.807,
      "step": 4370
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5540823340415955,
      "learning_rate": 9.971886646644096e-05,
      "loss": 2.7968,
      "step": 4375
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5787964463233948,
      "learning_rate": 9.971820907833065e-05,
      "loss": 2.8308,
      "step": 4380
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7056093215942383,
      "learning_rate": 9.971755092468843e-05,
      "loss": 2.7962,
      "step": 4385
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5760160088539124,
      "learning_rate": 9.971689200552445e-05,
      "loss": 2.7873,
      "step": 4390
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5419966578483582,
      "learning_rate": 9.971623232084886e-05,
      "loss": 2.8315,
      "step": 4395
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5885605812072754,
      "learning_rate": 9.971557187067179e-05,
      "loss": 2.882,
      "step": 4400
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5507292747497559,
      "learning_rate": 9.971491065500345e-05,
      "loss": 2.8348,
      "step": 4405
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5906220078468323,
      "learning_rate": 9.9714248673854e-05,
      "loss": 2.8267,
      "step": 4410
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5660029053688049,
      "learning_rate": 9.971358592723361e-05,
      "loss": 2.8509,
      "step": 4415
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5465315580368042,
      "learning_rate": 9.97129224151525e-05,
      "loss": 2.7666,
      "step": 4420
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5973290205001831,
      "learning_rate": 9.971225813762094e-05,
      "loss": 2.8791,
      "step": 4425
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6187855005264282,
      "learning_rate": 9.971159309464906e-05,
      "loss": 2.7705,
      "step": 4430
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9025347828865051,
      "learning_rate": 9.971092728624718e-05,
      "loss": 2.8761,
      "step": 4435
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6456098556518555,
      "learning_rate": 9.97102607124255e-05,
      "loss": 2.7025,
      "step": 4440
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6310580372810364,
      "learning_rate": 9.970959337319433e-05,
      "loss": 2.7857,
      "step": 4445
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5214359164237976,
      "learning_rate": 9.970892526856392e-05,
      "loss": 2.8176,
      "step": 4450
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5496059656143188,
      "learning_rate": 9.970825639854454e-05,
      "loss": 2.8101,
      "step": 4455
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5474091172218323,
      "learning_rate": 9.970758676314652e-05,
      "loss": 2.8666,
      "step": 4460
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6441124677658081,
      "learning_rate": 9.970691636238017e-05,
      "loss": 2.8616,
      "step": 4465
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.593958854675293,
      "learning_rate": 9.970624519625578e-05,
      "loss": 2.863,
      "step": 4470
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5348805785179138,
      "learning_rate": 9.970557326478371e-05,
      "loss": 2.7797,
      "step": 4475
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5528934597969055,
      "learning_rate": 9.97049005679743e-05,
      "loss": 2.786,
      "step": 4480
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5621671676635742,
      "learning_rate": 9.970422710583791e-05,
      "loss": 2.8778,
      "step": 4485
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5495939254760742,
      "learning_rate": 9.970355287838491e-05,
      "loss": 2.7939,
      "step": 4490
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5494716763496399,
      "learning_rate": 9.970287788562567e-05,
      "loss": 2.7599,
      "step": 4495
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5766279697418213,
      "learning_rate": 9.97022021275706e-05,
      "loss": 2.871,
      "step": 4500
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6260276436805725,
      "learning_rate": 9.97015256042301e-05,
      "loss": 2.7594,
      "step": 4505
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5766665935516357,
      "learning_rate": 9.97008483156146e-05,
      "loss": 2.9491,
      "step": 4510
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5482069849967957,
      "learning_rate": 9.970017026173448e-05,
      "loss": 2.8366,
      "step": 4515
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5957104563713074,
      "learning_rate": 9.969949144260022e-05,
      "loss": 2.7775,
      "step": 4520
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8788536190986633,
      "learning_rate": 9.969881185822227e-05,
      "loss": 2.7755,
      "step": 4525
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5543661713600159,
      "learning_rate": 9.969813150861111e-05,
      "loss": 2.8152,
      "step": 4530
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5647691488265991,
      "learning_rate": 9.969745039377717e-05,
      "loss": 2.7807,
      "step": 4535
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5769283771514893,
      "learning_rate": 9.969676851373098e-05,
      "loss": 2.8004,
      "step": 4540
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5356346368789673,
      "learning_rate": 9.969608586848301e-05,
      "loss": 2.8034,
      "step": 4545
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5801103115081787,
      "learning_rate": 9.969540245804377e-05,
      "loss": 2.8405,
      "step": 4550
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5575684905052185,
      "learning_rate": 9.96947182824238e-05,
      "loss": 2.8584,
      "step": 4555
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5410173535346985,
      "learning_rate": 9.969403334163364e-05,
      "loss": 2.7759,
      "step": 4560
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5318824052810669,
      "learning_rate": 9.969334763568383e-05,
      "loss": 2.7885,
      "step": 4565
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.523658812046051,
      "learning_rate": 9.969266116458491e-05,
      "loss": 2.853,
      "step": 4570
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5816494226455688,
      "learning_rate": 9.969197392834747e-05,
      "loss": 2.9108,
      "step": 4575
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.579142689704895,
      "learning_rate": 9.969128592698207e-05,
      "loss": 2.8303,
      "step": 4580
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5385864973068237,
      "learning_rate": 9.969059716049934e-05,
      "loss": 2.8232,
      "step": 4585
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5787963271141052,
      "learning_rate": 9.968990762890986e-05,
      "loss": 2.8329,
      "step": 4590
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5522583723068237,
      "learning_rate": 9.968921733222425e-05,
      "loss": 2.8434,
      "step": 4595
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.583321750164032,
      "learning_rate": 9.968852627045313e-05,
      "loss": 2.623,
      "step": 4600
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5671818852424622,
      "learning_rate": 9.968783444360716e-05,
      "loss": 2.8284,
      "step": 4605
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5284412503242493,
      "learning_rate": 9.968714185169698e-05,
      "loss": 2.7933,
      "step": 4610
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.507300853729248,
      "learning_rate": 9.968644849473326e-05,
      "loss": 2.7725,
      "step": 4615
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5488023161888123,
      "learning_rate": 9.968575437272667e-05,
      "loss": 2.7857,
      "step": 4620
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5575252175331116,
      "learning_rate": 9.968505948568792e-05,
      "loss": 2.7634,
      "step": 4625
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.573759913444519,
      "learning_rate": 9.968436383362766e-05,
      "loss": 2.7872,
      "step": 4630
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.56376051902771,
      "learning_rate": 9.968366741655663e-05,
      "loss": 2.7563,
      "step": 4635
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.535944938659668,
      "learning_rate": 9.968297023448558e-05,
      "loss": 2.7037,
      "step": 4640
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6488813161849976,
      "learning_rate": 9.96822722874252e-05,
      "loss": 2.714,
      "step": 4645
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5544204711914062,
      "learning_rate": 9.968157357538625e-05,
      "loss": 2.8622,
      "step": 4650
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5874863862991333,
      "learning_rate": 9.968087409837951e-05,
      "loss": 2.7744,
      "step": 4655
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5506981015205383,
      "learning_rate": 9.968017385641574e-05,
      "loss": 2.7766,
      "step": 4660
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.537937343120575,
      "learning_rate": 9.967947284950569e-05,
      "loss": 2.8563,
      "step": 4665
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7004063725471497,
      "learning_rate": 9.967877107766021e-05,
      "loss": 2.7378,
      "step": 4670
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.545401394367218,
      "learning_rate": 9.967806854089006e-05,
      "loss": 2.8209,
      "step": 4675
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5579882264137268,
      "learning_rate": 9.967736523920607e-05,
      "loss": 2.8242,
      "step": 4680
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.54048752784729,
      "learning_rate": 9.967666117261907e-05,
      "loss": 2.7923,
      "step": 4685
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5492881536483765,
      "learning_rate": 9.96759563411399e-05,
      "loss": 2.7733,
      "step": 4690
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5219313502311707,
      "learning_rate": 9.967525074477943e-05,
      "loss": 2.7983,
      "step": 4695
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5922059416770935,
      "learning_rate": 9.96745443835485e-05,
      "loss": 2.7228,
      "step": 4700
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5445477366447449,
      "learning_rate": 9.967383725745799e-05,
      "loss": 2.7869,
      "step": 4705
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5605489611625671,
      "learning_rate": 9.96731293665188e-05,
      "loss": 2.6417,
      "step": 4710
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5872758626937866,
      "learning_rate": 9.967242071074181e-05,
      "loss": 2.7561,
      "step": 4715
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5693061351776123,
      "learning_rate": 9.967171129013797e-05,
      "loss": 2.7679,
      "step": 4720
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5245131850242615,
      "learning_rate": 9.967100110471817e-05,
      "loss": 2.8069,
      "step": 4725
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6698682904243469,
      "learning_rate": 9.967029015449334e-05,
      "loss": 2.718,
      "step": 4730
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5695157647132874,
      "learning_rate": 9.966957843947444e-05,
      "loss": 2.7778,
      "step": 4735
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5457557439804077,
      "learning_rate": 9.966886595967243e-05,
      "loss": 2.669,
      "step": 4740
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6085060834884644,
      "learning_rate": 9.966815271509828e-05,
      "loss": 2.7612,
      "step": 4745
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5825067162513733,
      "learning_rate": 9.966743870576297e-05,
      "loss": 2.7551,
      "step": 4750
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5313762426376343,
      "learning_rate": 9.96667239316775e-05,
      "loss": 2.6435,
      "step": 4755
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5089870691299438,
      "learning_rate": 9.966600839285286e-05,
      "loss": 2.6244,
      "step": 4760
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.566623330116272,
      "learning_rate": 9.966529208930008e-05,
      "loss": 2.6657,
      "step": 4765
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5882644653320312,
      "learning_rate": 9.96645750210302e-05,
      "loss": 2.7522,
      "step": 4770
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5824774503707886,
      "learning_rate": 9.966385718805423e-05,
      "loss": 2.7479,
      "step": 4775
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5471545457839966,
      "learning_rate": 9.966313859038325e-05,
      "loss": 2.8032,
      "step": 4780
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5728386640548706,
      "learning_rate": 9.966241922802832e-05,
      "loss": 2.7007,
      "step": 4785
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5718958973884583,
      "learning_rate": 9.96616991010005e-05,
      "loss": 2.7808,
      "step": 4790
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5713360905647278,
      "learning_rate": 9.96609782093109e-05,
      "loss": 2.7496,
      "step": 4795
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5043028593063354,
      "learning_rate": 9.96602565529706e-05,
      "loss": 2.7409,
      "step": 4800
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5383256673812866,
      "learning_rate": 9.965953413199071e-05,
      "loss": 2.7359,
      "step": 4805
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0209953784942627,
      "learning_rate": 9.965881094638238e-05,
      "loss": 2.6813,
      "step": 4810
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5970777273178101,
      "learning_rate": 9.965808699615674e-05,
      "loss": 2.8105,
      "step": 4815
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5963553786277771,
      "learning_rate": 9.965736228132492e-05,
      "loss": 2.7847,
      "step": 4820
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5277134776115417,
      "learning_rate": 9.965663680189807e-05,
      "loss": 2.6918,
      "step": 4825
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5293276309967041,
      "learning_rate": 9.965591055788738e-05,
      "loss": 2.7735,
      "step": 4830
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5266904830932617,
      "learning_rate": 9.965518354930405e-05,
      "loss": 2.8158,
      "step": 4835
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5426038503646851,
      "learning_rate": 9.965445577615922e-05,
      "loss": 2.8088,
      "step": 4840
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5775620937347412,
      "learning_rate": 9.965372723846415e-05,
      "loss": 2.735,
      "step": 4845
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.551426112651825,
      "learning_rate": 9.965299793623003e-05,
      "loss": 2.8369,
      "step": 4850
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6450952887535095,
      "learning_rate": 9.96522678694681e-05,
      "loss": 2.6815,
      "step": 4855
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5293461680412292,
      "learning_rate": 9.965153703818959e-05,
      "loss": 2.7568,
      "step": 4860
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5016167759895325,
      "learning_rate": 9.965080544240575e-05,
      "loss": 2.7636,
      "step": 4865
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5262230634689331,
      "learning_rate": 9.965007308212786e-05,
      "loss": 2.776,
      "step": 4870
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5402688980102539,
      "learning_rate": 9.964933995736721e-05,
      "loss": 2.8093,
      "step": 4875
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6281560659408569,
      "learning_rate": 9.964860606813504e-05,
      "loss": 2.7251,
      "step": 4880
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5873793363571167,
      "learning_rate": 9.96478714144427e-05,
      "loss": 2.7274,
      "step": 4885
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7873474359512329,
      "learning_rate": 9.964713599630146e-05,
      "loss": 2.847,
      "step": 4890
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5547282695770264,
      "learning_rate": 9.964639981372267e-05,
      "loss": 2.6165,
      "step": 4895
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5535411834716797,
      "learning_rate": 9.964566286671766e-05,
      "loss": 2.7041,
      "step": 4900
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6280202865600586,
      "learning_rate": 9.964492515529777e-05,
      "loss": 2.7921,
      "step": 4905
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6017524003982544,
      "learning_rate": 9.964418667947438e-05,
      "loss": 2.8147,
      "step": 4910
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5106446146965027,
      "learning_rate": 9.964344743925884e-05,
      "loss": 2.7312,
      "step": 4915
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6104702353477478,
      "learning_rate": 9.964270743466253e-05,
      "loss": 2.8102,
      "step": 4920
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5727868676185608,
      "learning_rate": 9.964196666569686e-05,
      "loss": 2.7346,
      "step": 4925
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5588822364807129,
      "learning_rate": 9.964122513237322e-05,
      "loss": 2.7909,
      "step": 4930
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5001404881477356,
      "learning_rate": 9.964048283470305e-05,
      "loss": 2.7202,
      "step": 4935
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6987648606300354,
      "learning_rate": 9.963973977269776e-05,
      "loss": 2.7561,
      "step": 4940
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5384222269058228,
      "learning_rate": 9.963899594636879e-05,
      "loss": 2.7223,
      "step": 4945
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7603855133056641,
      "learning_rate": 9.96382513557276e-05,
      "loss": 2.7396,
      "step": 4950
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5654938817024231,
      "learning_rate": 9.963750600078565e-05,
      "loss": 2.8103,
      "step": 4955
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5289061665534973,
      "learning_rate": 9.963675988155443e-05,
      "loss": 2.7967,
      "step": 4960
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5846783518791199,
      "learning_rate": 9.963601299804541e-05,
      "loss": 2.7302,
      "step": 4965
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5168123841285706,
      "learning_rate": 9.963526535027012e-05,
      "loss": 2.8218,
      "step": 4970
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5388414859771729,
      "learning_rate": 9.963451693824003e-05,
      "loss": 2.7195,
      "step": 4975
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.591762125492096,
      "learning_rate": 9.963376776196669e-05,
      "loss": 2.7447,
      "step": 4980
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5367181301116943,
      "learning_rate": 9.963301782146163e-05,
      "loss": 2.6664,
      "step": 4985
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5322294235229492,
      "learning_rate": 9.96322671167364e-05,
      "loss": 2.6932,
      "step": 4990
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6408253312110901,
      "learning_rate": 9.963151564780256e-05,
      "loss": 2.7016,
      "step": 4995
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5248957872390747,
      "learning_rate": 9.963076341467165e-05,
      "loss": 2.7637,
      "step": 5000
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5231680274009705,
      "learning_rate": 9.963001041735531e-05,
      "loss": 2.7406,
      "step": 5005
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5237470865249634,
      "learning_rate": 9.962925665586509e-05,
      "loss": 2.7424,
      "step": 5010
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.561295211315155,
      "learning_rate": 9.962850213021262e-05,
      "loss": 2.7329,
      "step": 5015
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5227664113044739,
      "learning_rate": 9.96277468404095e-05,
      "loss": 2.8499,
      "step": 5020
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5117846131324768,
      "learning_rate": 9.962699078646735e-05,
      "loss": 2.81,
      "step": 5025
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.539286196231842,
      "learning_rate": 9.962623396839785e-05,
      "loss": 2.7591,
      "step": 5030
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5814435482025146,
      "learning_rate": 9.962547638621262e-05,
      "loss": 2.7402,
      "step": 5035
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5484329462051392,
      "learning_rate": 9.962471803992334e-05,
      "loss": 2.7494,
      "step": 5040
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6368593573570251,
      "learning_rate": 9.962395892954168e-05,
      "loss": 2.792,
      "step": 5045
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5759376287460327,
      "learning_rate": 9.962319905507934e-05,
      "loss": 2.692,
      "step": 5050
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5254651308059692,
      "learning_rate": 9.9622438416548e-05,
      "loss": 2.7008,
      "step": 5055
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5662387013435364,
      "learning_rate": 9.962167701395937e-05,
      "loss": 2.7567,
      "step": 5060
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6001858711242676,
      "learning_rate": 9.962091484732521e-05,
      "loss": 2.7894,
      "step": 5065
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5492300391197205,
      "learning_rate": 9.962015191665722e-05,
      "loss": 2.8107,
      "step": 5070
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5882088541984558,
      "learning_rate": 9.961938822196716e-05,
      "loss": 2.7519,
      "step": 5075
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5086878538131714,
      "learning_rate": 9.961862376326678e-05,
      "loss": 2.7251,
      "step": 5080
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5579080581665039,
      "learning_rate": 9.961785854056788e-05,
      "loss": 2.6696,
      "step": 5085
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5423374772071838,
      "learning_rate": 9.961709255388221e-05,
      "loss": 2.6632,
      "step": 5090
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5260607004165649,
      "learning_rate": 9.961632580322156e-05,
      "loss": 2.8024,
      "step": 5095
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.571349561214447,
      "learning_rate": 9.961555828859777e-05,
      "loss": 2.8285,
      "step": 5100
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5310987234115601,
      "learning_rate": 9.961479001002264e-05,
      "loss": 2.7752,
      "step": 5105
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5772765874862671,
      "learning_rate": 9.961402096750799e-05,
      "loss": 2.6274,
      "step": 5110
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5339385867118835,
      "learning_rate": 9.961325116106566e-05,
      "loss": 2.6201,
      "step": 5115
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49063020944595337,
      "learning_rate": 9.961248059070751e-05,
      "loss": 2.7844,
      "step": 5120
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5202596783638,
      "learning_rate": 9.961170925644542e-05,
      "loss": 2.7642,
      "step": 5125
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5302148461341858,
      "learning_rate": 9.961093715829127e-05,
      "loss": 2.688,
      "step": 5130
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5084593892097473,
      "learning_rate": 9.961016429625691e-05,
      "loss": 2.7705,
      "step": 5135
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5014408230781555,
      "learning_rate": 9.960939067035426e-05,
      "loss": 2.7312,
      "step": 5140
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5089525580406189,
      "learning_rate": 9.960861628059524e-05,
      "loss": 2.8446,
      "step": 5145
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.505311906337738,
      "learning_rate": 9.960784112699177e-05,
      "loss": 2.7223,
      "step": 5150
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5711766481399536,
      "learning_rate": 9.960706520955578e-05,
      "loss": 2.6587,
      "step": 5155
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5582578182220459,
      "learning_rate": 9.960628852829923e-05,
      "loss": 2.813,
      "step": 5160
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4905920624732971,
      "learning_rate": 9.960551108323406e-05,
      "loss": 2.6915,
      "step": 5165
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5299171209335327,
      "learning_rate": 9.960473287437224e-05,
      "loss": 2.7235,
      "step": 5170
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5168396830558777,
      "learning_rate": 9.960395390172579e-05,
      "loss": 2.7671,
      "step": 5175
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5302513837814331,
      "learning_rate": 9.960317416530665e-05,
      "loss": 2.7204,
      "step": 5180
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5288262367248535,
      "learning_rate": 9.960239366512686e-05,
      "loss": 2.7509,
      "step": 5185
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49829941987991333,
      "learning_rate": 9.960161240119843e-05,
      "loss": 2.6148,
      "step": 5190
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5136855244636536,
      "learning_rate": 9.96008303735334e-05,
      "loss": 2.7632,
      "step": 5195
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5380761027336121,
      "learning_rate": 9.960004758214379e-05,
      "loss": 2.6939,
      "step": 5200
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5969657897949219,
      "learning_rate": 9.959926402704165e-05,
      "loss": 2.7717,
      "step": 5205
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5580558180809021,
      "learning_rate": 9.959847970823908e-05,
      "loss": 2.7782,
      "step": 5210
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6690155267715454,
      "learning_rate": 9.959769462574813e-05,
      "loss": 2.7074,
      "step": 5215
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5459215641021729,
      "learning_rate": 9.959690877958088e-05,
      "loss": 2.7853,
      "step": 5220
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5472571849822998,
      "learning_rate": 9.959612216974946e-05,
      "loss": 2.7407,
      "step": 5225
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5172734260559082,
      "learning_rate": 9.959533479626596e-05,
      "loss": 2.6807,
      "step": 5230
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5462640523910522,
      "learning_rate": 9.95945466591425e-05,
      "loss": 2.6365,
      "step": 5235
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5166212916374207,
      "learning_rate": 9.959375775839124e-05,
      "loss": 2.8197,
      "step": 5240
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5435940027236938,
      "learning_rate": 9.95929680940243e-05,
      "loss": 2.7882,
      "step": 5245
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5087002515792847,
      "learning_rate": 9.959217766605384e-05,
      "loss": 2.6862,
      "step": 5250
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48126500844955444,
      "learning_rate": 9.959138647449205e-05,
      "loss": 2.6743,
      "step": 5255
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5846558809280396,
      "learning_rate": 9.95905945193511e-05,
      "loss": 2.71,
      "step": 5260
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5228373408317566,
      "learning_rate": 9.958980180064318e-05,
      "loss": 2.8978,
      "step": 5265
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5424765348434448,
      "learning_rate": 9.958900831838051e-05,
      "loss": 2.8534,
      "step": 5270
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5961100459098816,
      "learning_rate": 9.958821407257531e-05,
      "loss": 2.7497,
      "step": 5275
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4956708252429962,
      "learning_rate": 9.958741906323978e-05,
      "loss": 2.7419,
      "step": 5280
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5269626975059509,
      "learning_rate": 9.95866232903862e-05,
      "loss": 2.6809,
      "step": 5285
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5023139715194702,
      "learning_rate": 9.958582675402677e-05,
      "loss": 2.6911,
      "step": 5290
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5249695181846619,
      "learning_rate": 9.958502945417382e-05,
      "loss": 2.7571,
      "step": 5295
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.497692734003067,
      "learning_rate": 9.958423139083958e-05,
      "loss": 2.7346,
      "step": 5300
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5567381381988525,
      "learning_rate": 9.958343256403635e-05,
      "loss": 2.648,
      "step": 5305
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49906060099601746,
      "learning_rate": 9.958263297377643e-05,
      "loss": 2.6477,
      "step": 5310
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5152696371078491,
      "learning_rate": 9.958183262007215e-05,
      "loss": 2.6529,
      "step": 5315
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4817773699760437,
      "learning_rate": 9.95810315029358e-05,
      "loss": 2.6552,
      "step": 5320
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5372611880302429,
      "learning_rate": 9.958022962237974e-05,
      "loss": 2.7304,
      "step": 5325
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5765663385391235,
      "learning_rate": 9.957942697841632e-05,
      "loss": 2.7279,
      "step": 5330
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5257022976875305,
      "learning_rate": 9.957862357105785e-05,
      "loss": 2.7659,
      "step": 5335
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5349875688552856,
      "learning_rate": 9.957781940031676e-05,
      "loss": 2.7344,
      "step": 5340
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5753257870674133,
      "learning_rate": 9.957701446620543e-05,
      "loss": 2.6098,
      "step": 5345
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.512721836566925,
      "learning_rate": 9.957620876873621e-05,
      "loss": 2.6565,
      "step": 5350
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7526594400405884,
      "learning_rate": 9.957540230792152e-05,
      "loss": 2.7479,
      "step": 5355
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5500587821006775,
      "learning_rate": 9.957459508377381e-05,
      "loss": 2.7352,
      "step": 5360
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1481915712356567,
      "learning_rate": 9.957378709630548e-05,
      "loss": 2.7521,
      "step": 5365
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5022457838058472,
      "learning_rate": 9.957297834552898e-05,
      "loss": 2.713,
      "step": 5370
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5670760273933411,
      "learning_rate": 9.957216883145675e-05,
      "loss": 2.7669,
      "step": 5375
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5524647235870361,
      "learning_rate": 9.957135855410127e-05,
      "loss": 2.6799,
      "step": 5380
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5230464339256287,
      "learning_rate": 9.957054751347501e-05,
      "loss": 2.6514,
      "step": 5385
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5110558271408081,
      "learning_rate": 9.956973570959045e-05,
      "loss": 2.6728,
      "step": 5390
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5106539726257324,
      "learning_rate": 9.956892314246012e-05,
      "loss": 2.7938,
      "step": 5395
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5111415982246399,
      "learning_rate": 9.956810981209647e-05,
      "loss": 2.672,
      "step": 5400
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5018755197525024,
      "learning_rate": 9.956729571851211e-05,
      "loss": 2.6416,
      "step": 5405
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5006531476974487,
      "learning_rate": 9.956648086171949e-05,
      "loss": 2.699,
      "step": 5410
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.480538547039032,
      "learning_rate": 9.956566524173121e-05,
      "loss": 2.6907,
      "step": 5415
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5319927930831909,
      "learning_rate": 9.956484885855982e-05,
      "loss": 2.6637,
      "step": 5420
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5145319700241089,
      "learning_rate": 9.956403171221786e-05,
      "loss": 2.752,
      "step": 5425
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4778360426425934,
      "learning_rate": 9.956321380271796e-05,
      "loss": 2.7152,
      "step": 5430
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5014883875846863,
      "learning_rate": 9.956239513007268e-05,
      "loss": 2.6003,
      "step": 5435
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5141823291778564,
      "learning_rate": 9.956157569429464e-05,
      "loss": 2.6424,
      "step": 5440
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5105752348899841,
      "learning_rate": 9.956075549539643e-05,
      "loss": 2.6462,
      "step": 5445
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5390253663063049,
      "learning_rate": 9.955993453339072e-05,
      "loss": 2.7015,
      "step": 5450
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5435912609100342,
      "learning_rate": 9.955911280829011e-05,
      "loss": 2.7109,
      "step": 5455
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.511780321598053,
      "learning_rate": 9.95582903201073e-05,
      "loss": 2.706,
      "step": 5460
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.509019672870636,
      "learning_rate": 9.955746706885492e-05,
      "loss": 2.7421,
      "step": 5465
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5661734342575073,
      "learning_rate": 9.955664305454564e-05,
      "loss": 2.7085,
      "step": 5470
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5272811651229858,
      "learning_rate": 9.955581827719216e-05,
      "loss": 2.811,
      "step": 5475
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5376253724098206,
      "learning_rate": 9.955499273680719e-05,
      "loss": 2.7101,
      "step": 5480
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4983501434326172,
      "learning_rate": 9.955416643340343e-05,
      "loss": 2.7288,
      "step": 5485
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.521128237247467,
      "learning_rate": 9.95533393669936e-05,
      "loss": 2.6091,
      "step": 5490
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5185613632202148,
      "learning_rate": 9.955251153759044e-05,
      "loss": 2.7163,
      "step": 5495
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.476553350687027,
      "learning_rate": 9.955168294520672e-05,
      "loss": 2.6323,
      "step": 5500
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5548961758613586,
      "learning_rate": 9.955085358985514e-05,
      "loss": 2.6392,
      "step": 5505
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4915710389614105,
      "learning_rate": 9.955002347154851e-05,
      "loss": 2.7042,
      "step": 5510
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5378808379173279,
      "learning_rate": 9.954919259029962e-05,
      "loss": 2.6738,
      "step": 5515
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5679003596305847,
      "learning_rate": 9.954836094612123e-05,
      "loss": 2.6362,
      "step": 5520
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5377044677734375,
      "learning_rate": 9.954752853902619e-05,
      "loss": 2.607,
      "step": 5525
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5028848648071289,
      "learning_rate": 9.954669536902728e-05,
      "loss": 2.6227,
      "step": 5530
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5113602876663208,
      "learning_rate": 9.954586143613733e-05,
      "loss": 2.7084,
      "step": 5535
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5313628911972046,
      "learning_rate": 9.954502674036919e-05,
      "loss": 2.8071,
      "step": 5540
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4850282669067383,
      "learning_rate": 9.954419128173572e-05,
      "loss": 2.7298,
      "step": 5545
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5150644183158875,
      "learning_rate": 9.954335506024978e-05,
      "loss": 2.7897,
      "step": 5550
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48542383313179016,
      "learning_rate": 9.954251807592423e-05,
      "loss": 2.6255,
      "step": 5555
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49717721343040466,
      "learning_rate": 9.954168032877198e-05,
      "loss": 2.751,
      "step": 5560
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5661787986755371,
      "learning_rate": 9.95408418188059e-05,
      "loss": 2.6381,
      "step": 5565
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49290910363197327,
      "learning_rate": 9.954000254603894e-05,
      "loss": 2.7175,
      "step": 5570
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5620985627174377,
      "learning_rate": 9.9539162510484e-05,
      "loss": 2.6391,
      "step": 5575
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5258685946464539,
      "learning_rate": 9.953832171215401e-05,
      "loss": 2.5463,
      "step": 5580
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5371032953262329,
      "learning_rate": 9.953748015106193e-05,
      "loss": 2.6665,
      "step": 5585
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.51180499792099,
      "learning_rate": 9.953663782722069e-05,
      "loss": 2.6791,
      "step": 5590
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4978950023651123,
      "learning_rate": 9.953579474064329e-05,
      "loss": 2.7836,
      "step": 5595
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.49840787053108215,
      "learning_rate": 9.95349508913427e-05,
      "loss": 2.7102,
      "step": 5600
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5057438015937805,
      "learning_rate": 9.953410627933193e-05,
      "loss": 2.62,
      "step": 5605
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5040857195854187,
      "learning_rate": 9.953326090462397e-05,
      "loss": 2.6428,
      "step": 5610
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.498320996761322,
      "learning_rate": 9.953241476723181e-05,
      "loss": 2.6557,
      "step": 5615
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6283282041549683,
      "learning_rate": 9.953156786716851e-05,
      "loss": 2.6652,
      "step": 5620
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5339922308921814,
      "learning_rate": 9.953072020444711e-05,
      "loss": 2.7635,
      "step": 5625
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5136678814888,
      "learning_rate": 9.952987177908066e-05,
      "loss": 2.7086,
      "step": 5630
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4715080261230469,
      "learning_rate": 9.95290225910822e-05,
      "loss": 2.6675,
      "step": 5635
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4711659848690033,
      "learning_rate": 9.952817264046486e-05,
      "loss": 2.5351,
      "step": 5640
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5061845183372498,
      "learning_rate": 9.952732192724167e-05,
      "loss": 2.6558,
      "step": 5645
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5535271763801575,
      "learning_rate": 9.952647045142576e-05,
      "loss": 2.6328,
      "step": 5650
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.505882203578949,
      "learning_rate": 9.952561821303021e-05,
      "loss": 2.5517,
      "step": 5655
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47253355383872986,
      "learning_rate": 9.952476521206818e-05,
      "loss": 2.6813,
      "step": 5660
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5103576183319092,
      "learning_rate": 9.952391144855278e-05,
      "loss": 2.592,
      "step": 5665
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4871288239955902,
      "learning_rate": 9.952305692249717e-05,
      "loss": 2.7035,
      "step": 5670
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.546707034111023,
      "learning_rate": 9.952220163391451e-05,
      "loss": 2.5962,
      "step": 5675
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5110735893249512,
      "learning_rate": 9.952134558281793e-05,
      "loss": 2.7084,
      "step": 5680
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48624053597450256,
      "learning_rate": 9.952048876922066e-05,
      "loss": 2.6947,
      "step": 5685
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5251468420028687,
      "learning_rate": 9.951963119313588e-05,
      "loss": 2.597,
      "step": 5690
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5881426930427551,
      "learning_rate": 9.951877285457678e-05,
      "loss": 2.6173,
      "step": 5695
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5024569034576416,
      "learning_rate": 9.951791375355659e-05,
      "loss": 2.6946,
      "step": 5700
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5614703297615051,
      "learning_rate": 9.951705389008852e-05,
      "loss": 2.6263,
      "step": 5705
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4935329258441925,
      "learning_rate": 9.951619326418582e-05,
      "loss": 2.6885,
      "step": 5710
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5034992694854736,
      "learning_rate": 9.951533187586178e-05,
      "loss": 2.6331,
      "step": 5715
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47869011759757996,
      "learning_rate": 9.951446972512958e-05,
      "loss": 2.7134,
      "step": 5720
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49965810775756836,
      "learning_rate": 9.951360681200256e-05,
      "loss": 2.7143,
      "step": 5725
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5102935433387756,
      "learning_rate": 9.9512743136494e-05,
      "loss": 2.6319,
      "step": 5730
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5097643733024597,
      "learning_rate": 9.951187869861715e-05,
      "loss": 2.5698,
      "step": 5735
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5126561522483826,
      "learning_rate": 9.951101349838538e-05,
      "loss": 2.5891,
      "step": 5740
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5092840790748596,
      "learning_rate": 9.9510147535812e-05,
      "loss": 2.7026,
      "step": 5745
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6021207571029663,
      "learning_rate": 9.950928081091032e-05,
      "loss": 2.6857,
      "step": 5750
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4858684539794922,
      "learning_rate": 9.95084133236937e-05,
      "loss": 2.6455,
      "step": 5755
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.533289909362793,
      "learning_rate": 9.950754507417547e-05,
      "loss": 2.7041,
      "step": 5760
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.491449773311615,
      "learning_rate": 9.950667606236906e-05,
      "loss": 2.6527,
      "step": 5765
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5010853409767151,
      "learning_rate": 9.95058062882878e-05,
      "loss": 2.6802,
      "step": 5770
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5749230980873108,
      "learning_rate": 9.950493575194509e-05,
      "loss": 2.6241,
      "step": 5775
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4837588667869568,
      "learning_rate": 9.950406445335434e-05,
      "loss": 2.6887,
      "step": 5780
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48963940143585205,
      "learning_rate": 9.950319239252896e-05,
      "loss": 2.5256,
      "step": 5785
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5167273879051208,
      "learning_rate": 9.95023195694824e-05,
      "loss": 2.6493,
      "step": 5790
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5103415250778198,
      "learning_rate": 9.950144598422809e-05,
      "loss": 2.699,
      "step": 5795
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49510446190834045,
      "learning_rate": 9.950057163677944e-05,
      "loss": 2.5758,
      "step": 5800
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5099479556083679,
      "learning_rate": 9.949969652714997e-05,
      "loss": 2.6687,
      "step": 5805
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4803939759731293,
      "learning_rate": 9.949882065535312e-05,
      "loss": 2.6765,
      "step": 5810
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49504324793815613,
      "learning_rate": 9.949794402140239e-05,
      "loss": 2.6799,
      "step": 5815
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4860273003578186,
      "learning_rate": 9.949706662531127e-05,
      "loss": 2.6063,
      "step": 5820
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4922538995742798,
      "learning_rate": 9.949618846709327e-05,
      "loss": 2.6429,
      "step": 5825
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5439775586128235,
      "learning_rate": 9.949530954676194e-05,
      "loss": 2.7456,
      "step": 5830
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4888339340686798,
      "learning_rate": 9.949442986433076e-05,
      "loss": 2.687,
      "step": 5835
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5494887232780457,
      "learning_rate": 9.949354941981332e-05,
      "loss": 2.6055,
      "step": 5840
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47308313846588135,
      "learning_rate": 9.949266821322315e-05,
      "loss": 2.6283,
      "step": 5845
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49347397685050964,
      "learning_rate": 9.949178624457384e-05,
      "loss": 2.6372,
      "step": 5850
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5252546668052673,
      "learning_rate": 9.949090351387895e-05,
      "loss": 2.6766,
      "step": 5855
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5734273195266724,
      "learning_rate": 9.949002002115208e-05,
      "loss": 2.6759,
      "step": 5860
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5061395764350891,
      "learning_rate": 9.948913576640683e-05,
      "loss": 2.688,
      "step": 5865
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6558437347412109,
      "learning_rate": 9.948825074965682e-05,
      "loss": 2.6194,
      "step": 5870
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.514255166053772,
      "learning_rate": 9.948736497091569e-05,
      "loss": 2.6111,
      "step": 5875
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47513920068740845,
      "learning_rate": 9.948647843019705e-05,
      "loss": 2.6528,
      "step": 5880
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5274945497512817,
      "learning_rate": 9.948559112751456e-05,
      "loss": 2.6674,
      "step": 5885
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49536779522895813,
      "learning_rate": 9.948470306288189e-05,
      "loss": 2.6712,
      "step": 5890
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47468119859695435,
      "learning_rate": 9.948381423631273e-05,
      "loss": 2.6323,
      "step": 5895
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49464115500450134,
      "learning_rate": 9.948292464782072e-05,
      "loss": 2.7297,
      "step": 5900
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4860682189464569,
      "learning_rate": 9.948203429741961e-05,
      "loss": 2.6906,
      "step": 5905
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4750114679336548,
      "learning_rate": 9.948114318512307e-05,
      "loss": 2.7054,
      "step": 5910
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5355490446090698,
      "learning_rate": 9.948025131094482e-05,
      "loss": 2.7178,
      "step": 5915
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49242350459098816,
      "learning_rate": 9.947935867489862e-05,
      "loss": 2.5936,
      "step": 5920
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7153792381286621,
      "learning_rate": 9.94784652769982e-05,
      "loss": 2.6677,
      "step": 5925
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5207790732383728,
      "learning_rate": 9.947757111725732e-05,
      "loss": 2.6358,
      "step": 5930
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5126029849052429,
      "learning_rate": 9.947667619568974e-05,
      "loss": 2.6614,
      "step": 5935
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45663246512413025,
      "learning_rate": 9.947578051230925e-05,
      "loss": 2.659,
      "step": 5940
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47028765082359314,
      "learning_rate": 9.947488406712962e-05,
      "loss": 2.6448,
      "step": 5945
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.486018568277359,
      "learning_rate": 9.947398686016469e-05,
      "loss": 2.6298,
      "step": 5950
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4969647228717804,
      "learning_rate": 9.947308889142824e-05,
      "loss": 2.6389,
      "step": 5955
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4850294888019562,
      "learning_rate": 9.94721901609341e-05,
      "loss": 2.6313,
      "step": 5960
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5132781863212585,
      "learning_rate": 9.947129066869614e-05,
      "loss": 2.6457,
      "step": 5965
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5008549690246582,
      "learning_rate": 9.947039041472817e-05,
      "loss": 2.6082,
      "step": 5970
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.471832275390625,
      "learning_rate": 9.946948939904407e-05,
      "loss": 2.5996,
      "step": 5975
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46899718046188354,
      "learning_rate": 9.946858762165772e-05,
      "loss": 2.6334,
      "step": 5980
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5146117210388184,
      "learning_rate": 9.946768508258299e-05,
      "loss": 2.692,
      "step": 5985
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48759105801582336,
      "learning_rate": 9.946678178183377e-05,
      "loss": 2.6078,
      "step": 5990
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5333221554756165,
      "learning_rate": 9.946587771942401e-05,
      "loss": 2.6501,
      "step": 5995
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4601440727710724,
      "learning_rate": 9.946497289536758e-05,
      "loss": 2.704,
      "step": 6000
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4740259051322937,
      "learning_rate": 9.946406730967846e-05,
      "loss": 2.6049,
      "step": 6005
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5301644802093506,
      "learning_rate": 9.946316096237054e-05,
      "loss": 2.6083,
      "step": 6010
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48974570631980896,
      "learning_rate": 9.946225385345782e-05,
      "loss": 2.5458,
      "step": 6015
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.51124507188797,
      "learning_rate": 9.946134598295424e-05,
      "loss": 2.5091,
      "step": 6020
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45475339889526367,
      "learning_rate": 9.94604373508738e-05,
      "loss": 2.6739,
      "step": 6025
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8260575532913208,
      "learning_rate": 9.945952795723048e-05,
      "loss": 2.6665,
      "step": 6030
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4534899592399597,
      "learning_rate": 9.945861780203828e-05,
      "loss": 2.5989,
      "step": 6035
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48195523023605347,
      "learning_rate": 9.945770688531122e-05,
      "loss": 2.6109,
      "step": 6040
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5050069093704224,
      "learning_rate": 9.945679520706332e-05,
      "loss": 2.6108,
      "step": 6045
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5323428511619568,
      "learning_rate": 9.945588276730862e-05,
      "loss": 2.6584,
      "step": 6050
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4986659586429596,
      "learning_rate": 9.945496956606115e-05,
      "loss": 2.6554,
      "step": 6055
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5611687302589417,
      "learning_rate": 9.945405560333502e-05,
      "loss": 2.7148,
      "step": 6060
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6473665237426758,
      "learning_rate": 9.945314087914426e-05,
      "loss": 2.6533,
      "step": 6065
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4913814067840576,
      "learning_rate": 9.945222539350296e-05,
      "loss": 2.6279,
      "step": 6070
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4864082634449005,
      "learning_rate": 9.945130914642524e-05,
      "loss": 2.6014,
      "step": 6075
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5154943466186523,
      "learning_rate": 9.945039213792518e-05,
      "loss": 2.6442,
      "step": 6080
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5073487162590027,
      "learning_rate": 9.94494743680169e-05,
      "loss": 2.666,
      "step": 6085
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4942362904548645,
      "learning_rate": 9.944855583671455e-05,
      "loss": 2.6114,
      "step": 6090
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5035442113876343,
      "learning_rate": 9.944763654403226e-05,
      "loss": 2.6378,
      "step": 6095
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5317345857620239,
      "learning_rate": 9.94467164899842e-05,
      "loss": 2.5793,
      "step": 6100
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5874106287956238,
      "learning_rate": 9.944579567458452e-05,
      "loss": 2.6067,
      "step": 6105
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0532342195510864,
      "learning_rate": 9.944487409784741e-05,
      "loss": 2.6808,
      "step": 6110
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5187524557113647,
      "learning_rate": 9.944395175978705e-05,
      "loss": 2.6333,
      "step": 6115
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4909622073173523,
      "learning_rate": 9.944302866041764e-05,
      "loss": 2.6479,
      "step": 6120
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5116632580757141,
      "learning_rate": 9.944210479975342e-05,
      "loss": 2.5839,
      "step": 6125
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4872247576713562,
      "learning_rate": 9.944118017780857e-05,
      "loss": 2.6525,
      "step": 6130
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48204079270362854,
      "learning_rate": 9.944025479459735e-05,
      "loss": 2.7191,
      "step": 6135
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4488745927810669,
      "learning_rate": 9.943932865013404e-05,
      "loss": 2.509,
      "step": 6140
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4876132011413574,
      "learning_rate": 9.943840174443284e-05,
      "loss": 2.7668,
      "step": 6145
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4810884892940521,
      "learning_rate": 9.943747407750805e-05,
      "loss": 2.5987,
      "step": 6150
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5326345562934875,
      "learning_rate": 9.943654564937397e-05,
      "loss": 2.5928,
      "step": 6155
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5037736892700195,
      "learning_rate": 9.943561646004488e-05,
      "loss": 2.6348,
      "step": 6160
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4748021066188812,
      "learning_rate": 9.943468650953508e-05,
      "loss": 2.6273,
      "step": 6165
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5023642778396606,
      "learning_rate": 9.943375579785889e-05,
      "loss": 2.6364,
      "step": 6170
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8739017248153687,
      "learning_rate": 9.943282432503067e-05,
      "loss": 2.5945,
      "step": 6175
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5549447536468506,
      "learning_rate": 9.943189209106472e-05,
      "loss": 2.5287,
      "step": 6180
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5027145743370056,
      "learning_rate": 9.943095909597541e-05,
      "loss": 2.6572,
      "step": 6185
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5233054757118225,
      "learning_rate": 9.943002533977713e-05,
      "loss": 2.5225,
      "step": 6190
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5048193335533142,
      "learning_rate": 9.942909082248423e-05,
      "loss": 2.653,
      "step": 6195
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4964543282985687,
      "learning_rate": 9.942815554411109e-05,
      "loss": 2.5714,
      "step": 6200
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.487652987241745,
      "learning_rate": 9.942721950467214e-05,
      "loss": 2.6358,
      "step": 6205
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4709979295730591,
      "learning_rate": 9.942628270418177e-05,
      "loss": 2.6477,
      "step": 6210
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4704461395740509,
      "learning_rate": 9.942534514265442e-05,
      "loss": 2.644,
      "step": 6215
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4479375183582306,
      "learning_rate": 9.942440682010452e-05,
      "loss": 2.7008,
      "step": 6220
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.50529944896698,
      "learning_rate": 9.942346773654652e-05,
      "loss": 2.7119,
      "step": 6225
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4670368432998657,
      "learning_rate": 9.942252789199487e-05,
      "loss": 2.6707,
      "step": 6230
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4763373136520386,
      "learning_rate": 9.942158728646406e-05,
      "loss": 2.598,
      "step": 6235
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5019482374191284,
      "learning_rate": 9.942064591996855e-05,
      "loss": 2.7314,
      "step": 6240
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5228628516197205,
      "learning_rate": 9.941970379252284e-05,
      "loss": 2.6975,
      "step": 6245
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4956476092338562,
      "learning_rate": 9.941876090414146e-05,
      "loss": 2.5921,
      "step": 6250
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4914722144603729,
      "learning_rate": 9.941781725483889e-05,
      "loss": 2.7319,
      "step": 6255
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49626612663269043,
      "learning_rate": 9.941687284462969e-05,
      "loss": 2.6684,
      "step": 6260
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47868168354034424,
      "learning_rate": 9.941592767352839e-05,
      "loss": 2.6936,
      "step": 6265
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47672033309936523,
      "learning_rate": 9.941498174154954e-05,
      "loss": 2.6015,
      "step": 6270
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47058621048927307,
      "learning_rate": 9.94140350487077e-05,
      "loss": 2.6404,
      "step": 6275
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5296226739883423,
      "learning_rate": 9.941308759501747e-05,
      "loss": 2.597,
      "step": 6280
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4646286070346832,
      "learning_rate": 9.941213938049342e-05,
      "loss": 2.6089,
      "step": 6285
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5315638184547424,
      "learning_rate": 9.941119040515017e-05,
      "loss": 2.649,
      "step": 6290
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49352598190307617,
      "learning_rate": 9.941024066900229e-05,
      "loss": 2.6133,
      "step": 6295
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5096608996391296,
      "learning_rate": 9.940929017206445e-05,
      "loss": 2.5617,
      "step": 6300
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4620954990386963,
      "learning_rate": 9.940833891435124e-05,
      "loss": 2.6261,
      "step": 6305
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5154588222503662,
      "learning_rate": 9.940738689587735e-05,
      "loss": 2.6571,
      "step": 6310
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47830525040626526,
      "learning_rate": 9.940643411665742e-05,
      "loss": 2.6076,
      "step": 6315
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4852476119995117,
      "learning_rate": 9.940548057670612e-05,
      "loss": 2.58,
      "step": 6320
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4795626699924469,
      "learning_rate": 9.940452627603813e-05,
      "loss": 2.7263,
      "step": 6325
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4649480879306793,
      "learning_rate": 9.940357121466813e-05,
      "loss": 2.5775,
      "step": 6330
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.443804532289505,
      "learning_rate": 9.940261539261086e-05,
      "loss": 2.6391,
      "step": 6335
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5567120313644409,
      "learning_rate": 9.940165880988102e-05,
      "loss": 2.6674,
      "step": 6340
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48200955986976624,
      "learning_rate": 9.940070146649333e-05,
      "loss": 2.5898,
      "step": 6345
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4595739543437958,
      "learning_rate": 9.939974336246254e-05,
      "loss": 2.6071,
      "step": 6350
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5076342821121216,
      "learning_rate": 9.93987844978034e-05,
      "loss": 2.5588,
      "step": 6355
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48812124133110046,
      "learning_rate": 9.939782487253067e-05,
      "loss": 2.5993,
      "step": 6360
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49465906620025635,
      "learning_rate": 9.939686448665915e-05,
      "loss": 2.6227,
      "step": 6365
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5217050909996033,
      "learning_rate": 9.939590334020359e-05,
      "loss": 2.604,
      "step": 6370
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5447195172309875,
      "learning_rate": 9.939494143317881e-05,
      "loss": 2.512,
      "step": 6375
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44812533259391785,
      "learning_rate": 9.939397876559962e-05,
      "loss": 2.63,
      "step": 6380
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44782909750938416,
      "learning_rate": 9.939301533748084e-05,
      "loss": 2.6035,
      "step": 6385
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5258204340934753,
      "learning_rate": 9.939205114883731e-05,
      "loss": 2.6093,
      "step": 6390
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5773866176605225,
      "learning_rate": 9.939108619968386e-05,
      "loss": 2.5844,
      "step": 6395
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46234452724456787,
      "learning_rate": 9.939012049003537e-05,
      "loss": 2.6738,
      "step": 6400
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4827699661254883,
      "learning_rate": 9.93891540199067e-05,
      "loss": 2.5954,
      "step": 6405
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45628172159194946,
      "learning_rate": 9.938818678931271e-05,
      "loss": 2.5386,
      "step": 6410
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49428728222846985,
      "learning_rate": 9.938721879826832e-05,
      "loss": 2.6335,
      "step": 6415
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47251567244529724,
      "learning_rate": 9.938625004678844e-05,
      "loss": 2.615,
      "step": 6420
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5034798383712769,
      "learning_rate": 9.938528053488796e-05,
      "loss": 2.5573,
      "step": 6425
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4603208899497986,
      "learning_rate": 9.938431026258182e-05,
      "loss": 2.5775,
      "step": 6430
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44411569833755493,
      "learning_rate": 9.938333922988497e-05,
      "loss": 2.5434,
      "step": 6435
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4736400842666626,
      "learning_rate": 9.938236743681235e-05,
      "loss": 2.7272,
      "step": 6440
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4720572233200073,
      "learning_rate": 9.938139488337891e-05,
      "loss": 2.6722,
      "step": 6445
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5031664371490479,
      "learning_rate": 9.938042156959966e-05,
      "loss": 2.6989,
      "step": 6450
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46953800320625305,
      "learning_rate": 9.937944749548956e-05,
      "loss": 2.625,
      "step": 6455
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46520617604255676,
      "learning_rate": 9.937847266106359e-05,
      "loss": 2.5526,
      "step": 6460
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4614129662513733,
      "learning_rate": 9.937749706633682e-05,
      "loss": 2.636,
      "step": 6465
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45627981424331665,
      "learning_rate": 9.937652071132422e-05,
      "loss": 2.5955,
      "step": 6470
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4595586061477661,
      "learning_rate": 9.937554359604084e-05,
      "loss": 2.605,
      "step": 6475
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4893145263195038,
      "learning_rate": 9.937456572050172e-05,
      "loss": 2.6369,
      "step": 6480
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5343483090400696,
      "learning_rate": 9.937358708472193e-05,
      "loss": 2.6292,
      "step": 6485
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4517357051372528,
      "learning_rate": 9.937260768871654e-05,
      "loss": 2.6401,
      "step": 6490
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4946812391281128,
      "learning_rate": 9.93716275325006e-05,
      "loss": 2.7137,
      "step": 6495
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4635010361671448,
      "learning_rate": 9.937064661608924e-05,
      "loss": 2.5057,
      "step": 6500
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5068973898887634,
      "learning_rate": 9.936966493949752e-05,
      "loss": 2.6288,
      "step": 6505
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4592459499835968,
      "learning_rate": 9.936868250274059e-05,
      "loss": 2.5967,
      "step": 6510
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44604772329330444,
      "learning_rate": 9.936769930583359e-05,
      "loss": 2.6174,
      "step": 6515
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4490673243999481,
      "learning_rate": 9.93667153487916e-05,
      "loss": 2.6843,
      "step": 6520
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5230648517608643,
      "learning_rate": 9.936573063162984e-05,
      "loss": 2.6545,
      "step": 6525
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49705177545547485,
      "learning_rate": 9.936474515436342e-05,
      "loss": 2.5332,
      "step": 6530
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48039689660072327,
      "learning_rate": 9.936375891700753e-05,
      "loss": 2.6808,
      "step": 6535
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5146342515945435,
      "learning_rate": 9.936277191957737e-05,
      "loss": 2.5882,
      "step": 6540
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4540967345237732,
      "learning_rate": 9.936178416208813e-05,
      "loss": 2.5434,
      "step": 6545
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4543635845184326,
      "learning_rate": 9.9360795644555e-05,
      "loss": 2.5854,
      "step": 6550
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4559267461299896,
      "learning_rate": 9.935980636699322e-05,
      "loss": 2.5419,
      "step": 6555
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4426746964454651,
      "learning_rate": 9.935881632941801e-05,
      "loss": 2.601,
      "step": 6560
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4948815107345581,
      "learning_rate": 9.935782553184463e-05,
      "loss": 2.6735,
      "step": 6565
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4571368992328644,
      "learning_rate": 9.935683397428833e-05,
      "loss": 2.5908,
      "step": 6570
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5118866562843323,
      "learning_rate": 9.935584165676436e-05,
      "loss": 2.5961,
      "step": 6575
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.482772558927536,
      "learning_rate": 9.935484857928801e-05,
      "loss": 2.6865,
      "step": 6580
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4687405526638031,
      "learning_rate": 9.935385474187459e-05,
      "loss": 2.5817,
      "step": 6585
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4711369276046753,
      "learning_rate": 9.93528601445394e-05,
      "loss": 2.5427,
      "step": 6590
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4717005491256714,
      "learning_rate": 9.935186478729771e-05,
      "loss": 2.5211,
      "step": 6595
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4923033118247986,
      "learning_rate": 9.93508686701649e-05,
      "loss": 2.6404,
      "step": 6600
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47065800428390503,
      "learning_rate": 9.934987179315627e-05,
      "loss": 2.6482,
      "step": 6605
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4435714781284332,
      "learning_rate": 9.934887415628719e-05,
      "loss": 2.5154,
      "step": 6610
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47553834319114685,
      "learning_rate": 9.934787575957301e-05,
      "loss": 2.5145,
      "step": 6615
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4954455494880676,
      "learning_rate": 9.934687660302912e-05,
      "loss": 2.5873,
      "step": 6620
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4661914110183716,
      "learning_rate": 9.934587668667088e-05,
      "loss": 2.6096,
      "step": 6625
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45193225145339966,
      "learning_rate": 9.93448760105137e-05,
      "loss": 2.5979,
      "step": 6630
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45869526267051697,
      "learning_rate": 9.934387457457299e-05,
      "loss": 2.6026,
      "step": 6635
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47491827607154846,
      "learning_rate": 9.934287237886417e-05,
      "loss": 2.658,
      "step": 6640
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4344891905784607,
      "learning_rate": 9.934186942340265e-05,
      "loss": 2.5368,
      "step": 6645
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48386114835739136,
      "learning_rate": 9.934086570820392e-05,
      "loss": 2.609,
      "step": 6650
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4624238908290863,
      "learning_rate": 9.933986123328337e-05,
      "loss": 2.6158,
      "step": 6655
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5949715375900269,
      "learning_rate": 9.933885599865651e-05,
      "loss": 2.5235,
      "step": 6660
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6550095677375793,
      "learning_rate": 9.933785000433882e-05,
      "loss": 2.6235,
      "step": 6665
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4744337499141693,
      "learning_rate": 9.933684325034577e-05,
      "loss": 2.6371,
      "step": 6670
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4383944869041443,
      "learning_rate": 9.933583573669287e-05,
      "loss": 2.5887,
      "step": 6675
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4986136853694916,
      "learning_rate": 9.933482746339564e-05,
      "loss": 2.5876,
      "step": 6680
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47930264472961426,
      "learning_rate": 9.933381843046959e-05,
      "loss": 2.5204,
      "step": 6685
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45935702323913574,
      "learning_rate": 9.933280863793026e-05,
      "loss": 2.5259,
      "step": 6690
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4803405702114105,
      "learning_rate": 9.933179808579321e-05,
      "loss": 2.6937,
      "step": 6695
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49414077401161194,
      "learning_rate": 9.933078677407399e-05,
      "loss": 2.6232,
      "step": 6700
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49630555510520935,
      "learning_rate": 9.932977470278818e-05,
      "loss": 2.429,
      "step": 6705
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4609528183937073,
      "learning_rate": 9.932876187195135e-05,
      "loss": 2.6811,
      "step": 6710
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4363045394420624,
      "learning_rate": 9.932774828157911e-05,
      "loss": 2.6388,
      "step": 6715
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43772125244140625,
      "learning_rate": 9.932673393168705e-05,
      "loss": 2.5465,
      "step": 6720
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5012298226356506,
      "learning_rate": 9.932571882229081e-05,
      "loss": 2.6508,
      "step": 6725
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.508233904838562,
      "learning_rate": 9.932470295340599e-05,
      "loss": 2.5562,
      "step": 6730
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48446428775787354,
      "learning_rate": 9.932368632504826e-05,
      "loss": 2.6321,
      "step": 6735
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44289642572402954,
      "learning_rate": 9.932266893723325e-05,
      "loss": 2.5516,
      "step": 6740
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49040642380714417,
      "learning_rate": 9.932165078997666e-05,
      "loss": 2.5576,
      "step": 6745
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43649157881736755,
      "learning_rate": 9.932063188329412e-05,
      "loss": 2.5917,
      "step": 6750
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4274597764015198,
      "learning_rate": 9.931961221720136e-05,
      "loss": 2.5765,
      "step": 6755
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43549174070358276,
      "learning_rate": 9.931859179171405e-05,
      "loss": 2.5933,
      "step": 6760
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42989498376846313,
      "learning_rate": 9.931757060684792e-05,
      "loss": 2.5396,
      "step": 6765
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46395131945610046,
      "learning_rate": 9.93165486626187e-05,
      "loss": 2.5598,
      "step": 6770
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46919500827789307,
      "learning_rate": 9.93155259590421e-05,
      "loss": 2.6704,
      "step": 6775
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5224345922470093,
      "learning_rate": 9.931450249613389e-05,
      "loss": 2.565,
      "step": 6780
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4558461010456085,
      "learning_rate": 9.931347827390982e-05,
      "loss": 2.6595,
      "step": 6785
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7365400791168213,
      "learning_rate": 9.931245329238566e-05,
      "loss": 2.5487,
      "step": 6790
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4944753050804138,
      "learning_rate": 9.931142755157719e-05,
      "loss": 2.4828,
      "step": 6795
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4672546088695526,
      "learning_rate": 9.931040105150019e-05,
      "loss": 2.605,
      "step": 6800
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46416333317756653,
      "learning_rate": 9.93093737921705e-05,
      "loss": 2.5101,
      "step": 6805
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48406076431274414,
      "learning_rate": 9.93083457736039e-05,
      "loss": 2.6631,
      "step": 6810
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49696287512779236,
      "learning_rate": 9.930731699581626e-05,
      "loss": 2.5461,
      "step": 6815
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5621334314346313,
      "learning_rate": 9.930628745882339e-05,
      "loss": 2.5119,
      "step": 6820
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4359864592552185,
      "learning_rate": 9.930525716264114e-05,
      "loss": 2.554,
      "step": 6825
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4531484544277191,
      "learning_rate": 9.93042261072854e-05,
      "loss": 2.5884,
      "step": 6830
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49986356496810913,
      "learning_rate": 9.930319429277202e-05,
      "loss": 2.5677,
      "step": 6835
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4524780511856079,
      "learning_rate": 9.93021617191169e-05,
      "loss": 2.5771,
      "step": 6840
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5700281262397766,
      "learning_rate": 9.930112838633592e-05,
      "loss": 2.5869,
      "step": 6845
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49082717299461365,
      "learning_rate": 9.930009429444503e-05,
      "loss": 2.6131,
      "step": 6850
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4907187521457672,
      "learning_rate": 9.929905944346012e-05,
      "loss": 2.6446,
      "step": 6855
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45793184638023376,
      "learning_rate": 9.929802383339713e-05,
      "loss": 2.5532,
      "step": 6860
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4318350851535797,
      "learning_rate": 9.9296987464272e-05,
      "loss": 2.4623,
      "step": 6865
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44842439889907837,
      "learning_rate": 9.929595033610072e-05,
      "loss": 2.5765,
      "step": 6870
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46213942766189575,
      "learning_rate": 9.929491244889921e-05,
      "loss": 2.5366,
      "step": 6875
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46583133935928345,
      "learning_rate": 9.929387380268349e-05,
      "loss": 2.5826,
      "step": 6880
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5617339015007019,
      "learning_rate": 9.929283439746955e-05,
      "loss": 2.6368,
      "step": 6885
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4392379820346832,
      "learning_rate": 9.929179423327336e-05,
      "loss": 2.5849,
      "step": 6890
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4327978789806366,
      "learning_rate": 9.929075331011097e-05,
      "loss": 2.6288,
      "step": 6895
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49423307180404663,
      "learning_rate": 9.928971162799838e-05,
      "loss": 2.5319,
      "step": 6900
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4619885981082916,
      "learning_rate": 9.928866918695165e-05,
      "loss": 2.6142,
      "step": 6905
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48657846450805664,
      "learning_rate": 9.928762598698684e-05,
      "loss": 2.5853,
      "step": 6910
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45039063692092896,
      "learning_rate": 9.928658202811999e-05,
      "loss": 2.6324,
      "step": 6915
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46564215421676636,
      "learning_rate": 9.928553731036719e-05,
      "loss": 2.5925,
      "step": 6920
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4491654336452484,
      "learning_rate": 9.928449183374451e-05,
      "loss": 2.5316,
      "step": 6925
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4669044613838196,
      "learning_rate": 9.928344559826806e-05,
      "loss": 2.5353,
      "step": 6930
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45235103368759155,
      "learning_rate": 9.928239860395394e-05,
      "loss": 2.5842,
      "step": 6935
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4631016254425049,
      "learning_rate": 9.928135085081827e-05,
      "loss": 2.5857,
      "step": 6940
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41863447427749634,
      "learning_rate": 9.928030233887721e-05,
      "loss": 2.522,
      "step": 6945
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4659057855606079,
      "learning_rate": 9.927925306814686e-05,
      "loss": 2.5554,
      "step": 6950
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4856722354888916,
      "learning_rate": 9.927820303864341e-05,
      "loss": 2.5175,
      "step": 6955
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45341771841049194,
      "learning_rate": 9.927715225038303e-05,
      "loss": 2.5844,
      "step": 6960
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5942304134368896,
      "learning_rate": 9.927610070338187e-05,
      "loss": 2.5606,
      "step": 6965
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4762691259384155,
      "learning_rate": 9.927504839765615e-05,
      "loss": 2.5384,
      "step": 6970
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45530959963798523,
      "learning_rate": 9.927399533322206e-05,
      "loss": 2.581,
      "step": 6975
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44470521807670593,
      "learning_rate": 9.92729415100958e-05,
      "loss": 2.6214,
      "step": 6980
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4635991156101227,
      "learning_rate": 9.927188692829363e-05,
      "loss": 2.5737,
      "step": 6985
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46982499957084656,
      "learning_rate": 9.927083158783177e-05,
      "loss": 2.5209,
      "step": 6990
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4468192458152771,
      "learning_rate": 9.926977548872644e-05,
      "loss": 2.6244,
      "step": 6995
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4545411765575409,
      "learning_rate": 9.926871863099396e-05,
      "loss": 2.5032,
      "step": 7000
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.49754202365875244,
      "learning_rate": 9.926766101465056e-05,
      "loss": 2.5891,
      "step": 7005
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45873796939849854,
      "learning_rate": 9.926660263971254e-05,
      "loss": 2.4797,
      "step": 7010
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4941006898880005,
      "learning_rate": 9.92655435061962e-05,
      "loss": 2.561,
      "step": 7015
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4465232789516449,
      "learning_rate": 9.926448361411784e-05,
      "loss": 2.4988,
      "step": 7020
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44723638892173767,
      "learning_rate": 9.926342296349378e-05,
      "loss": 2.5399,
      "step": 7025
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4323752522468567,
      "learning_rate": 9.926236155434034e-05,
      "loss": 2.5962,
      "step": 7030
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44341418147087097,
      "learning_rate": 9.926129938667388e-05,
      "loss": 2.5295,
      "step": 7035
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44549980759620667,
      "learning_rate": 9.926023646051075e-05,
      "loss": 2.6355,
      "step": 7040
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4665738344192505,
      "learning_rate": 9.925917277586734e-05,
      "loss": 2.48,
      "step": 7045
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45731016993522644,
      "learning_rate": 9.925810833275998e-05,
      "loss": 2.5636,
      "step": 7050
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5238035917282104,
      "learning_rate": 9.925704313120508e-05,
      "loss": 2.6458,
      "step": 7055
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4327884614467621,
      "learning_rate": 9.925597717121906e-05,
      "loss": 2.5807,
      "step": 7060
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48271307349205017,
      "learning_rate": 9.925491045281832e-05,
      "loss": 2.5061,
      "step": 7065
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43220680952072144,
      "learning_rate": 9.925384297601928e-05,
      "loss": 2.6362,
      "step": 7070
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44794589281082153,
      "learning_rate": 9.925277474083835e-05,
      "loss": 2.5917,
      "step": 7075
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4481010437011719,
      "learning_rate": 9.925170574729204e-05,
      "loss": 2.6012,
      "step": 7080
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43885624408721924,
      "learning_rate": 9.925063599539677e-05,
      "loss": 2.5496,
      "step": 7085
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45461517572402954,
      "learning_rate": 9.924956548516901e-05,
      "loss": 2.6181,
      "step": 7090
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4422321617603302,
      "learning_rate": 9.924849421662528e-05,
      "loss": 2.6058,
      "step": 7095
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41687026619911194,
      "learning_rate": 9.924742218978203e-05,
      "loss": 2.5949,
      "step": 7100
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4457671344280243,
      "learning_rate": 9.924634940465577e-05,
      "loss": 2.4638,
      "step": 7105
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4755150079727173,
      "learning_rate": 9.924527586126305e-05,
      "loss": 2.5172,
      "step": 7110
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44020143151283264,
      "learning_rate": 9.924420155962037e-05,
      "loss": 2.5848,
      "step": 7115
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4535250961780548,
      "learning_rate": 9.924312649974427e-05,
      "loss": 2.6494,
      "step": 7120
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5285208821296692,
      "learning_rate": 9.924205068165134e-05,
      "loss": 2.5365,
      "step": 7125
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48362523317337036,
      "learning_rate": 9.924097410535811e-05,
      "loss": 2.526,
      "step": 7130
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.49188199639320374,
      "learning_rate": 9.923989677088116e-05,
      "loss": 2.5794,
      "step": 7135
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45113739371299744,
      "learning_rate": 9.92388186782371e-05,
      "loss": 2.538,
      "step": 7140
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4459697902202606,
      "learning_rate": 9.92377398274425e-05,
      "loss": 2.6359,
      "step": 7145
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47667649388313293,
      "learning_rate": 9.923666021851398e-05,
      "loss": 2.5871,
      "step": 7150
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4573513865470886,
      "learning_rate": 9.92355798514682e-05,
      "loss": 2.5938,
      "step": 7155
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48477187752723694,
      "learning_rate": 9.923449872632173e-05,
      "loss": 2.6083,
      "step": 7160
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.480961412191391,
      "learning_rate": 9.923341684309128e-05,
      "loss": 2.5504,
      "step": 7165
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4528869688510895,
      "learning_rate": 9.923233420179347e-05,
      "loss": 2.5409,
      "step": 7170
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4410327672958374,
      "learning_rate": 9.923125080244497e-05,
      "loss": 2.6092,
      "step": 7175
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43545448780059814,
      "learning_rate": 9.923016664506248e-05,
      "loss": 2.6139,
      "step": 7180
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47672194242477417,
      "learning_rate": 9.922908172966268e-05,
      "loss": 2.5725,
      "step": 7185
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4314936101436615,
      "learning_rate": 9.922799605626227e-05,
      "loss": 2.4434,
      "step": 7190
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46617764234542847,
      "learning_rate": 9.9226909624878e-05,
      "loss": 2.572,
      "step": 7195
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4467450976371765,
      "learning_rate": 9.922582243552654e-05,
      "loss": 2.5647,
      "step": 7200
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44466879963874817,
      "learning_rate": 9.922473448822469e-05,
      "loss": 2.5431,
      "step": 7205
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4341384172439575,
      "learning_rate": 9.922364578298916e-05,
      "loss": 2.5465,
      "step": 7210
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48069146275520325,
      "learning_rate": 9.922255631983673e-05,
      "loss": 2.5818,
      "step": 7215
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45424318313598633,
      "learning_rate": 9.922146609878418e-05,
      "loss": 2.5989,
      "step": 7220
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4503635764122009,
      "learning_rate": 9.922037511984829e-05,
      "loss": 2.52,
      "step": 7225
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4486410915851593,
      "learning_rate": 9.921928338304584e-05,
      "loss": 2.4725,
      "step": 7230
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4268917143344879,
      "learning_rate": 9.921819088839369e-05,
      "loss": 2.5622,
      "step": 7235
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42523452639579773,
      "learning_rate": 9.92170976359086e-05,
      "loss": 2.5336,
      "step": 7240
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4760739803314209,
      "learning_rate": 9.921600362560744e-05,
      "loss": 2.5523,
      "step": 7245
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45493513345718384,
      "learning_rate": 9.921490885750706e-05,
      "loss": 2.6555,
      "step": 7250
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4364657402038574,
      "learning_rate": 9.92138133316243e-05,
      "loss": 2.5396,
      "step": 7255
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42500635981559753,
      "learning_rate": 9.921271704797602e-05,
      "loss": 2.5029,
      "step": 7260
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4418283998966217,
      "learning_rate": 9.921162000657913e-05,
      "loss": 2.4767,
      "step": 7265
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42562904953956604,
      "learning_rate": 9.921052220745048e-05,
      "loss": 2.4791,
      "step": 7270
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4553951323032379,
      "learning_rate": 9.920942365060702e-05,
      "loss": 2.5741,
      "step": 7275
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4419585168361664,
      "learning_rate": 9.920832433606564e-05,
      "loss": 2.5487,
      "step": 7280
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45310381054878235,
      "learning_rate": 9.920722426384326e-05,
      "loss": 2.5813,
      "step": 7285
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43653249740600586,
      "learning_rate": 9.920612343395683e-05,
      "loss": 2.6036,
      "step": 7290
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5252856612205505,
      "learning_rate": 9.92050218464233e-05,
      "loss": 2.5176,
      "step": 7295
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45629024505615234,
      "learning_rate": 9.920391950125963e-05,
      "loss": 2.5372,
      "step": 7300
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45536187291145325,
      "learning_rate": 9.920281639848278e-05,
      "loss": 2.5576,
      "step": 7305
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4224608838558197,
      "learning_rate": 9.920171253810975e-05,
      "loss": 2.4816,
      "step": 7310
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4271993041038513,
      "learning_rate": 9.920060792015754e-05,
      "loss": 2.5442,
      "step": 7315
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4603693187236786,
      "learning_rate": 9.919950254464316e-05,
      "loss": 2.5246,
      "step": 7320
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.49064570665359497,
      "learning_rate": 9.919839641158361e-05,
      "loss": 2.5525,
      "step": 7325
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.468471884727478,
      "learning_rate": 9.919728952099593e-05,
      "loss": 2.6497,
      "step": 7330
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4951266348361969,
      "learning_rate": 9.919618187289717e-05,
      "loss": 2.588,
      "step": 7335
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.521508514881134,
      "learning_rate": 9.919507346730437e-05,
      "loss": 2.5782,
      "step": 7340
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4988766312599182,
      "learning_rate": 9.919396430423463e-05,
      "loss": 2.5598,
      "step": 7345
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45684102177619934,
      "learning_rate": 9.9192854383705e-05,
      "loss": 2.5775,
      "step": 7350
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4477953314781189,
      "learning_rate": 9.919174370573257e-05,
      "loss": 2.6618,
      "step": 7355
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4375406801700592,
      "learning_rate": 9.919063227033445e-05,
      "loss": 2.4707,
      "step": 7360
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4372744858264923,
      "learning_rate": 9.918952007752776e-05,
      "loss": 2.4719,
      "step": 7365
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5375922322273254,
      "learning_rate": 9.91884071273296e-05,
      "loss": 2.5474,
      "step": 7370
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5307102799415588,
      "learning_rate": 9.918729341975712e-05,
      "loss": 2.4635,
      "step": 7375
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.130879282951355,
      "learning_rate": 9.918617895482749e-05,
      "loss": 2.5747,
      "step": 7380
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4725744128227234,
      "learning_rate": 9.918506373255784e-05,
      "loss": 2.5504,
      "step": 7385
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4026028513908386,
      "learning_rate": 9.918394775296536e-05,
      "loss": 2.5246,
      "step": 7390
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5675021409988403,
      "learning_rate": 9.91828310160672e-05,
      "loss": 2.545,
      "step": 7395
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5162795186042786,
      "learning_rate": 9.918171352188062e-05,
      "loss": 2.5984,
      "step": 7400
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4789445400238037,
      "learning_rate": 9.918059527042275e-05,
      "loss": 2.6309,
      "step": 7405
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4239911735057831,
      "learning_rate": 9.917947626171086e-05,
      "loss": 2.5262,
      "step": 7410
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4225129783153534,
      "learning_rate": 9.917835649576217e-05,
      "loss": 2.5066,
      "step": 7415
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43272191286087036,
      "learning_rate": 9.91772359725939e-05,
      "loss": 2.5218,
      "step": 7420
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4284610152244568,
      "learning_rate": 9.917611469222333e-05,
      "loss": 2.5505,
      "step": 7425
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47917333245277405,
      "learning_rate": 9.917499265466772e-05,
      "loss": 2.5611,
      "step": 7430
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4351503551006317,
      "learning_rate": 9.917386985994432e-05,
      "loss": 2.5379,
      "step": 7435
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4314475655555725,
      "learning_rate": 9.917274630807047e-05,
      "loss": 2.4865,
      "step": 7440
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45712530612945557,
      "learning_rate": 9.917162199906341e-05,
      "loss": 2.6307,
      "step": 7445
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4570379853248596,
      "learning_rate": 9.917049693294049e-05,
      "loss": 2.5899,
      "step": 7450
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44030776619911194,
      "learning_rate": 9.916937110971902e-05,
      "loss": 2.628,
      "step": 7455
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4645618796348572,
      "learning_rate": 9.916824452941635e-05,
      "loss": 2.4584,
      "step": 7460
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43652772903442383,
      "learning_rate": 9.916711719204981e-05,
      "loss": 2.5404,
      "step": 7465
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4324008524417877,
      "learning_rate": 9.916598909763675e-05,
      "loss": 2.5334,
      "step": 7470
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.435794860124588,
      "learning_rate": 9.916486024619455e-05,
      "loss": 2.5792,
      "step": 7475
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42903581261634827,
      "learning_rate": 9.916373063774062e-05,
      "loss": 2.5936,
      "step": 7480
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.463533878326416,
      "learning_rate": 9.91626002722923e-05,
      "loss": 2.6078,
      "step": 7485
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45317718386650085,
      "learning_rate": 9.916146914986703e-05,
      "loss": 2.5507,
      "step": 7490
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4541989862918854,
      "learning_rate": 9.916033727048223e-05,
      "loss": 2.6097,
      "step": 7495
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44872036576271057,
      "learning_rate": 9.915920463415531e-05,
      "loss": 2.3673,
      "step": 7500
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5086636543273926,
      "learning_rate": 9.915807124090371e-05,
      "loss": 2.5364,
      "step": 7505
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4372543692588806,
      "learning_rate": 9.915693709074487e-05,
      "loss": 2.4741,
      "step": 7510
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4478601813316345,
      "learning_rate": 9.91558021836963e-05,
      "loss": 2.5054,
      "step": 7515
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4245615601539612,
      "learning_rate": 9.915466651977543e-05,
      "loss": 2.5769,
      "step": 7520
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4577113091945648,
      "learning_rate": 9.915353009899976e-05,
      "loss": 2.5284,
      "step": 7525
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4139074981212616,
      "learning_rate": 9.915239292138678e-05,
      "loss": 2.584,
      "step": 7530
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45214739441871643,
      "learning_rate": 9.915125498695402e-05,
      "loss": 2.6645,
      "step": 7535
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4410076141357422,
      "learning_rate": 9.9150116295719e-05,
      "loss": 2.5324,
      "step": 7540
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46784141659736633,
      "learning_rate": 9.914897684769921e-05,
      "loss": 2.5103,
      "step": 7545
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4975837469100952,
      "learning_rate": 9.914783664291225e-05,
      "loss": 2.5884,
      "step": 7550
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48534247279167175,
      "learning_rate": 9.914669568137565e-05,
      "loss": 2.5479,
      "step": 7555
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4127311110496521,
      "learning_rate": 9.914555396310698e-05,
      "loss": 2.5238,
      "step": 7560
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4457809329032898,
      "learning_rate": 9.914441148812382e-05,
      "loss": 2.6569,
      "step": 7565
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4202069640159607,
      "learning_rate": 9.914326825644376e-05,
      "loss": 2.5164,
      "step": 7570
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.622786819934845,
      "learning_rate": 9.91421242680844e-05,
      "loss": 2.4963,
      "step": 7575
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44363874197006226,
      "learning_rate": 9.914097952306338e-05,
      "loss": 2.57,
      "step": 7580
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44974684715270996,
      "learning_rate": 9.913983402139828e-05,
      "loss": 2.5986,
      "step": 7585
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42762845754623413,
      "learning_rate": 9.913868776310677e-05,
      "loss": 2.5447,
      "step": 7590
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47779008746147156,
      "learning_rate": 9.913754074820649e-05,
      "loss": 2.5048,
      "step": 7595
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6928739547729492,
      "learning_rate": 9.913639297671511e-05,
      "loss": 2.5754,
      "step": 7600
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5947225689888,
      "learning_rate": 9.91352444486503e-05,
      "loss": 2.4929,
      "step": 7605
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4378729462623596,
      "learning_rate": 9.913409516402974e-05,
      "loss": 2.5059,
      "step": 7610
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4311879277229309,
      "learning_rate": 9.913294512287112e-05,
      "loss": 2.5892,
      "step": 7615
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.511725127696991,
      "learning_rate": 9.913179432519215e-05,
      "loss": 2.5581,
      "step": 7620
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4317936301231384,
      "learning_rate": 9.913064277101055e-05,
      "loss": 2.5802,
      "step": 7625
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.425821989774704,
      "learning_rate": 9.912949046034406e-05,
      "loss": 2.5425,
      "step": 7630
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6782971620559692,
      "learning_rate": 9.912833739321042e-05,
      "loss": 2.5746,
      "step": 7635
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5352039337158203,
      "learning_rate": 9.912718356962738e-05,
      "loss": 2.6269,
      "step": 7640
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4434868097305298,
      "learning_rate": 9.91260289896127e-05,
      "loss": 2.6337,
      "step": 7645
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44488659501075745,
      "learning_rate": 9.912487365318417e-05,
      "loss": 2.5489,
      "step": 7650
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44559285044670105,
      "learning_rate": 9.912371756035957e-05,
      "loss": 2.5664,
      "step": 7655
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43834343552589417,
      "learning_rate": 9.912256071115671e-05,
      "loss": 2.4889,
      "step": 7660
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4872107207775116,
      "learning_rate": 9.91214031055934e-05,
      "loss": 2.5138,
      "step": 7665
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4114554822444916,
      "learning_rate": 9.912024474368745e-05,
      "loss": 2.5234,
      "step": 7670
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4369142949581146,
      "learning_rate": 9.911908562545671e-05,
      "loss": 2.5966,
      "step": 7675
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4713846743106842,
      "learning_rate": 9.911792575091903e-05,
      "loss": 2.597,
      "step": 7680
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4362868368625641,
      "learning_rate": 9.911676512009226e-05,
      "loss": 2.4709,
      "step": 7685
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42201676964759827,
      "learning_rate": 9.911560373299426e-05,
      "loss": 2.5289,
      "step": 7690
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41837775707244873,
      "learning_rate": 9.911444158964296e-05,
      "loss": 2.5363,
      "step": 7695
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4526817798614502,
      "learning_rate": 9.91132786900562e-05,
      "loss": 2.4776,
      "step": 7700
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4597223699092865,
      "learning_rate": 9.911211503425191e-05,
      "loss": 2.511,
      "step": 7705
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46648961305618286,
      "learning_rate": 9.911095062224798e-05,
      "loss": 2.5455,
      "step": 7710
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4529838263988495,
      "learning_rate": 9.91097854540624e-05,
      "loss": 2.5373,
      "step": 7715
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4725988805294037,
      "learning_rate": 9.910861952971304e-05,
      "loss": 2.5704,
      "step": 7720
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42108482122421265,
      "learning_rate": 9.910745284921791e-05,
      "loss": 2.5404,
      "step": 7725
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44580507278442383,
      "learning_rate": 9.910628541259495e-05,
      "loss": 2.5352,
      "step": 7730
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4159211218357086,
      "learning_rate": 9.91051172198621e-05,
      "loss": 2.5747,
      "step": 7735
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.463248610496521,
      "learning_rate": 9.910394827103742e-05,
      "loss": 2.4728,
      "step": 7740
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4523601830005646,
      "learning_rate": 9.910277856613886e-05,
      "loss": 2.505,
      "step": 7745
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4854637086391449,
      "learning_rate": 9.910160810518442e-05,
      "loss": 2.6133,
      "step": 7750
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4288189113140106,
      "learning_rate": 9.910043688819218e-05,
      "loss": 2.5131,
      "step": 7755
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42924144864082336,
      "learning_rate": 9.909926491518012e-05,
      "loss": 2.6231,
      "step": 7760
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4304296374320984,
      "learning_rate": 9.909809218616629e-05,
      "loss": 2.6357,
      "step": 7765
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44037508964538574,
      "learning_rate": 9.909691870116877e-05,
      "loss": 2.5913,
      "step": 7770
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4809196889400482,
      "learning_rate": 9.909574446020561e-05,
      "loss": 2.5163,
      "step": 7775
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45099735260009766,
      "learning_rate": 9.90945694632949e-05,
      "loss": 2.4726,
      "step": 7780
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43467795848846436,
      "learning_rate": 9.909339371045474e-05,
      "loss": 2.5192,
      "step": 7785
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46502915024757385,
      "learning_rate": 9.90922172017032e-05,
      "loss": 2.4644,
      "step": 7790
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4190245568752289,
      "learning_rate": 9.909103993705845e-05,
      "loss": 2.5421,
      "step": 7795
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4155759811401367,
      "learning_rate": 9.908986191653857e-05,
      "loss": 2.4777,
      "step": 7800
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.457491397857666,
      "learning_rate": 9.90886831401617e-05,
      "loss": 2.5429,
      "step": 7805
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4549599587917328,
      "learning_rate": 9.908750360794603e-05,
      "loss": 2.5347,
      "step": 7810
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4300247132778168,
      "learning_rate": 9.908632331990968e-05,
      "loss": 2.5631,
      "step": 7815
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4144454300403595,
      "learning_rate": 9.908514227607084e-05,
      "loss": 2.5674,
      "step": 7820
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4283376634120941,
      "learning_rate": 9.90839604764477e-05,
      "loss": 2.5533,
      "step": 7825
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48587173223495483,
      "learning_rate": 9.908277792105845e-05,
      "loss": 2.4832,
      "step": 7830
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4517517685890198,
      "learning_rate": 9.90815946099213e-05,
      "loss": 2.4651,
      "step": 7835
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4192942976951599,
      "learning_rate": 9.908041054305446e-05,
      "loss": 2.4558,
      "step": 7840
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43992483615875244,
      "learning_rate": 9.907922572047618e-05,
      "loss": 2.4994,
      "step": 7845
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4142637252807617,
      "learning_rate": 9.90780401422047e-05,
      "loss": 2.4901,
      "step": 7850
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42823973298072815,
      "learning_rate": 9.907685380825825e-05,
      "loss": 2.5658,
      "step": 7855
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.434349000453949,
      "learning_rate": 9.907566671865512e-05,
      "loss": 2.4438,
      "step": 7860
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6057260036468506,
      "learning_rate": 9.907447887341359e-05,
      "loss": 2.4983,
      "step": 7865
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42913541197776794,
      "learning_rate": 9.907329027255194e-05,
      "loss": 2.4853,
      "step": 7870
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45045357942581177,
      "learning_rate": 9.907210091608848e-05,
      "loss": 2.6283,
      "step": 7875
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4474749267101288,
      "learning_rate": 9.907091080404152e-05,
      "loss": 2.548,
      "step": 7880
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43322432041168213,
      "learning_rate": 9.906971993642937e-05,
      "loss": 2.4789,
      "step": 7885
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5895424485206604,
      "learning_rate": 9.906852831327038e-05,
      "loss": 2.4751,
      "step": 7890
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.441974401473999,
      "learning_rate": 9.906733593458289e-05,
      "loss": 2.482,
      "step": 7895
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4343891441822052,
      "learning_rate": 9.906614280038529e-05,
      "loss": 2.5238,
      "step": 7900
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4518453776836395,
      "learning_rate": 9.90649489106959e-05,
      "loss": 2.4087,
      "step": 7905
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4975084662437439,
      "learning_rate": 9.906375426553314e-05,
      "loss": 2.5929,
      "step": 7910
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41973111033439636,
      "learning_rate": 9.90625588649154e-05,
      "loss": 2.5237,
      "step": 7915
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4306008517742157,
      "learning_rate": 9.906136270886107e-05,
      "loss": 2.4476,
      "step": 7920
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.1563215255737305,
      "learning_rate": 9.90601657973886e-05,
      "loss": 2.4047,
      "step": 7925
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42872339487075806,
      "learning_rate": 9.905896813051636e-05,
      "loss": 2.4635,
      "step": 7930
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44585907459259033,
      "learning_rate": 9.905776970826286e-05,
      "loss": 2.5684,
      "step": 7935
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42978963255882263,
      "learning_rate": 9.905657053064651e-05,
      "loss": 2.4617,
      "step": 7940
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4475099444389343,
      "learning_rate": 9.905537059768578e-05,
      "loss": 2.5771,
      "step": 7945
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4373677670955658,
      "learning_rate": 9.905416990939915e-05,
      "loss": 2.5025,
      "step": 7950
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4509379267692566,
      "learning_rate": 9.905296846580512e-05,
      "loss": 2.4291,
      "step": 7955
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4464047849178314,
      "learning_rate": 9.905176626692217e-05,
      "loss": 2.5693,
      "step": 7960
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6174293160438538,
      "learning_rate": 9.905056331276884e-05,
      "loss": 2.6365,
      "step": 7965
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42413613200187683,
      "learning_rate": 9.904935960336359e-05,
      "loss": 2.4561,
      "step": 7970
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4160773754119873,
      "learning_rate": 9.904815513872503e-05,
      "loss": 2.5464,
      "step": 7975
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43968459963798523,
      "learning_rate": 9.904694991887164e-05,
      "loss": 2.5348,
      "step": 7980
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47595733404159546,
      "learning_rate": 9.904574394382203e-05,
      "loss": 2.571,
      "step": 7985
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42008063197135925,
      "learning_rate": 9.904453721359474e-05,
      "loss": 2.4933,
      "step": 7990
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4452323019504547,
      "learning_rate": 9.904332972820837e-05,
      "loss": 2.4725,
      "step": 7995
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4176363945007324,
      "learning_rate": 9.904212148768148e-05,
      "loss": 2.5932,
      "step": 8000
    }
  ],
  "logging_steps": 5,
  "max_steps": 126689,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 2.173076069626675e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}