{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 34236,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 2.1784449710473837,
      "learning_rate": 4.999999736861907e-05,
      "loss": 1.3118,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4897392967760945,
      "learning_rate": 4.999998947447681e-05,
      "loss": 1.2024,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5311854018449251,
      "learning_rate": 4.9999976317574895e-05,
      "loss": 1.1376,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3982158801348301,
      "learning_rate": 4.999995789791609e-05,
      "loss": 1.074,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3336193278462263,
      "learning_rate": 4.9999934215504286e-05,
      "loss": 1.0115,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3752137315104604,
      "learning_rate": 4.999990527034445e-05,
      "loss": 1.0576,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4094207690165101,
      "learning_rate": 4.999987106244268e-05,
      "loss": 1.1654,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5199782713779533,
      "learning_rate": 4.999983159180619e-05,
      "loss": 0.9992,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4566344917839809,
      "learning_rate": 4.999978685844327e-05,
      "loss": 1.074,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.564520831809233,
      "learning_rate": 4.999973686236335e-05,
      "loss": 1.021,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.468453331106892,
      "learning_rate": 4.9999681603576945e-05,
      "loss": 1.0803,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2428443229463324,
      "learning_rate": 4.999962108209569e-05,
      "loss": 1.0231,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.23791908522262,
      "learning_rate": 4.9999555297932345e-05,
      "loss": 0.9771,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2773848935764596,
      "learning_rate": 4.9999484251100726e-05,
      "loss": 1.1095,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3782292477996736,
      "learning_rate": 4.999940794161581e-05,
      "loss": 1.0616,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2952820131263536,
      "learning_rate": 4.999932636949365e-05,
      "loss": 1.0632,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4248329224299214,
      "learning_rate": 4.999923953475143e-05,
      "loss": 1.0455,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2220927446888543,
      "learning_rate": 4.999914743740742e-05,
      "loss": 1.0553,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2941886202299266,
      "learning_rate": 4.9999050077481014e-05,
      "loss": 1.0822,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3552724672192507,
      "learning_rate": 4.99989474549927e-05,
      "loss": 0.9869,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2325989312913683,
      "learning_rate": 4.9998839569964086e-05,
      "loss": 0.9796,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4295829729154879,
      "learning_rate": 4.999872642241788e-05,
      "loss": 1.0544,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1865757592832702,
      "learning_rate": 4.9998608012377904e-05,
      "loss": 1.0361,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2022374665421445,
      "learning_rate": 4.9998484339869085e-05,
      "loss": 1.0009,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2840172912471361,
      "learning_rate": 4.999835540491745e-05,
      "loss": 0.9948,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3908185779332902,
      "learning_rate": 4.999822120755015e-05,
      "loss": 1.031,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1783544616928932,
      "learning_rate": 4.9998081747795434e-05,
      "loss": 1.0133,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1582571514646922,
      "learning_rate": 4.999793702568265e-05,
      "loss": 0.9739,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.230808365640957,
      "learning_rate": 4.999778704124228e-05,
      "loss": 1.0328,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2306674242430162,
      "learning_rate": 4.999763179450588e-05,
      "loss": 1.0518,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.089913942169356,
      "learning_rate": 4.999747128550615e-05,
      "loss": 0.9716,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2127600303253867,
      "learning_rate": 4.999730551427686e-05,
      "loss": 1.0216,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2223076887165993,
      "learning_rate": 4.9997134480852916e-05,
      "loss": 1.0557,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1650424198208296,
      "learning_rate": 4.999695818527032e-05,
      "loss": 1.0516,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9089784717853382,
      "learning_rate": 4.999677662756618e-05,
      "loss": 1.0407,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1857033366836702,
      "learning_rate": 4.999658980777873e-05,
      "loss": 1.026,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1175077900584947,
      "learning_rate": 4.9996397725947276e-05,
      "loss": 1.0058,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2839942707097831,
      "learning_rate": 4.999620038211227e-05,
      "loss": 1.0985,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.198310511273475,
      "learning_rate": 4.9995997776315256e-05,
      "loss": 1.0306,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2046663657921814,
      "learning_rate": 4.9995789908598876e-05,
      "loss": 1.0083,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0217207631158867,
      "learning_rate": 4.999557677900689e-05,
      "loss": 1.0148,
      "step": 205
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.186093973343761,
      "learning_rate": 4.999535838758417e-05,
      "loss": 1.0204,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1090086047402792,
      "learning_rate": 4.999513473437668e-05,
      "loss": 1.0237,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1141811398253623,
      "learning_rate": 4.9994905819431516e-05,
      "loss": 1.0204,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3182671403308446,
      "learning_rate": 4.999467164279684e-05,
      "loss": 1.0291,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.386651938972161,
      "learning_rate": 4.999443220452198e-05,
      "loss": 1.0474,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.846959544880204,
      "learning_rate": 4.999418750465732e-05,
      "loss": 0.9256,
      "step": 235
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4501773981803823,
      "learning_rate": 4.999393754325438e-05,
      "loss": 0.9958,
      "step": 240
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1216837830840287,
      "learning_rate": 4.9993682320365783e-05,
      "loss": 0.9963,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.072695041562247,
      "learning_rate": 4.999342183604525e-05,
      "loss": 0.9837,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2854505754455945,
      "learning_rate": 4.9993156090347604e-05,
      "loss": 0.9735,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1201609223612168,
      "learning_rate": 4.999288508332881e-05,
      "loss": 1.0047,
      "step": 260
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1067023825942017,
      "learning_rate": 4.99926088150459e-05,
      "loss": 1.0462,
      "step": 265
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.488134858528184,
      "learning_rate": 4.999232728555705e-05,
      "loss": 0.9916,
      "step": 270
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1046296845118166,
      "learning_rate": 4.999204049492151e-05,
      "loss": 0.9228,
      "step": 275
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1179757334585674,
      "learning_rate": 4.999174844319965e-05,
      "loss": 1.0438,
      "step": 280
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2583407669600082,
      "learning_rate": 4.9991451130452964e-05,
      "loss": 0.9587,
      "step": 285
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.130825505538917,
      "learning_rate": 4.999114855674403e-05,
      "loss": 0.9418,
      "step": 290
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1276805384744066,
      "learning_rate": 4.999084072213654e-05,
      "loss": 0.9831,
      "step": 295
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0895304959445886,
      "learning_rate": 4.9990527626695306e-05,
      "loss": 0.9919,
      "step": 300
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3340398201279144,
      "learning_rate": 4.9990209270486235e-05,
      "loss": 1.0481,
      "step": 305
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2873488720246915,
      "learning_rate": 4.998988565357633e-05,
      "loss": 0.9639,
      "step": 310
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.170230482992979,
      "learning_rate": 4.9989556776033744e-05,
      "loss": 1.0525,
      "step": 315
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2297751507027324,
      "learning_rate": 4.998922263792768e-05,
      "loss": 1.0187,
      "step": 320
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2801466011614338,
      "learning_rate": 4.99888832393285e-05,
      "loss": 0.9391,
      "step": 325
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2169368970409447,
      "learning_rate": 4.9988538580307634e-05,
      "loss": 1.0129,
      "step": 330
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0959208225420087,
      "learning_rate": 4.9988188660937644e-05,
      "loss": 1.0192,
      "step": 335
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2706316396214556,
      "learning_rate": 4.99878334812922e-05,
      "loss": 0.9624,
      "step": 340
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1380901250678794,
      "learning_rate": 4.9987473041446056e-05,
      "loss": 1.0094,
      "step": 345
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.172421811889701,
      "learning_rate": 4.998710734147509e-05,
      "loss": 1.0305,
      "step": 350
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0737481983310084,
      "learning_rate": 4.99867363814563e-05,
      "loss": 0.9378,
      "step": 355
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2035643705297934,
      "learning_rate": 4.998636016146777e-05,
      "loss": 1.0645,
      "step": 360
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1937551163588551,
      "learning_rate": 4.998597868158869e-05,
      "loss": 0.9719,
      "step": 365
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1826140175264577,
      "learning_rate": 4.998559194189937e-05,
      "loss": 1.0146,
      "step": 370
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1042650143382684,
      "learning_rate": 4.998519994248123e-05,
      "loss": 1.0107,
      "step": 375
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2029194221209942,
      "learning_rate": 4.998480268341678e-05,
      "loss": 0.9828,
      "step": 380
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1072587809035566,
      "learning_rate": 4.9984400164789645e-05,
      "loss": 1.236,
      "step": 385
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0953844697196076,
      "learning_rate": 4.998399238668457e-05,
      "loss": 0.9292,
      "step": 390
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1716065545175716,
      "learning_rate": 4.99835793491874e-05,
      "loss": 1.0484,
      "step": 395
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0359817423985835,
      "learning_rate": 4.998316105238507e-05,
      "loss": 1.0815,
      "step": 400
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1579234526074056,
      "learning_rate": 4.998273749636564e-05,
      "loss": 1.0899,
      "step": 405
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2668886554383545,
      "learning_rate": 4.998230868121828e-05,
      "loss": 1.0541,
      "step": 410
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.222528950010076,
      "learning_rate": 4.9981874607033255e-05,
      "loss": 1.0427,
      "step": 415
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.048250239183132,
      "learning_rate": 4.998143527390194e-05,
      "loss": 0.9596,
      "step": 420
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.120426201376995,
      "learning_rate": 4.998099068191682e-05,
      "loss": 0.9472,
      "step": 425
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0439744398964554,
      "learning_rate": 4.9980540831171486e-05,
      "loss": 1.0614,
      "step": 430
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1171357434251363,
      "learning_rate": 4.998008572176064e-05,
      "loss": 0.9669,
      "step": 435
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.126080016712785,
      "learning_rate": 4.997962535378009e-05,
      "loss": 1.0044,
      "step": 440
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1291910970101615,
      "learning_rate": 4.997915972732674e-05,
      "loss": 1.0066,
      "step": 445
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0485393220686727,
      "learning_rate": 4.9978688842498614e-05,
      "loss": 1.0298,
      "step": 450
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1159726165725172,
      "learning_rate": 4.9978212699394835e-05,
      "loss": 0.9646,
      "step": 455
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1801373513766116,
      "learning_rate": 4.997773129811564e-05,
      "loss": 1.0419,
      "step": 460
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.328022904217585,
      "learning_rate": 4.997724463876235e-05,
      "loss": 0.995,
      "step": 465
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.553359191538908,
      "learning_rate": 4.997675272143745e-05,
      "loss": 1.0236,
      "step": 470
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0669585138225524,
      "learning_rate": 4.997625554624447e-05,
      "loss": 0.9219,
      "step": 475
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2514148117721813,
      "learning_rate": 4.997575311328807e-05,
      "loss": 1.0618,
      "step": 480
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0413844004800614,
      "learning_rate": 4.997524542267402e-05,
      "loss": 1.0033,
      "step": 485
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.360363412276393,
      "learning_rate": 4.9974732474509194e-05,
      "loss": 1.0419,
      "step": 490
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1644390899221422,
      "learning_rate": 4.997421426890158e-05,
      "loss": 1.0206,
      "step": 495
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1011003708276013,
      "learning_rate": 4.997369080596025e-05,
      "loss": 0.9966,
      "step": 500
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.18785372402119,
      "learning_rate": 4.997316208579541e-05,
      "loss": 1.043,
      "step": 505
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1598755032261203,
      "learning_rate": 4.9972628108518364e-05,
      "loss": 0.9685,
      "step": 510
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3077285598491475,
      "learning_rate": 4.997208887424152e-05,
      "loss": 0.9949,
      "step": 515
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1489137561146099,
      "learning_rate": 4.997154438307837e-05,
      "loss": 1.0071,
      "step": 520
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0181246713287804,
      "learning_rate": 4.9970994635143566e-05,
      "loss": 1.113,
      "step": 525
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2011072033512882,
      "learning_rate": 4.9970439630552826e-05,
      "loss": 1.0618,
      "step": 530
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.133566905118304,
      "learning_rate": 4.9969879369422964e-05,
      "loss": 0.9615,
      "step": 535
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2008484845480965,
      "learning_rate": 4.996931385187195e-05,
      "loss": 0.9305,
      "step": 540
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2055167711396377,
      "learning_rate": 4.9968743078018815e-05,
      "loss": 1.019,
      "step": 545
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2349141913988522,
      "learning_rate": 4.9968167047983715e-05,
      "loss": 1.0175,
      "step": 550
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.16231748566863,
      "learning_rate": 4.996758576188791e-05,
      "loss": 1.0421,
      "step": 555
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.134522538690211,
      "learning_rate": 4.996699921985378e-05,
      "loss": 1.0446,
      "step": 560
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0781459655801693,
      "learning_rate": 4.996640742200478e-05,
      "loss": 0.9381,
      "step": 565
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2137309722711722,
      "learning_rate": 4.9965810368465494e-05,
      "loss": 0.9967,
      "step": 570
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1529053128390343,
      "learning_rate": 4.996520805936161e-05,
      "loss": 0.9657,
      "step": 575
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0460006580720695,
      "learning_rate": 4.996460049481992e-05,
      "loss": 0.9482,
      "step": 580
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.238588161142535,
      "learning_rate": 4.996398767496833e-05,
      "loss": 1.0501,
      "step": 585
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2095871167802945,
      "learning_rate": 4.996336959993584e-05,
      "loss": 0.9922,
      "step": 590
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5591839757597055,
      "learning_rate": 4.9962746269852555e-05,
      "loss": 1.0473,
      "step": 595
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4018217429088933,
      "learning_rate": 4.99621176848497e-05,
      "loss": 1.0325,
      "step": 600
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2274248412044058,
      "learning_rate": 4.996148384505959e-05,
      "loss": 0.9981,
      "step": 605
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1372531103499968,
      "learning_rate": 4.9960844750615666e-05,
      "loss": 0.9938,
      "step": 610
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1484648375757214,
      "learning_rate": 4.9960200401652456e-05,
      "loss": 0.9234,
      "step": 615
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1663957851216613,
      "learning_rate": 4.995955079830561e-05,
      "loss": 1.0454,
      "step": 620
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.067926862773165,
      "learning_rate": 4.995889594071187e-05,
      "loss": 0.9921,
      "step": 625
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1268531360199512,
      "learning_rate": 4.995823582900909e-05,
      "loss": 0.8797,
      "step": 630
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.038312565787539,
      "learning_rate": 4.995757046333623e-05,
      "loss": 1.1061,
      "step": 635
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.120749068173633,
      "learning_rate": 4.995689984383337e-05,
      "loss": 1.0332,
      "step": 640
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1747213257891365,
      "learning_rate": 4.995622397064167e-05,
      "loss": 1.051,
      "step": 645
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4146223383616165,
      "learning_rate": 4.99555428439034e-05,
      "loss": 1.0477,
      "step": 650
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2217985044940483,
      "learning_rate": 4.9954856463761955e-05,
      "loss": 0.9759,
      "step": 655
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1476396571931295,
      "learning_rate": 4.9954164830361826e-05,
      "loss": 1.0341,
      "step": 660
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1468045837629024,
      "learning_rate": 4.9953467943848615e-05,
      "loss": 1.022,
      "step": 665
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1506220646964473,
      "learning_rate": 4.995276580436901e-05,
      "loss": 0.944,
      "step": 670
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2193283081896507,
      "learning_rate": 4.9952058412070827e-05,
      "loss": 1.0019,
      "step": 675
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6278010973094974,
      "learning_rate": 4.9951345767102975e-05,
      "loss": 0.969,
      "step": 680
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1542453383131253,
      "learning_rate": 4.995062786961547e-05,
      "loss": 0.9639,
      "step": 685
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.109667105016642,
      "learning_rate": 4.9949904719759456e-05,
      "loss": 0.984,
      "step": 690
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2279651658287232,
      "learning_rate": 4.994917631768714e-05,
      "loss": 1.0782,
      "step": 695
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2697221410971238,
      "learning_rate": 4.9948442663551875e-05,
      "loss": 1.0715,
      "step": 700
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0556788249673268,
      "learning_rate": 4.994770375750809e-05,
      "loss": 0.9582,
      "step": 705
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.212747883871362,
      "learning_rate": 4.9946959599711344e-05,
      "loss": 1.0071,
      "step": 710
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0986630962573964,
      "learning_rate": 4.994621019031829e-05,
      "loss": 1.0043,
      "step": 715
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2618896084074627,
      "learning_rate": 4.9945455529486674e-05,
      "loss": 0.9495,
      "step": 720
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0323216327931164,
      "learning_rate": 4.994469561737536e-05,
      "loss": 0.9683,
      "step": 725
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0364089610975664,
      "learning_rate": 4.994393045414434e-05,
      "loss": 0.9673,
      "step": 730
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0669230855911755,
      "learning_rate": 4.994316003995466e-05,
      "loss": 0.9519,
      "step": 735
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.11233721465925,
      "learning_rate": 4.9942384374968524e-05,
      "loss": 1.1105,
      "step": 740
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1407959492109718,
      "learning_rate": 4.9941603459349204e-05,
      "loss": 1.0423,
      "step": 745
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.125422017552011,
      "learning_rate": 4.99408172932611e-05,
      "loss": 0.9786,
      "step": 750
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.158106022415828,
      "learning_rate": 4.994002587686969e-05,
      "loss": 0.9698,
      "step": 755
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0939940993936046,
      "learning_rate": 4.9939229210341595e-05,
      "loss": 0.9609,
      "step": 760
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0622610266073254,
      "learning_rate": 4.993842729384452e-05,
      "loss": 1.0002,
      "step": 765
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1272829857252467,
      "learning_rate": 4.993762012754727e-05,
      "loss": 0.9943,
      "step": 770
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1726597080252255,
      "learning_rate": 4.993680771161976e-05,
      "loss": 0.9914,
      "step": 775
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1885807225605083,
      "learning_rate": 4.9935990046233013e-05,
      "loss": 1.04,
      "step": 780
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0252635112929571,
      "learning_rate": 4.9935167131559164e-05,
      "loss": 0.9601,
      "step": 785
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.070768351996017,
      "learning_rate": 4.993433896777144e-05,
      "loss": 0.936,
      "step": 790
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.979351135295802,
      "learning_rate": 4.9933505555044176e-05,
      "loss": 0.9474,
      "step": 795
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0923553433235567,
      "learning_rate": 4.9932666893552814e-05,
      "loss": 0.9527,
      "step": 800
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.063621339559813,
      "learning_rate": 4.993182298347391e-05,
      "loss": 0.8715,
      "step": 805
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0645635491907166,
      "learning_rate": 4.9930973824985106e-05,
      "loss": 1.0387,
      "step": 810
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4101887938562314,
      "learning_rate": 4.993011941826515e-05,
      "loss": 1.0784,
      "step": 815
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1403174474482154,
      "learning_rate": 4.992925976349393e-05,
      "loss": 0.9465,
      "step": 820
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.105964146430222,
      "learning_rate": 4.99283948608524e-05,
      "loss": 1.004,
      "step": 825
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0143092043613382,
      "learning_rate": 4.992752471052262e-05,
      "loss": 1.0008,
      "step": 830
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0563260851517375,
      "learning_rate": 4.992664931268778e-05,
      "loss": 0.9547,
      "step": 835
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1321506056379347,
      "learning_rate": 4.992576866753215e-05,
      "loss": 0.9449,
      "step": 840
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0550778724384844,
      "learning_rate": 4.992488277524112e-05,
      "loss": 0.9706,
      "step": 845
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1627731111385733,
      "learning_rate": 4.992399163600119e-05,
      "loss": 1.0122,
      "step": 850
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.084268912268043,
      "learning_rate": 4.9923095249999934e-05,
      "loss": 0.9857,
      "step": 855
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0544899608238885,
      "learning_rate": 4.992219361742606e-05,
      "loss": 1.0404,
      "step": 860
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1289084827453804,
      "learning_rate": 4.992128673846937e-05,
      "loss": 0.9843,
      "step": 865
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1240470360783996,
      "learning_rate": 4.992037461332079e-05,
      "loss": 0.9472,
      "step": 870
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1645030232116904,
      "learning_rate": 4.9919457242172294e-05,
      "loss": 1.052,
      "step": 875
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0237899559040131,
      "learning_rate": 4.991853462521703e-05,
      "loss": 0.9998,
      "step": 880
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1368450264431524,
      "learning_rate": 4.99176067626492e-05,
      "loss": 0.9078,
      "step": 885
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2101786105315597,
      "learning_rate": 4.991667365466415e-05,
      "loss": 0.8937,
      "step": 890
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9993742814448794,
      "learning_rate": 4.991573530145829e-05,
      "loss": 0.9655,
      "step": 895
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.114674160413653,
      "learning_rate": 4.9914791703229156e-05,
      "loss": 0.9457,
      "step": 900
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0911638319797456,
      "learning_rate": 4.9913842860175386e-05,
      "loss": 1.0958,
      "step": 905
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0083153225279762,
      "learning_rate": 4.9912888772496726e-05,
      "loss": 0.9445,
      "step": 910
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1273808617795962,
      "learning_rate": 4.9911929440394024e-05,
      "loss": 1.0286,
      "step": 915
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0090396888417874,
      "learning_rate": 4.991096486406922e-05,
      "loss": 0.9329,
      "step": 920
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0316018606241877,
      "learning_rate": 4.9909995043725367e-05,
      "loss": 1.0193,
      "step": 925
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0927576922289184,
      "learning_rate": 4.990901997956663e-05,
      "loss": 0.9225,
      "step": 930
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1525322010853176,
      "learning_rate": 4.990803967179827e-05,
      "loss": 1.0172,
      "step": 935
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2385687190577728,
      "learning_rate": 4.990705412062665e-05,
      "loss": 0.9959,
      "step": 940
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2865888272666233,
      "learning_rate": 4.9906063326259234e-05,
      "loss": 0.9272,
      "step": 945
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.042732327178097,
      "learning_rate": 4.99050672889046e-05,
      "loss": 0.9197,
      "step": 950
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2125210575633774,
      "learning_rate": 4.9904066008772425e-05,
      "loss": 1.0497,
      "step": 955
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0184920052524684,
      "learning_rate": 4.990305948607349e-05,
      "loss": 0.9117,
      "step": 960
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.023643684348291,
      "learning_rate": 4.9902047721019665e-05,
      "loss": 0.9475,
      "step": 965
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2246601761468199,
      "learning_rate": 4.990103071382396e-05,
      "loss": 1.0029,
      "step": 970
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1585569716875517,
      "learning_rate": 4.990000846470045e-05,
      "loss": 0.9906,
      "step": 975
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1231322075423755,
      "learning_rate": 4.989898097386433e-05,
      "loss": 0.9785,
      "step": 980
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.025858498315423,
      "learning_rate": 4.98979482415319e-05,
      "loss": 0.937,
      "step": 985
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.150967904443838,
      "learning_rate": 4.989691026792056e-05,
      "loss": 0.9761,
      "step": 990
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1880689515129315,
      "learning_rate": 4.9895867053248815e-05,
      "loss": 0.9922,
      "step": 995
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.066619733682898,
      "learning_rate": 4.989481859773628e-05,
      "loss": 1.0385,
      "step": 1000
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1150814501276582,
      "learning_rate": 4.9893764901603656e-05,
      "loss": 1.0158,
      "step": 1005
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0835925240891797,
      "learning_rate": 4.989270596507276e-05,
      "loss": 0.9909,
      "step": 1010
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1249097643403791,
      "learning_rate": 4.9891641788366504e-05,
      "loss": 0.9925,
      "step": 1015
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1070476901019226,
      "learning_rate": 4.989057237170892e-05,
      "loss": 0.9266,
      "step": 1020
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0295190981562214,
      "learning_rate": 4.988949771532512e-05,
      "loss": 0.8754,
      "step": 1025
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.034001470133943,
      "learning_rate": 4.988841781944134e-05,
      "loss": 1.0265,
      "step": 1030
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.259459341174265,
      "learning_rate": 4.9887332684284904e-05,
      "loss": 0.9695,
      "step": 1035
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0540381986986065,
      "learning_rate": 4.988624231008424e-05,
      "loss": 0.9158,
      "step": 1040
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9986687320766073,
      "learning_rate": 4.9885146697068895e-05,
      "loss": 0.9557,
      "step": 1045
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0571885372322036,
      "learning_rate": 4.98840458454695e-05,
      "loss": 0.909,
      "step": 1050
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3303470190955011,
      "learning_rate": 4.9882939755517796e-05,
      "loss": 0.929,
      "step": 1055
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4428684503209415,
      "learning_rate": 4.988182842744662e-05,
      "loss": 1.0033,
      "step": 1060
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.001470683681023,
      "learning_rate": 4.988071186148994e-05,
      "loss": 0.8846,
      "step": 1065
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5029225541016935,
      "learning_rate": 4.987959005788278e-05,
      "loss": 1.0021,
      "step": 1070
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2026572394839545,
      "learning_rate": 4.9878463016861294e-05,
      "loss": 1.0561,
      "step": 1075
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.113398975046172,
      "learning_rate": 4.987733073866275e-05,
      "loss": 0.9779,
      "step": 1080
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1850171783475074,
      "learning_rate": 4.98761932235255e-05,
      "loss": 0.9879,
      "step": 1085
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0889902195011716,
      "learning_rate": 4.9875050471688995e-05,
      "loss": 0.9488,
      "step": 1090
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.063107523673824,
      "learning_rate": 4.9873902483393804e-05,
      "loss": 0.9347,
      "step": 1095
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9517739473173996,
      "learning_rate": 4.987274925888159e-05,
      "loss": 0.9502,
      "step": 1100
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0496780085904533,
      "learning_rate": 4.9871590798395114e-05,
      "loss": 1.0158,
      "step": 1105
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9734234365963514,
      "learning_rate": 4.9870427102178245e-05,
      "loss": 0.9325,
      "step": 1110
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9700777836460857,
      "learning_rate": 4.9869258170475964e-05,
      "loss": 0.9613,
      "step": 1115
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0981548573974478,
      "learning_rate": 4.9868084003534324e-05,
      "loss": 0.8924,
      "step": 1120
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.071610990717182,
      "learning_rate": 4.9866904601600515e-05,
      "loss": 0.997,
      "step": 1125
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.049751015377668,
      "learning_rate": 4.9865719964922815e-05,
      "loss": 1.0079,
      "step": 1130
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.294098878584965,
      "learning_rate": 4.986453009375058e-05,
      "loss": 0.9611,
      "step": 1135
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0787102628401606,
      "learning_rate": 4.986333498833432e-05,
      "loss": 0.9487,
      "step": 1140
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1185282217701418,
      "learning_rate": 4.9862134648925606e-05,
      "loss": 1.0095,
      "step": 1145
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.008511461853383,
      "learning_rate": 4.986092907577711e-05,
      "loss": 0.9096,
      "step": 1150
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0146369090716436,
      "learning_rate": 4.985971826914263e-05,
      "loss": 0.8906,
      "step": 1155
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.054839433777709,
      "learning_rate": 4.9858502229277056e-05,
      "loss": 0.9606,
      "step": 1160
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.975493229451669,
      "learning_rate": 4.9857280956436366e-05,
      "loss": 0.9303,
      "step": 1165
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1929396164965356,
      "learning_rate": 4.9856054450877657e-05,
      "loss": 0.9367,
      "step": 1170
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9674508126117796,
      "learning_rate": 4.985482271285912e-05,
      "loss": 0.9015,
      "step": 1175
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0340864603119082,
      "learning_rate": 4.985358574264005e-05,
      "loss": 0.9937,
      "step": 1180
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1041655369749979,
      "learning_rate": 4.985234354048085e-05,
      "loss": 1.0129,
      "step": 1185
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1472324175635489,
      "learning_rate": 4.9851096106643003e-05,
      "loss": 0.9997,
      "step": 1190
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3921415331775695,
      "learning_rate": 4.984984344138911e-05,
      "loss": 0.8841,
      "step": 1195
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.077043943816124,
      "learning_rate": 4.984858554498287e-05,
      "loss": 0.9375,
      "step": 1200
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0989436703484008,
      "learning_rate": 4.984732241768909e-05,
      "loss": 0.9864,
      "step": 1205
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.166230423857481,
      "learning_rate": 4.984605405977367e-05,
      "loss": 0.994,
      "step": 1210
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2378507527837355,
      "learning_rate": 4.984478047150361e-05,
      "loss": 1.0142,
      "step": 1215
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1291435371399459,
      "learning_rate": 4.9843501653147005e-05,
      "loss": 1.0168,
      "step": 1220
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.136596067818321,
      "learning_rate": 4.984221760497307e-05,
      "loss": 0.9722,
      "step": 1225
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.030073369122721,
      "learning_rate": 4.9840928327252124e-05,
      "loss": 0.8917,
      "step": 1230
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0295089073150256,
      "learning_rate": 4.983963382025555e-05,
      "loss": 0.982,
      "step": 1235
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0028893637323986,
      "learning_rate": 4.983833408425586e-05,
      "loss": 0.9686,
      "step": 1240
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0108242129137919,
      "learning_rate": 4.9837029119526666e-05,
      "loss": 0.9802,
      "step": 1245
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.342055014654345,
      "learning_rate": 4.983571892634268e-05,
      "loss": 0.9146,
      "step": 1250
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9861838898768059,
      "learning_rate": 4.98344035049797e-05,
      "loss": 1.0945,
      "step": 1255
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0987386544632918,
      "learning_rate": 4.9833082855714655e-05,
      "loss": 0.9169,
      "step": 1260
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0240510509527712,
      "learning_rate": 4.9831756978825533e-05,
      "loss": 0.9588,
      "step": 1265
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.026861888806448,
      "learning_rate": 4.983042587459146e-05,
      "loss": 0.9212,
      "step": 1270
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0405422029154836,
      "learning_rate": 4.982908954329265e-05,
      "loss": 0.9657,
      "step": 1275
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0734083412242856,
      "learning_rate": 4.982774798521041e-05,
      "loss": 0.8937,
      "step": 1280
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.028849733659568,
      "learning_rate": 4.982640120062714e-05,
      "loss": 0.976,
      "step": 1285
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.5639846221607407,
      "learning_rate": 4.9825049189826364e-05,
      "loss": 0.9879,
      "step": 1290
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2306434216545792,
      "learning_rate": 4.98236919530927e-05,
      "loss": 1.0149,
      "step": 1295
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.998198552589239,
      "learning_rate": 4.982232949071185e-05,
      "loss": 1.0374,
      "step": 1300
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1037963291634068,
      "learning_rate": 4.982096180297063e-05,
      "loss": 0.9608,
      "step": 1305
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2580344419394551,
      "learning_rate": 4.9819588890156954e-05,
      "loss": 0.9281,
      "step": 1310
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9813615836076114,
      "learning_rate": 4.9818210752559836e-05,
      "loss": 1.0075,
      "step": 1315
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0726165269232262,
      "learning_rate": 4.9816827390469376e-05,
      "loss": 0.9427,
      "step": 1320
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1526945180857884,
      "learning_rate": 4.981543880417681e-05,
      "loss": 0.923,
      "step": 1325
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0887297803255238,
      "learning_rate": 4.981404499397442e-05,
      "loss": 0.9718,
      "step": 1330
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0129906993925826,
      "learning_rate": 4.981264596015565e-05,
      "loss": 0.9261,
      "step": 1335
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9977442889932219,
      "learning_rate": 4.9811241703014985e-05,
      "loss": 0.9647,
      "step": 1340
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1924499850171741,
      "learning_rate": 4.980983222284805e-05,
      "loss": 0.9626,
      "step": 1345
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9338597201109845,
      "learning_rate": 4.980841751995154e-05,
      "loss": 0.9113,
      "step": 1350
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9878861354812847,
      "learning_rate": 4.980699759462329e-05,
      "loss": 0.9838,
      "step": 1355
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.072420646942375,
      "learning_rate": 4.98055724471622e-05,
      "loss": 0.9378,
      "step": 1360
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0383692996758382,
      "learning_rate": 4.9804142077868255e-05,
      "loss": 1.0072,
      "step": 1365
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.964575336840462,
      "learning_rate": 4.9802706487042596e-05,
      "loss": 0.9501,
      "step": 1370
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0977595340605062,
      "learning_rate": 4.980126567498741e-05,
      "loss": 0.9698,
      "step": 1375
    },
    {
      "epoch": 0.16,
      "grad_norm": 43.58973051023571,
      "learning_rate": 4.979981964200601e-05,
      "loss": 1.0252,
      "step": 1380
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9996807116400318,
      "learning_rate": 4.979836838840279e-05,
      "loss": 1.0884,
      "step": 1385
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2340067790524,
      "learning_rate": 4.9796911914483266e-05,
      "loss": 0.9428,
      "step": 1390
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2437280129441672,
      "learning_rate": 4.979545022055405e-05,
      "loss": 0.8784,
      "step": 1395
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2031921231745022,
      "learning_rate": 4.979398330692282e-05,
      "loss": 0.9594,
      "step": 1400
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0385706074058578,
      "learning_rate": 4.979251117389839e-05,
      "loss": 0.9923,
      "step": 1405
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.070391407596474,
      "learning_rate": 4.979103382179067e-05,
      "loss": 0.8948,
      "step": 1410
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.313389868119059,
      "learning_rate": 4.9789551250910635e-05,
      "loss": 1.0072,
      "step": 1415
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0879150187440534,
      "learning_rate": 4.97880634615704e-05,
      "loss": 1.0532,
      "step": 1420
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9194567966273833,
      "learning_rate": 4.978657045408315e-05,
      "loss": 0.895,
      "step": 1425
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0708849906186182,
      "learning_rate": 4.978507222876318e-05,
      "loss": 0.9765,
      "step": 1430
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0456229104852912,
      "learning_rate": 4.9783568785925884e-05,
      "loss": 1.0524,
      "step": 1435
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.224747882054799,
      "learning_rate": 4.9782060125887754e-05,
      "loss": 0.9801,
      "step": 1440
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0562635138369583,
      "learning_rate": 4.978054624896638e-05,
      "loss": 0.9685,
      "step": 1445
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0770522445683612,
      "learning_rate": 4.9779027155480446e-05,
      "loss": 0.9567,
      "step": 1450
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0610209246694893,
      "learning_rate": 4.9777502845749734e-05,
      "loss": 1.0977,
      "step": 1455
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0007879688504147,
      "learning_rate": 4.977597332009514e-05,
      "loss": 1.0614,
      "step": 1460
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1286945667650947,
      "learning_rate": 4.977443857883862e-05,
      "loss": 0.9462,
      "step": 1465
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4756372869464596,
      "learning_rate": 4.977289862230328e-05,
      "loss": 0.8638,
      "step": 1470
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1007600667866084,
      "learning_rate": 4.977135345081328e-05,
      "loss": 1.0591,
      "step": 1475
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0628226523719524,
      "learning_rate": 4.976980306469391e-05,
      "loss": 0.9044,
      "step": 1480
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9976004764454622,
      "learning_rate": 4.976824746427153e-05,
      "loss": 0.9865,
      "step": 1485
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1673629310802622,
      "learning_rate": 4.976668664987361e-05,
      "loss": 0.9455,
      "step": 1490
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.019773002400684,
      "learning_rate": 4.9765120621828723e-05,
      "loss": 0.8515,
      "step": 1495
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9929101518268212,
      "learning_rate": 4.976354938046653e-05,
      "loss": 0.966,
      "step": 1500
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1659344987660303,
      "learning_rate": 4.97619729261178e-05,
      "loss": 0.8927,
      "step": 1505
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9800973379786984,
      "learning_rate": 4.976039125911438e-05,
      "loss": 0.9127,
      "step": 1510
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0293538916496288,
      "learning_rate": 4.975880437978925e-05,
      "loss": 0.8911,
      "step": 1515
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9786726951695742,
      "learning_rate": 4.975721228847644e-05,
      "loss": 0.9666,
      "step": 1520
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0950897672726212,
      "learning_rate": 4.9755614985511125e-05,
      "loss": 0.9644,
      "step": 1525
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1564485380567946,
      "learning_rate": 4.975401247122954e-05,
      "loss": 0.9354,
      "step": 1530
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.211567054566017,
      "learning_rate": 4.975240474596903e-05,
      "loss": 0.9813,
      "step": 1535
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0353263153354124,
      "learning_rate": 4.9750791810068037e-05,
      "loss": 0.9243,
      "step": 1540
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3556542000792884,
      "learning_rate": 4.974917366386611e-05,
      "loss": 1.0324,
      "step": 1545
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9965640833003246,
      "learning_rate": 4.974755030770388e-05,
      "loss": 1.0008,
      "step": 1550
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.060616779983905,
      "learning_rate": 4.974592174192308e-05,
      "loss": 0.9107,
      "step": 1555
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3131016948246124,
      "learning_rate": 4.974428796686655e-05,
      "loss": 0.903,
      "step": 1560
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0018412867219073,
      "learning_rate": 4.9742648982878206e-05,
      "loss": 0.9254,
      "step": 1565
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9838746414950001,
      "learning_rate": 4.974100479030307e-05,
      "loss": 0.941,
      "step": 1570
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9921677375009069,
      "learning_rate": 4.973935538948726e-05,
      "loss": 0.8576,
      "step": 1575
    },
    {
      "epoch": 0.18,
      "grad_norm": 32.02600340596563,
      "learning_rate": 4.9737700780778006e-05,
      "loss": 0.8981,
      "step": 1580
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0362957531909271,
      "learning_rate": 4.973604096452361e-05,
      "loss": 0.9904,
      "step": 1585
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0872834022022002,
      "learning_rate": 4.973437594107349e-05,
      "loss": 1.019,
      "step": 1590
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.671364234323081,
      "learning_rate": 4.9732705710778136e-05,
      "loss": 0.9141,
      "step": 1595
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0535153676037712,
      "learning_rate": 4.973103027398916e-05,
      "loss": 1.051,
      "step": 1600
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0544706958645702,
      "learning_rate": 4.972934963105925e-05,
      "loss": 1.0177,
      "step": 1605
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1073221246994636,
      "learning_rate": 4.9727663782342216e-05,
      "loss": 0.9306,
      "step": 1610
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.056906858082247,
      "learning_rate": 4.9725972728192925e-05,
      "loss": 1.0191,
      "step": 1615
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0980206496507552,
      "learning_rate": 4.9724276468967376e-05,
      "loss": 1.0461,
      "step": 1620
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9742009813970189,
      "learning_rate": 4.972257500502264e-05,
      "loss": 0.9959,
      "step": 1625
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1132119064493218,
      "learning_rate": 4.972086833671691e-05,
      "loss": 0.9713,
      "step": 1630
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9180554607431861,
      "learning_rate": 4.971915646440944e-05,
      "loss": 0.8513,
      "step": 1635
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2669533011597012,
      "learning_rate": 4.97174393884606e-05,
      "loss": 1.2273,
      "step": 1640
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9604477177346731,
      "learning_rate": 4.971571710923186e-05,
      "loss": 0.9914,
      "step": 1645
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0739417974073544,
      "learning_rate": 4.9713989627085774e-05,
      "loss": 0.964,
      "step": 1650
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.02137214500824,
      "learning_rate": 4.9712256942385995e-05,
      "loss": 1.0252,
      "step": 1655
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0150842290472988,
      "learning_rate": 4.971051905549726e-05,
      "loss": 0.9787,
      "step": 1660
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9700700274417345,
      "learning_rate": 4.9708775966785434e-05,
      "loss": 0.9451,
      "step": 1665
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.4431579105019137,
      "learning_rate": 4.9707027676617444e-05,
      "loss": 0.9389,
      "step": 1670
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0036855479909967,
      "learning_rate": 4.9705274185361325e-05,
      "loss": 0.8942,
      "step": 1675
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9898182747639732,
      "learning_rate": 4.97035154933862e-05,
      "loss": 0.9761,
      "step": 1680
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.004767069567346,
      "learning_rate": 4.9701751601062296e-05,
      "loss": 0.8977,
      "step": 1685
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0788304384765617,
      "learning_rate": 4.9699982508760934e-05,
      "loss": 0.9322,
      "step": 1690
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3660645241937865,
      "learning_rate": 4.969820821685452e-05,
      "loss": 0.9696,
      "step": 1695
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0565937040002569,
      "learning_rate": 4.969642872571656e-05,
      "loss": 1.0409,
      "step": 1700
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0440023633812632,
      "learning_rate": 4.9694644035721674e-05,
      "loss": 0.891,
      "step": 1705
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0819964185769326,
      "learning_rate": 4.969285414724554e-05,
      "loss": 0.9289,
      "step": 1710
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1084576623601878,
      "learning_rate": 4.9691059060664944e-05,
      "loss": 0.9735,
      "step": 1715
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0785946241396942,
      "learning_rate": 4.968925877635778e-05,
      "loss": 0.9522,
      "step": 1720
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0403368593941278,
      "learning_rate": 4.9687453294703037e-05,
      "loss": 0.9727,
      "step": 1725
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0205488654226593,
      "learning_rate": 4.968564261608077e-05,
      "loss": 0.8656,
      "step": 1730
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6384143210213529,
      "learning_rate": 4.968382674087215e-05,
      "loss": 0.9569,
      "step": 1735
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7801143627607843,
      "learning_rate": 4.968200566945944e-05,
      "loss": 0.9265,
      "step": 1740
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1349526474514298,
      "learning_rate": 4.9680179402226e-05,
      "loss": 1.0261,
      "step": 1745
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.144111915916139,
      "learning_rate": 4.967834793955627e-05,
      "loss": 0.9004,
      "step": 1750
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9958760712213459,
      "learning_rate": 4.9676511281835795e-05,
      "loss": 0.9929,
      "step": 1755
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.100461026765821,
      "learning_rate": 4.96746694294512e-05,
      "loss": 0.9216,
      "step": 1760
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9914756050641614,
      "learning_rate": 4.967282238279024e-05,
      "loss": 0.9949,
      "step": 1765
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1073400438840872,
      "learning_rate": 4.967097014224171e-05,
      "loss": 0.994,
      "step": 1770
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2147150611970752,
      "learning_rate": 4.9669112708195555e-05,
      "loss": 0.9717,
      "step": 1775
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9268310524106047,
      "learning_rate": 4.966725008104276e-05,
      "loss": 0.9517,
      "step": 1780
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9372467545249018,
      "learning_rate": 4.966538226117543e-05,
      "loss": 0.939,
      "step": 1785
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0761847075054334,
      "learning_rate": 4.966350924898677e-05,
      "loss": 0.9365,
      "step": 1790
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9995303594538358,
      "learning_rate": 4.966163104487107e-05,
      "loss": 0.9657,
      "step": 1795
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2779518847116105,
      "learning_rate": 4.96597476492237e-05,
      "loss": 0.9172,
      "step": 1800
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3880637551350963,
      "learning_rate": 4.965785906244114e-05,
      "loss": 0.9239,
      "step": 1805
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9177267522535165,
      "learning_rate": 4.965596528492096e-05,
      "loss": 0.8447,
      "step": 1810
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9420407257960208,
      "learning_rate": 4.965406631706182e-05,
      "loss": 0.8569,
      "step": 1815
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0271277300742228,
      "learning_rate": 4.965216215926347e-05,
      "loss": 0.8694,
      "step": 1820
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0650930656875617,
      "learning_rate": 4.9650252811926755e-05,
      "loss": 0.8789,
      "step": 1825
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.098138271404505,
      "learning_rate": 4.964833827545362e-05,
      "loss": 1.0128,
      "step": 1830
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8880726674631114,
      "learning_rate": 4.9646418550247086e-05,
      "loss": 0.9438,
      "step": 1835
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.37137181096444,
      "learning_rate": 4.9644493636711275e-05,
      "loss": 0.9536,
      "step": 1840
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1349121207619406,
      "learning_rate": 4.9642563535251405e-05,
      "loss": 0.8945,
      "step": 1845
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0199324476720972,
      "learning_rate": 4.9640628246273776e-05,
      "loss": 0.8518,
      "step": 1850
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.0796366441547445,
      "learning_rate": 4.96386877701858e-05,
      "loss": 0.9347,
      "step": 1855
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0265880593292798,
      "learning_rate": 4.9636742107395965e-05,
      "loss": 1.0344,
      "step": 1860
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1115498681795417,
      "learning_rate": 4.963479125831384e-05,
      "loss": 0.998,
      "step": 1865
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9983932174222161,
      "learning_rate": 4.963283522335011e-05,
      "loss": 1.0467,
      "step": 1870
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0069162672897751,
      "learning_rate": 4.963087400291654e-05,
      "loss": 0.9391,
      "step": 1875
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9449861144085916,
      "learning_rate": 4.9628907597425986e-05,
      "loss": 0.9381,
      "step": 1880
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0023292009166762,
      "learning_rate": 4.9626936007292394e-05,
      "loss": 0.8568,
      "step": 1885
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9288546742514908,
      "learning_rate": 4.962495923293081e-05,
      "loss": 0.9373,
      "step": 1890
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1206214411358377,
      "learning_rate": 4.962297727475736e-05,
      "loss": 1.012,
      "step": 1895
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.102011598389579,
      "learning_rate": 4.9620990133189274e-05,
      "loss": 0.94,
      "step": 1900
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.102609050951811,
      "learning_rate": 4.9618997808644864e-05,
      "loss": 0.9391,
      "step": 1905
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0654246307959698,
      "learning_rate": 4.961700030154353e-05,
      "loss": 0.9658,
      "step": 1910
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1005104987219085,
      "learning_rate": 4.961499761230576e-05,
      "loss": 0.9468,
      "step": 1915
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0256121111487757,
      "learning_rate": 4.9612989741353164e-05,
      "loss": 1.0237,
      "step": 1920
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0468017919184678,
      "learning_rate": 4.9610976689108415e-05,
      "loss": 0.9045,
      "step": 1925
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9989011872789753,
      "learning_rate": 4.9608958455995265e-05,
      "loss": 1.0025,
      "step": 1930
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0599468757943245,
      "learning_rate": 4.9606935042438584e-05,
      "loss": 0.9578,
      "step": 1935
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.994261797120377,
      "learning_rate": 4.960490644886433e-05,
      "loss": 0.9877,
      "step": 1940
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.999539882035706,
      "learning_rate": 4.9602872675699516e-05,
      "loss": 0.982,
      "step": 1945
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0461348478709427,
      "learning_rate": 4.960083372337231e-05,
      "loss": 0.9658,
      "step": 1950
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1214366738174446,
      "learning_rate": 4.95987895923119e-05,
      "loss": 0.9433,
      "step": 1955
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9896133914145191,
      "learning_rate": 4.959674028294862e-05,
      "loss": 1.0174,
      "step": 1960
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0654497666503353,
      "learning_rate": 4.959468579571386e-05,
      "loss": 0.9731,
      "step": 1965
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9682466593387831,
      "learning_rate": 4.9592626131040105e-05,
      "loss": 0.9353,
      "step": 1970
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0461286941244134,
      "learning_rate": 4.959056128936095e-05,
      "loss": 0.9267,
      "step": 1975
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.217230793575659,
      "learning_rate": 4.958849127111105e-05,
      "loss": 0.9779,
      "step": 1980
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.008087374556498,
      "learning_rate": 4.958641607672619e-05,
      "loss": 1.0081,
      "step": 1985
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0930387058212025,
      "learning_rate": 4.9584335706643194e-05,
      "loss": 0.9487,
      "step": 1990
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9531729301746176,
      "learning_rate": 4.958225016130002e-05,
      "loss": 0.9946,
      "step": 1995
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9619034951743967,
      "learning_rate": 4.9580159441135685e-05,
      "loss": 0.9251,
      "step": 2000
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9904024793521159,
      "learning_rate": 4.957806354659031e-05,
      "loss": 0.9031,
      "step": 2005
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0499680746726634,
      "learning_rate": 4.9575962478105115e-05,
      "loss": 0.9011,
      "step": 2010
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9945935916217389,
      "learning_rate": 4.9573856236122383e-05,
      "loss": 0.875,
      "step": 2015
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9743685311839844,
      "learning_rate": 4.9571744821085494e-05,
      "loss": 0.9096,
      "step": 2020
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0497111678386803,
      "learning_rate": 4.956962823343895e-05,
      "loss": 0.8738,
      "step": 2025
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0932387309732174,
      "learning_rate": 4.956750647362829e-05,
      "loss": 0.8564,
      "step": 2030
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.025830754286173,
      "learning_rate": 4.956537954210018e-05,
      "loss": 0.9633,
      "step": 2035
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.000027276228554,
      "learning_rate": 4.9563247439302344e-05,
      "loss": 0.8779,
      "step": 2040
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9625342626885018,
      "learning_rate": 4.956111016568363e-05,
      "loss": 0.9348,
      "step": 2045
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9763805549131147,
      "learning_rate": 4.955896772169395e-05,
      "loss": 0.9944,
      "step": 2050
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9910882147777211,
      "learning_rate": 4.955682010778432e-05,
      "loss": 0.9646,
      "step": 2055
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0309272557339653,
      "learning_rate": 4.9554667324406814e-05,
      "loss": 1.0429,
      "step": 2060
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.075680518698965,
      "learning_rate": 4.9552509372014636e-05,
      "loss": 0.8867,
      "step": 2065
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0732233966703328,
      "learning_rate": 4.955034625106204e-05,
      "loss": 0.8747,
      "step": 2070
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0703269000267739,
      "learning_rate": 4.954817796200441e-05,
      "loss": 0.9328,
      "step": 2075
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1682895759835192,
      "learning_rate": 4.954600450529817e-05,
      "loss": 0.9911,
      "step": 2080
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0605161246329127,
      "learning_rate": 4.954382588140086e-05,
      "loss": 1.0787,
      "step": 2085
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0746688585425606,
      "learning_rate": 4.9541642090771115e-05,
      "loss": 0.9707,
      "step": 2090
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1166360666019006,
      "learning_rate": 4.953945313386864e-05,
      "loss": 0.9907,
      "step": 2095
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.101683961882563,
      "learning_rate": 4.953725901115422e-05,
      "loss": 0.953,
      "step": 2100
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.056657955523902,
      "learning_rate": 4.953505972308976e-05,
      "loss": 0.9325,
      "step": 2105
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.999828227759176,
      "learning_rate": 4.9532855270138226e-05,
      "loss": 0.9419,
      "step": 2110
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9860568170600005,
      "learning_rate": 4.953064565276368e-05,
      "loss": 0.9516,
      "step": 2115
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9898163712376775,
      "learning_rate": 4.9528430871431264e-05,
      "loss": 0.9693,
      "step": 2120
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0773087335738685,
      "learning_rate": 4.952621092660722e-05,
      "loss": 0.9071,
      "step": 2125
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0436769253979625,
      "learning_rate": 4.952398581875886e-05,
      "loss": 1.004,
      "step": 2130
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9757969393954983,
      "learning_rate": 4.95217555483546e-05,
      "loss": 0.9222,
      "step": 2135
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0186056430216128,
      "learning_rate": 4.9519520115863937e-05,
      "loss": 0.9371,
      "step": 2140
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1110737550470071,
      "learning_rate": 4.9517279521757446e-05,
      "loss": 1.0049,
      "step": 2145
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.103422546430909,
      "learning_rate": 4.9515033766506805e-05,
      "loss": 0.7888,
      "step": 2150
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.929146499878184,
      "learning_rate": 4.951278285058476e-05,
      "loss": 0.9972,
      "step": 2155
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9297137090132164,
      "learning_rate": 4.951052677446515e-05,
      "loss": 0.9387,
      "step": 2160
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0057194407826662,
      "learning_rate": 4.950826553862292e-05,
      "loss": 0.9708,
      "step": 2165
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9909416049658488,
      "learning_rate": 4.950599914353406e-05,
      "loss": 0.9934,
      "step": 2170
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0313929731620215,
      "learning_rate": 4.950372758967569e-05,
      "loss": 0.9364,
      "step": 2175
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9842558788029845,
      "learning_rate": 4.950145087752598e-05,
      "loss": 1.0124,
      "step": 2180
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9856973064084651,
      "learning_rate": 4.949916900756422e-05,
      "loss": 0.9041,
      "step": 2185
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0166774424926104,
      "learning_rate": 4.949688198027075e-05,
      "loss": 0.9168,
      "step": 2190
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0145061405636069,
      "learning_rate": 4.949458979612702e-05,
      "loss": 0.9105,
      "step": 2195
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9488445041446063,
      "learning_rate": 4.9492292455615565e-05,
      "loss": 0.9902,
      "step": 2200
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1884658052352395,
      "learning_rate": 4.9489989959219986e-05,
      "loss": 0.9966,
      "step": 2205
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0515928059240325,
      "learning_rate": 4.9487682307424985e-05,
      "loss": 1.0284,
      "step": 2210
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9705709426071323,
      "learning_rate": 4.9485369500716365e-05,
      "loss": 0.9415,
      "step": 2215
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4392749705612937,
      "learning_rate": 4.948305153958098e-05,
      "loss": 0.9081,
      "step": 2220
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0914357539028203,
      "learning_rate": 4.948072842450678e-05,
      "loss": 1.0203,
      "step": 2225
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9517975268601816,
      "learning_rate": 4.9478400155982815e-05,
      "loss": 0.9065,
      "step": 2230
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9857692625038519,
      "learning_rate": 4.947606673449922e-05,
      "loss": 0.9281,
      "step": 2235
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1965974391031806,
      "learning_rate": 4.9473728160547184e-05,
      "loss": 0.9814,
      "step": 2240
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.931918323998934,
      "learning_rate": 4.947138443461901e-05,
      "loss": 0.9632,
      "step": 2245
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2424534680701529,
      "learning_rate": 4.9469035557208074e-05,
      "loss": 0.8815,
      "step": 2250
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0804964338036775,
      "learning_rate": 4.946668152880884e-05,
      "loss": 1.0169,
      "step": 2255
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0050442196221274,
      "learning_rate": 4.9464322349916866e-05,
      "loss": 0.9392,
      "step": 2260
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0672729123381384,
      "learning_rate": 4.946195802102878e-05,
      "loss": 0.9886,
      "step": 2265
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1050264427087864,
      "learning_rate": 4.9459588542642276e-05,
      "loss": 0.8766,
      "step": 2270
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0385687804876034,
      "learning_rate": 4.9457213915256186e-05,
      "loss": 0.879,
      "step": 2275
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0480805729433877,
      "learning_rate": 4.9454834139370373e-05,
      "loss": 0.9988,
      "step": 2280
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2752768724008028,
      "learning_rate": 4.9452449215485814e-05,
      "loss": 0.8835,
      "step": 2285
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9903481586303367,
      "learning_rate": 4.9450059144104554e-05,
      "loss": 0.9346,
      "step": 2290
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4711399334225232,
      "learning_rate": 4.944766392572974e-05,
      "loss": 0.8156,
      "step": 2295
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9916844541286457,
      "learning_rate": 4.9445263560865574e-05,
      "loss": 0.8965,
      "step": 2300
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0361221929417161,
      "learning_rate": 4.944285805001737e-05,
      "loss": 1.0175,
      "step": 2305
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0799656489302991,
      "learning_rate": 4.9440447393691505e-05,
      "loss": 0.9811,
      "step": 2310
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1561924038632971,
      "learning_rate": 4.943803159239546e-05,
      "loss": 0.9925,
      "step": 2315
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0217733016841795,
      "learning_rate": 4.943561064663777e-05,
      "loss": 1.0075,
      "step": 2320
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9296018093283421,
      "learning_rate": 4.943318455692808e-05,
      "loss": 0.9273,
      "step": 2325
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.947181231161687,
      "learning_rate": 4.943075332377711e-05,
      "loss": 1.004,
      "step": 2330
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9775957360955153,
      "learning_rate": 4.942831694769665e-05,
      "loss": 0.9509,
      "step": 2335
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0304155163978852,
      "learning_rate": 4.942587542919959e-05,
      "loss": 1.0683,
      "step": 2340
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.028559900699405,
      "learning_rate": 4.9423428768799895e-05,
      "loss": 0.9023,
      "step": 2345
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9355822387484952,
      "learning_rate": 4.942097696701261e-05,
      "loss": 1.0111,
      "step": 2350
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0343094695838864,
      "learning_rate": 4.941852002435387e-05,
      "loss": 0.9462,
      "step": 2355
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9490484407437914,
      "learning_rate": 4.941605794134087e-05,
      "loss": 0.8872,
      "step": 2360
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9733407846727499,
      "learning_rate": 4.9413590718491927e-05,
      "loss": 0.9536,
      "step": 2365
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0885703766584243,
      "learning_rate": 4.94111183563264e-05,
      "loss": 0.8668,
      "step": 2370
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0291312285654912,
      "learning_rate": 4.9408640855364766e-05,
      "loss": 0.9678,
      "step": 2375
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0465668480489048,
      "learning_rate": 4.940615821612854e-05,
      "loss": 0.8852,
      "step": 2380
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9576448277741111,
      "learning_rate": 4.940367043914037e-05,
      "loss": 0.8695,
      "step": 2385
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9521806529133873,
      "learning_rate": 4.940117752492394e-05,
      "loss": 0.9553,
      "step": 2390
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9965331170260615,
      "learning_rate": 4.9398679474004036e-05,
      "loss": 0.9443,
      "step": 2395
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9929617469278481,
      "learning_rate": 4.939617628690654e-05,
      "loss": 0.9898,
      "step": 2400
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.022046218347804,
      "learning_rate": 4.939366796415838e-05,
      "loss": 0.9837,
      "step": 2405
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9161635143646696,
      "learning_rate": 4.939115450628759e-05,
      "loss": 0.8949,
      "step": 2410
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.001534067477496,
      "learning_rate": 4.9388635913823276e-05,
      "loss": 0.9111,
      "step": 2415
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1066495535991843,
      "learning_rate": 4.9386112187295645e-05,
      "loss": 0.8868,
      "step": 2420
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9884705917124958,
      "learning_rate": 4.938358332723595e-05,
      "loss": 0.975,
      "step": 2425
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.072196623371505,
      "learning_rate": 4.9381049334176556e-05,
      "loss": 0.9471,
      "step": 2430
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0872776991462638,
      "learning_rate": 4.937851020865088e-05,
      "loss": 1.0266,
      "step": 2435
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0518911165357627,
      "learning_rate": 4.937596595119345e-05,
      "loss": 0.9534,
      "step": 2440
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0892624142061076,
      "learning_rate": 4.9373416562339835e-05,
      "loss": 0.9071,
      "step": 2445
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.232676411555252,
      "learning_rate": 4.9370862042626745e-05,
      "loss": 0.9999,
      "step": 2450
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0878922768218178,
      "learning_rate": 4.936830239259189e-05,
      "loss": 0.9959,
      "step": 2455
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2065797810654832,
      "learning_rate": 4.936573761277414e-05,
      "loss": 0.9064,
      "step": 2460
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0430843034159922,
      "learning_rate": 4.936316770371339e-05,
      "loss": 0.9976,
      "step": 2465
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0791384462173006,
      "learning_rate": 4.936059266595064e-05,
      "loss": 0.915,
      "step": 2470
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9768206190883695,
      "learning_rate": 4.935801250002795e-05,
      "loss": 0.9264,
      "step": 2475
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9290093480571731,
      "learning_rate": 4.935542720648849e-05,
      "loss": 0.9211,
      "step": 2480
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9576971219770666,
      "learning_rate": 4.9352836785876475e-05,
      "loss": 1.0057,
      "step": 2485
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9383325318591361,
      "learning_rate": 4.9350241238737225e-05,
      "loss": 0.961,
      "step": 2490
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0544191874502755,
      "learning_rate": 4.934764056561713e-05,
      "loss": 0.9107,
      "step": 2495
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.09184960623287,
      "learning_rate": 4.934503476706366e-05,
      "loss": 0.8734,
      "step": 2500
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.626087550208894,
      "learning_rate": 4.9342423843625354e-05,
      "loss": 0.8435,
      "step": 2505
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9188646760831191,
      "learning_rate": 4.933980779585183e-05,
      "loss": 0.8747,
      "step": 2510
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9750772856183894,
      "learning_rate": 4.933718662429383e-05,
      "loss": 0.8752,
      "step": 2515
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9119480198871687,
      "learning_rate": 4.9334560329503105e-05,
      "loss": 1.3787,
      "step": 2520
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.544445095992932,
      "learning_rate": 4.933192891203253e-05,
      "loss": 0.945,
      "step": 2525
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.081224532231963,
      "learning_rate": 4.9329292372436044e-05,
      "loss": 0.9645,
      "step": 2530
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.920134814037843,
      "learning_rate": 4.932665071126866e-05,
      "loss": 0.9326,
      "step": 2535
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0525978932870466,
      "learning_rate": 4.932400392908649e-05,
      "loss": 0.9884,
      "step": 2540
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1325733499830895,
      "learning_rate": 4.93213520264467e-05,
      "loss": 0.9399,
      "step": 2545
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9859079362445832,
      "learning_rate": 4.9318695003907545e-05,
      "loss": 0.9954,
      "step": 2550
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0165748185591408,
      "learning_rate": 4.931603286202835e-05,
      "loss": 0.869,
      "step": 2555
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9232422414048324,
      "learning_rate": 4.931336560136953e-05,
      "loss": 0.9703,
      "step": 2560
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9815186857455945,
      "learning_rate": 4.931069322249258e-05,
      "loss": 0.9752,
      "step": 2565
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9246440863691903,
      "learning_rate": 4.930801572596005e-05,
      "loss": 1.0154,
      "step": 2570
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0722734879138376,
      "learning_rate": 4.930533311233557e-05,
      "loss": 0.9113,
      "step": 2575
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9414323488721418,
      "learning_rate": 4.930264538218389e-05,
      "loss": 0.8482,
      "step": 2580
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9559687725252016,
      "learning_rate": 4.929995253607078e-05,
      "loss": 0.8771,
      "step": 2585
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9884286336038881,
      "learning_rate": 4.929725457456312e-05,
      "loss": 0.9949,
      "step": 2590
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9826661070910763,
      "learning_rate": 4.929455149822887e-05,
      "loss": 0.93,
      "step": 2595
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0180709164209465,
      "learning_rate": 4.929184330763704e-05,
      "loss": 0.9295,
      "step": 2600
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0084177859809078,
      "learning_rate": 4.9289130003357744e-05,
      "loss": 1.0116,
      "step": 2605
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.949883707096381,
      "learning_rate": 4.928641158596215e-05,
      "loss": 0.8502,
      "step": 2610
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1068546345812,
      "learning_rate": 4.928368805602252e-05,
      "loss": 1.0092,
      "step": 2615
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.954898595831912,
      "learning_rate": 4.9280959414112184e-05,
      "loss": 0.9578,
      "step": 2620
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8662476745851277,
      "learning_rate": 4.927822566080556e-05,
      "loss": 0.849,
      "step": 2625
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8849945901540008,
      "learning_rate": 4.927548679667811e-05,
      "loss": 0.9345,
      "step": 2630
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0043542735829336,
      "learning_rate": 4.9272742822306404e-05,
      "loss": 0.8519,
      "step": 2635
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0100590956998168,
      "learning_rate": 4.9269993738268084e-05,
      "loss": 0.9497,
      "step": 2640
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.082575179811648,
      "learning_rate": 4.9267239545141855e-05,
      "loss": 0.9621,
      "step": 2645
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0870659005442584,
      "learning_rate": 4.926448024350751e-05,
      "loss": 0.9921,
      "step": 2650
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.118395019353692,
      "learning_rate": 4.92617158339459e-05,
      "loss": 0.9255,
      "step": 2655
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0147065011669647,
      "learning_rate": 4.9258946317038967e-05,
      "loss": 0.9186,
      "step": 2660
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.939720111574292,
      "learning_rate": 4.9256171693369724e-05,
      "loss": 0.8523,
      "step": 2665
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0430188436900594,
      "learning_rate": 4.925339196352226e-05,
      "loss": 0.9459,
      "step": 2670
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9607539094185692,
      "learning_rate": 4.925060712808174e-05,
      "loss": 0.7983,
      "step": 2675
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0311514516159483,
      "learning_rate": 4.924781718763438e-05,
      "loss": 0.9423,
      "step": 2680
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9630614717407954,
      "learning_rate": 4.9245022142767526e-05,
      "loss": 0.9417,
      "step": 2685
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0202763718220864,
      "learning_rate": 4.924222199406954e-05,
      "loss": 0.8712,
      "step": 2690
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0158369048337754,
      "learning_rate": 4.923941674212989e-05,
      "loss": 0.8647,
      "step": 2695
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9795044626481724,
      "learning_rate": 4.92366063875391e-05,
      "loss": 0.9445,
      "step": 2700
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9642520944769797,
      "learning_rate": 4.9233790930888805e-05,
      "loss": 0.9583,
      "step": 2705
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9322141000631244,
      "learning_rate": 4.923097037277166e-05,
      "loss": 0.8844,
      "step": 2710
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9296945733326111,
      "learning_rate": 4.9228144713781435e-05,
      "loss": 0.869,
      "step": 2715
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9674562115919055,
      "learning_rate": 4.9225313954512965e-05,
      "loss": 0.9717,
      "step": 2720
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1958624855560076,
      "learning_rate": 4.922247809556215e-05,
      "loss": 0.9778,
      "step": 2725
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8583557213963918,
      "learning_rate": 4.9219637137525963e-05,
      "loss": 0.9669,
      "step": 2730
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1631100570882231,
      "learning_rate": 4.921679108100246e-05,
      "loss": 0.8419,
      "step": 2735
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.084321427743312,
      "learning_rate": 4.921393992659077e-05,
      "loss": 1.0262,
      "step": 2740
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.362756035361668,
      "learning_rate": 4.921108367489108e-05,
      "loss": 0.9827,
      "step": 2745
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9256269790194894,
      "learning_rate": 4.920822232650467e-05,
      "loss": 0.9296,
      "step": 2750
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1396124516737005,
      "learning_rate": 4.9205355882033885e-05,
      "loss": 0.944,
      "step": 2755
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9678768380588751,
      "learning_rate": 4.9202484342082135e-05,
      "loss": 0.9527,
      "step": 2760
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9661086011155546,
      "learning_rate": 4.919960770725391e-05,
      "loss": 0.8902,
      "step": 2765
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3251674666991091,
      "learning_rate": 4.919672597815478e-05,
      "loss": 0.999,
      "step": 2770
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9726361573108259,
      "learning_rate": 4.919383915539136e-05,
      "loss": 0.9536,
      "step": 2775
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.90487447873456,
      "learning_rate": 4.919094723957138e-05,
      "loss": 0.9986,
      "step": 2780
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9626231464935668,
      "learning_rate": 4.9188050231303604e-05,
      "loss": 0.9595,
      "step": 2785
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9374856877133149,
      "learning_rate": 4.918514813119789e-05,
      "loss": 0.916,
      "step": 2790
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9561303093764307,
      "learning_rate": 4.918224093986516e-05,
      "loss": 0.8875,
      "step": 2795
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4089821387226624,
      "learning_rate": 4.917932865791739e-05,
      "loss": 0.967,
      "step": 2800
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9469865846642878,
      "learning_rate": 4.917641128596767e-05,
      "loss": 0.9069,
      "step": 2805
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9949370084932259,
      "learning_rate": 4.917348882463013e-05,
      "loss": 1.0126,
      "step": 2810
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.200759951729825,
      "learning_rate": 4.9170561274519976e-05,
      "loss": 0.9478,
      "step": 2815
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9928901061210181,
      "learning_rate": 4.916762863625349e-05,
      "loss": 0.8878,
      "step": 2820
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9582105762009233,
      "learning_rate": 4.916469091044803e-05,
      "loss": 0.9834,
      "step": 2825
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1540783618074437,
      "learning_rate": 4.916174809772199e-05,
      "loss": 0.8523,
      "step": 2830
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0531775113786088,
      "learning_rate": 4.9158800198694896e-05,
      "loss": 0.9445,
      "step": 2835
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.073871804610859,
      "learning_rate": 4.9155847213987305e-05,
      "loss": 0.8917,
      "step": 2840
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1677878364422891,
      "learning_rate": 4.9152889144220835e-05,
      "loss": 0.8572,
      "step": 2845
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9954259885389227,
      "learning_rate": 4.9149925990018204e-05,
      "loss": 0.9245,
      "step": 2850
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1752106055774592,
      "learning_rate": 4.914695775200318e-05,
      "loss": 0.9651,
      "step": 2855
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1768655767032141,
      "learning_rate": 4.914398443080062e-05,
      "loss": 1.0641,
      "step": 2860
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9555298034078296,
      "learning_rate": 4.9141006027036417e-05,
      "loss": 0.9058,
      "step": 2865
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9715196417521234,
      "learning_rate": 4.913802254133758e-05,
      "loss": 0.9832,
      "step": 2870
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9869689106405393,
      "learning_rate": 4.913503397433215e-05,
      "loss": 0.9757,
      "step": 2875
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0222448994283775,
      "learning_rate": 4.913204032664926e-05,
      "loss": 0.94,
      "step": 2880
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3685013944103575,
      "learning_rate": 4.91290415989191e-05,
      "loss": 1.015,
      "step": 2885
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0819621741081182,
      "learning_rate": 4.9126037791772936e-05,
      "loss": 0.9726,
      "step": 2890
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9188569988661278,
      "learning_rate": 4.9123028905843096e-05,
      "loss": 0.8222,
      "step": 2895
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9847904938147446,
      "learning_rate": 4.912001494176298e-05,
      "loss": 0.9818,
      "step": 2900
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9722499551191345,
      "learning_rate": 4.911699590016707e-05,
      "loss": 0.9852,
      "step": 2905
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9794396170425321,
      "learning_rate": 4.911397178169089e-05,
      "loss": 0.8712,
      "step": 2910
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9277676450289588,
      "learning_rate": 4.9110942586971063e-05,
      "loss": 0.8835,
      "step": 2915
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9652657650925404,
      "learning_rate": 4.910790831664527e-05,
      "loss": 0.9498,
      "step": 2920
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0296350726273003,
      "learning_rate": 4.9104868971352234e-05,
      "loss": 0.9344,
      "step": 2925
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.151515006554475,
      "learning_rate": 4.9101824551731794e-05,
      "loss": 0.8435,
      "step": 2930
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0499957918181868,
      "learning_rate": 4.909877505842482e-05,
      "loss": 0.9727,
      "step": 2935
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.083112909131835,
      "learning_rate": 4.909572049207326e-05,
      "loss": 0.8516,
      "step": 2940
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.075195671744196,
      "learning_rate": 4.9092660853320136e-05,
      "loss": 0.9314,
      "step": 2945
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9238139779363551,
      "learning_rate": 4.908959614280953e-05,
      "loss": 0.9429,
      "step": 2950
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0908803571726293,
      "learning_rate": 4.908652636118661e-05,
      "loss": 0.8782,
      "step": 2955
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9479817659551019,
      "learning_rate": 4.908345150909758e-05,
      "loss": 0.9838,
      "step": 2960
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9766407253343226,
      "learning_rate": 4.9080371587189733e-05,
      "loss": 0.8407,
      "step": 2965
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.011893092811412,
      "learning_rate": 4.9077286596111424e-05,
      "loss": 0.8957,
      "step": 2970
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9792082715502184,
      "learning_rate": 4.9074196536512086e-05,
      "loss": 1.1728,
      "step": 2975
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9113771488229973,
      "learning_rate": 4.90711014090422e-05,
      "loss": 0.9546,
      "step": 2980
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0896247441785774,
      "learning_rate": 4.906800121435332e-05,
      "loss": 0.9244,
      "step": 2985
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.08651788115613,
      "learning_rate": 4.9064895953098086e-05,
      "loss": 0.9743,
      "step": 2990
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9835695542501453,
      "learning_rate": 4.9061785625930166e-05,
      "loss": 0.8719,
      "step": 2995
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0188446261250015,
      "learning_rate": 4.9058670233504334e-05,
      "loss": 0.8452,
      "step": 3000
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8559846064638893,
      "learning_rate": 4.9055549776476396e-05,
      "loss": 0.871,
      "step": 3005
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9977916618614925,
      "learning_rate": 4.905242425550326e-05,
      "loss": 0.8807,
      "step": 3010
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9105788014199488,
      "learning_rate": 4.904929367124286e-05,
      "loss": 0.8189,
      "step": 3015
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0117513405476202,
      "learning_rate": 4.904615802435423e-05,
      "loss": 0.9234,
      "step": 3020
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0312616631987472,
      "learning_rate": 4.904301731549746e-05,
      "loss": 0.8749,
      "step": 3025
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0374045678241164,
      "learning_rate": 4.9039871545333695e-05,
      "loss": 0.9879,
      "step": 3030
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9251833838287623,
      "learning_rate": 4.903672071452515e-05,
      "loss": 0.8905,
      "step": 3035
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9808020454280866,
      "learning_rate": 4.9033564823735115e-05,
      "loss": 0.9036,
      "step": 3040
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8634500576183282,
      "learning_rate": 4.903040387362794e-05,
      "loss": 0.937,
      "step": 3045
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0030832358770914,
      "learning_rate": 4.902723786486903e-05,
      "loss": 0.9796,
      "step": 3050
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.07901100788446,
      "learning_rate": 4.9024066798124856e-05,
      "loss": 0.8604,
      "step": 3055
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.027171957887921,
      "learning_rate": 4.902089067406298e-05,
      "loss": 0.8031,
      "step": 3060
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9953708063939386,
      "learning_rate": 4.9017709493352e-05,
      "loss": 0.9195,
      "step": 3065
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.158676130786448,
      "learning_rate": 4.901452325666159e-05,
      "loss": 0.9301,
      "step": 3070
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0047225791013188,
      "learning_rate": 4.9011331964662474e-05,
      "loss": 0.9188,
      "step": 3075
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.017150424414422,
      "learning_rate": 4.9008135618026475e-05,
      "loss": 0.9713,
      "step": 3080
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0631614941745693,
      "learning_rate": 4.900493421742644e-05,
      "loss": 0.8969,
      "step": 3085
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1453441720230393,
      "learning_rate": 4.9001727763536304e-05,
      "loss": 0.9167,
      "step": 3090
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.962536915609097,
      "learning_rate": 4.8998516257031056e-05,
      "loss": 0.912,
      "step": 3095
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9384833307069022,
      "learning_rate": 4.899529969858675e-05,
      "loss": 0.8796,
      "step": 3100
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9738294990297953,
      "learning_rate": 4.899207808888051e-05,
      "loss": 0.8597,
      "step": 3105
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9731458004211785,
      "learning_rate": 4.898885142859052e-05,
      "loss": 0.8853,
      "step": 3110
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9671123438525385,
      "learning_rate": 4.8985619718396025e-05,
      "loss": 0.9604,
      "step": 3115
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1739490131600088,
      "learning_rate": 4.8982382958977334e-05,
      "loss": 0.9938,
      "step": 3120
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9570476646533199,
      "learning_rate": 4.897914115101581e-05,
      "loss": 0.9317,
      "step": 3125
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9875332516891884,
      "learning_rate": 4.89758942951939e-05,
      "loss": 0.9771,
      "step": 3130
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0100905293852893,
      "learning_rate": 4.897264239219509e-05,
      "loss": 1.0268,
      "step": 3135
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0275794718432414,
      "learning_rate": 4.896938544270394e-05,
      "loss": 0.9127,
      "step": 3140
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0412756847375793,
      "learning_rate": 4.8966123447406084e-05,
      "loss": 0.8887,
      "step": 3145
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.987992081513018,
      "learning_rate": 4.89628564069882e-05,
      "loss": 0.9161,
      "step": 3150
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8437278171894443,
      "learning_rate": 4.895958432213803e-05,
      "loss": 0.8649,
      "step": 3155
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1196960218204823,
      "learning_rate": 4.8956307193544396e-05,
      "loss": 1.0025,
      "step": 3160
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0905613402252354,
      "learning_rate": 4.895302502189715e-05,
      "loss": 0.9304,
      "step": 3165
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9790350423791192,
      "learning_rate": 4.894973780788722e-05,
      "loss": 0.9091,
      "step": 3170
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.951724969988976,
      "learning_rate": 4.894644555220662e-05,
      "loss": 0.9091,
      "step": 3175
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8808331608565764,
      "learning_rate": 4.89431482555484e-05,
      "loss": 0.8943,
      "step": 3180
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0398444619224043,
      "learning_rate": 4.893984591860666e-05,
      "loss": 0.9017,
      "step": 3185
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1761590418893446,
      "learning_rate": 4.8936538542076596e-05,
      "loss": 0.914,
      "step": 3190
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1598503289608948,
      "learning_rate": 4.893322612665442e-05,
      "loss": 0.899,
      "step": 3195
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1788779225336012,
      "learning_rate": 4.892990867303746e-05,
      "loss": 0.849,
      "step": 3200
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.09993036242583,
      "learning_rate": 4.892658618192406e-05,
      "loss": 1.0156,
      "step": 3205
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9245927037778428,
      "learning_rate": 4.8923258654013625e-05,
      "loss": 0.7657,
      "step": 3210
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1620986771023685,
      "learning_rate": 4.891992609000666e-05,
      "loss": 0.9115,
      "step": 3215
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.960263948912723,
      "learning_rate": 4.89165884906047e-05,
      "loss": 0.8494,
      "step": 3220
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.068493760356761,
      "learning_rate": 4.8913245856510325e-05,
      "loss": 0.8763,
      "step": 3225
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.323625170395326,
      "learning_rate": 4.8909898188427214e-05,
      "loss": 0.8659,
      "step": 3230
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0375389700327822,
      "learning_rate": 4.8906545487060086e-05,
      "loss": 0.8975,
      "step": 3235
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3120119678153748,
      "learning_rate": 4.890318775311471e-05,
      "loss": 0.9708,
      "step": 3240
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0162533650075878,
      "learning_rate": 4.8899824987297925e-05,
      "loss": 0.9788,
      "step": 3245
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0115847758646506,
      "learning_rate": 4.8896457190317635e-05,
      "loss": 0.9105,
      "step": 3250
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9620865967537976,
      "learning_rate": 4.889308436288279e-05,
      "loss": 0.9193,
      "step": 3255
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.895107993395294,
      "learning_rate": 4.8889706505703407e-05,
      "loss": 0.865,
      "step": 3260
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9613138314308484,
      "learning_rate": 4.8886323619490566e-05,
      "loss": 0.936,
      "step": 3265
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8906143795908595,
      "learning_rate": 4.888293570495639e-05,
      "loss": 0.8952,
      "step": 3270
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9234029374816892,
      "learning_rate": 4.887954276281409e-05,
      "loss": 0.962,
      "step": 3275
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9402627059485674,
      "learning_rate": 4.887614479377789e-05,
      "loss": 0.9338,
      "step": 3280
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9667260247135531,
      "learning_rate": 4.88727417985631e-05,
      "loss": 0.89,
      "step": 3285
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8707815571078674,
      "learning_rate": 4.886933377788612e-05,
      "loss": 0.8951,
      "step": 3290
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9174998711832358,
      "learning_rate": 4.886592073246434e-05,
      "loss": 0.8957,
      "step": 3295
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8704654666554288,
      "learning_rate": 4.886250266301625e-05,
      "loss": 0.9659,
      "step": 3300
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.923399218666802,
      "learning_rate": 4.885907957026139e-05,
      "loss": 0.9264,
      "step": 3305
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9726739535206362,
      "learning_rate": 4.885565145492036e-05,
      "loss": 0.9814,
      "step": 3310
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9081310006502049,
      "learning_rate": 4.885221831771482e-05,
      "loss": 0.8522,
      "step": 3315
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0783138772420493,
      "learning_rate": 4.884878015936747e-05,
      "loss": 0.9699,
      "step": 3320
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9963254199382147,
      "learning_rate": 4.884533698060208e-05,
      "loss": 1.0089,
      "step": 3325
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2552027299542712,
      "learning_rate": 4.884188878214347e-05,
      "loss": 0.8451,
      "step": 3330
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9285096007241638,
      "learning_rate": 4.883843556471754e-05,
      "loss": 0.8354,
      "step": 3335
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8832948838268647,
      "learning_rate": 4.8834977329051225e-05,
      "loss": 0.9583,
      "step": 3340
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8645112372948729,
      "learning_rate": 4.8831514075872495e-05,
      "loss": 0.934,
      "step": 3345
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0339029443434846,
      "learning_rate": 4.882804580591043e-05,
      "loss": 0.9797,
      "step": 3350
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0538097924029293,
      "learning_rate": 4.882457251989513e-05,
      "loss": 0.9737,
      "step": 3355
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1217270662925884,
      "learning_rate": 4.882109421855775e-05,
      "loss": 0.9868,
      "step": 3360
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9589544546099569,
      "learning_rate": 4.881761090263051e-05,
      "loss": 0.995,
      "step": 3365
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.045924708142707,
      "learning_rate": 4.8814122572846686e-05,
      "loss": 0.8722,
      "step": 3370
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.042472714033629,
      "learning_rate": 4.881062922994061e-05,
      "loss": 0.9242,
      "step": 3375
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8493291573005518,
      "learning_rate": 4.880713087464767e-05,
      "loss": 0.8903,
      "step": 3380
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9657697879521427,
      "learning_rate": 4.88036275077043e-05,
      "loss": 0.9369,
      "step": 3385
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9737366223994887,
      "learning_rate": 4.8800119129848e-05,
      "loss": 1.0918,
      "step": 3390
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9672779862477917,
      "learning_rate": 4.879660574181733e-05,
      "loss": 0.9547,
      "step": 3395
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9281581147091152,
      "learning_rate": 4.879308734435186e-05,
      "loss": 0.8319,
      "step": 3400
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9629420634129155,
      "learning_rate": 4.878956393819229e-05,
      "loss": 0.896,
      "step": 3405
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9143774121504513,
      "learning_rate": 4.878603552408032e-05,
      "loss": 0.9587,
      "step": 3410
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.921423590452113,
      "learning_rate": 4.8782502102758705e-05,
      "loss": 0.9646,
      "step": 3415
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9414297550730459,
      "learning_rate": 4.8778963674971276e-05,
      "loss": 0.9735,
      "step": 3420
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.035230794090253,
      "learning_rate": 4.877542024146292e-05,
      "loss": 0.8706,
      "step": 3425
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9719924579135688,
      "learning_rate": 4.877187180297955e-05,
      "loss": 0.9009,
      "step": 3430
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9060580029938262,
      "learning_rate": 4.876831836026816e-05,
      "loss": 0.9422,
      "step": 3435
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0369044673459387,
      "learning_rate": 4.876475991407679e-05,
      "loss": 0.942,
      "step": 3440
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1387762914430315,
      "learning_rate": 4.876119646515452e-05,
      "loss": 0.8789,
      "step": 3445
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.037051145404901,
      "learning_rate": 4.875762801425149e-05,
      "loss": 0.9871,
      "step": 3450
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4205212192870267,
      "learning_rate": 4.875405456211891e-05,
      "loss": 0.9601,
      "step": 3455
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9113096528951976,
      "learning_rate": 4.875047610950903e-05,
      "loss": 0.8599,
      "step": 3460
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.938545710586039,
      "learning_rate": 4.874689265717514e-05,
      "loss": 0.8487,
      "step": 3465
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9578703197917051,
      "learning_rate": 4.874330420587159e-05,
      "loss": 0.9043,
      "step": 3470
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8878961559275605,
      "learning_rate": 4.8739710756353806e-05,
      "loss": 0.8568,
      "step": 3475
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0774414056263133,
      "learning_rate": 4.873611230937824e-05,
      "loss": 0.9369,
      "step": 3480
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.848770373189615,
      "learning_rate": 4.8732508865702383e-05,
      "loss": 0.9162,
      "step": 3485
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1187796611034349,
      "learning_rate": 4.872890042608482e-05,
      "loss": 0.9479,
      "step": 3490
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9799458708066113,
      "learning_rate": 4.8725286991285166e-05,
      "loss": 0.918,
      "step": 3495
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9758562791483246,
      "learning_rate": 4.872166856206407e-05,
      "loss": 0.8546,
      "step": 3500
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9417872835004778,
      "learning_rate": 4.871804513918327e-05,
      "loss": 0.8721,
      "step": 3505
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9778738679536334,
      "learning_rate": 4.871441672340551e-05,
      "loss": 1.0115,
      "step": 3510
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9285846263492525,
      "learning_rate": 4.8710783315494634e-05,
      "loss": 0.8275,
      "step": 3515
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.014052487657275,
      "learning_rate": 4.8707144916215495e-05,
      "loss": 0.9113,
      "step": 3520
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9787743356118391,
      "learning_rate": 4.870350152633403e-05,
      "loss": 0.8502,
      "step": 3525
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2207431321000475,
      "learning_rate": 4.86998531466172e-05,
      "loss": 0.9187,
      "step": 3530
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1431518242301262,
      "learning_rate": 4.869619977783302e-05,
      "loss": 0.8886,
      "step": 3535
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9623796712042414,
      "learning_rate": 4.869254142075057e-05,
      "loss": 0.9358,
      "step": 3540
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9647785006447487,
      "learning_rate": 4.868887807613999e-05,
      "loss": 0.8993,
      "step": 3545
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.992140586359378,
      "learning_rate": 4.868520974477242e-05,
      "loss": 0.9473,
      "step": 3550
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1316075323489232,
      "learning_rate": 4.8681536427420106e-05,
      "loss": 0.8848,
      "step": 3555
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.002368008144341,
      "learning_rate": 4.8677858124856304e-05,
      "loss": 0.9134,
      "step": 3560
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9906151117851623,
      "learning_rate": 4.867417483785534e-05,
      "loss": 0.9978,
      "step": 3565
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9840590733033028,
      "learning_rate": 4.867048656719261e-05,
      "loss": 0.7565,
      "step": 3570
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9736154818193262,
      "learning_rate": 4.866679331364449e-05,
      "loss": 0.9821,
      "step": 3575
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9857477854974904,
      "learning_rate": 4.8663095077988476e-05,
      "loss": 0.8801,
      "step": 3580
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9848764944902861,
      "learning_rate": 4.865939186100308e-05,
      "loss": 0.9331,
      "step": 3585
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.868983713213606,
      "learning_rate": 4.8655683663467866e-05,
      "loss": 0.9015,
      "step": 3590
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9097275723858047,
      "learning_rate": 4.865197048616345e-05,
      "loss": 0.9275,
      "step": 3595
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9232703938802409,
      "learning_rate": 4.8648252329871494e-05,
      "loss": 0.9712,
      "step": 3600
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0454061209137948,
      "learning_rate": 4.864452919537471e-05,
      "loss": 0.8645,
      "step": 3605
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9503720222629888,
      "learning_rate": 4.8640801083456855e-05,
      "loss": 0.9811,
      "step": 3610
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.002101452538871,
      "learning_rate": 4.863706799490273e-05,
      "loss": 0.97,
      "step": 3615
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0758078400103814,
      "learning_rate": 4.8633329930498205e-05,
      "loss": 0.9431,
      "step": 3620
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1113795208235766,
      "learning_rate": 4.862958689103017e-05,
      "loss": 0.9679,
      "step": 3625
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3153025648107326,
      "learning_rate": 4.862583887728658e-05,
      "loss": 0.8778,
      "step": 3630
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0315773528641883,
      "learning_rate": 4.862208589005642e-05,
      "loss": 0.9315,
      "step": 3635
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2002600160762993,
      "learning_rate": 4.861832793012975e-05,
      "loss": 0.9482,
      "step": 3640
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.015632637034598,
      "learning_rate": 4.861456499829764e-05,
      "loss": 0.9335,
      "step": 3645
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7865503244245898,
      "learning_rate": 4.8610797095352244e-05,
      "loss": 0.9218,
      "step": 3650
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.93604473107611,
      "learning_rate": 4.8607024222086736e-05,
      "loss": 0.8848,
      "step": 3655
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9335809782245598,
      "learning_rate": 4.8603246379295345e-05,
      "loss": 0.8826,
      "step": 3660
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8557833747486344,
      "learning_rate": 4.8599463567773354e-05,
      "loss": 0.8693,
      "step": 3665
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2620023598165786,
      "learning_rate": 4.859567578831708e-05,
      "loss": 0.9407,
      "step": 3670
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9554266700919634,
      "learning_rate": 4.8591883041723886e-05,
      "loss": 0.8954,
      "step": 3675
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9436274921228281,
      "learning_rate": 4.8588085328792185e-05,
      "loss": 0.9836,
      "step": 3680
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1258174269715784,
      "learning_rate": 4.858428265032145e-05,
      "loss": 0.9217,
      "step": 3685
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9403918675581807,
      "learning_rate": 4.858047500711217e-05,
      "loss": 0.9027,
      "step": 3690
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9414728823176005,
      "learning_rate": 4.857666239996589e-05,
      "loss": 0.9704,
      "step": 3695
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8516341972896481,
      "learning_rate": 4.8572844829685215e-05,
      "loss": 0.9258,
      "step": 3700
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9603444658951118,
      "learning_rate": 4.856902229707378e-05,
      "loss": 0.864,
      "step": 3705
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8887753326166056,
      "learning_rate": 4.8565194802936274e-05,
      "loss": 0.8712,
      "step": 3710
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9899098582615348,
      "learning_rate": 4.8561362348078406e-05,
      "loss": 0.9132,
      "step": 3715
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0404052953709504,
      "learning_rate": 4.8557524933306976e-05,
      "loss": 0.8879,
      "step": 3720
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.15102975947701,
      "learning_rate": 4.8553682559429774e-05,
      "loss": 1.0017,
      "step": 3725
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9237196737475427,
      "learning_rate": 4.854983522725568e-05,
      "loss": 0.9869,
      "step": 3730
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8964349023778845,
      "learning_rate": 4.8545982937594576e-05,
      "loss": 0.9358,
      "step": 3735
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9484986614369278,
      "learning_rate": 4.854212569125742e-05,
      "loss": 0.904,
      "step": 3740
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0694168360652927,
      "learning_rate": 4.853826348905621e-05,
      "loss": 0.9153,
      "step": 3745
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9569871125779592,
      "learning_rate": 4.853439633180398e-05,
      "loss": 0.9858,
      "step": 3750
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1165116866806482,
      "learning_rate": 4.85305242203148e-05,
      "loss": 0.9694,
      "step": 3755
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0063975761554997,
      "learning_rate": 4.852664715540378e-05,
      "loss": 0.9284,
      "step": 3760
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9712906839689703,
      "learning_rate": 4.8522765137887104e-05,
      "loss": 0.993,
      "step": 3765
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9378860687864021,
      "learning_rate": 4.851887816858196e-05,
      "loss": 0.8751,
      "step": 3770
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0293388298169437,
      "learning_rate": 4.851498624830661e-05,
      "loss": 0.9277,
      "step": 3775
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.964803385452915,
      "learning_rate": 4.851108937788034e-05,
      "loss": 0.9359,
      "step": 3780
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9552239954787749,
      "learning_rate": 4.8507187558123466e-05,
      "loss": 0.9062,
      "step": 3785
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9825883172444517,
      "learning_rate": 4.850328078985739e-05,
      "loss": 0.8924,
      "step": 3790
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4606665057540582,
      "learning_rate": 4.84993690739045e-05,
      "loss": 0.9281,
      "step": 3795
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8861706171878336,
      "learning_rate": 4.849545241108827e-05,
      "loss": 0.8769,
      "step": 3800
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0164064103564012,
      "learning_rate": 4.8491530802233195e-05,
      "loss": 0.9701,
      "step": 3805
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0018902462532118,
      "learning_rate": 4.848760424816482e-05,
      "loss": 0.9851,
      "step": 3810
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0023656745784852,
      "learning_rate": 4.848367274970972e-05,
      "loss": 0.8762,
      "step": 3815
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9686707426613198,
      "learning_rate": 4.8479736307695514e-05,
      "loss": 0.9294,
      "step": 3820
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9814802177232217,
      "learning_rate": 4.8475794922950855e-05,
      "loss": 0.9429,
      "step": 3825
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9630873007715853,
      "learning_rate": 4.8471848596305466e-05,
      "loss": 1.0083,
      "step": 3830
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8831676045344481,
      "learning_rate": 4.846789732859009e-05,
      "loss": 0.8681,
      "step": 3835
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9343165035221371,
      "learning_rate": 4.84639411206365e-05,
      "loss": 0.9011,
      "step": 3840
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8607232413760687,
      "learning_rate": 4.845997997327751e-05,
      "loss": 0.9791,
      "step": 3845
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9253966585760002,
      "learning_rate": 4.8456013887347007e-05,
      "loss": 1.0433,
      "step": 3850
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0537063093602257,
      "learning_rate": 4.8452042863679867e-05,
      "loss": 0.9764,
      "step": 3855
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0574085332855863,
      "learning_rate": 4.844806690311206e-05,
      "loss": 0.911,
      "step": 3860
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9166911162048151,
      "learning_rate": 4.844408600648055e-05,
      "loss": 0.9462,
      "step": 3865
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9639905473307447,
      "learning_rate": 4.8440100174623354e-05,
      "loss": 0.9236,
      "step": 3870
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.185921639828869,
      "learning_rate": 4.843610940837954e-05,
      "loss": 0.9108,
      "step": 3875
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9631774554939261,
      "learning_rate": 4.843211370858921e-05,
      "loss": 0.8965,
      "step": 3880
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1035487993577413,
      "learning_rate": 4.842811307609349e-05,
      "loss": 0.9307,
      "step": 3885
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9889554691312329,
      "learning_rate": 4.8424107511734564e-05,
      "loss": 0.881,
      "step": 3890
    },
    {
      "epoch": 0.46,
      "grad_norm": 32.859193072098826,
      "learning_rate": 4.8420097016355636e-05,
      "loss": 1.0766,
      "step": 3895
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9531821021610071,
      "learning_rate": 4.841608159080097e-05,
      "loss": 0.8961,
      "step": 3900
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1455809758226858,
      "learning_rate": 4.841206123591584e-05,
      "loss": 0.9413,
      "step": 3905
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9362217231198446,
      "learning_rate": 4.8408035952546585e-05,
      "loss": 0.944,
      "step": 3910
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8591659987270633,
      "learning_rate": 4.840400574154056e-05,
      "loss": 0.9388,
      "step": 3915
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.998473689677426,
      "learning_rate": 4.839997060374617e-05,
      "loss": 0.9462,
      "step": 3920
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9057666617989911,
      "learning_rate": 4.8395930540012856e-05,
      "loss": 0.9196,
      "step": 3925
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9816403390756506,
      "learning_rate": 4.83918855511911e-05,
      "loss": 0.8476,
      "step": 3930
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9620236884016036,
      "learning_rate": 4.83878356381324e-05,
      "loss": 0.9618,
      "step": 3935
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9579678045099858,
      "learning_rate": 4.8383780801689316e-05,
      "loss": 0.8265,
      "step": 3940
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0364290194181867,
      "learning_rate": 4.837972104271543e-05,
      "loss": 0.9419,
      "step": 3945
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9267244060802147,
      "learning_rate": 4.837565636206536e-05,
      "loss": 0.9367,
      "step": 3950
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0341605920182162,
      "learning_rate": 4.837158676059478e-05,
      "loss": 0.8827,
      "step": 3955
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9497141268959044,
      "learning_rate": 4.8367512239160356e-05,
      "loss": 0.9438,
      "step": 3960
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9076414951706951,
      "learning_rate": 4.8363432798619837e-05,
      "loss": 0.9458,
      "step": 3965
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9072736861592536,
      "learning_rate": 4.835934843983199e-05,
      "loss": 0.8802,
      "step": 3970
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2747160746450632,
      "learning_rate": 4.835525916365661e-05,
      "loss": 0.9166,
      "step": 3975
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1995159217567877,
      "learning_rate": 4.835116497095452e-05,
      "loss": 1.0289,
      "step": 3980
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0060549362142395,
      "learning_rate": 4.834706586258762e-05,
      "loss": 0.9123,
      "step": 3985
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1419137891389273,
      "learning_rate": 4.834296183941878e-05,
      "loss": 0.9299,
      "step": 3990
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0463373724405032,
      "learning_rate": 4.833885290231196e-05,
      "loss": 0.95,
      "step": 3995
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9563839536977482,
      "learning_rate": 4.833473905213214e-05,
      "loss": 0.8681,
      "step": 4000
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7736234244855686,
      "learning_rate": 4.833062028974531e-05,
      "loss": 0.9081,
      "step": 4005
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0449183588325297,
      "learning_rate": 4.832649661601853e-05,
      "loss": 0.896,
      "step": 4010
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.050794809840981,
      "learning_rate": 4.8322368031819866e-05,
      "loss": 0.9457,
      "step": 4015
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8857422867920175,
      "learning_rate": 4.8318234538018436e-05,
      "loss": 0.8709,
      "step": 4020
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9857144285928388,
      "learning_rate": 4.831409613548437e-05,
      "loss": 0.9058,
      "step": 4025
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.089070194398636,
      "learning_rate": 4.830995282508886e-05,
      "loss": 0.9154,
      "step": 4030
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.100980387334043,
      "learning_rate": 4.8305804607704105e-05,
      "loss": 0.9042,
      "step": 4035
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1858151008226716,
      "learning_rate": 4.830165148420337e-05,
      "loss": 0.927,
      "step": 4040
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0286130018827226,
      "learning_rate": 4.8297493455460895e-05,
      "loss": 0.954,
      "step": 4045
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0712085938152927,
      "learning_rate": 4.829333052235202e-05,
      "loss": 0.8351,
      "step": 4050
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9593806141720698,
      "learning_rate": 4.828916268575308e-05,
      "loss": 0.9098,
      "step": 4055
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8772398047170469,
      "learning_rate": 4.828498994654143e-05,
      "loss": 0.9087,
      "step": 4060
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9115033207309853,
      "learning_rate": 4.8280812305595494e-05,
      "loss": 0.8425,
      "step": 4065
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.038099343316853,
      "learning_rate": 4.827662976379469e-05,
      "loss": 0.923,
      "step": 4070
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9044012139299706,
      "learning_rate": 4.8272442322019514e-05,
      "loss": 0.9471,
      "step": 4075
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9995296892135088,
      "learning_rate": 4.826824998115145e-05,
      "loss": 0.8989,
      "step": 4080
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9668511561268096,
      "learning_rate": 4.826405274207303e-05,
      "loss": 0.9637,
      "step": 4085
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9369342435071403,
      "learning_rate": 4.8259850605667824e-05,
      "loss": 0.9217,
      "step": 4090
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8873437758274902,
      "learning_rate": 4.825564357282041e-05,
      "loss": 0.9025,
      "step": 4095
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.093349441361928,
      "learning_rate": 4.8251431644416434e-05,
      "loss": 0.8621,
      "step": 4100
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.905289402143519,
      "learning_rate": 4.824721482134253e-05,
      "loss": 0.8614,
      "step": 4105
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3170466529933793,
      "learning_rate": 4.82429931044864e-05,
      "loss": 0.9297,
      "step": 4110
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9723323292609203,
      "learning_rate": 4.8238766494736756e-05,
      "loss": 0.8245,
      "step": 4115
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1199257421067867,
      "learning_rate": 4.8234534992983324e-05,
      "loss": 0.8882,
      "step": 4120
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2417072458316452,
      "learning_rate": 4.823029860011692e-05,
      "loss": 0.9395,
      "step": 4125
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.909759254924053,
      "learning_rate": 4.8226057317029305e-05,
      "loss": 0.9428,
      "step": 4130
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9896126573154411,
      "learning_rate": 4.822181114461334e-05,
      "loss": 0.9567,
      "step": 4135
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.10468734439763,
      "learning_rate": 4.821756008376289e-05,
      "loss": 0.8813,
      "step": 4140
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0242063783201032,
      "learning_rate": 4.821330413537283e-05,
      "loss": 0.9361,
      "step": 4145
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0106293242206477,
      "learning_rate": 4.82090433003391e-05,
      "loss": 0.9634,
      "step": 4150
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.013843054008336,
      "learning_rate": 4.820477757955864e-05,
      "loss": 0.9401,
      "step": 4155
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0293826642364385,
      "learning_rate": 4.820050697392943e-05,
      "loss": 0.9863,
      "step": 4160
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.887095006123414,
      "learning_rate": 4.819623148435048e-05,
      "loss": 0.9072,
      "step": 4165
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8721552648061809,
      "learning_rate": 4.8191951111721824e-05,
      "loss": 0.8849,
      "step": 4170
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5979987854315105,
      "learning_rate": 4.818766585694452e-05,
      "loss": 0.999,
      "step": 4175
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9340425058718178,
      "learning_rate": 4.818337572092067e-05,
      "loss": 0.8784,
      "step": 4180
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9430409666393136,
      "learning_rate": 4.8179080704553386e-05,
      "loss": 0.8564,
      "step": 4185
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9335305379224227,
      "learning_rate": 4.817478080874681e-05,
      "loss": 0.9266,
      "step": 4190
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8789693682753087,
      "learning_rate": 4.817047603440613e-05,
      "loss": 0.931,
      "step": 4195
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9770230379694931,
      "learning_rate": 4.816616638243753e-05,
      "loss": 0.8626,
      "step": 4200
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.974435575411324,
      "learning_rate": 4.816185185374824e-05,
      "loss": 0.9195,
      "step": 4205
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0296248792216185,
      "learning_rate": 4.8157532449246524e-05,
      "loss": 0.8555,
      "step": 4210
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9829890377233504,
      "learning_rate": 4.8153208169841646e-05,
      "loss": 0.8907,
      "step": 4215
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0199595843789433,
      "learning_rate": 4.814887901644392e-05,
      "loss": 0.9138,
      "step": 4220
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9120296230173331,
      "learning_rate": 4.814454498996468e-05,
      "loss": 0.9015,
      "step": 4225
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9967938476619388,
      "learning_rate": 4.8140206091316284e-05,
      "loss": 0.9581,
      "step": 4230
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9148132145145298,
      "learning_rate": 4.813586232141212e-05,
      "loss": 0.9081,
      "step": 4235
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8808678011367258,
      "learning_rate": 4.813151368116658e-05,
      "loss": 0.9289,
      "step": 4240
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8910130972642033,
      "learning_rate": 4.8127160171495115e-05,
      "loss": 0.9786,
      "step": 4245
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9160860753616116,
      "learning_rate": 4.8122801793314174e-05,
      "loss": 0.8792,
      "step": 4250
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2075292463237108,
      "learning_rate": 4.8118438547541254e-05,
      "loss": 0.9563,
      "step": 4255
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.103142596488639,
      "learning_rate": 4.811407043509486e-05,
      "loss": 0.9339,
      "step": 4260
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0403493518530094,
      "learning_rate": 4.810969745689451e-05,
      "loss": 0.8931,
      "step": 4265
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4496385194282688,
      "learning_rate": 4.8105319613860776e-05,
      "loss": 0.8533,
      "step": 4270
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0841434371550498,
      "learning_rate": 4.810093690691524e-05,
      "loss": 0.9292,
      "step": 4275
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9015486947572188,
      "learning_rate": 4.80965493369805e-05,
      "loss": 0.8138,
      "step": 4280
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0521073120390954,
      "learning_rate": 4.809215690498019e-05,
      "loss": 0.8947,
      "step": 4285
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8972151213895544,
      "learning_rate": 4.808775961183897e-05,
      "loss": 0.8401,
      "step": 4290
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0161410560430715,
      "learning_rate": 4.8083357458482505e-05,
      "loss": 0.8835,
      "step": 4295
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0165586013998356,
      "learning_rate": 4.8078950445837494e-05,
      "loss": 0.9276,
      "step": 4300
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9651747648409673,
      "learning_rate": 4.807453857483167e-05,
      "loss": 0.7909,
      "step": 4305
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.964488308075261,
      "learning_rate": 4.8070121846393766e-05,
      "loss": 0.8243,
      "step": 4310
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8766598782448897,
      "learning_rate": 4.806570026145356e-05,
      "loss": 0.9263,
      "step": 4315
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8604352610375053,
      "learning_rate": 4.8061273820941845e-05,
      "loss": 0.8852,
      "step": 4320
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0340822406917236,
      "learning_rate": 4.8056842525790415e-05,
      "loss": 0.7946,
      "step": 4325
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9541468500778845,
      "learning_rate": 4.805240637693212e-05,
      "loss": 0.8285,
      "step": 4330
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8857830983714304,
      "learning_rate": 4.8047965375300806e-05,
      "loss": 0.899,
      "step": 4335
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9166475002622385,
      "learning_rate": 4.804351952183136e-05,
      "loss": 0.8427,
      "step": 4340
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9182539569609515,
      "learning_rate": 4.803906881745968e-05,
      "loss": 0.8439,
      "step": 4345
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0236257744641755,
      "learning_rate": 4.803461326312267e-05,
      "loss": 0.9148,
      "step": 4350
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.880806259464865,
      "learning_rate": 4.8030152859758284e-05,
      "loss": 0.8697,
      "step": 4355
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.903662387966118,
      "learning_rate": 4.8025687608305494e-05,
      "loss": 0.9598,
      "step": 4360
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1530261481497712,
      "learning_rate": 4.802121750970425e-05,
      "loss": 0.8588,
      "step": 4365
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0126756005036486,
      "learning_rate": 4.8016742564895595e-05,
      "loss": 0.9082,
      "step": 4370
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0059117894191174,
      "learning_rate": 4.801226277482153e-05,
      "loss": 0.9754,
      "step": 4375
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9557827124825298,
      "learning_rate": 4.8007778140425084e-05,
      "loss": 0.9483,
      "step": 4380
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.994421892397985,
      "learning_rate": 4.800328866265035e-05,
      "loss": 0.9208,
      "step": 4385
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8896209300235811,
      "learning_rate": 4.7998794342442385e-05,
      "loss": 0.8685,
      "step": 4390
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9706044832348802,
      "learning_rate": 4.79942951807473e-05,
      "loss": 0.951,
      "step": 4395
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9421807432329478,
      "learning_rate": 4.798979117851222e-05,
      "loss": 0.9295,
      "step": 4400
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9521430855453626,
      "learning_rate": 4.7985282336685277e-05,
      "loss": 0.8864,
      "step": 4405
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.154205685917419,
      "learning_rate": 4.798076865621564e-05,
      "loss": 0.8945,
      "step": 4410
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.92249520650557,
      "learning_rate": 4.797625013805348e-05,
      "loss": 0.9049,
      "step": 4415
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9648365961624972,
      "learning_rate": 4.7971726783149985e-05,
      "loss": 0.9322,
      "step": 4420
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9710002033355778,
      "learning_rate": 4.7967198592457376e-05,
      "loss": 0.853,
      "step": 4425
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9733922923210124,
      "learning_rate": 4.796266556692889e-05,
      "loss": 0.9371,
      "step": 4430
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8548254886728215,
      "learning_rate": 4.7958127707518765e-05,
      "loss": 0.8413,
      "step": 4435
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9504784552774678,
      "learning_rate": 4.7953585015182266e-05,
      "loss": 0.9082,
      "step": 4440
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8843052821219075,
      "learning_rate": 4.79490374908757e-05,
      "loss": 0.9312,
      "step": 4445
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9917184205598669,
      "learning_rate": 4.794448513555634e-05,
      "loss": 0.93,
      "step": 4450
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8953364964928886,
      "learning_rate": 4.793992795018253e-05,
      "loss": 0.8632,
      "step": 4455
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9121544104994831,
      "learning_rate": 4.793536593571358e-05,
      "loss": 0.8949,
      "step": 4460
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0186207045283104,
      "learning_rate": 4.7930799093109855e-05,
      "loss": 0.9303,
      "step": 4465
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9036844654904278,
      "learning_rate": 4.792622742333273e-05,
      "loss": 0.9547,
      "step": 4470
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9890914415279327,
      "learning_rate": 4.792165092734458e-05,
      "loss": 0.8666,
      "step": 4475
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0815879001314463,
      "learning_rate": 4.79170696061088e-05,
      "loss": 0.9001,
      "step": 4480
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0436150477506996,
      "learning_rate": 4.791248346058983e-05,
      "loss": 0.9092,
      "step": 4485
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9496671598384474,
      "learning_rate": 4.790789249175307e-05,
      "loss": 0.9871,
      "step": 4490
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9440921396640157,
      "learning_rate": 4.7903296700564994e-05,
      "loss": 0.8873,
      "step": 4495
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9228445487502488,
      "learning_rate": 4.789869608799304e-05,
      "loss": 0.9328,
      "step": 4500
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.935063192093136,
      "learning_rate": 4.7894090655005705e-05,
      "loss": 0.9226,
      "step": 4505
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9415307777230048,
      "learning_rate": 4.7889480402572474e-05,
      "loss": 0.9135,
      "step": 4510
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0131784606673566,
      "learning_rate": 4.788486533166384e-05,
      "loss": 0.8674,
      "step": 4515
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9766246881415149,
      "learning_rate": 4.788024544325135e-05,
      "loss": 0.8623,
      "step": 4520
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9175019038199792,
      "learning_rate": 4.787562073830753e-05,
      "loss": 0.9839,
      "step": 4525
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9077423233786132,
      "learning_rate": 4.787099121780591e-05,
      "loss": 0.9163,
      "step": 4530
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7487036207781055,
      "learning_rate": 4.7866356882721074e-05,
      "loss": 0.8959,
      "step": 4535
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.929568282711205,
      "learning_rate": 4.786171773402859e-05,
      "loss": 0.8877,
      "step": 4540
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.007750144587037,
      "learning_rate": 4.785707377270505e-05,
      "loss": 1.0033,
      "step": 4545
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.921980687483201,
      "learning_rate": 4.785242499972805e-05,
      "loss": 0.844,
      "step": 4550
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0735634987598275,
      "learning_rate": 4.784777141607622e-05,
      "loss": 0.9172,
      "step": 4555
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8729592164633109,
      "learning_rate": 4.784311302272917e-05,
      "loss": 0.847,
      "step": 4560
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0716544675804034,
      "learning_rate": 4.783844982066755e-05,
      "loss": 0.9323,
      "step": 4565
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8876233822448721,
      "learning_rate": 4.783378181087301e-05,
      "loss": 0.8679,
      "step": 4570
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9759926093199263,
      "learning_rate": 4.782910899432822e-05,
      "loss": 0.816,
      "step": 4575
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.912771700478242,
      "learning_rate": 4.782443137201685e-05,
      "loss": 0.8517,
      "step": 4580
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9077966050043738,
      "learning_rate": 4.781974894492361e-05,
      "loss": 0.8785,
      "step": 4585
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9772236585209746,
      "learning_rate": 4.781506171403416e-05,
      "loss": 0.9017,
      "step": 4590
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8587431172774337,
      "learning_rate": 4.7810369680335244e-05,
      "loss": 0.9044,
      "step": 4595
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.051849428066615,
      "learning_rate": 4.780567284481457e-05,
      "loss": 0.8639,
      "step": 4600
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.029216872114914,
      "learning_rate": 4.780097120846088e-05,
      "loss": 0.9181,
      "step": 4605
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9255108873033635,
      "learning_rate": 4.77962647722639e-05,
      "loss": 0.8954,
      "step": 4610
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.252840975415005,
      "learning_rate": 4.779155353721441e-05,
      "loss": 0.8463,
      "step": 4615
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8421827663973196,
      "learning_rate": 4.778683750430415e-05,
      "loss": 0.9074,
      "step": 4620
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9593702747479365,
      "learning_rate": 4.7782116674525914e-05,
      "loss": 0.8486,
      "step": 4625
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9407666945429404,
      "learning_rate": 4.7777391048873465e-05,
      "loss": 0.9361,
      "step": 4630
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9990179060227713,
      "learning_rate": 4.777266062834162e-05,
      "loss": 0.9305,
      "step": 4635
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9161714977758391,
      "learning_rate": 4.7767925413926166e-05,
      "loss": 1.0146,
      "step": 4640
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9159481848966394,
      "learning_rate": 4.776318540662392e-05,
      "loss": 0.9511,
      "step": 4645
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6947797501387168,
      "learning_rate": 4.7758440607432706e-05,
      "loss": 0.8602,
      "step": 4650
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9121919980966544,
      "learning_rate": 4.775369101735135e-05,
      "loss": 0.9218,
      "step": 4655
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9284442379877902,
      "learning_rate": 4.7748936637379694e-05,
      "loss": 0.9223,
      "step": 4660
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9426241791268728,
      "learning_rate": 4.774417746851858e-05,
      "loss": 0.8775,
      "step": 4665
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9048623332101565,
      "learning_rate": 4.773941351176987e-05,
      "loss": 0.9031,
      "step": 4670
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9501726490455027,
      "learning_rate": 4.773464476813642e-05,
      "loss": 0.9741,
      "step": 4675
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9451190277500853,
      "learning_rate": 4.77298712386221e-05,
      "loss": 0.8654,
      "step": 4680
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8322069942575794,
      "learning_rate": 4.7725092924231794e-05,
      "loss": 0.8728,
      "step": 4685
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.985024297194413,
      "learning_rate": 4.772030982597138e-05,
      "loss": 0.9418,
      "step": 4690
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1569374921834679,
      "learning_rate": 4.7715521944847755e-05,
      "loss": 0.9027,
      "step": 4695
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.935350101482852,
      "learning_rate": 4.7710729281868824e-05,
      "loss": 0.9695,
      "step": 4700
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9342995362666469,
      "learning_rate": 4.770593183804348e-05,
      "loss": 1.0218,
      "step": 4705
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9751550781951213,
      "learning_rate": 4.770112961438164e-05,
      "loss": 0.9267,
      "step": 4710
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9928310477183011,
      "learning_rate": 4.769632261189423e-05,
      "loss": 0.9102,
      "step": 4715
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9004968422207804,
      "learning_rate": 4.769151083159315e-05,
      "loss": 0.8548,
      "step": 4720
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8602921643959474,
      "learning_rate": 4.768669427449137e-05,
      "loss": 0.8487,
      "step": 4725
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9374211564514093,
      "learning_rate": 4.76818729416028e-05,
      "loss": 0.8868,
      "step": 4730
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9086038133032283,
      "learning_rate": 4.767704683394238e-05,
      "loss": 0.9158,
      "step": 4735
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9777676541258469,
      "learning_rate": 4.7672215952526064e-05,
      "loss": 0.8774,
      "step": 4740
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9131377351677685,
      "learning_rate": 4.76673802983708e-05,
      "loss": 0.8715,
      "step": 4745
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2818015794324125,
      "learning_rate": 4.766253987249454e-05,
      "loss": 0.9264,
      "step": 4750
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8879275727418001,
      "learning_rate": 4.765769467591625e-05,
      "loss": 0.8928,
      "step": 4755
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8730931288349222,
      "learning_rate": 4.76528447096559e-05,
      "loss": 0.9204,
      "step": 4760
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9797363287638522,
      "learning_rate": 4.7647989974734445e-05,
      "loss": 1.0487,
      "step": 4765
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9139472801202914,
      "learning_rate": 4.764313047217386e-05,
      "loss": 0.8991,
      "step": 4770
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8151305238101016,
      "learning_rate": 4.763826620299714e-05,
      "loss": 0.8295,
      "step": 4775
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9020269228062616,
      "learning_rate": 4.763339716822824e-05,
      "loss": 0.8683,
      "step": 4780
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9680056504033638,
      "learning_rate": 4.7628523368892155e-05,
      "loss": 0.8972,
      "step": 4785
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.006194949570781,
      "learning_rate": 4.7623644806014865e-05,
      "loss": 0.9575,
      "step": 4790
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9761220891959065,
      "learning_rate": 4.761876148062337e-05,
      "loss": 1.0198,
      "step": 4795
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9089820236597586,
      "learning_rate": 4.761387339374565e-05,
      "loss": 0.8853,
      "step": 4800
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8835758111217743,
      "learning_rate": 4.7608980546410705e-05,
      "loss": 0.95,
      "step": 4805
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9362541763841585,
      "learning_rate": 4.760408293964852e-05,
      "loss": 0.9253,
      "step": 4810
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1263601048794918,
      "learning_rate": 4.759918057449011e-05,
      "loss": 0.8761,
      "step": 4815
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9118611897844395,
      "learning_rate": 4.7594273451967455e-05,
      "loss": 0.897,
      "step": 4820
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9483346946621739,
      "learning_rate": 4.7589361573113573e-05,
      "loss": 0.8852,
      "step": 4825
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0592288762146533,
      "learning_rate": 4.758444493896246e-05,
      "loss": 0.8483,
      "step": 4830
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9621172975665744,
      "learning_rate": 4.757952355054911e-05,
      "loss": 0.9341,
      "step": 4835
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9914306692414427,
      "learning_rate": 4.7574597408909535e-05,
      "loss": 0.9061,
      "step": 4840
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9106002064523849,
      "learning_rate": 4.7569666515080735e-05,
      "loss": 0.8839,
      "step": 4845
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1394912907978014,
      "learning_rate": 4.756473087010073e-05,
      "loss": 0.8607,
      "step": 4850
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9325749129810041,
      "learning_rate": 4.7559790475008505e-05,
      "loss": 0.8703,
      "step": 4855
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9514193475932677,
      "learning_rate": 4.7554845330844066e-05,
      "loss": 1.0,
      "step": 4860
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9472240188171506,
      "learning_rate": 4.754989543864844e-05,
      "loss": 0.9111,
      "step": 4865
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8495427060955298,
      "learning_rate": 4.7544940799463606e-05,
      "loss": 0.9019,
      "step": 4870
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9074393459645913,
      "learning_rate": 4.753998141433258e-05,
      "loss": 0.8653,
      "step": 4875
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9589345459468107,
      "learning_rate": 4.753501728429936e-05,
      "loss": 0.9159,
      "step": 4880
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.031657021745517,
      "learning_rate": 4.7530048410408956e-05,
      "loss": 0.9071,
      "step": 4885
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8088513800368576,
      "learning_rate": 4.752507479370735e-05,
      "loss": 0.9179,
      "step": 4890
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.997198203419029,
      "learning_rate": 4.752009643524156e-05,
      "loss": 0.8878,
      "step": 4895
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.01836760046325,
      "learning_rate": 4.751511333605957e-05,
      "loss": 0.848,
      "step": 4900
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0784323710187127,
      "learning_rate": 4.7510125497210386e-05,
      "loss": 1.0026,
      "step": 4905
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8034495194620102,
      "learning_rate": 4.7505132919743985e-05,
      "loss": 0.8813,
      "step": 4910
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9489501992658349,
      "learning_rate": 4.750013560471137e-05,
      "loss": 0.8547,
      "step": 4915
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8420252740468569,
      "learning_rate": 4.749513355316452e-05,
      "loss": 0.8621,
      "step": 4920
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9326413415392528,
      "learning_rate": 4.749012676615642e-05,
      "loss": 0.9095,
      "step": 4925
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0359317637757601,
      "learning_rate": 4.7485115244741054e-05,
      "loss": 0.9154,
      "step": 4930
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0627702479440768,
      "learning_rate": 4.748009898997341e-05,
      "loss": 0.8251,
      "step": 4935
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9363955667423713,
      "learning_rate": 4.747507800290944e-05,
      "loss": 0.8728,
      "step": 4940
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8783367661148899,
      "learning_rate": 4.7470052284606126e-05,
      "loss": 0.8952,
      "step": 4945
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8945666331789611,
      "learning_rate": 4.746502183612144e-05,
      "loss": 0.8836,
      "step": 4950
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0774006235838147,
      "learning_rate": 4.745998665851433e-05,
      "loss": 0.902,
      "step": 4955
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9349831367340914,
      "learning_rate": 4.745494675284477e-05,
      "loss": 0.9048,
      "step": 4960
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9343428599788585,
      "learning_rate": 4.744990212017371e-05,
      "loss": 0.896,
      "step": 4965
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0767793624774202,
      "learning_rate": 4.744485276156307e-05,
      "loss": 0.9706,
      "step": 4970
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.022127402159067,
      "learning_rate": 4.743979867807584e-05,
      "loss": 0.9905,
      "step": 4975
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0514513769724285,
      "learning_rate": 4.7434739870775926e-05,
      "loss": 1.0015,
      "step": 4980
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8905449188591552,
      "learning_rate": 4.7429676340728265e-05,
      "loss": 0.952,
      "step": 4985
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.998220478427225,
      "learning_rate": 4.742460808899878e-05,
      "loss": 0.9352,
      "step": 4990
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0630463119634113,
      "learning_rate": 4.74195351166544e-05,
      "loss": 0.8896,
      "step": 4995
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9953669765411227,
      "learning_rate": 4.7414457424763036e-05,
      "loss": 0.8861,
      "step": 5000
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.174672433352205,
      "learning_rate": 4.7409375014393596e-05,
      "loss": 0.8438,
      "step": 5005
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1098446664691177,
      "learning_rate": 4.740428788661598e-05,
      "loss": 0.9717,
      "step": 5010
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0471451752218335,
      "learning_rate": 4.739919604250108e-05,
      "loss": 0.8306,
      "step": 5015
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.950781312544515,
      "learning_rate": 4.739409948312078e-05,
      "loss": 0.8309,
      "step": 5020
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0125908425171948,
      "learning_rate": 4.7388998209547966e-05,
      "loss": 0.9146,
      "step": 5025
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8133748718770355,
      "learning_rate": 4.7383892222856507e-05,
      "loss": 0.789,
      "step": 5030
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9024164148250481,
      "learning_rate": 4.737878152412126e-05,
      "loss": 0.9415,
      "step": 5035
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9055245550422416,
      "learning_rate": 4.737366611441809e-05,
      "loss": 0.9178,
      "step": 5040
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.01711755576948,
      "learning_rate": 4.736854599482384e-05,
      "loss": 0.91,
      "step": 5045
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9376320461704247,
      "learning_rate": 4.736342116641635e-05,
      "loss": 0.9189,
      "step": 5050
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9350308591729168,
      "learning_rate": 4.735829163027445e-05,
      "loss": 0.8618,
      "step": 5055
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8757662379212904,
      "learning_rate": 4.7353157387477954e-05,
      "loss": 0.9307,
      "step": 5060
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9366293955471897,
      "learning_rate": 4.734801843910769e-05,
      "loss": 0.9464,
      "step": 5065
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9084759807917344,
      "learning_rate": 4.734287478624545e-05,
      "loss": 0.8634,
      "step": 5070
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0675134198899723,
      "learning_rate": 4.733772642997402e-05,
      "loss": 0.8878,
      "step": 5075
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9687610193245658,
      "learning_rate": 4.73325733713772e-05,
      "loss": 0.9911,
      "step": 5080
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9667590688322949,
      "learning_rate": 4.732741561153974e-05,
      "loss": 0.9058,
      "step": 5085
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9219803235931052,
      "learning_rate": 4.732225315154743e-05,
      "loss": 0.8915,
      "step": 5090
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9623469186831466,
      "learning_rate": 4.7317085992487e-05,
      "loss": 0.9821,
      "step": 5095
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9352372212174855,
      "learning_rate": 4.73119141354462e-05,
      "loss": 0.8408,
      "step": 5100
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8491799245817572,
      "learning_rate": 4.730673758151376e-05,
      "loss": 0.8709,
      "step": 5105
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9280866922102348,
      "learning_rate": 4.73015563317794e-05,
      "loss": 0.8633,
      "step": 5110
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8808344319929562,
      "learning_rate": 4.729637038733382e-05,
      "loss": 0.9068,
      "step": 5115
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9420756531187319,
      "learning_rate": 4.729117974926873e-05,
      "loss": 0.9112,
      "step": 5120
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9110611714148382,
      "learning_rate": 4.7285984418676796e-05,
      "loss": 0.9096,
      "step": 5125
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.018586607557034,
      "learning_rate": 4.72807843966517e-05,
      "loss": 0.9809,
      "step": 5130
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9931577086155187,
      "learning_rate": 4.7275579684288104e-05,
      "loss": 0.8507,
      "step": 5135
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0013243282244393,
      "learning_rate": 4.727037028268164e-05,
      "loss": 0.9356,
      "step": 5140
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9245783918713356,
      "learning_rate": 4.726515619292896e-05,
      "loss": 0.8777,
      "step": 5145
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9083869557073138,
      "learning_rate": 4.725993741612767e-05,
      "loss": 0.8558,
      "step": 5150
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9495953702450738,
      "learning_rate": 4.7254713953376384e-05,
      "loss": 0.9251,
      "step": 5155
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0355595589739228,
      "learning_rate": 4.72494858057747e-05,
      "loss": 0.9259,
      "step": 5160
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9838685115091735,
      "learning_rate": 4.7244252974423185e-05,
      "loss": 0.9668,
      "step": 5165
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0456732532342636,
      "learning_rate": 4.7239015460423414e-05,
      "loss": 0.8755,
      "step": 5170
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8736368841801216,
      "learning_rate": 4.723377326487794e-05,
      "loss": 0.846,
      "step": 5175
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0678483944355903,
      "learning_rate": 4.722852638889029e-05,
      "loss": 0.9019,
      "step": 5180
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.031726819181275,
      "learning_rate": 4.7223274833565e-05,
      "loss": 0.8859,
      "step": 5185
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8978785546916876,
      "learning_rate": 4.721801860000757e-05,
      "loss": 1.0386,
      "step": 5190
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9677446475986918,
      "learning_rate": 4.721275768932449e-05,
      "loss": 0.8884,
      "step": 5195
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0338790914888178,
      "learning_rate": 4.720749210262324e-05,
      "loss": 0.9328,
      "step": 5200
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8701627091871788,
      "learning_rate": 4.7202221841012275e-05,
      "loss": 0.919,
      "step": 5205
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8883313644836678,
      "learning_rate": 4.7196946905601055e-05,
      "loss": 0.8951,
      "step": 5210
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0195922728084343,
      "learning_rate": 4.7191667297499994e-05,
      "loss": 0.8481,
      "step": 5215
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8879610077544813,
      "learning_rate": 4.7186383017820516e-05,
      "loss": 0.869,
      "step": 5220
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9838993840626952,
      "learning_rate": 4.718109406767502e-05,
      "loss": 0.9522,
      "step": 5225
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0250758583408783,
      "learning_rate": 4.7175800448176865e-05,
      "loss": 0.9395,
      "step": 5230
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8760290609510584,
      "learning_rate": 4.7170502160440424e-05,
      "loss": 0.8169,
      "step": 5235
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9721790601477689,
      "learning_rate": 4.7165199205581053e-05,
      "loss": 0.8023,
      "step": 5240
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9638278814892691,
      "learning_rate": 4.715989158471507e-05,
      "loss": 0.922,
      "step": 5245
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.899404320245155,
      "learning_rate": 4.715457929895978e-05,
      "loss": 0.9962,
      "step": 5250
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.04560341680004,
      "learning_rate": 4.714926234943348e-05,
      "loss": 0.9187,
      "step": 5255
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0588051794471998,
      "learning_rate": 4.714394073725544e-05,
      "loss": 0.836,
      "step": 5260
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0045174364708946,
      "learning_rate": 4.7138614463545924e-05,
      "loss": 0.8425,
      "step": 5265
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0476991024630613,
      "learning_rate": 4.7133283529426165e-05,
      "loss": 0.9722,
      "step": 5270
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0333987022084203,
      "learning_rate": 4.712794793601838e-05,
      "loss": 0.8874,
      "step": 5275
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.068849767404288,
      "learning_rate": 4.712260768444576e-05,
      "loss": 1.0005,
      "step": 5280
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.895126643617668,
      "learning_rate": 4.7117262775832496e-05,
      "loss": 0.8608,
      "step": 5285
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0016050519456048,
      "learning_rate": 4.711191321130374e-05,
      "loss": 0.9331,
      "step": 5290
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9054625464875337,
      "learning_rate": 4.7106558991985634e-05,
      "loss": 0.8671,
      "step": 5295
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8700732350470702,
      "learning_rate": 4.710120011900529e-05,
      "loss": 0.886,
      "step": 5300
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8504473513118219,
      "learning_rate": 4.7095836593490815e-05,
      "loss": 0.7917,
      "step": 5305
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9123168341760761,
      "learning_rate": 4.709046841657129e-05,
      "loss": 0.8171,
      "step": 5310
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2653412760434275,
      "learning_rate": 4.708509558937676e-05,
      "loss": 0.9549,
      "step": 5315
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9261057097853282,
      "learning_rate": 4.707971811303827e-05,
      "loss": 0.9926,
      "step": 5320
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9694522719348073,
      "learning_rate": 4.707433598868784e-05,
      "loss": 0.9287,
      "step": 5325
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9941257419055902,
      "learning_rate": 4.706894921745845e-05,
      "loss": 0.9856,
      "step": 5330
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0869813931558723,
      "learning_rate": 4.706355780048408e-05,
      "loss": 0.9224,
      "step": 5335
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8492471857242857,
      "learning_rate": 4.705816173889969e-05,
      "loss": 0.9393,
      "step": 5340
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9840307181844872,
      "learning_rate": 4.705276103384119e-05,
      "loss": 0.8486,
      "step": 5345
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9955482658021549,
      "learning_rate": 4.7047355686445485e-05,
      "loss": 0.898,
      "step": 5350
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.900983276408742,
      "learning_rate": 4.7041945697850466e-05,
      "loss": 0.8927,
      "step": 5355
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9271025297172,
      "learning_rate": 4.7036531069194996e-05,
      "loss": 0.8635,
      "step": 5360
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9285434480936893,
      "learning_rate": 4.70311118016189e-05,
      "loss": 0.8862,
      "step": 5365
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9325358849171663,
      "learning_rate": 4.7025687896262994e-05,
      "loss": 0.9087,
      "step": 5370
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9153431473739005,
      "learning_rate": 4.702025935426907e-05,
      "loss": 0.9348,
      "step": 5375
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.943839638144225,
      "learning_rate": 4.7014826176779896e-05,
      "loss": 0.9862,
      "step": 5380
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9471473629703717,
      "learning_rate": 4.7009388364939203e-05,
      "loss": 0.7952,
      "step": 5385
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8740480776148996,
      "learning_rate": 4.7003945919891715e-05,
      "loss": 0.9124,
      "step": 5390
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9255818551861128,
      "learning_rate": 4.699849884278312e-05,
      "loss": 0.8174,
      "step": 5395
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9610244958701634,
      "learning_rate": 4.699304713476009e-05,
      "loss": 0.8935,
      "step": 5400
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9652297207776812,
      "learning_rate": 4.698759079697026e-05,
      "loss": 0.8923,
      "step": 5405
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9383503284105464,
      "learning_rate": 4.698212983056225e-05,
      "loss": 0.834,
      "step": 5410
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8760602065533816,
      "learning_rate": 4.697666423668565e-05,
      "loss": 0.8266,
      "step": 5415
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9210700781335122,
      "learning_rate": 4.6971194016491025e-05,
      "loss": 0.9142,
      "step": 5420
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0983602314379115,
      "learning_rate": 4.696571917112991e-05,
      "loss": 0.841,
      "step": 5425
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8851036441317692,
      "learning_rate": 4.696023970175483e-05,
      "loss": 0.8705,
      "step": 5430
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9511684237588188,
      "learning_rate": 4.695475560951925e-05,
      "loss": 0.8897,
      "step": 5435
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.885053507235492,
      "learning_rate": 4.694926689557764e-05,
      "loss": 0.8006,
      "step": 5440
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9849892194518868,
      "learning_rate": 4.6943773561085435e-05,
      "loss": 0.9323,
      "step": 5445
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.910899058845763,
      "learning_rate": 4.6938275607199033e-05,
      "loss": 0.8622,
      "step": 5450
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9777009686317241,
      "learning_rate": 4.6932773035075816e-05,
      "loss": 0.9072,
      "step": 5455
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8716996270901086,
      "learning_rate": 4.6927265845874136e-05,
      "loss": 0.873,
      "step": 5460
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9909763148258036,
      "learning_rate": 4.69217540407533e-05,
      "loss": 1.0057,
      "step": 5465
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9720374658668953,
      "learning_rate": 4.6916237620873617e-05,
      "loss": 0.8676,
      "step": 5470
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8910963485197985,
      "learning_rate": 4.6910716587396344e-05,
      "loss": 0.8518,
      "step": 5475
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9569929361626156,
      "learning_rate": 4.690519094148371e-05,
      "loss": 0.9391,
      "step": 5480
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0511680915658494,
      "learning_rate": 4.6899660684298936e-05,
      "loss": 0.8968,
      "step": 5485
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0801385418230076,
      "learning_rate": 4.689412581700618e-05,
      "loss": 0.9384,
      "step": 5490
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8927714251323101,
      "learning_rate": 4.688858634077061e-05,
      "loss": 0.898,
      "step": 5495
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1309981861996445,
      "learning_rate": 4.688304225675834e-05,
      "loss": 0.9083,
      "step": 5500
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9220005244397378,
      "learning_rate": 4.687749356613643e-05,
      "loss": 0.9628,
      "step": 5505
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8450685429645087,
      "learning_rate": 4.6871940270072976e-05,
      "loss": 0.8543,
      "step": 5510
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8743105335085044,
      "learning_rate": 4.686638236973698e-05,
      "loss": 0.9632,
      "step": 5515
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8788870715691519,
      "learning_rate": 4.6860819866298456e-05,
      "loss": 0.8929,
      "step": 5520
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8715849901477508,
      "learning_rate": 4.6855252760928356e-05,
      "loss": 0.9212,
      "step": 5525
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8434996407999087,
      "learning_rate": 4.6849681054798615e-05,
      "loss": 0.9428,
      "step": 5530
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8768448790280514,
      "learning_rate": 4.6844104749082144e-05,
      "loss": 0.9217,
      "step": 5535
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8608543898621194,
      "learning_rate": 4.68385238449528e-05,
      "loss": 0.9223,
      "step": 5540
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8828757282033372,
      "learning_rate": 4.683293834358544e-05,
      "loss": 0.9527,
      "step": 5545
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9530063799041077,
      "learning_rate": 4.6827348246155855e-05,
      "loss": 0.9122,
      "step": 5550
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9231426689889262,
      "learning_rate": 4.682175355384082e-05,
      "loss": 0.9776,
      "step": 5555
    },
    {
      "epoch": 0.65,
      "grad_norm": 62.51705595956571,
      "learning_rate": 4.6816154267818084e-05,
      "loss": 0.9379,
      "step": 5560
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1773638956250652,
      "learning_rate": 4.6810550389266356e-05,
      "loss": 0.9749,
      "step": 5565
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9525301740186433,
      "learning_rate": 4.680494191936531e-05,
      "loss": 0.8828,
      "step": 5570
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9835525247928325,
      "learning_rate": 4.679932885929558e-05,
      "loss": 0.8223,
      "step": 5575
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9308323914965436,
      "learning_rate": 4.679371121023877e-05,
      "loss": 0.794,
      "step": 5580
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0642197192722602,
      "learning_rate": 4.678808897337747e-05,
      "loss": 0.8937,
      "step": 5585
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.863722934745398,
      "learning_rate": 4.678246214989522e-05,
      "loss": 0.9496,
      "step": 5590
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8946901888023797,
      "learning_rate": 4.6776830740976505e-05,
      "loss": 0.8944,
      "step": 5595
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9255928319484048,
      "learning_rate": 4.6771194747806826e-05,
      "loss": 0.9524,
      "step": 5600
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9496776135844713,
      "learning_rate": 4.676555417157259e-05,
      "loss": 0.9024,
      "step": 5605
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.826806279167785,
      "learning_rate": 4.67599090134612e-05,
      "loss": 0.9267,
      "step": 5610
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9945564753833892,
      "learning_rate": 4.675425927466104e-05,
      "loss": 0.897,
      "step": 5615
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1326767874904067,
      "learning_rate": 4.674860495636143e-05,
      "loss": 1.0306,
      "step": 5620
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.192659378890715,
      "learning_rate": 4.6742946059752654e-05,
      "loss": 0.8268,
      "step": 5625
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.865472885386012,
      "learning_rate": 4.673728258602599e-05,
      "loss": 0.9125,
      "step": 5630
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8859023700638049,
      "learning_rate": 4.673161453637363e-05,
      "loss": 0.7883,
      "step": 5635
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9828220946204883,
      "learning_rate": 4.672594191198878e-05,
      "loss": 0.7975,
      "step": 5640
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0218056914595839,
      "learning_rate": 4.672026471406558e-05,
      "loss": 0.8704,
      "step": 5645
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9962228419117912,
      "learning_rate": 4.671458294379914e-05,
      "loss": 0.8443,
      "step": 5650
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.916842238976931,
      "learning_rate": 4.670889660238553e-05,
      "loss": 0.9196,
      "step": 5655
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.896973578985922,
      "learning_rate": 4.670320569102179e-05,
      "loss": 0.9462,
      "step": 5660
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.869017141870717,
      "learning_rate": 4.669751021090592e-05,
      "loss": 0.8841,
      "step": 5665
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.94360156088277,
      "learning_rate": 4.6691810163236855e-05,
      "loss": 0.8789,
      "step": 5670
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9073554986575458,
      "learning_rate": 4.668610554921454e-05,
      "loss": 0.8325,
      "step": 5675
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9257045965587248,
      "learning_rate": 4.668039637003984e-05,
      "loss": 0.8651,
      "step": 5680
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8828285742161036,
      "learning_rate": 4.66746826269146e-05,
      "loss": 0.8829,
      "step": 5685
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1408332564085044,
      "learning_rate": 4.666896432104164e-05,
      "loss": 0.9076,
      "step": 5690
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8766537384116836,
      "learning_rate": 4.66632414536247e-05,
      "loss": 0.9158,
      "step": 5695
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8636114269587559,
      "learning_rate": 4.665751402586852e-05,
      "loss": 0.844,
      "step": 5700
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1465206874810836,
      "learning_rate": 4.665178203897877e-05,
      "loss": 0.9747,
      "step": 5705
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9935397619995754,
      "learning_rate": 4.6646045494162115e-05,
      "loss": 0.9081,
      "step": 5710
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0416302205511918,
      "learning_rate": 4.664030439262613e-05,
      "loss": 0.9922,
      "step": 5715
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9765975302035652,
      "learning_rate": 4.663455873557939e-05,
      "loss": 0.8822,
      "step": 5720
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9113509057044953,
      "learning_rate": 4.6628808524231413e-05,
      "loss": 0.9486,
      "step": 5725
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0315351882329007,
      "learning_rate": 4.6623053759792696e-05,
      "loss": 0.9191,
      "step": 5730
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8948879757766348,
      "learning_rate": 4.661729444347465e-05,
      "loss": 0.8486,
      "step": 5735
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8766710372397554,
      "learning_rate": 4.661153057648969e-05,
      "loss": 0.8894,
      "step": 5740
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.82706258378984,
      "learning_rate": 4.660576216005116e-05,
      "loss": 0.8505,
      "step": 5745
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.930694026529663,
      "learning_rate": 4.6599989195373385e-05,
      "loss": 0.8854,
      "step": 5750
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1264148491246995,
      "learning_rate": 4.659421168367162e-05,
      "loss": 0.8439,
      "step": 5755
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9425710713723578,
      "learning_rate": 4.6588429626162095e-05,
      "loss": 0.9722,
      "step": 5760
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9296863266965247,
      "learning_rate": 4.6582643024062e-05,
      "loss": 0.9878,
      "step": 5765
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9717643081597708,
      "learning_rate": 4.6576851878589475e-05,
      "loss": 0.9861,
      "step": 5770
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0128093772015851,
      "learning_rate": 4.657105619096361e-05,
      "loss": 0.848,
      "step": 5775
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8858755147245149,
      "learning_rate": 4.656525596240446e-05,
      "loss": 0.9132,
      "step": 5780
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0529894716410615,
      "learning_rate": 4.655945119413304e-05,
      "loss": 0.9049,
      "step": 5785
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9423077431089073,
      "learning_rate": 4.6553641887371305e-05,
      "loss": 0.8939,
      "step": 5790
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9289900362286485,
      "learning_rate": 4.6547828043342186e-05,
      "loss": 0.8498,
      "step": 5795
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8434000431799341,
      "learning_rate": 4.654200966326955e-05,
      "loss": 0.8566,
      "step": 5800
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9383065891406847,
      "learning_rate": 4.653618674837823e-05,
      "loss": 0.875,
      "step": 5805
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0008310804902216,
      "learning_rate": 4.6530359299894006e-05,
      "loss": 0.8826,
      "step": 5810
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.888970604341223,
      "learning_rate": 4.652452731904362e-05,
      "loss": 0.9341,
      "step": 5815
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9564059476156773,
      "learning_rate": 4.651869080705477e-05,
      "loss": 0.8431,
      "step": 5820
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.91077967654913,
      "learning_rate": 4.65128497651561e-05,
      "loss": 0.867,
      "step": 5825
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2406196675947683,
      "learning_rate": 4.650700419457721e-05,
      "loss": 0.9075,
      "step": 5830
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8850204148321628,
      "learning_rate": 4.6501154096548646e-05,
      "loss": 0.8777,
      "step": 5835
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8883847420156631,
      "learning_rate": 4.6495299472301925e-05,
      "loss": 0.8567,
      "step": 5840
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9476475233766783,
      "learning_rate": 4.64894403230695e-05,
      "loss": 0.841,
      "step": 5845
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9487764754969552,
      "learning_rate": 4.6483576650084795e-05,
      "loss": 0.9376,
      "step": 5850
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9165964621495063,
      "learning_rate": 4.647770845458216e-05,
      "loss": 0.9088,
      "step": 5855
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9974674944880536,
      "learning_rate": 4.647183573779692e-05,
      "loss": 0.8475,
      "step": 5860
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8980672638517232,
      "learning_rate": 4.646595850096534e-05,
      "loss": 0.8466,
      "step": 5865
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8751879975488112,
      "learning_rate": 4.6460076745324646e-05,
      "loss": 0.9086,
      "step": 5870
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9058933220665744,
      "learning_rate": 4.6454190472113e-05,
      "loss": 0.9135,
      "step": 5875
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9606409834423856,
      "learning_rate": 4.644829968256953e-05,
      "loss": 0.943,
      "step": 5880
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2415993852198044,
      "learning_rate": 4.644240437793431e-05,
      "loss": 0.8554,
      "step": 5885
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9317260883498478,
      "learning_rate": 4.6436504559448354e-05,
      "loss": 0.9151,
      "step": 5890
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2617341544231906,
      "learning_rate": 4.643060022835365e-05,
      "loss": 1.0054,
      "step": 5895
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0409449577750085,
      "learning_rate": 4.642469138589311e-05,
      "loss": 0.8763,
      "step": 5900
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8450256402510777,
      "learning_rate": 4.641877803331061e-05,
      "loss": 0.8358,
      "step": 5905
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9507390060939424,
      "learning_rate": 4.641286017185098e-05,
      "loss": 0.9093,
      "step": 5910
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8747175940849005,
      "learning_rate": 4.6406937802759984e-05,
      "loss": 0.91,
      "step": 5915
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9070793302110012,
      "learning_rate": 4.640101092728434e-05,
      "loss": 0.8121,
      "step": 5920
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.949786764506466,
      "learning_rate": 4.6395079546671725e-05,
      "loss": 0.9231,
      "step": 5925
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1435484446328472,
      "learning_rate": 4.6389143662170755e-05,
      "loss": 0.9155,
      "step": 5930
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9058030053546251,
      "learning_rate": 4.6383203275030996e-05,
      "loss": 0.9596,
      "step": 5935
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9352875952310603,
      "learning_rate": 4.6377258386502956e-05,
      "loss": 1.0381,
      "step": 5940
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9408422310720795,
      "learning_rate": 4.6371308997838107e-05,
      "loss": 0.9698,
      "step": 5945
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.066883861343165,
      "learning_rate": 4.6365355110288845e-05,
      "loss": 0.918,
      "step": 5950
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.881656248780669,
      "learning_rate": 4.635939672510853e-05,
      "loss": 0.9496,
      "step": 5955
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.906309309915082,
      "learning_rate": 4.635343384355147e-05,
      "loss": 0.8641,
      "step": 5960
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9486582240345701,
      "learning_rate": 4.6347466466872916e-05,
      "loss": 0.9471,
      "step": 5965
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9448056393967973,
      "learning_rate": 4.634149459632905e-05,
      "loss": 0.8951,
      "step": 5970
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8680934354031703,
      "learning_rate": 4.633551823317702e-05,
      "loss": 0.8387,
      "step": 5975
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0298181055442204,
      "learning_rate": 4.6329537378674914e-05,
      "loss": 0.8783,
      "step": 5980
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1161018549931019,
      "learning_rate": 4.6323552034081773e-05,
      "loss": 0.8897,
      "step": 5985
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8837510703914684,
      "learning_rate": 4.631756220065756e-05,
      "loss": 0.9294,
      "step": 5990
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8860897108704975,
      "learning_rate": 4.6311567879663206e-05,
      "loss": 0.867,
      "step": 5995
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9029700041572566,
      "learning_rate": 4.6305569072360576e-05,
      "loss": 0.8449,
      "step": 6000
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9428846895181254,
      "learning_rate": 4.629956578001248e-05,
      "loss": 0.9825,
      "step": 6005
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9153696985371471,
      "learning_rate": 4.6293558003882676e-05,
      "loss": 0.804,
      "step": 6010
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8585207867675193,
      "learning_rate": 4.6287545745235874e-05,
      "loss": 0.878,
      "step": 6015
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.21832783563486,
      "learning_rate": 4.62815290053377e-05,
      "loss": 0.9089,
      "step": 6020
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9377488059388995,
      "learning_rate": 4.627550778545474e-05,
      "loss": 1.0253,
      "step": 6025
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9992394738506685,
      "learning_rate": 4.626948208685454e-05,
      "loss": 0.8695,
      "step": 6030
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8940893498337988,
      "learning_rate": 4.6263451910805564e-05,
      "loss": 0.9124,
      "step": 6035
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9109467535072565,
      "learning_rate": 4.6257417258577226e-05,
      "loss": 0.8536,
      "step": 6040
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.87354951406432,
      "learning_rate": 4.625137813143989e-05,
      "loss": 0.8777,
      "step": 6045
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9320667565405548,
      "learning_rate": 4.624533453066485e-05,
      "loss": 0.9779,
      "step": 6050
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9184223567517243,
      "learning_rate": 4.623928645752434e-05,
      "loss": 0.9153,
      "step": 6055
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9068618601226535,
      "learning_rate": 4.6233233913291564e-05,
      "loss": 0.9139,
      "step": 6060
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8831840608811291,
      "learning_rate": 4.622717689924062e-05,
      "loss": 0.8134,
      "step": 6065
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9251633590782754,
      "learning_rate": 4.6221115416646594e-05,
      "loss": 0.9339,
      "step": 6070
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9243066351426512,
      "learning_rate": 4.621504946678548e-05,
      "loss": 0.8668,
      "step": 6075
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8835072500508083,
      "learning_rate": 4.620897905093423e-05,
      "loss": 0.9419,
      "step": 6080
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1830639824657618,
      "learning_rate": 4.6202904170370725e-05,
      "loss": 0.92,
      "step": 6085
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0377641678174525,
      "learning_rate": 4.61968248263738e-05,
      "loss": 0.8622,
      "step": 6090
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8221855416618498,
      "learning_rate": 4.619074102022321e-05,
      "loss": 0.879,
      "step": 6095
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1236408060018133,
      "learning_rate": 4.6184652753199654e-05,
      "loss": 0.8666,
      "step": 6100
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0546708431225267,
      "learning_rate": 4.6178560026584796e-05,
      "loss": 0.9434,
      "step": 6105
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1096496852030064,
      "learning_rate": 4.617246284166121e-05,
      "loss": 0.975,
      "step": 6110
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9842139713889004,
      "learning_rate": 4.616636119971241e-05,
      "loss": 0.8923,
      "step": 6115
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1695387595408357,
      "learning_rate": 4.6160255102022865e-05,
      "loss": 0.8549,
      "step": 6120
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9674107209626502,
      "learning_rate": 4.6154144549877966e-05,
      "loss": 0.8989,
      "step": 6125
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1585208366143323,
      "learning_rate": 4.614802954456405e-05,
      "loss": 0.8816,
      "step": 6130
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9408702981592619,
      "learning_rate": 4.61419100873684e-05,
      "loss": 0.8636,
      "step": 6135
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8983370202177836,
      "learning_rate": 4.613578617957921e-05,
      "loss": 0.8763,
      "step": 6140
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.097332295548343,
      "learning_rate": 4.6129657822485637e-05,
      "loss": 0.8812,
      "step": 6145
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8769493817593639,
      "learning_rate": 4.6123525017377757e-05,
      "loss": 0.8373,
      "step": 6150
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9922450337690748,
      "learning_rate": 4.611738776554659e-05,
      "loss": 0.8961,
      "step": 6155
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8890048430127167,
      "learning_rate": 4.611124606828411e-05,
      "loss": 0.8995,
      "step": 6160
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8768629434253079,
      "learning_rate": 4.610509992688319e-05,
      "loss": 0.8798,
      "step": 6165
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0440007922113836,
      "learning_rate": 4.6098949342637655e-05,
      "loss": 0.8847,
      "step": 6170
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0012655568302622,
      "learning_rate": 4.6092794316842266e-05,
      "loss": 0.9017,
      "step": 6175
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9532168138198784,
      "learning_rate": 4.6086634850792743e-05,
      "loss": 0.8492,
      "step": 6180
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8829541753778574,
      "learning_rate": 4.608047094578569e-05,
      "loss": 0.8476,
      "step": 6185
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8695714693195097,
      "learning_rate": 4.60743026031187e-05,
      "loss": 0.8765,
      "step": 6190
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9128963621393668,
      "learning_rate": 4.606812982409025e-05,
      "loss": 0.9509,
      "step": 6195
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8859489839261778,
      "learning_rate": 4.606195260999979e-05,
      "loss": 0.8819,
      "step": 6200
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9390059600209083,
      "learning_rate": 4.6055770962147684e-05,
      "loss": 0.9154,
      "step": 6205
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9243235446324867,
      "learning_rate": 4.604958488183523e-05,
      "loss": 0.8912,
      "step": 6210
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9729508229223045,
      "learning_rate": 4.6043394370364665e-05,
      "loss": 0.8596,
      "step": 6215
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8484031459132618,
      "learning_rate": 4.603719942903916e-05,
      "loss": 0.9308,
      "step": 6220
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9665400919268459,
      "learning_rate": 4.603100005916281e-05,
      "loss": 0.9341,
      "step": 6225
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8432324874952188,
      "learning_rate": 4.6024796262040656e-05,
      "loss": 0.9389,
      "step": 6230
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0460759951319094,
      "learning_rate": 4.601858803897865e-05,
      "loss": 0.9347,
      "step": 6235
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.816847428520859,
      "learning_rate": 4.6012375391283704e-05,
      "loss": 0.8765,
      "step": 6240
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.035963830805095,
      "learning_rate": 4.6006158320263623e-05,
      "loss": 0.903,
      "step": 6245
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9084377151112936,
      "learning_rate": 4.5999936827227174e-05,
      "loss": 0.9154,
      "step": 6250
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9251730841669199,
      "learning_rate": 4.5993710913484065e-05,
      "loss": 0.9767,
      "step": 6255
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9461081359999874,
      "learning_rate": 4.5987480580344886e-05,
      "loss": 0.8985,
      "step": 6260
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9366044223690984,
      "learning_rate": 4.598124582912121e-05,
      "loss": 0.9089,
      "step": 6265
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8806746000216982,
      "learning_rate": 4.5975006661125515e-05,
      "loss": 0.8571,
      "step": 6270
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8890877537842857,
      "learning_rate": 4.59687630776712e-05,
      "loss": 0.8431,
      "step": 6275
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.875535656849131,
      "learning_rate": 4.596251508007261e-05,
      "loss": 0.8789,
      "step": 6280
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8622018079377948,
      "learning_rate": 4.5956262669645026e-05,
      "loss": 0.8664,
      "step": 6285
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8807128797583802,
      "learning_rate": 4.595000584770462e-05,
      "loss": 0.9086,
      "step": 6290
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8197993515527898,
      "learning_rate": 4.594374461556854e-05,
      "loss": 0.8242,
      "step": 6295
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9044114149932014,
      "learning_rate": 4.593747897455484e-05,
      "loss": 0.84,
      "step": 6300
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.844962428245253,
      "learning_rate": 4.593120892598249e-05,
      "loss": 0.9456,
      "step": 6305
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9007442506130924,
      "learning_rate": 4.592493447117141e-05,
      "loss": 0.8419,
      "step": 6310
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9295044964217902,
      "learning_rate": 4.591865561144244e-05,
      "loss": 0.9004,
      "step": 6315
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9626186549600156,
      "learning_rate": 4.5912372348117336e-05,
      "loss": 0.8287,
      "step": 6320
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9007240148017548,
      "learning_rate": 4.59060846825188e-05,
      "loss": 0.9189,
      "step": 6325
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9268166833165955,
      "learning_rate": 4.589979261597045e-05,
      "loss": 0.8387,
      "step": 6330
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2251278176367635,
      "learning_rate": 4.5893496149796834e-05,
      "loss": 0.882,
      "step": 6335
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8273194127826791,
      "learning_rate": 4.588719528532342e-05,
      "loss": 0.7887,
      "step": 6340
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.000549915201785,
      "learning_rate": 4.58808900238766e-05,
      "loss": 0.985,
      "step": 6345
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9529477105878277,
      "learning_rate": 4.5874580366783706e-05,
      "loss": 0.88,
      "step": 6350
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8475429884907233,
      "learning_rate": 4.586826631537299e-05,
      "loss": 0.8077,
      "step": 6355
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8431966618407057,
      "learning_rate": 4.586194787097362e-05,
      "loss": 0.8776,
      "step": 6360
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8818010759884471,
      "learning_rate": 4.58556250349157e-05,
      "loss": 0.8218,
      "step": 6365
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9719406565621362,
      "learning_rate": 4.584929780853023e-05,
      "loss": 0.9744,
      "step": 6370
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8169917993462518,
      "learning_rate": 4.5842966193149196e-05,
      "loss": 0.8829,
      "step": 6375
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8964425215861298,
      "learning_rate": 4.583663019010545e-05,
      "loss": 0.9191,
      "step": 6380
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9222675607785201,
      "learning_rate": 4.5830289800732775e-05,
      "loss": 0.8921,
      "step": 6385
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9528886187326482,
      "learning_rate": 4.582394502636591e-05,
      "loss": 0.8363,
      "step": 6390
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8767551135711331,
      "learning_rate": 4.581759586834049e-05,
      "loss": 0.9017,
      "step": 6395
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8796793033406571,
      "learning_rate": 4.5811242327993066e-05,
      "loss": 0.9381,
      "step": 6400
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.869895599396756,
      "learning_rate": 4.5804884406661143e-05,
      "loss": 0.961,
      "step": 6405
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8944114677302881,
      "learning_rate": 4.579852210568313e-05,
      "loss": 0.9182,
      "step": 6410
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0548227460991775,
      "learning_rate": 4.579215542639834e-05,
      "loss": 0.9077,
      "step": 6415
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9862335460732246,
      "learning_rate": 4.5785784370147036e-05,
      "loss": 0.8815,
      "step": 6420
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9832163639675155,
      "learning_rate": 4.5779408938270396e-05,
      "loss": 1.0411,
      "step": 6425
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9217683195009689,
      "learning_rate": 4.577302913211051e-05,
      "loss": 0.8684,
      "step": 6430
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.093573529908637,
      "learning_rate": 4.5766644953010406e-05,
      "loss": 0.9057,
      "step": 6435
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9141308948145531,
      "learning_rate": 4.5760256402314e-05,
      "loss": 0.9413,
      "step": 6440
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8536925340397831,
      "learning_rate": 4.575386348136617e-05,
      "loss": 0.9157,
      "step": 6445
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.012324444749008,
      "learning_rate": 4.574746619151267e-05,
      "loss": 0.8648,
      "step": 6450
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8389768064145428,
      "learning_rate": 4.5741064534100206e-05,
      "loss": 0.8937,
      "step": 6455
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7900443792939897,
      "learning_rate": 4.573465851047641e-05,
      "loss": 0.909,
      "step": 6460
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8103292562631329,
      "learning_rate": 4.57282481219898e-05,
      "loss": 0.9096,
      "step": 6465
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8431917074010248,
      "learning_rate": 4.572183336998983e-05,
      "loss": 0.8685,
      "step": 6470
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0345249342212555,
      "learning_rate": 4.571541425582688e-05,
      "loss": 0.8977,
      "step": 6475
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8997773810571221,
      "learning_rate": 4.570899078085223e-05,
      "loss": 0.9831,
      "step": 6480
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.133684431280131,
      "learning_rate": 4.5702562946418096e-05,
      "loss": 0.9852,
      "step": 6485
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8515355959743727,
      "learning_rate": 4.5696130753877605e-05,
      "loss": 0.8296,
      "step": 6490
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.881822819644578,
      "learning_rate": 4.568969420458481e-05,
      "loss": 0.8074,
      "step": 6495
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9981784732119469,
      "learning_rate": 4.568325329989466e-05,
      "loss": 0.9014,
      "step": 6500
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9708027233798908,
      "learning_rate": 4.567680804116302e-05,
      "loss": 0.9014,
      "step": 6505
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9602401756058483,
      "learning_rate": 4.567035842974672e-05,
      "loss": 0.9401,
      "step": 6510
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8881231641217178,
      "learning_rate": 4.566390446700343e-05,
      "loss": 0.9916,
      "step": 6515
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.012553516599443,
      "learning_rate": 4.565744615429181e-05,
      "loss": 0.8627,
      "step": 6520
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9668416999748121,
      "learning_rate": 4.565098349297139e-05,
      "loss": 0.9217,
      "step": 6525
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1165722314674247,
      "learning_rate": 4.564451648440262e-05,
      "loss": 0.8758,
      "step": 6530
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0443461673657293,
      "learning_rate": 4.5638045129946884e-05,
      "loss": 0.8174,
      "step": 6535
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0005526034038386,
      "learning_rate": 4.563156943096647e-05,
      "loss": 0.8884,
      "step": 6540
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8738596141005056,
      "learning_rate": 4.5625089388824574e-05,
      "loss": 0.7904,
      "step": 6545
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8205869197196017,
      "learning_rate": 4.5618605004885304e-05,
      "loss": 0.822,
      "step": 6550
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.993592490390536,
      "learning_rate": 4.561211628051372e-05,
      "loss": 0.8492,
      "step": 6555
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0526138718755598,
      "learning_rate": 4.560562321707574e-05,
      "loss": 0.8476,
      "step": 6560
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0509429220233961,
      "learning_rate": 4.559912581593823e-05,
      "loss": 0.8446,
      "step": 6565
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9261331908819719,
      "learning_rate": 4.559262407846896e-05,
      "loss": 0.8301,
      "step": 6570
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8656335970010844,
      "learning_rate": 4.5586118006036624e-05,
      "loss": 0.8711,
      "step": 6575
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0044467466466536,
      "learning_rate": 4.55796076000108e-05,
      "loss": 0.9174,
      "step": 6580
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8599510579625476,
      "learning_rate": 4.557309286176202e-05,
      "loss": 0.9533,
      "step": 6585
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9141088247845843,
      "learning_rate": 4.556657379266168e-05,
      "loss": 0.9352,
      "step": 6590
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8477613746736062,
      "learning_rate": 4.556005039408212e-05,
      "loss": 0.8875,
      "step": 6595
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.867459635396512,
      "learning_rate": 4.5553522667396596e-05,
      "loss": 0.8872,
      "step": 6600
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9365938449215916,
      "learning_rate": 4.5546990613979254e-05,
      "loss": 0.941,
      "step": 6605
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9362390934799447,
      "learning_rate": 4.5540454235205155e-05,
      "loss": 0.9718,
      "step": 6610
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8240309388980643,
      "learning_rate": 4.553391353245028e-05,
      "loss": 0.8478,
      "step": 6615
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9416357888311052,
      "learning_rate": 4.552736850709152e-05,
      "loss": 0.8707,
      "step": 6620
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9642286942572582,
      "learning_rate": 4.5520819160506663e-05,
      "loss": 0.8754,
      "step": 6625
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9399405270703323,
      "learning_rate": 4.551426549407442e-05,
      "loss": 0.9623,
      "step": 6630
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.920686265338966,
      "learning_rate": 4.55077075091744e-05,
      "loss": 0.8966,
      "step": 6635
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8855940071389286,
      "learning_rate": 4.5501145207187146e-05,
      "loss": 1.0745,
      "step": 6640
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8723032412287673,
      "learning_rate": 4.5494578589494067e-05,
      "loss": 0.8585,
      "step": 6645
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8200793561196714,
      "learning_rate": 4.548800765747752e-05,
      "loss": 0.9757,
      "step": 6650
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8927281867615093,
      "learning_rate": 4.548143241252075e-05,
      "loss": 0.9107,
      "step": 6655
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9292826126834808,
      "learning_rate": 4.547485285600791e-05,
      "loss": 0.8477,
      "step": 6660
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9861904227565962,
      "learning_rate": 4.546826898932408e-05,
      "loss": 0.8929,
      "step": 6665
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9585159916124865,
      "learning_rate": 4.546168081385522e-05,
      "loss": 0.8093,
      "step": 6670
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9577613091747206,
      "learning_rate": 4.545508833098822e-05,
      "loss": 0.977,
      "step": 6675
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9192788328338644,
      "learning_rate": 4.544849154211085e-05,
      "loss": 0.8832,
      "step": 6680
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.987998066896299,
      "learning_rate": 4.5441890448611826e-05,
      "loss": 0.7768,
      "step": 6685
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1912815115222373,
      "learning_rate": 4.543528505188073e-05,
      "loss": 0.8507,
      "step": 6690
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9393612929571797,
      "learning_rate": 4.542867535330807e-05,
      "loss": 0.8959,
      "step": 6695
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9845511721012468,
      "learning_rate": 4.542206135428526e-05,
      "loss": 0.8748,
      "step": 6700
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8903440465894508,
      "learning_rate": 4.541544305620462e-05,
      "loss": 0.9035,
      "step": 6705
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9548946271504166,
      "learning_rate": 4.5408820460459354e-05,
      "loss": 0.9212,
      "step": 6710
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9030897660349191,
      "learning_rate": 4.540219356844362e-05,
      "loss": 0.8025,
      "step": 6715
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8424287819579624,
      "learning_rate": 4.5395562381552414e-05,
      "loss": 0.7828,
      "step": 6720
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8832524513732642,
      "learning_rate": 4.5388926901181696e-05,
      "loss": 0.9107,
      "step": 6725
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7678228777709971,
      "learning_rate": 4.538228712872828e-05,
      "loss": 0.8686,
      "step": 6730
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9443007432414015,
      "learning_rate": 4.5375643065589926e-05,
      "loss": 0.8986,
      "step": 6735
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8166760834680121,
      "learning_rate": 4.5368994713165274e-05,
      "loss": 0.8385,
      "step": 6740
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0589750260582995,
      "learning_rate": 4.536234207285387e-05,
      "loss": 0.872,
      "step": 6745
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9670038000292133,
      "learning_rate": 4.535568514605617e-05,
      "loss": 0.7944,
      "step": 6750
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9249180768610487,
      "learning_rate": 4.534902393417352e-05,
      "loss": 0.9245,
      "step": 6755
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9667722628382301,
      "learning_rate": 4.534235843860818e-05,
      "loss": 0.9252,
      "step": 6760
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9510002135775978,
      "learning_rate": 4.5335688660763295e-05,
      "loss": 0.8906,
      "step": 6765
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9595636576431867,
      "learning_rate": 4.532901460204295e-05,
      "loss": 0.885,
      "step": 6770
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8287853742632969,
      "learning_rate": 4.5322336263852074e-05,
      "loss": 0.8593,
      "step": 6775
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9499119473730897,
      "learning_rate": 4.5315653647596535e-05,
      "loss": 0.8761,
      "step": 6780
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.906177355917264,
      "learning_rate": 4.530896675468311e-05,
      "loss": 0.8467,
      "step": 6785
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9552272331103624,
      "learning_rate": 4.530227558651944e-05,
      "loss": 0.9359,
      "step": 6790
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9837491128468989,
      "learning_rate": 4.52955801445141e-05,
      "loss": 0.898,
      "step": 6795
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0734456804213968,
      "learning_rate": 4.528888043007654e-05,
      "loss": 0.8603,
      "step": 6800
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8189829717059002,
      "learning_rate": 4.528217644461713e-05,
      "loss": 0.8911,
      "step": 6805
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0087337719968905,
      "learning_rate": 4.527546818954712e-05,
      "loss": 0.973,
      "step": 6810
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9127711731357061,
      "learning_rate": 4.526875566627867e-05,
      "loss": 0.8793,
      "step": 6815
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8172810337144737,
      "learning_rate": 4.5262038876224845e-05,
      "loss": 0.8434,
      "step": 6820
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0137957503406376,
      "learning_rate": 4.525531782079959e-05,
      "loss": 0.9221,
      "step": 6825
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8665790946319983,
      "learning_rate": 4.5248592501417766e-05,
      "loss": 0.9085,
      "step": 6830
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7968224632190943,
      "learning_rate": 4.524186291949511e-05,
      "loss": 0.818,
      "step": 6835
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8617178046624142,
      "learning_rate": 4.523512907644828e-05,
      "loss": 0.8725,
      "step": 6840
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8616638385384007,
      "learning_rate": 4.522839097369482e-05,
      "loss": 0.8988,
      "step": 6845
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9129405302298863,
      "learning_rate": 4.522164861265317e-05,
      "loss": 0.8525,
      "step": 6850
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.993762251834204,
      "learning_rate": 4.521490199474266e-05,
      "loss": 0.9097,
      "step": 6855
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8627959591614259,
      "learning_rate": 4.5208151121383546e-05,
      "loss": 0.9018,
      "step": 6860
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8772948923706035,
      "learning_rate": 4.520139599399693e-05,
      "loss": 0.8907,
      "step": 6865
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9281018046141196,
      "learning_rate": 4.519463661400485e-05,
      "loss": 0.8834,
      "step": 6870
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0879450505596755,
      "learning_rate": 4.518787298283023e-05,
      "loss": 0.8029,
      "step": 6875
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.964585574353663,
      "learning_rate": 4.518110510189688e-05,
      "loss": 0.833,
      "step": 6880
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0101205874621093,
      "learning_rate": 4.517433297262951e-05,
      "loss": 0.9045,
      "step": 6885
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9226063392721335,
      "learning_rate": 4.516755659645372e-05,
      "loss": 0.872,
      "step": 6890
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8649322144361529,
      "learning_rate": 4.5160775974796024e-05,
      "loss": 0.8576,
      "step": 6895
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8894395684373384,
      "learning_rate": 4.5153991109083793e-05,
      "loss": 0.9085,
      "step": 6900
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0986583258385774,
      "learning_rate": 4.514720200074532e-05,
      "loss": 0.8539,
      "step": 6905
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.821887475104247,
      "learning_rate": 4.5140408651209786e-05,
      "loss": 0.8554,
      "step": 6910
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8520470635574051,
      "learning_rate": 4.513361106190727e-05,
      "loss": 0.8537,
      "step": 6915
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9575081196572843,
      "learning_rate": 4.512680923426872e-05,
      "loss": 0.8286,
      "step": 6920
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8687385291363767,
      "learning_rate": 4.512000316972601e-05,
      "loss": 0.8684,
      "step": 6925
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8291128300019188,
      "learning_rate": 4.5113192869711867e-05,
      "loss": 0.866,
      "step": 6930
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9040886126399529,
      "learning_rate": 4.510637833565995e-05,
      "loss": 0.892,
      "step": 6935
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9257901574732298,
      "learning_rate": 4.509955956900477e-05,
      "loss": 0.8653,
      "step": 6940
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8725862551744366,
      "learning_rate": 4.509273657118176e-05,
      "loss": 0.8321,
      "step": 6945
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9536244255736042,
      "learning_rate": 4.508590934362724e-05,
      "loss": 0.8459,
      "step": 6950
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9536386407407413,
      "learning_rate": 4.50790778877784e-05,
      "loss": 0.9171,
      "step": 6955
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2092264578524776,
      "learning_rate": 4.5072242205073344e-05,
      "loss": 0.9571,
      "step": 6960
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9257429332429866,
      "learning_rate": 4.5065402296951045e-05,
      "loss": 0.8044,
      "step": 6965
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.937915435848398,
      "learning_rate": 4.505855816485137e-05,
      "loss": 0.8464,
      "step": 6970
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8423602784480106,
      "learning_rate": 4.50517098102151e-05,
      "loss": 0.8331,
      "step": 6975
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9221368086856745,
      "learning_rate": 4.5044857234483875e-05,
      "loss": 0.8868,
      "step": 6980
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8778849955736138,
      "learning_rate": 4.5038000439100225e-05,
      "loss": 0.8345,
      "step": 6985
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.851903570309572,
      "learning_rate": 4.5031139425507594e-05,
      "loss": 0.9198,
      "step": 6990
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.849994639818449,
      "learning_rate": 4.502427419515029e-05,
      "loss": 0.8434,
      "step": 6995
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9802392034633585,
      "learning_rate": 4.501740474947351e-05,
      "loss": 0.7763,
      "step": 7000
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9829221380810969,
      "learning_rate": 4.501053108992336e-05,
      "loss": 0.8766,
      "step": 7005
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0364018301876563,
      "learning_rate": 4.50036532179468e-05,
      "loss": 0.947,
      "step": 7010
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.894787156679939,
      "learning_rate": 4.49967711349917e-05,
      "loss": 0.8863,
      "step": 7015
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8963814443599233,
      "learning_rate": 4.498988484250681e-05,
      "loss": 0.8614,
      "step": 7020
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9396352911688199,
      "learning_rate": 4.498299434194177e-05,
      "loss": 0.9108,
      "step": 7025
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1286919734957201,
      "learning_rate": 4.49760996347471e-05,
      "loss": 0.843,
      "step": 7030
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8928371685122503,
      "learning_rate": 4.496920072237421e-05,
      "loss": 0.8339,
      "step": 7035
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8248669470936781,
      "learning_rate": 4.496229760627539e-05,
      "loss": 0.9174,
      "step": 7040
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8887586721124235,
      "learning_rate": 4.495539028790382e-05,
      "loss": 0.8729,
      "step": 7045
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8319933863371728,
      "learning_rate": 4.494847876871357e-05,
      "loss": 0.857,
      "step": 7050
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8338766329165384,
      "learning_rate": 4.494156305015957e-05,
      "loss": 0.7854,
      "step": 7055
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0364781233200655,
      "learning_rate": 4.493464313369767e-05,
      "loss": 0.8082,
      "step": 7060
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9106671310578824,
      "learning_rate": 4.4927719020784575e-05,
      "loss": 0.9232,
      "step": 7065
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9719011131736461,
      "learning_rate": 4.492079071287789e-05,
      "loss": 0.8632,
      "step": 7070
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8383111219368286,
      "learning_rate": 4.491385821143608e-05,
      "loss": 0.8703,
      "step": 7075
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9990005730927887,
      "learning_rate": 4.490692151791853e-05,
      "loss": 0.8291,
      "step": 7080
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8941197465996347,
      "learning_rate": 4.489998063378547e-05,
      "loss": 0.8375,
      "step": 7085
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9295576392869906,
      "learning_rate": 4.489303556049804e-05,
      "loss": 0.8976,
      "step": 7090
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0573081770403099,
      "learning_rate": 4.488608629951825e-05,
      "loss": 0.9799,
      "step": 7095
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9466852685840387,
      "learning_rate": 4.487913285230898e-05,
      "loss": 0.9285,
      "step": 7100
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8787468187624394,
      "learning_rate": 4.4872175220334024e-05,
      "loss": 0.8556,
      "step": 7105
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9993758420568105,
      "learning_rate": 4.4865213405058014e-05,
      "loss": 0.9783,
      "step": 7110
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9959396538377233,
      "learning_rate": 4.4858247407946496e-05,
      "loss": 0.8926,
      "step": 7115
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8170150853699522,
      "learning_rate": 4.4851277230465894e-05,
      "loss": 0.9501,
      "step": 7120
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9171553468384273,
      "learning_rate": 4.4844302874083486e-05,
      "loss": 0.9088,
      "step": 7125
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7953580046271129,
      "learning_rate": 4.4837324340267464e-05,
      "loss": 0.8915,
      "step": 7130
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8540800082337093,
      "learning_rate": 4.483034163048686e-05,
      "loss": 0.7784,
      "step": 7135
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8279915498440072,
      "learning_rate": 4.4823354746211635e-05,
      "loss": 0.8532,
      "step": 7140
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9172370024750921,
      "learning_rate": 4.4816363688912576e-05,
      "loss": 0.8394,
      "step": 7145
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8689893295016209,
      "learning_rate": 4.4809368460061396e-05,
      "loss": 0.926,
      "step": 7150
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.98822179973456,
      "learning_rate": 4.480236906113066e-05,
      "loss": 0.8924,
      "step": 7155
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.91796060807761,
      "learning_rate": 4.4795365493593785e-05,
      "loss": 0.7828,
      "step": 7160
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9321675939799655,
      "learning_rate": 4.478835775892514e-05,
      "loss": 0.817,
      "step": 7165
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.84580744467292,
      "learning_rate": 4.47813458585999e-05,
      "loss": 0.9439,
      "step": 7170
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.076125532310432,
      "learning_rate": 4.477432979409414e-05,
      "loss": 0.8454,
      "step": 7175
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8852067963073116,
      "learning_rate": 4.476730956688483e-05,
      "loss": 0.8063,
      "step": 7180
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7138234884476036,
      "learning_rate": 4.47602851784498e-05,
      "loss": 0.8239,
      "step": 7185
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8755414679994341,
      "learning_rate": 4.4753256630267746e-05,
      "loss": 0.9468,
      "step": 7190
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8677059920952861,
      "learning_rate": 4.4746223923818264e-05,
      "loss": 0.9976,
      "step": 7195
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7956263078427499,
      "learning_rate": 4.47391870605818e-05,
      "loss": 0.8508,
      "step": 7200
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.965560728475822,
      "learning_rate": 4.473214604203969e-05,
      "loss": 0.8297,
      "step": 7205
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.863538707034915,
      "learning_rate": 4.4725100869674155e-05,
      "loss": 0.933,
      "step": 7210
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9302430461321485,
      "learning_rate": 4.4718051544968266e-05,
      "loss": 0.8753,
      "step": 7215
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8806079690272463,
      "learning_rate": 4.4710998069405975e-05,
      "loss": 0.8898,
      "step": 7220
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0550357765235905,
      "learning_rate": 4.470394044447212e-05,
      "loss": 0.9357,
      "step": 7225
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8581816349941516,
      "learning_rate": 4.469687867165241e-05,
      "loss": 0.8513,
      "step": 7230
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9317271016868308,
      "learning_rate": 4.468981275243342e-05,
      "loss": 0.9328,
      "step": 7235
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9039812407732404,
      "learning_rate": 4.468274268830259e-05,
      "loss": 0.8282,
      "step": 7240
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9215722972687789,
      "learning_rate": 4.467566848074825e-05,
      "loss": 0.8609,
      "step": 7245
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.926232072579969,
      "learning_rate": 4.4668590131259594e-05,
      "loss": 0.8559,
      "step": 7250
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8941784287881752,
      "learning_rate": 4.4661507641326685e-05,
      "loss": 0.8695,
      "step": 7255
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9449107847537849,
      "learning_rate": 4.4654421012440476e-05,
      "loss": 0.8083,
      "step": 7260
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8335736185175803,
      "learning_rate": 4.464733024609275e-05,
      "loss": 0.9429,
      "step": 7265
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0711296982828185,
      "learning_rate": 4.464023534377622e-05,
      "loss": 0.9065,
      "step": 7270
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9088398525717711,
      "learning_rate": 4.46331363069844e-05,
      "loss": 0.8549,
      "step": 7275
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9450818322946247,
      "learning_rate": 4.462603313721174e-05,
      "loss": 0.9091,
      "step": 7280
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8363897736840603,
      "learning_rate": 4.461892583595353e-05,
      "loss": 0.8449,
      "step": 7285
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8339214324087547,
      "learning_rate": 4.461181440470592e-05,
      "loss": 0.8904,
      "step": 7290
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.906352616047358,
      "learning_rate": 4.460469884496595e-05,
      "loss": 0.8437,
      "step": 7295
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8389865386863039,
      "learning_rate": 4.4597579158231505e-05,
      "loss": 0.8968,
      "step": 7300
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8889636566786414,
      "learning_rate": 4.4590455346001375e-05,
      "loss": 0.8225,
      "step": 7305
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8750682110510425,
      "learning_rate": 4.458332740977519e-05,
      "loss": 0.9207,
      "step": 7310
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9855576560554214,
      "learning_rate": 4.4576195351053444e-05,
      "loss": 0.8669,
      "step": 7315
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.993406261887127,
      "learning_rate": 4.4569059171337514e-05,
      "loss": 0.8795,
      "step": 7320
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8275597830126532,
      "learning_rate": 4.4561918872129646e-05,
      "loss": 0.7501,
      "step": 7325
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.852750929980776,
      "learning_rate": 4.4554774454932944e-05,
      "loss": 0.8946,
      "step": 7330
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9669793933891,
      "learning_rate": 4.454762592125139e-05,
      "loss": 0.8546,
      "step": 7335
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4807142343419046,
      "learning_rate": 4.454047327258982e-05,
      "loss": 0.9621,
      "step": 7340
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8685218519116757,
      "learning_rate": 4.4533316510453936e-05,
      "loss": 0.8769,
      "step": 7345
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8838099072632095,
      "learning_rate": 4.452615563635032e-05,
      "loss": 0.8063,
      "step": 7350
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8384367798190346,
      "learning_rate": 4.451899065178641e-05,
      "loss": 0.8492,
      "step": 7355
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0879312532034804,
      "learning_rate": 4.4511821558270505e-05,
      "loss": 0.8746,
      "step": 7360
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8753858386142155,
      "learning_rate": 4.4504648357311774e-05,
      "loss": 0.8384,
      "step": 7365
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8975403180372352,
      "learning_rate": 4.449747105042026e-05,
      "loss": 0.8702,
      "step": 7370
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8824452326774359,
      "learning_rate": 4.449028963910685e-05,
      "loss": 0.8306,
      "step": 7375
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9338620618287319,
      "learning_rate": 4.4483104124883324e-05,
      "loss": 0.8762,
      "step": 7380
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8917119009385744,
      "learning_rate": 4.447591450926229e-05,
      "loss": 0.8558,
      "step": 7385
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8499078539903893,
      "learning_rate": 4.4468720793757236e-05,
      "loss": 0.8949,
      "step": 7390
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9651890648776259,
      "learning_rate": 4.446152297988252e-05,
      "loss": 0.908,
      "step": 7395
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9031133732989215,
      "learning_rate": 4.445432106915337e-05,
      "loss": 0.8266,
      "step": 7400
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8490767265607513,
      "learning_rate": 4.444711506308584e-05,
      "loss": 0.8585,
      "step": 7405
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9354401790724719,
      "learning_rate": 4.4439904963196894e-05,
      "loss": 0.8587,
      "step": 7410
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8416776015267864,
      "learning_rate": 4.443269077100431e-05,
      "loss": 0.8578,
      "step": 7415
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9068181958709155,
      "learning_rate": 4.4425472488026774e-05,
      "loss": 0.8348,
      "step": 7420
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8552300375235387,
      "learning_rate": 4.44182501157838e-05,
      "loss": 0.8834,
      "step": 7425
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1822478721503025,
      "learning_rate": 4.4411023655795766e-05,
      "loss": 0.9594,
      "step": 7430
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9337057371607408,
      "learning_rate": 4.440379310958392e-05,
      "loss": 0.952,
      "step": 7435
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8881575925688209,
      "learning_rate": 4.439655847867038e-05,
      "loss": 0.887,
      "step": 7440
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8060294523314736,
      "learning_rate": 4.4389319764578095e-05,
      "loss": 0.7746,
      "step": 7445
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8504355014812155,
      "learning_rate": 4.43820769688309e-05,
      "loss": 0.8395,
      "step": 7450
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8735277695511526,
      "learning_rate": 4.4374830092953487e-05,
      "loss": 0.9266,
      "step": 7455
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9524538225370099,
      "learning_rate": 4.436757913847138e-05,
      "loss": 0.8189,
      "step": 7460
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9359354689109824,
      "learning_rate": 4.436032410691099e-05,
      "loss": 0.9346,
      "step": 7465
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.847021754575933,
      "learning_rate": 4.4353064999799583e-05,
      "loss": 0.8342,
      "step": 7470
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1582344510460938,
      "learning_rate": 4.4345801818665266e-05,
      "loss": 0.9424,
      "step": 7475
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9179071600486363,
      "learning_rate": 4.4338534565037026e-05,
      "loss": 0.8427,
      "step": 7480
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8281592959524418,
      "learning_rate": 4.433126324044469e-05,
      "loss": 0.7833,
      "step": 7485
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9500621494862264,
      "learning_rate": 4.4323987846418946e-05,
      "loss": 0.9238,
      "step": 7490
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9553499030283522,
      "learning_rate": 4.431670838449134e-05,
      "loss": 0.8584,
      "step": 7495
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9655582350564316,
      "learning_rate": 4.4309424856194285e-05,
      "loss": 0.8825,
      "step": 7500
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8487596128740992,
      "learning_rate": 4.430213726306103e-05,
      "loss": 0.9031,
      "step": 7505
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9083099187740944,
      "learning_rate": 4.429484560662569e-05,
      "loss": 0.9605,
      "step": 7510
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9626390301041929,
      "learning_rate": 4.428754988842324e-05,
      "loss": 0.8377,
      "step": 7515
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8429245794037871,
      "learning_rate": 4.42802501099895e-05,
      "loss": 0.9388,
      "step": 7520
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8515202658200619,
      "learning_rate": 4.427294627286116e-05,
      "loss": 0.906,
      "step": 7525
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8423680410003936,
      "learning_rate": 4.426563837857574e-05,
      "loss": 0.8997,
      "step": 7530
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8455909980356767,
      "learning_rate": 4.425832642867164e-05,
      "loss": 0.8226,
      "step": 7535
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0586815596505421,
      "learning_rate": 4.42510104246881e-05,
      "loss": 0.8735,
      "step": 7540
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7989108896023916,
      "learning_rate": 4.4243690368165206e-05,
      "loss": 0.7778,
      "step": 7545
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9034705300027844,
      "learning_rate": 4.423636626064391e-05,
      "loss": 0.8303,
      "step": 7550
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8738189650258167,
      "learning_rate": 4.4229038103666035e-05,
      "loss": 0.8344,
      "step": 7555
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9029087738200049,
      "learning_rate": 4.4221705898774196e-05,
      "loss": 0.8146,
      "step": 7560
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.802702294189799,
      "learning_rate": 4.421436964751193e-05,
      "loss": 0.9148,
      "step": 7565
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9196903884412462,
      "learning_rate": 4.420702935142358e-05,
      "loss": 0.8627,
      "step": 7570
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9192381209596086,
      "learning_rate": 4.419968501205436e-05,
      "loss": 0.7788,
      "step": 7575
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.875387380395048,
      "learning_rate": 4.4192336630950335e-05,
      "loss": 0.8942,
      "step": 7580
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8370622170085728,
      "learning_rate": 4.418498420965841e-05,
      "loss": 0.8686,
      "step": 7585
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.976379331606312,
      "learning_rate": 4.417762774972634e-05,
      "loss": 0.905,
      "step": 7590
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8351051453852988,
      "learning_rate": 4.417026725270275e-05,
      "loss": 0.8752,
      "step": 7595
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8640799042921292,
      "learning_rate": 4.41629027201371e-05,
      "loss": 0.8371,
      "step": 7600
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.918241552380792,
      "learning_rate": 4.415553415357969e-05,
      "loss": 0.8618,
      "step": 7605
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8356063467697566,
      "learning_rate": 4.414816155458169e-05,
      "loss": 0.8498,
      "step": 7610
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8696933172220548,
      "learning_rate": 4.4140784924695113e-05,
      "loss": 0.9145,
      "step": 7615
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8086781616429023,
      "learning_rate": 4.413340426547281e-05,
      "loss": 0.9038,
      "step": 7620
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8995082272653837,
      "learning_rate": 4.412601957846848e-05,
      "loss": 0.9295,
      "step": 7625
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8417526696960922,
      "learning_rate": 4.411863086523669e-05,
      "loss": 0.7611,
      "step": 7630
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8964104192476045,
      "learning_rate": 4.4111238127332846e-05,
      "loss": 0.7623,
      "step": 7635
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8097264311749593,
      "learning_rate": 4.410384136631317e-05,
      "loss": 0.7836,
      "step": 7640
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8553596703542771,
      "learning_rate": 4.409644058373479e-05,
      "loss": 0.81,
      "step": 7645
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9262993151525223,
      "learning_rate": 4.4089035781155624e-05,
      "loss": 0.9227,
      "step": 7650
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8414048889190111,
      "learning_rate": 4.408162696013447e-05,
      "loss": 0.9514,
      "step": 7655
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9945279706822402,
      "learning_rate": 4.407421412223097e-05,
      "loss": 0.9017,
      "step": 7660
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8764404857676572,
      "learning_rate": 4.406679726900559e-05,
      "loss": 0.8853,
      "step": 7665
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9700127757280103,
      "learning_rate": 4.4059376402019667e-05,
      "loss": 0.9016,
      "step": 7670
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9122234958658786,
      "learning_rate": 4.405195152283536e-05,
      "loss": 0.7754,
      "step": 7675
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8377198184403536,
      "learning_rate": 4.4044522633015696e-05,
      "loss": 0.8706,
      "step": 7680
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9509993766392825,
      "learning_rate": 4.403708973412453e-05,
      "loss": 0.9026,
      "step": 7685
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9211180166402848,
      "learning_rate": 4.402965282772655e-05,
      "loss": 0.8235,
      "step": 7690
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8863883737860377,
      "learning_rate": 4.4022211915387324e-05,
      "loss": 0.8747,
      "step": 7695
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9679860825554182,
      "learning_rate": 4.4014766998673235e-05,
      "loss": 0.8785,
      "step": 7700
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0920781342037216,
      "learning_rate": 4.400731807915151e-05,
      "loss": 0.8382,
      "step": 7705
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0631999306433024,
      "learning_rate": 4.399986515839024e-05,
      "loss": 0.941,
      "step": 7710
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9738722037707661,
      "learning_rate": 4.399240823795832e-05,
      "loss": 0.9302,
      "step": 7715
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0510927259912366,
      "learning_rate": 4.398494731942553e-05,
      "loss": 0.8237,
      "step": 7720
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9407111049629848,
      "learning_rate": 4.397748240436246e-05,
      "loss": 0.9747,
      "step": 7725
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7885052263345511,
      "learning_rate": 4.3970013494340554e-05,
      "loss": 0.8429,
      "step": 7730
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9167767667033337,
      "learning_rate": 4.39625405909321e-05,
      "loss": 0.8546,
      "step": 7735
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8899655500112281,
      "learning_rate": 4.395506369571022e-05,
      "loss": 0.8293,
      "step": 7740
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9261869430632528,
      "learning_rate": 4.394758281024888e-05,
      "loss": 0.9227,
      "step": 7745
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.604827012798249,
      "learning_rate": 4.3940097936122885e-05,
      "loss": 0.8483,
      "step": 7750
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9843785975297291,
      "learning_rate": 4.393260907490787e-05,
      "loss": 0.8694,
      "step": 7755
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8153554723451127,
      "learning_rate": 4.392511622818034e-05,
      "loss": 0.8588,
      "step": 7760
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8989001263785075,
      "learning_rate": 4.39176193975176e-05,
      "loss": 0.9211,
      "step": 7765
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8703437261717202,
      "learning_rate": 4.391011858449782e-05,
      "loss": 0.8237,
      "step": 7770
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8786086110597998,
      "learning_rate": 4.390261379069999e-05,
      "loss": 0.8792,
      "step": 7775
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8811646614688738,
      "learning_rate": 4.389510501770395e-05,
      "loss": 0.8552,
      "step": 7780
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9546398739304482,
      "learning_rate": 4.388759226709038e-05,
      "loss": 0.8962,
      "step": 7785
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2749201624682613,
      "learning_rate": 4.38800755404408e-05,
      "loss": 0.773,
      "step": 7790
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9685431429058441,
      "learning_rate": 4.3872554839337556e-05,
      "loss": 1.0673,
      "step": 7795
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9578475078534364,
      "learning_rate": 4.386503016536382e-05,
      "loss": 0.8263,
      "step": 7800
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8811396651535431,
      "learning_rate": 4.385750152010362e-05,
      "loss": 0.7402,
      "step": 7805
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9070274254113131,
      "learning_rate": 4.384996890514183e-05,
      "loss": 0.9536,
      "step": 7810
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8982033015188957,
      "learning_rate": 4.3842432322064125e-05,
      "loss": 0.8949,
      "step": 7815
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8615567369881546,
      "learning_rate": 4.383489177245705e-05,
      "loss": 0.8619,
      "step": 7820
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8434363800544157,
      "learning_rate": 4.382734725790796e-05,
      "loss": 0.8753,
      "step": 7825
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1535678646719105,
      "learning_rate": 4.381979878000506e-05,
      "loss": 0.941,
      "step": 7830
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9130487069363806,
      "learning_rate": 4.381224634033738e-05,
      "loss": 0.8545,
      "step": 7835
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9812289780991902,
      "learning_rate": 4.380468994049478e-05,
      "loss": 0.9435,
      "step": 7840
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8292687856191864,
      "learning_rate": 4.379712958206798e-05,
      "loss": 0.8656,
      "step": 7845
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0232887613760528,
      "learning_rate": 4.37895652666485e-05,
      "loss": 0.9256,
      "step": 7850
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8886137181697363,
      "learning_rate": 4.378199699582871e-05,
      "loss": 0.8973,
      "step": 7855
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9205648031929387,
      "learning_rate": 4.377442477120182e-05,
      "loss": 0.9087,
      "step": 7860
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8785491269263982,
      "learning_rate": 4.3766848594361854e-05,
      "loss": 0.8481,
      "step": 7865
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9416529397448189,
      "learning_rate": 4.3759268466903675e-05,
      "loss": 0.8888,
      "step": 7870
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8939233801619114,
      "learning_rate": 4.3751684390422984e-05,
      "loss": 0.9231,
      "step": 7875
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.975428929451796,
      "learning_rate": 4.37440963665163e-05,
      "loss": 0.8259,
      "step": 7880
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.037004013318425,
      "learning_rate": 4.3736504396781e-05,
      "loss": 0.868,
      "step": 7885
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9111686711396364,
      "learning_rate": 4.372890848281525e-05,
      "loss": 0.8528,
      "step": 7890
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8140355105059677,
      "learning_rate": 4.3721308626218085e-05,
      "loss": 0.8923,
      "step": 7895
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9158453273251607,
      "learning_rate": 4.3713704828589346e-05,
      "loss": 0.8618,
      "step": 7900
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9220709512441392,
      "learning_rate": 4.370609709152972e-05,
      "loss": 0.8994,
      "step": 7905
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9473518294180306,
      "learning_rate": 4.369848541664071e-05,
      "loss": 0.878,
      "step": 7910
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8482481449234285,
      "learning_rate": 4.369086980552466e-05,
      "loss": 0.8266,
      "step": 7915
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.993146400834012,
      "learning_rate": 4.368325025978472e-05,
      "loss": 0.9275,
      "step": 7920
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.887792878713791,
      "learning_rate": 4.36756267810249e-05,
      "loss": 0.7967,
      "step": 7925
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8569225427737016,
      "learning_rate": 4.366799937085002e-05,
      "loss": 0.7997,
      "step": 7930
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.830616619635648,
      "learning_rate": 4.366036803086573e-05,
      "loss": 0.8733,
      "step": 7935
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0218893664993307,
      "learning_rate": 4.365273276267849e-05,
      "loss": 0.7861,
      "step": 7940
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8766447865684245,
      "learning_rate": 4.3645093567895626e-05,
      "loss": 1.2592,
      "step": 7945
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7944329052758466,
      "learning_rate": 4.363745044812526e-05,
      "loss": 0.7989,
      "step": 7950
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8536075854356092,
      "learning_rate": 4.362980340497635e-05,
      "loss": 0.8326,
      "step": 7955
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7784297862170187,
      "learning_rate": 4.3622152440058674e-05,
      "loss": 0.7916,
      "step": 7960
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8848112222364299,
      "learning_rate": 4.361449755498285e-05,
      "loss": 0.8864,
      "step": 7965
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8524859394050325,
      "learning_rate": 4.36068387513603e-05,
      "loss": 0.8436,
      "step": 7970
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8654517155959764,
      "learning_rate": 4.359917603080329e-05,
      "loss": 0.8255,
      "step": 7975
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9139644605327134,
      "learning_rate": 4.35915093949249e-05,
      "loss": 0.9451,
      "step": 7980
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.902218266135654,
      "learning_rate": 4.358383884533903e-05,
      "loss": 0.946,
      "step": 7985
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.91475779195718,
      "learning_rate": 4.3576164383660436e-05,
      "loss": 0.9333,
      "step": 7990
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.905054290621748,
      "learning_rate": 4.356848601150464e-05,
      "loss": 0.8524,
      "step": 7995
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8029074963074335,
      "learning_rate": 4.3560803730488044e-05,
      "loss": 0.8801,
      "step": 8000
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.831531752281927,
      "learning_rate": 4.355311754222784e-05,
      "loss": 0.8057,
      "step": 8005
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8249539916397222,
      "learning_rate": 4.354542744834204e-05,
      "loss": 0.8134,
      "step": 8010
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.856510386137243,
      "learning_rate": 4.353773345044951e-05,
      "loss": 0.8655,
      "step": 8015
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0670913060531149,
      "learning_rate": 4.353003555016991e-05,
      "loss": 0.9247,
      "step": 8020
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9756111892275695,
      "learning_rate": 4.3522333749123716e-05,
      "loss": 0.9578,
      "step": 8025
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8848304769493224,
      "learning_rate": 4.3514628048932255e-05,
      "loss": 0.8264,
      "step": 8030
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8748368542147439,
      "learning_rate": 4.350691845121765e-05,
      "loss": 0.9501,
      "step": 8035
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8974351797393816,
      "learning_rate": 4.349920495760285e-05,
      "loss": 0.8997,
      "step": 8040
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2611093166185807,
      "learning_rate": 4.349148756971163e-05,
      "loss": 0.8797,
      "step": 8045
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.079256272851702,
      "learning_rate": 4.3483766289168584e-05,
      "loss": 0.8593,
      "step": 8050
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8340106284949618,
      "learning_rate": 4.347604111759911e-05,
      "loss": 0.8293,
      "step": 8055
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7998464253661705,
      "learning_rate": 4.346831205662944e-05,
      "loss": 0.8679,
      "step": 8060
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9589135364911668,
      "learning_rate": 4.3460579107886645e-05,
      "loss": 0.929,
      "step": 8065
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3926542476798163,
      "learning_rate": 4.345284227299857e-05,
      "loss": 0.9015,
      "step": 8070
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0190252572650633,
      "learning_rate": 4.3445101553593905e-05,
      "loss": 0.9291,
      "step": 8075
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8653309228283771,
      "learning_rate": 4.3437356951302146e-05,
      "loss": 0.8695,
      "step": 8080
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7860877367668022,
      "learning_rate": 4.342960846775363e-05,
      "loss": 0.8471,
      "step": 8085
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8785542500215634,
      "learning_rate": 4.342185610457947e-05,
      "loss": 0.8249,
      "step": 8090
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0210060558161365,
      "learning_rate": 4.3414099863411636e-05,
      "loss": 0.8873,
      "step": 8095
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9000806358746681,
      "learning_rate": 4.340633974588289e-05,
      "loss": 0.7748,
      "step": 8100
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8143824814684562,
      "learning_rate": 4.3398575753626836e-05,
      "loss": 0.8731,
      "step": 8105
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8370597099782764,
      "learning_rate": 4.339080788827785e-05,
      "loss": 0.8549,
      "step": 8110
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8872156968522008,
      "learning_rate": 4.338303615147117e-05,
      "loss": 0.8865,
      "step": 8115
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9910013496879678,
      "learning_rate": 4.337526054484281e-05,
      "loss": 0.8821,
      "step": 8120
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8928308931971602,
      "learning_rate": 4.336748107002964e-05,
      "loss": 0.9145,
      "step": 8125
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9530466754362821,
      "learning_rate": 4.335969772866929e-05,
      "loss": 0.875,
      "step": 8130
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9007972002630497,
      "learning_rate": 4.335191052240026e-05,
      "loss": 0.9073,
      "step": 8135
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8315730303923892,
      "learning_rate": 4.334411945286183e-05,
      "loss": 0.9835,
      "step": 8140
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9965317077573741,
      "learning_rate": 4.3336324521694104e-05,
      "loss": 0.8098,
      "step": 8145
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8585706172071598,
      "learning_rate": 4.3328525730538e-05,
      "loss": 0.9347,
      "step": 8150
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8972964846110897,
      "learning_rate": 4.332072308103523e-05,
      "loss": 0.8488,
      "step": 8155
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8364032291090727,
      "learning_rate": 4.3312916574828346e-05,
      "loss": 0.821,
      "step": 8160
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9005046384498866,
      "learning_rate": 4.33051062135607e-05,
      "loss": 1.0117,
      "step": 8165
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8845071675603398,
      "learning_rate": 4.329729199887644e-05,
      "loss": 0.8134,
      "step": 8170
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7576112709340472,
      "learning_rate": 4.328947393242056e-05,
      "loss": 0.807,
      "step": 8175
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8334840626733158,
      "learning_rate": 4.328165201583884e-05,
      "loss": 0.8181,
      "step": 8180
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8981205993340662,
      "learning_rate": 4.327382625077787e-05,
      "loss": 0.938,
      "step": 8185
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8008175152417177,
      "learning_rate": 4.3265996638885054e-05,
      "loss": 0.8813,
      "step": 8190
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9899264131841488,
      "learning_rate": 4.325816318180861e-05,
      "loss": 0.9119,
      "step": 8195
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.986877206037316,
      "learning_rate": 4.325032588119757e-05,
      "loss": 0.8776,
      "step": 8200
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.830813507739178,
      "learning_rate": 4.324248473870176e-05,
      "loss": 0.8436,
      "step": 8205
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9872635561319187,
      "learning_rate": 4.323463975597182e-05,
      "loss": 0.9274,
      "step": 8210
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9530456433579224,
      "learning_rate": 4.322679093465921e-05,
      "loss": 0.9176,
      "step": 8215
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9718604669934869,
      "learning_rate": 4.3218938276416176e-05,
      "loss": 0.9145,
      "step": 8220
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8867516134392972,
      "learning_rate": 4.32110817828958e-05,
      "loss": 0.8624,
      "step": 8225
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.858527411325421,
      "learning_rate": 4.320322145575194e-05,
      "loss": 0.8577,
      "step": 8230
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8929595856361262,
      "learning_rate": 4.319535729663929e-05,
      "loss": 0.8205,
      "step": 8235
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8736540470074058,
      "learning_rate": 4.3187489307213344e-05,
      "loss": 0.8287,
      "step": 8240
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8279054855564584,
      "learning_rate": 4.317961748913038e-05,
      "loss": 0.8817,
      "step": 8245
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.060861338637028,
      "learning_rate": 4.3171741844047506e-05,
      "loss": 0.8359,
      "step": 8250
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9268994828497623,
      "learning_rate": 4.316386237362262e-05,
      "loss": 0.8926,
      "step": 8255
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9288009933447495,
      "learning_rate": 4.315597907951444e-05,
      "loss": 0.8847,
      "step": 8260
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9456603140696496,
      "learning_rate": 4.314809196338249e-05,
      "loss": 0.9044,
      "step": 8265
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9409157423481455,
      "learning_rate": 4.3140201026887076e-05,
      "loss": 0.9362,
      "step": 8270
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0359051877183574,
      "learning_rate": 4.3132306271689325e-05,
      "loss": 0.918,
      "step": 8275
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8632678728285839,
      "learning_rate": 4.3124407699451174e-05,
      "loss": 0.9411,
      "step": 8280
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9006400051978153,
      "learning_rate": 4.3116505311835356e-05,
      "loss": 0.9037,
      "step": 8285
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9222179607791998,
      "learning_rate": 4.3108599110505395e-05,
      "loss": 0.8709,
      "step": 8290
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9194497962870436,
      "learning_rate": 4.310068909712564e-05,
      "loss": 0.8563,
      "step": 8295
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8560794466219108,
      "learning_rate": 4.309277527336122e-05,
      "loss": 0.8541,
      "step": 8300
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9134584307097554,
      "learning_rate": 4.3084857640878085e-05,
      "loss": 0.937,
      "step": 8305
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8631699793571805,
      "learning_rate": 4.307693620134298e-05,
      "loss": 0.8839,
      "step": 8310
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7938715044186743,
      "learning_rate": 4.306901095642345e-05,
      "loss": 0.7466,
      "step": 8315
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.908922549759451,
      "learning_rate": 4.306108190778785e-05,
      "loss": 0.8871,
      "step": 8320
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9613961768025776,
      "learning_rate": 4.305314905710531e-05,
      "loss": 0.8913,
      "step": 8325
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8218487721111957,
      "learning_rate": 4.304521240604579e-05,
      "loss": 1.0229,
      "step": 8330
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9226161424710482,
      "learning_rate": 4.3037271956280034e-05,
      "loss": 0.7477,
      "step": 8335
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.059244778426306,
      "learning_rate": 4.3029327709479606e-05,
      "loss": 0.8783,
      "step": 8340
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.990054633319157,
      "learning_rate": 4.3021379667316825e-05,
      "loss": 0.8801,
      "step": 8345
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8454553073110684,
      "learning_rate": 4.301342783146486e-05,
      "loss": 0.9038,
      "step": 8350
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0051622442985124,
      "learning_rate": 4.300547220359764e-05,
      "loss": 0.8841,
      "step": 8355
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0767182244496156,
      "learning_rate": 4.2997512785389924e-05,
      "loss": 0.865,
      "step": 8360
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9068668624822273,
      "learning_rate": 4.2989549578517235e-05,
      "loss": 0.9434,
      "step": 8365
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9601550402458588,
      "learning_rate": 4.298158258465592e-05,
      "loss": 0.8319,
      "step": 8370
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1524707054789232,
      "learning_rate": 4.297361180548313e-05,
      "loss": 0.9629,
      "step": 8375
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.86720359093305,
      "learning_rate": 4.296563724267677e-05,
      "loss": 0.8412,
      "step": 8380
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0625063481591932,
      "learning_rate": 4.295765889791558e-05,
      "loss": 0.8373,
      "step": 8385
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.804245606488221,
      "learning_rate": 4.2949676772879096e-05,
      "loss": 0.8035,
      "step": 8390
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9797327685527931,
      "learning_rate": 4.294169086924763e-05,
      "loss": 0.9291,
      "step": 8395
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9440667048256564,
      "learning_rate": 4.293370118870229e-05,
      "loss": 0.8395,
      "step": 8400
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8492610007357806,
      "learning_rate": 4.2925707732925e-05,
      "loss": 0.8207,
      "step": 8405
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.891723107223008,
      "learning_rate": 4.2917710503598456e-05,
      "loss": 0.8967,
      "step": 8410
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0062080236250128,
      "learning_rate": 4.290970950240617e-05,
      "loss": 0.9174,
      "step": 8415
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9200574284219023,
      "learning_rate": 4.290170473103243e-05,
      "loss": 0.9054,
      "step": 8420
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.868072689426447,
      "learning_rate": 4.289369619116233e-05,
      "loss": 0.8697,
      "step": 8425
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0437361215521352,
      "learning_rate": 4.288568388448174e-05,
      "loss": 0.8899,
      "step": 8430
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8937510612281192,
      "learning_rate": 4.2877667812677346e-05,
      "loss": 0.8259,
      "step": 8435
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9579781125129077,
      "learning_rate": 4.286964797743661e-05,
      "loss": 0.8266,
      "step": 8440
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9845783571596185,
      "learning_rate": 4.286162438044779e-05,
      "loss": 0.9319,
      "step": 8445
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0235380371029066,
      "learning_rate": 4.285359702339994e-05,
      "loss": 0.9022,
      "step": 8450
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8718725903336981,
      "learning_rate": 4.2845565907982897e-05,
      "loss": 0.8037,
      "step": 8455
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8978180694702957,
      "learning_rate": 4.2837531035887305e-05,
      "loss": 0.8338,
      "step": 8460
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8214721327663154,
      "learning_rate": 4.282949240880458e-05,
      "loss": 0.788,
      "step": 8465
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8217165450763345,
      "learning_rate": 4.282145002842695e-05,
      "loss": 0.8033,
      "step": 8470
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8353144923625703,
      "learning_rate": 4.281340389644739e-05,
      "loss": 0.8433,
      "step": 8475
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9516764231486345,
      "learning_rate": 4.280535401455973e-05,
      "loss": 0.8923,
      "step": 8480
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8359667390068116,
      "learning_rate": 4.279730038445854e-05,
      "loss": 0.9234,
      "step": 8485
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9398406561365003,
      "learning_rate": 4.27892430078392e-05,
      "loss": 0.8771,
      "step": 8490
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0075501297873979,
      "learning_rate": 4.278118188639786e-05,
      "loss": 0.8377,
      "step": 8495
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8543631271581719,
      "learning_rate": 4.2773117021831466e-05,
      "loss": 0.8342,
      "step": 8500
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9249497184350577,
      "learning_rate": 4.276504841583778e-05,
      "loss": 0.7878,
      "step": 8505
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9304974218438378,
      "learning_rate": 4.2756976070115305e-05,
      "loss": 0.8765,
      "step": 8510
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9237755546258369,
      "learning_rate": 4.274889998636338e-05,
      "loss": 0.9027,
      "step": 8515
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8725459720640374,
      "learning_rate": 4.274082016628208e-05,
      "loss": 0.8785,
      "step": 8520
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.865420587506024,
      "learning_rate": 4.27327366115723e-05,
      "loss": 0.8557,
      "step": 8525
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8530479992793641,
      "learning_rate": 4.272464932393572e-05,
      "loss": 0.9093,
      "step": 8530
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8523136600293201,
      "learning_rate": 4.271655830507478e-05,
      "loss": 0.8154,
      "step": 8535
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9319590427323424,
      "learning_rate": 4.2708463556692745e-05,
      "loss": 0.981,
      "step": 8540
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9393820145931006,
      "learning_rate": 4.270036508049364e-05,
      "loss": 0.9124,
      "step": 8545
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8400295085421112,
      "learning_rate": 4.269226287818228e-05,
      "loss": 0.9175,
      "step": 8550
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8476841916976854,
      "learning_rate": 4.268415695146425e-05,
      "loss": 0.8423,
      "step": 8555
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9642996969885996,
      "learning_rate": 4.267604730204594e-05,
      "loss": 0.8078,
      "step": 8560
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9277646533991848,
      "learning_rate": 4.266793393163452e-05,
      "loss": 0.6013,
      "step": 8565
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0706887324499244,
      "learning_rate": 4.265981684193794e-05,
      "loss": 0.6044,
      "step": 8570
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8273192611142373,
      "learning_rate": 4.265169603466493e-05,
      "loss": 0.6383,
      "step": 8575
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8205728595780964,
      "learning_rate": 4.2643571511525006e-05,
      "loss": 0.6131,
      "step": 8580
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.865032770485709,
      "learning_rate": 4.263544327422846e-05,
      "loss": 0.5971,
      "step": 8585
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9061162295993133,
      "learning_rate": 4.2627311324486377e-05,
      "loss": 0.7053,
      "step": 8590
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8434590515527532,
      "learning_rate": 4.261917566401061e-05,
      "loss": 0.6075,
      "step": 8595
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9029329507315093,
      "learning_rate": 4.261103629451381e-05,
      "loss": 0.5954,
      "step": 8600
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8740159041699627,
      "learning_rate": 4.26028932177094e-05,
      "loss": 0.6419,
      "step": 8605
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9079979043161515,
      "learning_rate": 4.259474643531157e-05,
      "loss": 0.6654,
      "step": 8610
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.894369058539577,
      "learning_rate": 4.258659594903532e-05,
      "loss": 0.6042,
      "step": 8615
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8826620387477946,
      "learning_rate": 4.25784417605964e-05,
      "loss": 0.6314,
      "step": 8620
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0169609325560889,
      "learning_rate": 4.2570283871711355e-05,
      "loss": 0.5917,
      "step": 8625
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1388351557100063,
      "learning_rate": 4.256212228409752e-05,
      "loss": 0.6313,
      "step": 8630
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0177762666025187,
      "learning_rate": 4.255395699947297e-05,
      "loss": 0.6291,
      "step": 8635
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8301895445486615,
      "learning_rate": 4.25457880195566e-05,
      "loss": 0.6603,
      "step": 8640
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.938875998709611,
      "learning_rate": 4.253761534606806e-05,
      "loss": 0.5926,
      "step": 8645
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8747794937997334,
      "learning_rate": 4.2529438980727784e-05,
      "loss": 0.5593,
      "step": 8650
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9189268262888526,
      "learning_rate": 4.252125892525698e-05,
      "loss": 0.6634,
      "step": 8655
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0710265091378568,
      "learning_rate": 4.251307518137765e-05,
      "loss": 0.6206,
      "step": 8660
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8851240901140472,
      "learning_rate": 4.250488775081254e-05,
      "loss": 0.6182,
      "step": 8665
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9909197448091167,
      "learning_rate": 4.2496696635285185e-05,
      "loss": 0.6392,
      "step": 8670
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7932300897557871,
      "learning_rate": 4.248850183651992e-05,
      "loss": 0.6653,
      "step": 8675
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.938322781180203,
      "learning_rate": 4.248030335624183e-05,
      "loss": 0.7057,
      "step": 8680
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8229756810065206,
      "learning_rate": 4.247210119617679e-05,
      "loss": 0.5989,
      "step": 8685
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8822519759321903,
      "learning_rate": 4.246389535805141e-05,
      "loss": 0.6396,
      "step": 8690
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8691081495744949,
      "learning_rate": 4.245568584359313e-05,
      "loss": 0.6192,
      "step": 8695
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8599436694083061,
      "learning_rate": 4.244747265453014e-05,
      "loss": 0.6125,
      "step": 8700
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0259563736331991,
      "learning_rate": 4.2439255792591384e-05,
      "loss": 0.5917,
      "step": 8705
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9156046669276074,
      "learning_rate": 4.2431035259506626e-05,
      "loss": 0.5313,
      "step": 8710
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8705457318769528,
      "learning_rate": 4.242281105700634e-05,
      "loss": 0.569,
      "step": 8715
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9337200241141412,
      "learning_rate": 4.241458318682183e-05,
      "loss": 0.5953,
      "step": 8720
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9166056678744614,
      "learning_rate": 4.240635165068515e-05,
      "loss": 0.5987,
      "step": 8725
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9914530288041218,
      "learning_rate": 4.23981164503291e-05,
      "loss": 0.5812,
      "step": 8730
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8363380284298336,
      "learning_rate": 4.23898775874873e-05,
      "loss": 0.6235,
      "step": 8735
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8469659350724175,
      "learning_rate": 4.238163506389412e-05,
      "loss": 0.5904,
      "step": 8740
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0554835077197375,
      "learning_rate": 4.237338888128468e-05,
      "loss": 0.566,
      "step": 8745
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9815381006067133,
      "learning_rate": 4.23651390413949e-05,
      "loss": 0.5417,
      "step": 8750
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8777158051005103,
      "learning_rate": 4.2356885545961456e-05,
      "loss": 0.6009,
      "step": 8755
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9689041668086507,
      "learning_rate": 4.234862839672178e-05,
      "loss": 0.6056,
      "step": 8760
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8421991341881123,
      "learning_rate": 4.234036759541411e-05,
      "loss": 0.5487,
      "step": 8765
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8918366158803973,
      "learning_rate": 4.233210314377742e-05,
      "loss": 0.5595,
      "step": 8770
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1488261033917695,
      "learning_rate": 4.232383504355147e-05,
      "loss": 0.5579,
      "step": 8775
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8756084810754023,
      "learning_rate": 4.2315563296476775e-05,
      "loss": 0.5558,
      "step": 8780
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0034840453434597,
      "learning_rate": 4.230728790429462e-05,
      "loss": 0.6144,
      "step": 8785
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.057321079114524,
      "learning_rate": 4.229900886874708e-05,
      "loss": 0.5941,
      "step": 8790
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8978693265667533,
      "learning_rate": 4.2290726191576954e-05,
      "loss": 0.5667,
      "step": 8795
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8279063536947038,
      "learning_rate": 4.2282439874527854e-05,
      "loss": 0.5823,
      "step": 8800
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8383008864951491,
      "learning_rate": 4.227414991934413e-05,
      "loss": 0.6906,
      "step": 8805
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7895799159407739,
      "learning_rate": 4.22658563277709e-05,
      "loss": 0.5837,
      "step": 8810
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7682059172445806,
      "learning_rate": 4.225755910155407e-05,
      "loss": 0.5899,
      "step": 8815
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3960673280094005,
      "learning_rate": 4.224925824244025e-05,
      "loss": 0.6074,
      "step": 8820
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9385953158411825,
      "learning_rate": 4.22409537521769e-05,
      "loss": 0.5001,
      "step": 8825
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0166282802883184,
      "learning_rate": 4.223264563251219e-05,
      "loss": 0.5527,
      "step": 8830
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9022081026016874,
      "learning_rate": 4.222433388519505e-05,
      "loss": 0.5848,
      "step": 8835
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8869404967605953,
      "learning_rate": 4.221601851197523e-05,
      "loss": 0.6249,
      "step": 8840
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8681790209968652,
      "learning_rate": 4.220769951460315e-05,
      "loss": 0.5946,
      "step": 8845
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8831074405654262,
      "learning_rate": 4.219937689483009e-05,
      "loss": 0.6251,
      "step": 8850
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8372940849502932,
      "learning_rate": 4.219105065440802e-05,
      "loss": 0.6001,
      "step": 8855
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9066301464572489,
      "learning_rate": 4.218272079508972e-05,
      "loss": 0.7102,
      "step": 8860
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.1079515746735977,
      "learning_rate": 4.217438731862871e-05,
      "loss": 0.6173,
      "step": 8865
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9234636985975533,
      "learning_rate": 4.2166050226779254e-05,
      "loss": 0.6437,
      "step": 8870
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0418391095017363,
      "learning_rate": 4.2157709521296415e-05,
      "loss": 0.6029,
      "step": 8875
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8635321632060303,
      "learning_rate": 4.2149365203936e-05,
      "loss": 0.6495,
      "step": 8880
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.804516687821039,
      "learning_rate": 4.2141017276454574e-05,
      "loss": 0.5862,
      "step": 8885
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9365539590127896,
      "learning_rate": 4.213266574060946e-05,
      "loss": 0.6732,
      "step": 8890
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9018465454943395,
      "learning_rate": 4.212431059815874e-05,
      "loss": 0.568,
      "step": 8895
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8982739941406391,
      "learning_rate": 4.2115951850861266e-05,
      "loss": 0.6291,
      "step": 8900
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.1873884308517382,
      "learning_rate": 4.210758950047664e-05,
      "loss": 0.6416,
      "step": 8905
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8430312157736384,
      "learning_rate": 4.2099223548765224e-05,
      "loss": 0.6515,
      "step": 8910
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0910697163500291,
      "learning_rate": 4.209085399748814e-05,
      "loss": 0.5151,
      "step": 8915
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8671703016367165,
      "learning_rate": 4.2082480848407256e-05,
      "loss": 0.5851,
      "step": 8920
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9542144961145261,
      "learning_rate": 4.207410410328523e-05,
      "loss": 0.5596,
      "step": 8925
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.1689561120146947,
      "learning_rate": 4.206572376388543e-05,
      "loss": 0.6648,
      "step": 8930
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8376895251130306,
      "learning_rate": 4.205733983197201e-05,
      "loss": 0.5707,
      "step": 8935
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7675738075828713,
      "learning_rate": 4.204895230930989e-05,
      "loss": 0.63,
      "step": 8940
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.937283923962815,
      "learning_rate": 4.204056119766473e-05,
      "loss": 0.6893,
      "step": 8945
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9141561528968974,
      "learning_rate": 4.2032166498802924e-05,
      "loss": 0.6336,
      "step": 8950
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9018641662825158,
      "learning_rate": 4.202376821449167e-05,
      "loss": 0.6107,
      "step": 8955
    },
    {
      "epoch": 1.05,
      "grad_norm": 9.240946444757233,
      "learning_rate": 4.2015366346498876e-05,
      "loss": 0.5445,
      "step": 8960
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8463706600609651,
      "learning_rate": 4.200696089659324e-05,
      "loss": 0.5881,
      "step": 8965
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8810863042510195,
      "learning_rate": 4.1998551866544176e-05,
      "loss": 0.5918,
      "step": 8970
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8878701771813601,
      "learning_rate": 4.1990139258121896e-05,
      "loss": 0.5974,
      "step": 8975
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.97833524704983,
      "learning_rate": 4.198172307309733e-05,
      "loss": 0.6416,
      "step": 8980
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9418290733556817,
      "learning_rate": 4.197330331324217e-05,
      "loss": 0.5702,
      "step": 8985
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8526099423601223,
      "learning_rate": 4.196487998032888e-05,
      "loss": 0.6007,
      "step": 8990
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8234540809248805,
      "learning_rate": 4.195645307613064e-05,
      "loss": 0.6498,
      "step": 8995
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8638309536728099,
      "learning_rate": 4.194802260242141e-05,
      "loss": 0.5696,
      "step": 9000
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8702117797112877,
      "learning_rate": 4.1939588560975895e-05,
      "loss": 0.5776,
      "step": 9005
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8827369710038147,
      "learning_rate": 4.1931150953569545e-05,
      "loss": 0.6699,
      "step": 9010
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8918531483354363,
      "learning_rate": 4.1922709781978576e-05,
      "loss": 0.6426,
      "step": 9015
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8137736136300915,
      "learning_rate": 4.191426504797992e-05,
      "loss": 0.5947,
      "step": 9020
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0294362932822392,
      "learning_rate": 4.1905816753351305e-05,
      "loss": 0.5357,
      "step": 9025
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8762244052287632,
      "learning_rate": 4.189736489987117e-05,
      "loss": 0.6312,
      "step": 9030
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8386457480979533,
      "learning_rate": 4.188890948931873e-05,
      "loss": 0.5845,
      "step": 9035
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8535662324621395,
      "learning_rate": 4.1880450523473934e-05,
      "loss": 0.6403,
      "step": 9040
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8490395229400508,
      "learning_rate": 4.187198800411748e-05,
      "loss": 0.6169,
      "step": 9045
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8176561930436175,
      "learning_rate": 4.1863521933030814e-05,
      "loss": 0.6224,
      "step": 9050
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8769277862052794,
      "learning_rate": 4.1855052311996143e-05,
      "loss": 0.6115,
      "step": 9055
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9966852282859848,
      "learning_rate": 4.18465791427964e-05,
      "loss": 0.6389,
      "step": 9060
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8727974424722675,
      "learning_rate": 4.1838102427215284e-05,
      "loss": 0.5951,
      "step": 9065
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9241233613092971,
      "learning_rate": 4.182962216703723e-05,
      "loss": 0.607,
      "step": 9070
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9418839716192525,
      "learning_rate": 4.182113836404742e-05,
      "loss": 0.6727,
      "step": 9075
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9842083814523354,
      "learning_rate": 4.181265102003178e-05,
      "loss": 0.6622,
      "step": 9080
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8108174508816137,
      "learning_rate": 4.1804160136776996e-05,
      "loss": 0.6423,
      "step": 9085
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8227960799903279,
      "learning_rate": 4.1795665716070474e-05,
      "loss": 0.636,
      "step": 9090
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8200827037169234,
      "learning_rate": 4.1787167759700396e-05,
      "loss": 0.5591,
      "step": 9095
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9142772491569335,
      "learning_rate": 4.177866626945565e-05,
      "loss": 0.5679,
      "step": 9100
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.944257216292202,
      "learning_rate": 4.17701612471259e-05,
      "loss": 0.6504,
      "step": 9105
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8774004676351174,
      "learning_rate": 4.176165269450155e-05,
      "loss": 0.6472,
      "step": 9110
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9121387145443576,
      "learning_rate": 4.175314061337372e-05,
      "loss": 0.5886,
      "step": 9115
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8289642292304378,
      "learning_rate": 4.174462500553431e-05,
      "loss": 0.5988,
      "step": 9120
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.1576133661813495,
      "learning_rate": 4.173610587277594e-05,
      "loss": 0.6673,
      "step": 9125
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8408511980336993,
      "learning_rate": 4.172758321689197e-05,
      "loss": 0.5441,
      "step": 9130
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8962694132061438,
      "learning_rate": 4.1719057039676515e-05,
      "loss": 0.6548,
      "step": 9135
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9271376576225812,
      "learning_rate": 4.171052734292442e-05,
      "loss": 0.5554,
      "step": 9140
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9028916892519055,
      "learning_rate": 4.1701994128431274e-05,
      "loss": 0.6887,
      "step": 9145
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8868096108743126,
      "learning_rate": 4.169345739799342e-05,
      "loss": 0.5993,
      "step": 9150
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0278611265858315,
      "learning_rate": 4.168491715340792e-05,
      "loss": 0.5831,
      "step": 9155
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.928107869559645,
      "learning_rate": 4.167637339647258e-05,
      "loss": 0.6587,
      "step": 9160
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8726913467580857,
      "learning_rate": 4.166782612898596e-05,
      "loss": 0.5525,
      "step": 9165
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.909095473410213,
      "learning_rate": 4.1659275352747346e-05,
      "loss": 0.6084,
      "step": 9170
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8915410194765684,
      "learning_rate": 4.165072106955676e-05,
      "loss": 0.5754,
      "step": 9175
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0573317389146333,
      "learning_rate": 4.1642163281214984e-05,
      "loss": 0.5563,
      "step": 9180
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9594329532730426,
      "learning_rate": 4.163360198952351e-05,
      "loss": 0.5826,
      "step": 9185
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.914275718877347,
      "learning_rate": 4.1625037196284575e-05,
      "loss": 0.6283,
      "step": 9190
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9293038106149837,
      "learning_rate": 4.1616468903301166e-05,
      "loss": 0.5936,
      "step": 9195
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9802699630248849,
      "learning_rate": 4.1607897112377e-05,
      "loss": 0.647,
      "step": 9200
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9616383156693995,
      "learning_rate": 4.159932182531653e-05,
      "loss": 0.6315,
      "step": 9205
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8562848646733909,
      "learning_rate": 4.159074304392493e-05,
      "loss": 0.6008,
      "step": 9210
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8766904592822528,
      "learning_rate": 4.1582160770008135e-05,
      "loss": 0.6655,
      "step": 9215
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8803698710118834,
      "learning_rate": 4.15735750053728e-05,
      "loss": 0.6639,
      "step": 9220
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8872627668714872,
      "learning_rate": 4.156498575182633e-05,
      "loss": 0.6478,
      "step": 9225
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9215102380152673,
      "learning_rate": 4.155639301117683e-05,
      "loss": 0.6187,
      "step": 9230
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8941727018771481,
      "learning_rate": 4.1547796785233184e-05,
      "loss": 0.6569,
      "step": 9235
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.8455878025031203,
      "learning_rate": 4.153919707580497e-05,
      "loss": 0.5984,
      "step": 9240
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3255126938280823,
      "learning_rate": 4.153059388470253e-05,
      "loss": 0.601,
      "step": 9245
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9585304710125648,
      "learning_rate": 4.152198721373691e-05,
      "loss": 0.5188,
      "step": 9250
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.928525307708516,
      "learning_rate": 4.151337706471993e-05,
      "loss": 0.6191,
      "step": 9255
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0062232763206145,
      "learning_rate": 4.1504763439464086e-05,
      "loss": 0.572,
      "step": 9260
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0251831560539522,
      "learning_rate": 4.149614633978266e-05,
      "loss": 0.5353,
      "step": 9265
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0153972607329083,
      "learning_rate": 4.1487525767489635e-05,
      "loss": 0.5528,
      "step": 9270
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0575186158980403,
      "learning_rate": 4.147890172439973e-05,
      "loss": 0.6301,
      "step": 9275
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9252518054199487,
      "learning_rate": 4.1470274212328396e-05,
      "loss": 0.6703,
      "step": 9280
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9186694597928409,
      "learning_rate": 4.146164323309182e-05,
      "loss": 0.6048,
      "step": 9285
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9385692278078165,
      "learning_rate": 4.14530087885069e-05,
      "loss": 0.6208,
      "step": 9290
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8684047001595284,
      "learning_rate": 4.144437088039129e-05,
      "loss": 0.5817,
      "step": 9295
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9773036752922679,
      "learning_rate": 4.143572951056336e-05,
      "loss": 0.6249,
      "step": 9300
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8527892871727347,
      "learning_rate": 4.14270846808422e-05,
      "loss": 0.5567,
      "step": 9305
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0247760836400146,
      "learning_rate": 4.141843639304765e-05,
      "loss": 0.6064,
      "step": 9310
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8344753696442245,
      "learning_rate": 4.140978464900025e-05,
      "loss": 0.5822,
      "step": 9315
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9618762819662215,
      "learning_rate": 4.140112945052129e-05,
      "loss": 0.5971,
      "step": 9320
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9571653438072899,
      "learning_rate": 4.1392470799432784e-05,
      "loss": 0.5972,
      "step": 9325
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8194742217294482,
      "learning_rate": 4.138380869755747e-05,
      "loss": 0.6727,
      "step": 9330
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8311485080490008,
      "learning_rate": 4.137514314671879e-05,
      "loss": 0.6181,
      "step": 9335
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9556187434309861,
      "learning_rate": 4.1366474148740964e-05,
      "loss": 0.61,
      "step": 9340
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8833476520157852,
      "learning_rate": 4.1357801705448885e-05,
      "loss": 0.8532,
      "step": 9345
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9063824733888173,
      "learning_rate": 4.134912581866821e-05,
      "loss": 0.6533,
      "step": 9350
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8287359909143839,
      "learning_rate": 4.134044649022528e-05,
      "loss": 0.5994,
      "step": 9355
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.022126103378075,
      "learning_rate": 4.13317637219472e-05,
      "loss": 0.5835,
      "step": 9360
    },
    {
      "epoch": 1.09,
      "grad_norm": 4.530502518214563,
      "learning_rate": 4.1323077515661794e-05,
      "loss": 0.6326,
      "step": 9365
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8019002699371386,
      "learning_rate": 4.131438787319759e-05,
      "loss": 0.6269,
      "step": 9370
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9288439870395792,
      "learning_rate": 4.130569479638383e-05,
      "loss": 0.6165,
      "step": 9375
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8976834896314357,
      "learning_rate": 4.1296998287050524e-05,
      "loss": 0.6369,
      "step": 9380
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8527835679808553,
      "learning_rate": 4.128829834702838e-05,
      "loss": 0.6508,
      "step": 9385
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8422118491219953,
      "learning_rate": 4.1279594978148796e-05,
      "loss": 0.5052,
      "step": 9390
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9007388888373373,
      "learning_rate": 4.127088818224395e-05,
      "loss": 0.6108,
      "step": 9395
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0024797564172927,
      "learning_rate": 4.1262177961146706e-05,
      "loss": 0.6461,
      "step": 9400
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.1111025670090133,
      "learning_rate": 4.125346431669065e-05,
      "loss": 0.6582,
      "step": 9405
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.829441850559287,
      "learning_rate": 4.1244747250710114e-05,
      "loss": 0.6998,
      "step": 9410
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9620225771555422,
      "learning_rate": 4.123602676504011e-05,
      "loss": 0.6588,
      "step": 9415
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.758821533454097,
      "learning_rate": 4.12273028615164e-05,
      "loss": 0.6214,
      "step": 9420
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9005743815041396,
      "learning_rate": 4.1218575541975465e-05,
      "loss": 0.6235,
      "step": 9425
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9638660573830756,
      "learning_rate": 4.1209844808254485e-05,
      "loss": 0.6536,
      "step": 9430
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9096082315522549,
      "learning_rate": 4.120111066219138e-05,
      "loss": 0.6032,
      "step": 9435
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9228745839604217,
      "learning_rate": 4.119237310562477e-05,
      "loss": 0.5797,
      "step": 9440
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8215471996763309,
      "learning_rate": 4.1183632140394014e-05,
      "loss": 0.5996,
      "step": 9445
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0194203421439125,
      "learning_rate": 4.1174887768339164e-05,
      "loss": 0.621,
      "step": 9450
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8182899096337711,
      "learning_rate": 4.1166139991301014e-05,
      "loss": 0.6232,
      "step": 9455
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8334896884227421,
      "learning_rate": 4.115738881112105e-05,
      "loss": 0.5143,
      "step": 9460
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.983983227144839,
      "learning_rate": 4.1148634229641495e-05,
      "loss": 0.6021,
      "step": 9465
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8348851332798832,
      "learning_rate": 4.113987624870528e-05,
      "loss": 0.6391,
      "step": 9470
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9105741906663594,
      "learning_rate": 4.1131114870156054e-05,
      "loss": 0.6757,
      "step": 9475
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8280699053178365,
      "learning_rate": 4.1122350095838174e-05,
      "loss": 0.5671,
      "step": 9480
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.857219718278141,
      "learning_rate": 4.1113581927596724e-05,
      "loss": 0.6313,
      "step": 9485
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9016560303307284,
      "learning_rate": 4.110481036727748e-05,
      "loss": 0.5797,
      "step": 9490
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8285068014702185,
      "learning_rate": 4.1096035416726966e-05,
      "loss": 0.606,
      "step": 9495
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8810691364569447,
      "learning_rate": 4.108725707779238e-05,
      "loss": 0.6305,
      "step": 9500
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8538779516110756,
      "learning_rate": 4.1078475352321675e-05,
      "loss": 0.6377,
      "step": 9505
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.932471400423201,
      "learning_rate": 4.1069690242163484e-05,
      "loss": 0.5733,
      "step": 9510
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9081992259361201,
      "learning_rate": 4.1060901749167175e-05,
      "loss": 0.6013,
      "step": 9515
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8750173420278271,
      "learning_rate": 4.1052109875182804e-05,
      "loss": 0.6356,
      "step": 9520
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2485698544680992,
      "learning_rate": 4.1043314622061165e-05,
      "loss": 0.6439,
      "step": 9525
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9083552100444398,
      "learning_rate": 4.103451599165374e-05,
      "loss": 0.6134,
      "step": 9530
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9421113651564124,
      "learning_rate": 4.102571398581274e-05,
      "loss": 0.6521,
      "step": 9535
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2557029200993437,
      "learning_rate": 4.101690860639108e-05,
      "loss": 0.5907,
      "step": 9540
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9665697226056439,
      "learning_rate": 4.100809985524238e-05,
      "loss": 0.6242,
      "step": 9545
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9617080770593286,
      "learning_rate": 4.099928773422098e-05,
      "loss": 0.562,
      "step": 9550
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9148415802633895,
      "learning_rate": 4.099047224518192e-05,
      "loss": 0.6158,
      "step": 9555
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0141716049378553,
      "learning_rate": 4.0981653389980954e-05,
      "loss": 0.653,
      "step": 9560
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9622663943050648,
      "learning_rate": 4.097283117047455e-05,
      "loss": 0.588,
      "step": 9565
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9922851758966393,
      "learning_rate": 4.096400558851986e-05,
      "loss": 0.5943,
      "step": 9570
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0328846038388495,
      "learning_rate": 4.0955176645974774e-05,
      "loss": 0.593,
      "step": 9575
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8832446055522517,
      "learning_rate": 4.0946344344697874e-05,
      "loss": 0.5124,
      "step": 9580
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8088913061832707,
      "learning_rate": 4.0937508686548455e-05,
      "loss": 0.6461,
      "step": 9585
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9261208965877685,
      "learning_rate": 4.092866967338651e-05,
      "loss": 0.5735,
      "step": 9590
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0427701686523054,
      "learning_rate": 4.091982730707275e-05,
      "loss": 0.5435,
      "step": 9595
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7612729987856766,
      "learning_rate": 4.091098158946859e-05,
      "loss": 0.5958,
      "step": 9600
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8885194063013234,
      "learning_rate": 4.090213252243612e-05,
      "loss": 0.6246,
      "step": 9605
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8562718735726935,
      "learning_rate": 4.08932801078382e-05,
      "loss": 0.6134,
      "step": 9610
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.839932526258277,
      "learning_rate": 4.088442434753833e-05,
      "loss": 0.5738,
      "step": 9615
    },
    {
      "epoch": 1.12,
      "grad_norm": 13.248673901344128,
      "learning_rate": 4.087556524340075e-05,
      "loss": 0.6389,
      "step": 9620
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9464387608328463,
      "learning_rate": 4.086670279729038e-05,
      "loss": 0.5768,
      "step": 9625
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7600374099784892,
      "learning_rate": 4.085783701107288e-05,
      "loss": 0.6513,
      "step": 9630
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.068162854657603,
      "learning_rate": 4.084896788661458e-05,
      "loss": 0.623,
      "step": 9635
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8999060240232348,
      "learning_rate": 4.084009542578251e-05,
      "loss": 0.5755,
      "step": 9640
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0708600562994768,
      "learning_rate": 4.083121963044444e-05,
      "loss": 0.6702,
      "step": 9645
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7671671241956116,
      "learning_rate": 4.08223405024688e-05,
      "loss": 0.5828,
      "step": 9650
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9056359957694474,
      "learning_rate": 4.081345804372475e-05,
      "loss": 0.589,
      "step": 9655
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9139667473825068,
      "learning_rate": 4.080457225608214e-05,
      "loss": 0.5492,
      "step": 9660
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9227650272336652,
      "learning_rate": 4.079568314141152e-05,
      "loss": 0.5915,
      "step": 9665
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9151880775080945,
      "learning_rate": 4.078679070158413e-05,
      "loss": 0.5721,
      "step": 9670
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8697274735123237,
      "learning_rate": 4.077789493847194e-05,
      "loss": 0.6507,
      "step": 9675
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9371052089851482,
      "learning_rate": 4.07689958539476e-05,
      "loss": 0.678,
      "step": 9680
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.044862678211282,
      "learning_rate": 4.076009344988444e-05,
      "loss": 0.6084,
      "step": 9685
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8312392235480726,
      "learning_rate": 4.075118772815654e-05,
      "loss": 0.6451,
      "step": 9690
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9860466179648407,
      "learning_rate": 4.074227869063861e-05,
      "loss": 0.5837,
      "step": 9695
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7249402738180497,
      "learning_rate": 4.073336633920613e-05,
      "loss": 0.6021,
      "step": 9700
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8472237106522102,
      "learning_rate": 4.072445067573523e-05,
      "loss": 0.5836,
      "step": 9705
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8751154208796319,
      "learning_rate": 4.0715531702102735e-05,
      "loss": 0.6976,
      "step": 9710
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9176567732963984,
      "learning_rate": 4.070660942018622e-05,
      "loss": 0.6185,
      "step": 9715
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8201043298299713,
      "learning_rate": 4.0697683831863877e-05,
      "loss": 0.589,
      "step": 9720
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8525492515407452,
      "learning_rate": 4.068875493901467e-05,
      "loss": 0.5547,
      "step": 9725
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9868701328291947,
      "learning_rate": 4.0679822743518206e-05,
      "loss": 0.5843,
      "step": 9730
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.926300738253284,
      "learning_rate": 4.06708872472548e-05,
      "loss": 0.5719,
      "step": 9735
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8928032034763357,
      "learning_rate": 4.066194845210549e-05,
      "loss": 0.5763,
      "step": 9740
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9702162928445587,
      "learning_rate": 4.065300635995197e-05,
      "loss": 0.5954,
      "step": 9745
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.081141134487754,
      "learning_rate": 4.064406097267665e-05,
      "loss": 0.5845,
      "step": 9750
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.904794270196421,
      "learning_rate": 4.0635112292162615e-05,
      "loss": 0.5764,
      "step": 9755
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9085037644608931,
      "learning_rate": 4.062616032029367e-05,
      "loss": 0.6563,
      "step": 9760
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8150858058827991,
      "learning_rate": 4.06172050589543e-05,
      "loss": 0.671,
      "step": 9765
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8996870804957054,
      "learning_rate": 4.060824651002967e-05,
      "loss": 0.6075,
      "step": 9770
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8213482296231155,
      "learning_rate": 4.059928467540566e-05,
      "loss": 0.5613,
      "step": 9775
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9006394954800155,
      "learning_rate": 4.059031955696882e-05,
      "loss": 0.6062,
      "step": 9780
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8907361452518608,
      "learning_rate": 4.058135115660641e-05,
      "loss": 0.5952,
      "step": 9785
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.849663574160607,
      "learning_rate": 4.057237947620636e-05,
      "loss": 0.6055,
      "step": 9790
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8471132332767213,
      "learning_rate": 4.0563404517657324e-05,
      "loss": 0.6151,
      "step": 9795
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9101727179278354,
      "learning_rate": 4.05544262828486e-05,
      "loss": 0.6292,
      "step": 9800
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7981621928808994,
      "learning_rate": 4.054544477367022e-05,
      "loss": 0.5964,
      "step": 9805
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9594345392819724,
      "learning_rate": 4.053645999201287e-05,
      "loss": 0.6154,
      "step": 9810
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.189914031503265,
      "learning_rate": 4.052747193976795e-05,
      "loss": 0.5879,
      "step": 9815
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0107693337040382,
      "learning_rate": 4.051848061882755e-05,
      "loss": 0.6197,
      "step": 9820
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8918712702288432,
      "learning_rate": 4.050948603108441e-05,
      "loss": 0.611,
      "step": 9825
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9043076750860054,
      "learning_rate": 4.050048817843201e-05,
      "loss": 0.572,
      "step": 9830
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8321686514270472,
      "learning_rate": 4.049148706276447e-05,
      "loss": 0.639,
      "step": 9835
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9551571579166243,
      "learning_rate": 4.048248268597664e-05,
      "loss": 0.5846,
      "step": 9840
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8420750522382193,
      "learning_rate": 4.047347504996403e-05,
      "loss": 0.6793,
      "step": 9845
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8413534977306244,
      "learning_rate": 4.046446415662283e-05,
      "loss": 0.5947,
      "step": 9850
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9801330995809016,
      "learning_rate": 4.0455450007849945e-05,
      "loss": 0.6505,
      "step": 9855
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.047270367782501,
      "learning_rate": 4.0446432605542926e-05,
      "loss": 0.6357,
      "step": 9860
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9162854258787997,
      "learning_rate": 4.0437411951600056e-05,
      "loss": 0.6805,
      "step": 9865
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8031869014755787,
      "learning_rate": 4.042838804792026e-05,
      "loss": 0.6598,
      "step": 9870
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9469068665864222,
      "learning_rate": 4.041936089640317e-05,
      "loss": 0.6192,
      "step": 9875
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9110327010793229,
      "learning_rate": 4.041033049894909e-05,
      "loss": 0.5754,
      "step": 9880
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8798608223896566,
      "learning_rate": 4.0401296857459026e-05,
      "loss": 0.6593,
      "step": 9885
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8952185728910895,
      "learning_rate": 4.039225997383464e-05,
      "loss": 0.6579,
      "step": 9890
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8969629395565243,
      "learning_rate": 4.038321984997829e-05,
      "loss": 0.6985,
      "step": 9895
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8302735968430788,
      "learning_rate": 4.037417648779304e-05,
      "loss": 0.5187,
      "step": 9900
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1303999185258016,
      "learning_rate": 4.0365129889182584e-05,
      "loss": 0.6846,
      "step": 9905
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9772078663141259,
      "learning_rate": 4.0356080056051345e-05,
      "loss": 0.6823,
      "step": 9910
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9535922750694439,
      "learning_rate": 4.034702699030439e-05,
      "loss": 0.6919,
      "step": 9915
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9398525330317363,
      "learning_rate": 4.03379706938475e-05,
      "loss": 0.5385,
      "step": 9920
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9215923704026867,
      "learning_rate": 4.0328911168587125e-05,
      "loss": 0.6222,
      "step": 9925
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8069367191223502,
      "learning_rate": 4.031984841643037e-05,
      "loss": 0.6438,
      "step": 9930
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.903549102288383,
      "learning_rate": 4.0310782439285054e-05,
      "loss": 0.6276,
      "step": 9935
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8979396615731928,
      "learning_rate": 4.030171323905965e-05,
      "loss": 0.5848,
      "step": 9940
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8753038102570072,
      "learning_rate": 4.029264081766333e-05,
      "loss": 0.5724,
      "step": 9945
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0475434911700867,
      "learning_rate": 4.0283565177005924e-05,
      "loss": 0.6489,
      "step": 9950
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.848606080971198,
      "learning_rate": 4.027448631899796e-05,
      "loss": 0.6598,
      "step": 9955
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8882713828216475,
      "learning_rate": 4.026540424555062e-05,
      "loss": 0.5622,
      "step": 9960
    },
    {
      "epoch": 1.16,
      "grad_norm": 2.2082566947531976,
      "learning_rate": 4.025631895857579e-05,
      "loss": 0.6491,
      "step": 9965
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0495436080584797,
      "learning_rate": 4.0247230459986004e-05,
      "loss": 0.6334,
      "step": 9970
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9350201595609539,
      "learning_rate": 4.02381387516945e-05,
      "loss": 0.6555,
      "step": 9975
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8879965573607457,
      "learning_rate": 4.0229043835615155e-05,
      "loss": 0.588,
      "step": 9980
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8918351647989582,
      "learning_rate": 4.0219945713662576e-05,
      "loss": 0.613,
      "step": 9985
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7840567649434786,
      "learning_rate": 4.021084438775199e-05,
      "loss": 0.6332,
      "step": 9990
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0335261984915007,
      "learning_rate": 4.0201739859799325e-05,
      "loss": 0.5649,
      "step": 9995
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8861762512251794,
      "learning_rate": 4.0192632131721185e-05,
      "loss": 0.5838,
      "step": 10000
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7712705175084428,
      "learning_rate": 4.018352120543485e-05,
      "loss": 0.5988,
      "step": 10005
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0029455847776978,
      "learning_rate": 4.017440708285824e-05,
      "loss": 0.5617,
      "step": 10010
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0963691107536784,
      "learning_rate": 4.0165289765909996e-05,
      "loss": 0.6162,
      "step": 10015
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9950886489154548,
      "learning_rate": 4.01561692565094e-05,
      "loss": 0.6229,
      "step": 10020
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9440204288452254,
      "learning_rate": 4.0147045556576415e-05,
      "loss": 0.6334,
      "step": 10025
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9964489671959494,
      "learning_rate": 4.013791866803167e-05,
      "loss": 0.6313,
      "step": 10030
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9249501924597987,
      "learning_rate": 4.0128788592796484e-05,
      "loss": 0.6486,
      "step": 10035
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9213352087336827,
      "learning_rate": 4.0119655332792826e-05,
      "loss": 0.6733,
      "step": 10040
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9568225596954095,
      "learning_rate": 4.011051888994335e-05,
      "loss": 0.6222,
      "step": 10045
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8569036072099062,
      "learning_rate": 4.010137926617136e-05,
      "loss": 0.6248,
      "step": 10050
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9383601287942281,
      "learning_rate": 4.009223646340084e-05,
      "loss": 0.6061,
      "step": 10055
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.903270509340819,
      "learning_rate": 4.008309048355646e-05,
      "loss": 0.6312,
      "step": 10060
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9724861070811751,
      "learning_rate": 4.0073941328563546e-05,
      "loss": 0.6176,
      "step": 10065
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9085347640631056,
      "learning_rate": 4.006478900034808e-05,
      "loss": 0.6114,
      "step": 10070
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8590173109493656,
      "learning_rate": 4.005563350083673e-05,
      "loss": 0.5889,
      "step": 10075
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8568817119257549,
      "learning_rate": 4.004647483195682e-05,
      "loss": 0.6356,
      "step": 10080
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8751475523043187,
      "learning_rate": 4.0037312995636336e-05,
      "loss": 0.6386,
      "step": 10085
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8681531696645705,
      "learning_rate": 4.002814799380397e-05,
      "loss": 0.612,
      "step": 10090
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0569052005853135,
      "learning_rate": 4.001897982838902e-05,
      "loss": 0.6351,
      "step": 10095
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9015183523209024,
      "learning_rate": 4.00098085013215e-05,
      "loss": 0.596,
      "step": 10100
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8442722392585974,
      "learning_rate": 4.0000634014532054e-05,
      "loss": 0.5995,
      "step": 10105
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8850762982309546,
      "learning_rate": 3.999145636995203e-05,
      "loss": 0.6084,
      "step": 10110
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0571424102878644,
      "learning_rate": 3.998227556951341e-05,
      "loss": 0.6794,
      "step": 10115
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8628784934756678,
      "learning_rate": 3.997309161514883e-05,
      "loss": 0.6222,
      "step": 10120
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9588455777465492,
      "learning_rate": 3.996390450879163e-05,
      "loss": 0.5995,
      "step": 10125
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8900907773977574,
      "learning_rate": 3.995471425237578e-05,
      "loss": 0.5684,
      "step": 10130
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8591476511074881,
      "learning_rate": 3.9945520847835935e-05,
      "loss": 0.576,
      "step": 10135
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7951726521016427,
      "learning_rate": 3.9936324297107396e-05,
      "loss": 0.5558,
      "step": 10140
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0588470665638625,
      "learning_rate": 3.9927124602126135e-05,
      "loss": 0.5434,
      "step": 10145
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2723282099950253,
      "learning_rate": 3.991792176482879e-05,
      "loss": 0.582,
      "step": 10150
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.075406036860975,
      "learning_rate": 3.990871578715264e-05,
      "loss": 0.662,
      "step": 10155
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8328623678809383,
      "learning_rate": 3.989950667103566e-05,
      "loss": 0.6482,
      "step": 10160
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.847580232745473,
      "learning_rate": 3.989029441841645e-05,
      "loss": 0.5367,
      "step": 10165
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8589010882682073,
      "learning_rate": 3.9881079031234295e-05,
      "loss": 0.5862,
      "step": 10170
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8610734142118764,
      "learning_rate": 3.987186051142911e-05,
      "loss": 0.6158,
      "step": 10175
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.86627951786458,
      "learning_rate": 3.9862638860941524e-05,
      "loss": 0.6112,
      "step": 10180
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0667526656804673,
      "learning_rate": 3.985341408171277e-05,
      "loss": 0.6354,
      "step": 10185
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8711287886263307,
      "learning_rate": 3.984418617568475e-05,
      "loss": 0.6355,
      "step": 10190
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9881251887237487,
      "learning_rate": 3.983495514480007e-05,
      "loss": 0.622,
      "step": 10195
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8792489533541943,
      "learning_rate": 3.982572099100192e-05,
      "loss": 0.5141,
      "step": 10200
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9152498687376003,
      "learning_rate": 3.981648371623421e-05,
      "loss": 0.6207,
      "step": 10205
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7989876504398636,
      "learning_rate": 3.980724332244147e-05,
      "loss": 0.5269,
      "step": 10210
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8797185298604385,
      "learning_rate": 3.9797999811568916e-05,
      "loss": 0.5778,
      "step": 10215
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9223928804353906,
      "learning_rate": 3.978875318556238e-05,
      "loss": 0.6445,
      "step": 10220
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9364308011543939,
      "learning_rate": 3.97795034463684e-05,
      "loss": 0.6594,
      "step": 10225
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9220385812390476,
      "learning_rate": 3.977025059593413e-05,
      "loss": 0.6346,
      "step": 10230
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9701765029649853,
      "learning_rate": 3.976099463620738e-05,
      "loss": 0.6367,
      "step": 10235
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8867570256512409,
      "learning_rate": 3.975173556913665e-05,
      "loss": 0.6055,
      "step": 10240
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8125509656428566,
      "learning_rate": 3.974247339667106e-05,
      "loss": 0.6225,
      "step": 10245
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0892926497453927,
      "learning_rate": 3.973320812076039e-05,
      "loss": 0.5704,
      "step": 10250
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.026750775890507,
      "learning_rate": 3.972393974335508e-05,
      "loss": 0.6553,
      "step": 10255
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1427438785669062,
      "learning_rate": 3.971466826640622e-05,
      "loss": 0.6834,
      "step": 10260
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8839448364089717,
      "learning_rate": 3.9705393691865566e-05,
      "loss": 0.5522,
      "step": 10265
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8791945443161713,
      "learning_rate": 3.969611602168549e-05,
      "loss": 0.5669,
      "step": 10270
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9515027738904531,
      "learning_rate": 3.968683525781906e-05,
      "loss": 0.6297,
      "step": 10275
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4876496267656212,
      "learning_rate": 3.967755140221996e-05,
      "loss": 0.589,
      "step": 10280
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1247622916072126,
      "learning_rate": 3.966826445684255e-05,
      "loss": 0.633,
      "step": 10285
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8302846402144587,
      "learning_rate": 3.965897442364182e-05,
      "loss": 0.6365,
      "step": 10290
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9908935689677275,
      "learning_rate": 3.964968130457341e-05,
      "loss": 0.6323,
      "step": 10295
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9340391113359591,
      "learning_rate": 3.964038510159364e-05,
      "loss": 0.6151,
      "step": 10300
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.2718292438791907,
      "learning_rate": 3.963108581665945e-05,
      "loss": 0.6114,
      "step": 10305
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7785590263663952,
      "learning_rate": 3.962178345172844e-05,
      "loss": 0.5637,
      "step": 10310
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9645013541347448,
      "learning_rate": 3.961247800875885e-05,
      "loss": 0.5801,
      "step": 10315
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8583297905620315,
      "learning_rate": 3.9603169489709567e-05,
      "loss": 0.6214,
      "step": 10320
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9282947415255546,
      "learning_rate": 3.959385789654014e-05,
      "loss": 0.6309,
      "step": 10325
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9368910532197497,
      "learning_rate": 3.9584543231210766e-05,
      "loss": 0.582,
      "step": 10330
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8091883054913099,
      "learning_rate": 3.9575225495682266e-05,
      "loss": 0.6053,
      "step": 10335
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9250748071469435,
      "learning_rate": 3.956590469191612e-05,
      "loss": 0.6578,
      "step": 10340
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8699557441268959,
      "learning_rate": 3.955658082187447e-05,
      "loss": 0.6516,
      "step": 10345
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8144798873280884,
      "learning_rate": 3.954725388752006e-05,
      "loss": 0.5899,
      "step": 10350
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8413361061870538,
      "learning_rate": 3.953792389081635e-05,
      "loss": 0.5566,
      "step": 10355
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.002226599985344,
      "learning_rate": 3.9528590833727366e-05,
      "loss": 0.5435,
      "step": 10360
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1095065513935778,
      "learning_rate": 3.951925471821781e-05,
      "loss": 0.5638,
      "step": 10365
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9947145888090904,
      "learning_rate": 3.950991554625307e-05,
      "loss": 0.59,
      "step": 10370
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1471816219045725,
      "learning_rate": 3.9500573319799104e-05,
      "loss": 0.6571,
      "step": 10375
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2056447335181821,
      "learning_rate": 3.949122804082257e-05,
      "loss": 0.6385,
      "step": 10380
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9236499290260043,
      "learning_rate": 3.9481879711290735e-05,
      "loss": 0.5761,
      "step": 10385
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9871575598379568,
      "learning_rate": 3.9472528333171535e-05,
      "loss": 0.652,
      "step": 10390
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.788541974001009,
      "learning_rate": 3.94631739084335e-05,
      "loss": 0.5995,
      "step": 10395
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1523334655894832,
      "learning_rate": 3.9453816439045876e-05,
      "loss": 0.5945,
      "step": 10400
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.063804037800662,
      "learning_rate": 3.944445592697848e-05,
      "loss": 0.6398,
      "step": 10405
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9686012869197632,
      "learning_rate": 3.943509237420181e-05,
      "loss": 0.6056,
      "step": 10410
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9297495694452663,
      "learning_rate": 3.9425725782686984e-05,
      "loss": 0.6175,
      "step": 10415
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1319803102042276,
      "learning_rate": 3.9416356154405775e-05,
      "loss": 0.6641,
      "step": 10420
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9009336255386489,
      "learning_rate": 3.940698349133059e-05,
      "loss": 0.666,
      "step": 10425
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8854415142953846,
      "learning_rate": 3.939760779543447e-05,
      "loss": 0.6813,
      "step": 10430
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0041367685322249,
      "learning_rate": 3.938822906869108e-05,
      "loss": 0.6322,
      "step": 10435
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1876986370837992,
      "learning_rate": 3.937884731307477e-05,
      "loss": 0.6451,
      "step": 10440
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8686221873500196,
      "learning_rate": 3.936946253056047e-05,
      "loss": 0.571,
      "step": 10445
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8786593514516138,
      "learning_rate": 3.936007472312381e-05,
      "loss": 0.6693,
      "step": 10450
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8636998162821604,
      "learning_rate": 3.935068389274098e-05,
      "loss": 0.5903,
      "step": 10455
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9589885303494541,
      "learning_rate": 3.9341290041388884e-05,
      "loss": 0.5917,
      "step": 10460
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0097558423849307,
      "learning_rate": 3.9331893171045e-05,
      "loss": 0.5752,
      "step": 10465
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9040003496213147,
      "learning_rate": 3.932249328368748e-05,
      "loss": 0.6171,
      "step": 10470
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9117511057238641,
      "learning_rate": 3.93130903812951e-05,
      "loss": 0.5481,
      "step": 10475
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9177250458951397,
      "learning_rate": 3.9303684465847254e-05,
      "loss": 0.6379,
      "step": 10480
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8738506459598745,
      "learning_rate": 3.929427553932402e-05,
      "loss": 0.6947,
      "step": 10485
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.2560711170573873,
      "learning_rate": 3.9284863603706036e-05,
      "loss": 0.6615,
      "step": 10490
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9556202554831419,
      "learning_rate": 3.927544866097464e-05,
      "loss": 0.6716,
      "step": 10495
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0206437800131474,
      "learning_rate": 3.9266030713111755e-05,
      "loss": 0.5995,
      "step": 10500
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9605446928738075,
      "learning_rate": 3.925660976209997e-05,
      "loss": 0.6041,
      "step": 10505
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9548800749168715,
      "learning_rate": 3.924718580992251e-05,
      "loss": 0.5636,
      "step": 10510
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9932209464059008,
      "learning_rate": 3.923775885856319e-05,
      "loss": 0.6875,
      "step": 10515
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9048080570269783,
      "learning_rate": 3.922832891000649e-05,
      "loss": 0.6603,
      "step": 10520
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0231057550763838,
      "learning_rate": 3.921889596623751e-05,
      "loss": 0.5725,
      "step": 10525
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9322082392496664,
      "learning_rate": 3.9209460029242e-05,
      "loss": 0.5851,
      "step": 10530
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9579507146406717,
      "learning_rate": 3.920002110100629e-05,
      "loss": 0.7321,
      "step": 10535
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8083429849855337,
      "learning_rate": 3.9190579183517406e-05,
      "loss": 0.636,
      "step": 10540
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8156973969412518,
      "learning_rate": 3.9181134278762964e-05,
      "loss": 0.5845,
      "step": 10545
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.879778222064731,
      "learning_rate": 3.91716863887312e-05,
      "loss": 0.6019,
      "step": 10550
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8923596962786368,
      "learning_rate": 3.916223551541101e-05,
      "loss": 0.5634,
      "step": 10555
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9155326327793609,
      "learning_rate": 3.915278166079189e-05,
      "loss": 0.6537,
      "step": 10560
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0492039727253646,
      "learning_rate": 3.914332482686399e-05,
      "loss": 0.6017,
      "step": 10565
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8001464314523077,
      "learning_rate": 3.913386501561806e-05,
      "loss": 0.6189,
      "step": 10570
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8678962272468027,
      "learning_rate": 3.9124402229045495e-05,
      "loss": 0.5916,
      "step": 10575
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7826536308145431,
      "learning_rate": 3.911493646913831e-05,
      "loss": 0.5379,
      "step": 10580
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9353426230920516,
      "learning_rate": 3.910546773788914e-05,
      "loss": 0.673,
      "step": 10585
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7802279526029655,
      "learning_rate": 3.909599603729126e-05,
      "loss": 0.5673,
      "step": 10590
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.803304245282809,
      "learning_rate": 3.9086521369338556e-05,
      "loss": 0.5585,
      "step": 10595
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8199001979000007,
      "learning_rate": 3.9077043736025556e-05,
      "loss": 0.539,
      "step": 10600
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7984079430611349,
      "learning_rate": 3.9067563139347386e-05,
      "loss": 0.6036,
      "step": 10605
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8371379939357879,
      "learning_rate": 3.905807958129982e-05,
      "loss": 0.6438,
      "step": 10610
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0017476169312525,
      "learning_rate": 3.9048593063879246e-05,
      "loss": 0.6572,
      "step": 10615
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8923742455765313,
      "learning_rate": 3.903910358908267e-05,
      "loss": 0.583,
      "step": 10620
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.827146707829598,
      "learning_rate": 3.9029611158907745e-05,
      "loss": 0.6236,
      "step": 10625
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8940512986771159,
      "learning_rate": 3.902011577535269e-05,
      "loss": 0.638,
      "step": 10630
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8542815063650978,
      "learning_rate": 3.901061744041643e-05,
      "loss": 0.6362,
      "step": 10635
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7677566757686113,
      "learning_rate": 3.900111615609842e-05,
      "loss": 0.6269,
      "step": 10640
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8962289926396972,
      "learning_rate": 3.899161192439881e-05,
      "loss": 0.5695,
      "step": 10645
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8544061172511264,
      "learning_rate": 3.8982104747318316e-05,
      "loss": 0.6068,
      "step": 10650
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5333065249275686,
      "learning_rate": 3.897259462685832e-05,
      "loss": 0.5868,
      "step": 10655
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8123447402985269,
      "learning_rate": 3.896308156502079e-05,
      "loss": 0.5808,
      "step": 10660
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7606482859001378,
      "learning_rate": 3.895356556380833e-05,
      "loss": 0.6026,
      "step": 10665
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8720049492312372,
      "learning_rate": 3.894404662522416e-05,
      "loss": 0.6163,
      "step": 10670
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0972196656470814,
      "learning_rate": 3.893452475127211e-05,
      "loss": 0.6978,
      "step": 10675
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8357015413947352,
      "learning_rate": 3.892499994395663e-05,
      "loss": 0.6325,
      "step": 10680
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8376635428915669,
      "learning_rate": 3.891547220528281e-05,
      "loss": 0.5576,
      "step": 10685
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9301395880749401,
      "learning_rate": 3.890594153725632e-05,
      "loss": 0.6492,
      "step": 10690
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8991462800998806,
      "learning_rate": 3.8896407941883475e-05,
      "loss": 0.5933,
      "step": 10695
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.135268620694561,
      "learning_rate": 3.888687142117119e-05,
      "loss": 0.6041,
      "step": 10700
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7950192411597671,
      "learning_rate": 3.8877331977127016e-05,
      "loss": 0.549,
      "step": 10705
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9886058872581209,
      "learning_rate": 3.886778961175909e-05,
      "loss": 0.6432,
      "step": 10710
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9020636491058205,
      "learning_rate": 3.885824432707619e-05,
      "loss": 0.5564,
      "step": 10715
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9417537441195247,
      "learning_rate": 3.884869612508769e-05,
      "loss": 0.594,
      "step": 10720
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8395106968537629,
      "learning_rate": 3.8839145007803595e-05,
      "loss": 0.5911,
      "step": 10725
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9534682639358851,
      "learning_rate": 3.882959097723452e-05,
      "loss": 0.6634,
      "step": 10730
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8537875219359679,
      "learning_rate": 3.882003403539167e-05,
      "loss": 0.5958,
      "step": 10735
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9305107670073952,
      "learning_rate": 3.881047418428689e-05,
      "loss": 0.6929,
      "step": 10740
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9670989101416944,
      "learning_rate": 3.880091142593264e-05,
      "loss": 0.5983,
      "step": 10745
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8448567402070611,
      "learning_rate": 3.8791345762341964e-05,
      "loss": 0.6476,
      "step": 10750
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8342757687344932,
      "learning_rate": 3.878177719552854e-05,
      "loss": 0.5962,
      "step": 10755
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9821085080167582,
      "learning_rate": 3.877220572750666e-05,
      "loss": 0.6087,
      "step": 10760
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8143332953793175,
      "learning_rate": 3.8762631360291216e-05,
      "loss": 0.6719,
      "step": 10765
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8939371738689622,
      "learning_rate": 3.87530540958977e-05,
      "loss": 0.6668,
      "step": 10770
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9984507961597239,
      "learning_rate": 3.8743473936342237e-05,
      "loss": 0.7245,
      "step": 10775
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0366333910297614,
      "learning_rate": 3.8733890883641546e-05,
      "loss": 0.5783,
      "step": 10780
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9395341069698825,
      "learning_rate": 3.872430493981296e-05,
      "loss": 0.654,
      "step": 10785
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8717797602352988,
      "learning_rate": 3.871471610687443e-05,
      "loss": 0.593,
      "step": 10790
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8405064538772097,
      "learning_rate": 3.8705124386844496e-05,
      "loss": 0.6237,
      "step": 10795
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.011690964651043,
      "learning_rate": 3.869552978174232e-05,
      "loss": 0.5654,
      "step": 10800
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8274264177970064,
      "learning_rate": 3.8685932293587667e-05,
      "loss": 0.735,
      "step": 10805
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7492990356826792,
      "learning_rate": 3.867633192440091e-05,
      "loss": 0.6513,
      "step": 10810
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9556706424766879,
      "learning_rate": 3.8666728676203014e-05,
      "loss": 0.6411,
      "step": 10815
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9250297568023449,
      "learning_rate": 3.865712255101558e-05,
      "loss": 0.5837,
      "step": 10820
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9267620841402138,
      "learning_rate": 3.864751355086079e-05,
      "loss": 0.7109,
      "step": 10825
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8955447348850617,
      "learning_rate": 3.8637901677761444e-05,
      "loss": 0.6102,
      "step": 10830
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9267094887891539,
      "learning_rate": 3.8628286933740944e-05,
      "loss": 0.6228,
      "step": 10835
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.042260881163461,
      "learning_rate": 3.8618669320823275e-05,
      "loss": 0.6555,
      "step": 10840
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7754079249431357,
      "learning_rate": 3.860904884103307e-05,
      "loss": 0.5808,
      "step": 10845
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7762541189855575,
      "learning_rate": 3.8599425496395524e-05,
      "loss": 0.6904,
      "step": 10850
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8438497344552657,
      "learning_rate": 3.8589799288936455e-05,
      "loss": 0.6961,
      "step": 10855
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8486304572198412,
      "learning_rate": 3.8580170220682296e-05,
      "loss": 0.6287,
      "step": 10860
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9356315348099948,
      "learning_rate": 3.857053829366004e-05,
      "loss": 0.6161,
      "step": 10865
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8007096208982735,
      "learning_rate": 3.856090350989734e-05,
      "loss": 0.5484,
      "step": 10870
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9925025701344553,
      "learning_rate": 3.8551265871422396e-05,
      "loss": 0.6214,
      "step": 10875
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9825450554788376,
      "learning_rate": 3.854162538026403e-05,
      "loss": 0.6641,
      "step": 10880
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9996087481642618,
      "learning_rate": 3.853198203845168e-05,
      "loss": 0.6026,
      "step": 10885
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9274024570030142,
      "learning_rate": 3.8522335848015354e-05,
      "loss": 0.5601,
      "step": 10890
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8811789207679005,
      "learning_rate": 3.85126868109857e-05,
      "loss": 0.5875,
      "step": 10895
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9871747760850451,
      "learning_rate": 3.8503034929393925e-05,
      "loss": 0.5681,
      "step": 10900
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0026799254752736,
      "learning_rate": 3.8493380205271853e-05,
      "loss": 0.6066,
      "step": 10905
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9995022492168373,
      "learning_rate": 3.848372264065191e-05,
      "loss": 0.6323,
      "step": 10910
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8492706133660896,
      "learning_rate": 3.84740622375671e-05,
      "loss": 0.569,
      "step": 10915
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8581076132316923,
      "learning_rate": 3.846439899805105e-05,
      "loss": 0.6762,
      "step": 10920
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9502228297935466,
      "learning_rate": 3.845473292413798e-05,
      "loss": 0.5787,
      "step": 10925
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8637341736551217,
      "learning_rate": 3.844506401786269e-05,
      "loss": 0.5803,
      "step": 10930
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8524150662452137,
      "learning_rate": 3.843539228126058e-05,
      "loss": 0.5958,
      "step": 10935
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8270053635687636,
      "learning_rate": 3.842571771636768e-05,
      "loss": 0.6677,
      "step": 10940
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0086486412209947,
      "learning_rate": 3.841604032522055e-05,
      "loss": 0.6658,
      "step": 10945
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9352842090904804,
      "learning_rate": 3.840636010985641e-05,
      "loss": 0.5932,
      "step": 10950
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0806432789409284,
      "learning_rate": 3.8396677072313034e-05,
      "loss": 0.6332,
      "step": 10955
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7999354005122973,
      "learning_rate": 3.8386991214628797e-05,
      "loss": 0.5391,
      "step": 10960
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8218652169517799,
      "learning_rate": 3.837730253884269e-05,
      "loss": 0.5802,
      "step": 10965
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9829707526277693,
      "learning_rate": 3.836761104699427e-05,
      "loss": 0.6717,
      "step": 10970
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.999222528590918,
      "learning_rate": 3.83579167411237e-05,
      "loss": 0.5802,
      "step": 10975
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.860445495174055,
      "learning_rate": 3.834821962327173e-05,
      "loss": 0.6537,
      "step": 10980
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8125634256790278,
      "learning_rate": 3.8338519695479716e-05,
      "loss": 0.6635,
      "step": 10985
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9595748660746783,
      "learning_rate": 3.8328816959789575e-05,
      "loss": 0.6035,
      "step": 10990
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9214771561884445,
      "learning_rate": 3.831911141824385e-05,
      "loss": 0.5591,
      "step": 10995
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.857980786523769,
      "learning_rate": 3.830940307288566e-05,
      "loss": 0.5479,
      "step": 11000
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7538894567415371,
      "learning_rate": 3.829969192575871e-05,
      "loss": 0.5774,
      "step": 11005
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8062194948061725,
      "learning_rate": 3.828997797890729e-05,
      "loss": 0.6194,
      "step": 11010
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9548394974181096,
      "learning_rate": 3.8280261234376287e-05,
      "loss": 0.6614,
      "step": 11015
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8897333022950585,
      "learning_rate": 3.82705416942112e-05,
      "loss": 0.6274,
      "step": 11020
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8977298770015277,
      "learning_rate": 3.8260819360458066e-05,
      "loss": 0.6162,
      "step": 11025
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9066455645379055,
      "learning_rate": 3.8251094235163554e-05,
      "loss": 0.6113,
      "step": 11030
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9580061450012253,
      "learning_rate": 3.824136632037491e-05,
      "loss": 0.5868,
      "step": 11035
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9456190707715768,
      "learning_rate": 3.8231635618139936e-05,
      "loss": 0.6277,
      "step": 11040
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8788374367071842,
      "learning_rate": 3.822190213050708e-05,
      "loss": 0.667,
      "step": 11045
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8557056891283095,
      "learning_rate": 3.821216585952532e-05,
      "loss": 0.631,
      "step": 11050
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8657693530524959,
      "learning_rate": 3.820242680724425e-05,
      "loss": 0.6108,
      "step": 11055
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9362997076936507,
      "learning_rate": 3.8192684975714044e-05,
      "loss": 0.6881,
      "step": 11060
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9086232257851192,
      "learning_rate": 3.8182940366985453e-05,
      "loss": 0.575,
      "step": 11065
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9798969018535946,
      "learning_rate": 3.817319298310984e-05,
      "loss": 0.5777,
      "step": 11070
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9204227526266643,
      "learning_rate": 3.81634428261391e-05,
      "loss": 0.6161,
      "step": 11075
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9031456475026092,
      "learning_rate": 3.815368989812577e-05,
      "loss": 0.5984,
      "step": 11080
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8332886586657042,
      "learning_rate": 3.814393420112293e-05,
      "loss": 0.5782,
      "step": 11085
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.948327368803431,
      "learning_rate": 3.813417573718425e-05,
      "loss": 0.6042,
      "step": 11090
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9302673907811143,
      "learning_rate": 3.8124414508364e-05,
      "loss": 0.6084,
      "step": 11095
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0149846346708356,
      "learning_rate": 3.811465051671703e-05,
      "loss": 0.7347,
      "step": 11100
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8853276720694164,
      "learning_rate": 3.810488376429875e-05,
      "loss": 0.6183,
      "step": 11105
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8997255542636341,
      "learning_rate": 3.809511425316516e-05,
      "loss": 0.5893,
      "step": 11110
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.2493794398722728,
      "learning_rate": 3.8085341985372847e-05,
      "loss": 0.638,
      "step": 11115
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9648031344636927,
      "learning_rate": 3.807556696297898e-05,
      "loss": 0.5742,
      "step": 11120
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7729946615076482,
      "learning_rate": 3.80657891880413e-05,
      "loss": 0.6159,
      "step": 11125
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8102769410124832,
      "learning_rate": 3.805600866261814e-05,
      "loss": 0.5689,
      "step": 11130
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9167695446361862,
      "learning_rate": 3.804622538876839e-05,
      "loss": 0.5909,
      "step": 11135
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.885746496310935,
      "learning_rate": 3.803643936855154e-05,
      "loss": 0.6393,
      "step": 11140
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8345147567446465,
      "learning_rate": 3.8026650604027634e-05,
      "loss": 0.6152,
      "step": 11145
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8324694114308352,
      "learning_rate": 3.8016859097257346e-05,
      "loss": 0.5446,
      "step": 11150
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8321645576957094,
      "learning_rate": 3.8007064850301856e-05,
      "loss": 0.6058,
      "step": 11155
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8846630905388356,
      "learning_rate": 3.7997267865222966e-05,
      "loss": 0.5836,
      "step": 11160
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8318205805952699,
      "learning_rate": 3.7987468144083036e-05,
      "loss": 0.5852,
      "step": 11165
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8646265924602047,
      "learning_rate": 3.7977665688945034e-05,
      "loss": 0.6084,
      "step": 11170
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0135804808549047,
      "learning_rate": 3.7967860501872466e-05,
      "loss": 0.6305,
      "step": 11175
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0572248931514374,
      "learning_rate": 3.795805258492942e-05,
      "loss": 0.5791,
      "step": 11180
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8241309161367708,
      "learning_rate": 3.794824194018057e-05,
      "loss": 0.6332,
      "step": 11185
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0207797991250498,
      "learning_rate": 3.793842856969115e-05,
      "loss": 0.5828,
      "step": 11190
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8703248082964679,
      "learning_rate": 3.7928612475526994e-05,
      "loss": 0.5701,
      "step": 11195
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9909917811865638,
      "learning_rate": 3.791879365975449e-05,
      "loss": 0.6104,
      "step": 11200
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9724696243898826,
      "learning_rate": 3.79089721244406e-05,
      "loss": 0.7083,
      "step": 11205
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8660167582718417,
      "learning_rate": 3.7899147871652854e-05,
      "loss": 0.6314,
      "step": 11210
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8612859220817484,
      "learning_rate": 3.788932090345936e-05,
      "loss": 0.6172,
      "step": 11215
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8656672706644697,
      "learning_rate": 3.78794912219288e-05,
      "loss": 0.6021,
      "step": 11220
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8465723235668507,
      "learning_rate": 3.786965882913043e-05,
      "loss": 0.5641,
      "step": 11225
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9442867577106012,
      "learning_rate": 3.785982372713407e-05,
      "loss": 0.6582,
      "step": 11230
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8928565402328836,
      "learning_rate": 3.7849985918010104e-05,
      "loss": 0.566,
      "step": 11235
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9336485202547289,
      "learning_rate": 3.7840145403829504e-05,
      "loss": 0.5899,
      "step": 11240
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8785527733013121,
      "learning_rate": 3.783030218666379e-05,
      "loss": 0.6675,
      "step": 11245
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9097826967379423,
      "learning_rate": 3.782045626858508e-05,
      "loss": 0.6363,
      "step": 11250
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8987518233115371,
      "learning_rate": 3.781060765166603e-05,
      "loss": 0.5632,
      "step": 11255
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7868921879200586,
      "learning_rate": 3.780075633797987e-05,
      "loss": 0.5834,
      "step": 11260
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0211945683796622,
      "learning_rate": 3.7790902329600425e-05,
      "loss": 0.5981,
      "step": 11265
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8419824706101989,
      "learning_rate": 3.7781045628602055e-05,
      "loss": 0.6016,
      "step": 11270
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9685983278017648,
      "learning_rate": 3.7771186237059684e-05,
      "loss": 0.7299,
      "step": 11275
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8817880606433147,
      "learning_rate": 3.776132415704885e-05,
      "loss": 0.5684,
      "step": 11280
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8953681497446253,
      "learning_rate": 3.7751459390645606e-05,
      "loss": 0.6323,
      "step": 11285
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.92791057615013,
      "learning_rate": 3.774159193992659e-05,
      "loss": 0.6532,
      "step": 11290
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7796992976893938,
      "learning_rate": 3.773172180696899e-05,
      "loss": 0.5357,
      "step": 11295
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0095280570162695,
      "learning_rate": 3.77218489938506e-05,
      "loss": 0.6254,
      "step": 11300
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8728540668103351,
      "learning_rate": 3.7711973502649734e-05,
      "loss": 0.5855,
      "step": 11305
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8208463354991954,
      "learning_rate": 3.770209533544528e-05,
      "loss": 0.6229,
      "step": 11310
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.913216067875161,
      "learning_rate": 3.769221449431672e-05,
      "loss": 0.5345,
      "step": 11315
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8384641919175935,
      "learning_rate": 3.768233098134405e-05,
      "loss": 0.6026,
      "step": 11320
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1479452158066363,
      "learning_rate": 3.767244479860786e-05,
      "loss": 0.6281,
      "step": 11325
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8902637630830832,
      "learning_rate": 3.7662555948189306e-05,
      "loss": 0.6356,
      "step": 11330
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9125532587337507,
      "learning_rate": 3.765266443217007e-05,
      "loss": 0.5757,
      "step": 11335
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9114758609594183,
      "learning_rate": 3.7642770252632445e-05,
      "loss": 0.6894,
      "step": 11340
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9808822816853947,
      "learning_rate": 3.7632873411659255e-05,
      "loss": 0.6391,
      "step": 11345
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8492305095279656,
      "learning_rate": 3.7622973911333876e-05,
      "loss": 0.6428,
      "step": 11350
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9750796246164326,
      "learning_rate": 3.761307175374026e-05,
      "loss": 0.5861,
      "step": 11355
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8629630109832057,
      "learning_rate": 3.760316694096293e-05,
      "loss": 0.6064,
      "step": 11360
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3550431161513656,
      "learning_rate": 3.7593259475086935e-05,
      "loss": 0.6905,
      "step": 11365
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9469540128688306,
      "learning_rate": 3.7583349358197904e-05,
      "loss": 0.6067,
      "step": 11370
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8695447191434387,
      "learning_rate": 3.7573436592382026e-05,
      "loss": 0.6105,
      "step": 11375
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9728299524926329,
      "learning_rate": 3.7563521179726044e-05,
      "loss": 0.5891,
      "step": 11380
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9792883660167887,
      "learning_rate": 3.755360312231726e-05,
      "loss": 0.6005,
      "step": 11385
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9105457541277419,
      "learning_rate": 3.75436824222435e-05,
      "loss": 0.6469,
      "step": 11390
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7922867610863673,
      "learning_rate": 3.753375908159321e-05,
      "loss": 0.6414,
      "step": 11395
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.111971860056968,
      "learning_rate": 3.752383310245534e-05,
      "loss": 0.6063,
      "step": 11400
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9304507725403305,
      "learning_rate": 3.7513904486919415e-05,
      "loss": 0.6509,
      "step": 11405
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8469496890401091,
      "learning_rate": 3.7503973237075514e-05,
      "loss": 0.6037,
      "step": 11410
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9445851104822968,
      "learning_rate": 3.749403935501427e-05,
      "loss": 0.644,
      "step": 11415
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8120529458493533,
      "learning_rate": 3.748410284282686e-05,
      "loss": 0.7007,
      "step": 11420
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8413557511647501,
      "learning_rate": 3.7474163702605034e-05,
      "loss": 0.6032,
      "step": 11425
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8352067275623862,
      "learning_rate": 3.7464221936441094e-05,
      "loss": 0.5946,
      "step": 11430
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0183922714166689,
      "learning_rate": 3.745427754642787e-05,
      "loss": 0.6426,
      "step": 11435
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.902175681401637,
      "learning_rate": 3.7444330534658755e-05,
      "loss": 0.573,
      "step": 11440
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9848095718516258,
      "learning_rate": 3.743438090322772e-05,
      "loss": 0.556,
      "step": 11445
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9348176258943024,
      "learning_rate": 3.742442865422925e-05,
      "loss": 0.6395,
      "step": 11450
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8557819391771571,
      "learning_rate": 3.741447378975841e-05,
      "loss": 0.5686,
      "step": 11455
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9832075277554719,
      "learning_rate": 3.740451631191079e-05,
      "loss": 0.6982,
      "step": 11460
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.848202394046599,
      "learning_rate": 3.739455622278255e-05,
      "loss": 0.5942,
      "step": 11465
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8952627283268336,
      "learning_rate": 3.7384593524470396e-05,
      "loss": 0.581,
      "step": 11470
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8229072416704125,
      "learning_rate": 3.7374628219071576e-05,
      "loss": 0.5735,
      "step": 11475
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.886607252396971,
      "learning_rate": 3.7364660308683894e-05,
      "loss": 0.6148,
      "step": 11480
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8676237723911505,
      "learning_rate": 3.735468979540569e-05,
      "loss": 0.6434,
      "step": 11485
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.848116892329449,
      "learning_rate": 3.7344716681335887e-05,
      "loss": 0.6368,
      "step": 11490
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9443951647533891,
      "learning_rate": 3.73347409685739e-05,
      "loss": 0.6804,
      "step": 11495
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8141488078395794,
      "learning_rate": 3.732476265921974e-05,
      "loss": 0.6425,
      "step": 11500
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9608420311948448,
      "learning_rate": 3.731478175537393e-05,
      "loss": 0.6219,
      "step": 11505
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8621877078630583,
      "learning_rate": 3.730479825913758e-05,
      "loss": 0.6431,
      "step": 11510
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.826002432148367,
      "learning_rate": 3.7294812172612284e-05,
      "loss": 0.6545,
      "step": 11515
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.846160540751185,
      "learning_rate": 3.728482349790025e-05,
      "loss": 0.5828,
      "step": 11520
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0051060833180396,
      "learning_rate": 3.7274832237104185e-05,
      "loss": 0.615,
      "step": 11525
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8096020240629066,
      "learning_rate": 3.726483839232736e-05,
      "loss": 0.6059,
      "step": 11530
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8649721951013966,
      "learning_rate": 3.725484196567358e-05,
      "loss": 0.6594,
      "step": 11535
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8220253964395031,
      "learning_rate": 3.7244842959247196e-05,
      "loss": 0.6212,
      "step": 11540
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9547267956323815,
      "learning_rate": 3.7234841375153105e-05,
      "loss": 0.5733,
      "step": 11545
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.021393747095624,
      "learning_rate": 3.7224837215496746e-05,
      "loss": 0.7078,
      "step": 11550
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9635282878003277,
      "learning_rate": 3.7214830482384095e-05,
      "loss": 0.6321,
      "step": 11555
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9400741543897787,
      "learning_rate": 3.720482117792169e-05,
      "loss": 0.5791,
      "step": 11560
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8244965274555158,
      "learning_rate": 3.719480930421657e-05,
      "loss": 0.6157,
      "step": 11565
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0930063113874906,
      "learning_rate": 3.7184794863376354e-05,
      "loss": 0.5673,
      "step": 11570
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0914630542998986,
      "learning_rate": 3.717477785750919e-05,
      "loss": 0.6337,
      "step": 11575
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.892354492440155,
      "learning_rate": 3.716475828872375e-05,
      "loss": 0.5459,
      "step": 11580
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8611819198520095,
      "learning_rate": 3.7154736159129265e-05,
      "loss": 0.5331,
      "step": 11585
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8068972894930494,
      "learning_rate": 3.714471147083551e-05,
      "loss": 0.6029,
      "step": 11590
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8843196201964665,
      "learning_rate": 3.7134684225952766e-05,
      "loss": 0.6172,
      "step": 11595
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9012176991567481,
      "learning_rate": 3.712465442659188e-05,
      "loss": 0.5888,
      "step": 11600
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.419713424001114,
      "learning_rate": 3.711462207486424e-05,
      "loss": 0.6504,
      "step": 11605
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9966863116420186,
      "learning_rate": 3.710458717288176e-05,
      "loss": 0.6061,
      "step": 11610
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8515444217587302,
      "learning_rate": 3.7094549722756866e-05,
      "loss": 0.6732,
      "step": 11615
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9083326272134167,
      "learning_rate": 3.708450972660258e-05,
      "loss": 0.6058,
      "step": 11620
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0100396309954474,
      "learning_rate": 3.7074467186532405e-05,
      "loss": 0.6137,
      "step": 11625
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8813512492341774,
      "learning_rate": 3.706442210466041e-05,
      "loss": 1.3633,
      "step": 11630
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8262883146845641,
      "learning_rate": 3.7054374483101194e-05,
      "loss": 0.6067,
      "step": 11635
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8063940645429819,
      "learning_rate": 3.7044324323969875e-05,
      "loss": 0.6256,
      "step": 11640
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8502441867849484,
      "learning_rate": 3.7034271629382125e-05,
      "loss": 0.6471,
      "step": 11645
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9558059639172778,
      "learning_rate": 3.702421640145414e-05,
      "loss": 0.6065,
      "step": 11650
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8964591128487243,
      "learning_rate": 3.7014158642302645e-05,
      "loss": 0.6249,
      "step": 11655
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9410443055703813,
      "learning_rate": 3.700409835404491e-05,
      "loss": 0.6858,
      "step": 11660
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8359924847753424,
      "learning_rate": 3.699403553879873e-05,
      "loss": 0.6574,
      "step": 11665
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8506670110351549,
      "learning_rate": 3.6983970198682435e-05,
      "loss": 0.6281,
      "step": 11670
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.864007163969048,
      "learning_rate": 3.6973902335814886e-05,
      "loss": 0.6158,
      "step": 11675
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9014929669141521,
      "learning_rate": 3.696383195231546e-05,
      "loss": 0.6167,
      "step": 11680
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8483195813673494,
      "learning_rate": 3.695375905030409e-05,
      "loss": 0.6211,
      "step": 11685
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7976878432407867,
      "learning_rate": 3.694368363190123e-05,
      "loss": 0.6019,
      "step": 11690
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.946242048073552,
      "learning_rate": 3.6933605699227855e-05,
      "loss": 0.6295,
      "step": 11695
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8219336558297399,
      "learning_rate": 3.692352525440548e-05,
      "loss": 0.6224,
      "step": 11700
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8923286866315998,
      "learning_rate": 3.6913442299556135e-05,
      "loss": 0.6115,
      "step": 11705
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8435919575200298,
      "learning_rate": 3.690335683680239e-05,
      "loss": 0.6032,
      "step": 11710
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0458727498746518,
      "learning_rate": 3.689326886826735e-05,
      "loss": 0.5331,
      "step": 11715
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8457603664996752,
      "learning_rate": 3.688317839607463e-05,
      "loss": 0.6043,
      "step": 11720
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9141727510462546,
      "learning_rate": 3.6873085422348385e-05,
      "loss": 0.602,
      "step": 11725
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9706587815811856,
      "learning_rate": 3.686298994921329e-05,
      "loss": 0.5999,
      "step": 11730
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8392066019989034,
      "learning_rate": 3.685289197879454e-05,
      "loss": 0.6388,
      "step": 11735
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9963421838539251,
      "learning_rate": 3.684279151321788e-05,
      "loss": 0.5961,
      "step": 11740
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8536922835244994,
      "learning_rate": 3.683268855460955e-05,
      "loss": 0.6118,
      "step": 11745
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7901944889689158,
      "learning_rate": 3.682258310509633e-05,
      "loss": 0.6362,
      "step": 11750
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9220205609995663,
      "learning_rate": 3.6812475166805526e-05,
      "loss": 0.6294,
      "step": 11755
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8741271739185323,
      "learning_rate": 3.680236474186497e-05,
      "loss": 0.6089,
      "step": 11760
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9070119288566725,
      "learning_rate": 3.6792251832403e-05,
      "loss": 0.664,
      "step": 11765
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8534567714319213,
      "learning_rate": 3.67821364405485e-05,
      "loss": 0.6307,
      "step": 11770
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8874654347954261,
      "learning_rate": 3.6772018568430864e-05,
      "loss": 0.6057,
      "step": 11775
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9480911478991431,
      "learning_rate": 3.676189821818e-05,
      "loss": 0.5985,
      "step": 11780
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8174653350358779,
      "learning_rate": 3.675177539192636e-05,
      "loss": 0.5718,
      "step": 11785
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7888549229450207,
      "learning_rate": 3.674165009180091e-05,
      "loss": 0.5888,
      "step": 11790
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9299848532603948,
      "learning_rate": 3.673152231993512e-05,
      "loss": 0.6406,
      "step": 11795
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8579765079823473,
      "learning_rate": 3.672139207846099e-05,
      "loss": 0.599,
      "step": 11800
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9839779105297691,
      "learning_rate": 3.671125936951105e-05,
      "loss": 0.6279,
      "step": 11805
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8922445801315554,
      "learning_rate": 3.670112419521833e-05,
      "loss": 0.5724,
      "step": 11810
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9473545153433485,
      "learning_rate": 3.669098655771641e-05,
      "loss": 0.6988,
      "step": 11815
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9594339530288059,
      "learning_rate": 3.668084645913935e-05,
      "loss": 0.6917,
      "step": 11820
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.752388039245179,
      "learning_rate": 3.667070390162176e-05,
      "loss": 0.5212,
      "step": 11825
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.945652079162958,
      "learning_rate": 3.666055888729874e-05,
      "loss": 0.5946,
      "step": 11830
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8740843589928866,
      "learning_rate": 3.665041141830594e-05,
      "loss": 0.6667,
      "step": 11835
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8873866105458915,
      "learning_rate": 3.66402614967795e-05,
      "loss": 0.6325,
      "step": 11840
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9989105772306284,
      "learning_rate": 3.6630109124856077e-05,
      "loss": 0.6198,
      "step": 11845
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9745653244874164,
      "learning_rate": 3.6619954304672865e-05,
      "loss": 0.7203,
      "step": 11850
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8955360500291366,
      "learning_rate": 3.660979703836755e-05,
      "loss": 0.6057,
      "step": 11855
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0822483631246398,
      "learning_rate": 3.6599637328078346e-05,
      "loss": 0.6273,
      "step": 11860
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8862916401341311,
      "learning_rate": 3.658947517594398e-05,
      "loss": 0.6566,
      "step": 11865
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7865539902489639,
      "learning_rate": 3.657931058410369e-05,
      "loss": 0.6186,
      "step": 11870
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7606360957566586,
      "learning_rate": 3.656914355469724e-05,
      "loss": 0.5791,
      "step": 11875
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0426033911520904,
      "learning_rate": 3.655897408986487e-05,
      "loss": 0.6493,
      "step": 11880
    },
    {
      "epoch": 1.39,
      "grad_norm": 3.3394917940746476,
      "learning_rate": 3.6548802191747385e-05,
      "loss": 0.6398,
      "step": 11885
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8190767899160933,
      "learning_rate": 3.6538627862486064e-05,
      "loss": 0.7146,
      "step": 11890
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8416893474339635,
      "learning_rate": 3.65284511042227e-05,
      "loss": 0.6684,
      "step": 11895
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8594889718555592,
      "learning_rate": 3.6518271919099635e-05,
      "loss": 0.6567,
      "step": 11900
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8430123737106848,
      "learning_rate": 3.6508090309259674e-05,
      "loss": 0.6225,
      "step": 11905
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8647792785300874,
      "learning_rate": 3.649790627684616e-05,
      "loss": 0.5574,
      "step": 11910
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9818801422694434,
      "learning_rate": 3.6487719824002926e-05,
      "loss": 0.58,
      "step": 11915
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9276693398318523,
      "learning_rate": 3.647753095287434e-05,
      "loss": 0.613,
      "step": 11920
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9167940833193173,
      "learning_rate": 3.646733966560527e-05,
      "loss": 0.6399,
      "step": 11925
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8515885891556684,
      "learning_rate": 3.6457145964341064e-05,
      "loss": 0.636,
      "step": 11930
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1826619378287282,
      "learning_rate": 3.6446949851227636e-05,
      "loss": 0.6507,
      "step": 11935
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8151826147835061,
      "learning_rate": 3.6436751328411346e-05,
      "loss": 0.6634,
      "step": 11940
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8933784656186476,
      "learning_rate": 3.642655039803911e-05,
      "loss": 0.5745,
      "step": 11945
    },
    {
      "epoch": 1.4,
      "grad_norm": 18.275452782991643,
      "learning_rate": 3.6416347062258325e-05,
      "loss": 0.6362,
      "step": 11950
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9538021653732811,
      "learning_rate": 3.640614132321689e-05,
      "loss": 0.6611,
      "step": 11955
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.267132202895838,
      "learning_rate": 3.639593318306323e-05,
      "loss": 0.6567,
      "step": 11960
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9172124975874861,
      "learning_rate": 3.638572264394626e-05,
      "loss": 0.6367,
      "step": 11965
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8664333095838165,
      "learning_rate": 3.637550970801543e-05,
      "loss": 0.5852,
      "step": 11970
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9156970694456673,
      "learning_rate": 3.636529437742063e-05,
      "loss": 0.6654,
      "step": 11975
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8622107729700692,
      "learning_rate": 3.635507665431231e-05,
      "loss": 0.614,
      "step": 11980
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.799167859462027,
      "learning_rate": 3.6344856540841424e-05,
      "loss": 0.6015,
      "step": 11985
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9432145201955199,
      "learning_rate": 3.633463403915939e-05,
      "loss": 0.5937,
      "step": 11990
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8942013630608001,
      "learning_rate": 3.632440915141816e-05,
      "loss": 0.7099,
      "step": 11995
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8970834872690424,
      "learning_rate": 3.6314181879770195e-05,
      "loss": 0.6072,
      "step": 12000
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8670091301488518,
      "learning_rate": 3.630395222636842e-05,
      "loss": 0.6866,
      "step": 12005
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9950770407163555,
      "learning_rate": 3.629372019336629e-05,
      "loss": 0.688,
      "step": 12010
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0092870679045602,
      "learning_rate": 3.628348578291776e-05,
      "loss": 0.6212,
      "step": 12015
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8970458337270661,
      "learning_rate": 3.6273248997177284e-05,
      "loss": 0.5511,
      "step": 12020
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9267648123052946,
      "learning_rate": 3.626300983829981e-05,
      "loss": 0.641,
      "step": 12025
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9212166095013694,
      "learning_rate": 3.625276830844078e-05,
      "loss": 0.6525,
      "step": 12030
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8567839607925971,
      "learning_rate": 3.624252440975615e-05,
      "loss": 0.6262,
      "step": 12035
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8906713133012847,
      "learning_rate": 3.623227814440237e-05,
      "loss": 0.6247,
      "step": 12040
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1224718084593563,
      "learning_rate": 3.622202951453637e-05,
      "loss": 0.5882,
      "step": 12045
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.220008869058968,
      "learning_rate": 3.621177852231563e-05,
      "loss": 0.6168,
      "step": 12050
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9706766043476424,
      "learning_rate": 3.620152516989805e-05,
      "loss": 0.617,
      "step": 12055
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8153129469677494,
      "learning_rate": 3.619126945944209e-05,
      "loss": 0.6571,
      "step": 12060
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0215444864061816,
      "learning_rate": 3.618101139310669e-05,
      "loss": 0.6171,
      "step": 12065
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8934896089793263,
      "learning_rate": 3.617075097305126e-05,
      "loss": 0.5959,
      "step": 12070
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9843694449696093,
      "learning_rate": 3.6160488201435745e-05,
      "loss": 0.5846,
      "step": 12075
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8697273081394036,
      "learning_rate": 3.6150223080420555e-05,
      "loss": 0.5674,
      "step": 12080
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.800288367189229,
      "learning_rate": 3.6139955612166615e-05,
      "loss": 0.5886,
      "step": 12085
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9040934975175596,
      "learning_rate": 3.612968579883532e-05,
      "loss": 0.642,
      "step": 12090
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9528880927170442,
      "learning_rate": 3.611941364258859e-05,
      "loss": 0.616,
      "step": 12095
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9271861118502893,
      "learning_rate": 3.610913914558881e-05,
      "loss": 0.6094,
      "step": 12100
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8450485138889783,
      "learning_rate": 3.609886230999886e-05,
      "loss": 0.6819,
      "step": 12105
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9521220678078501,
      "learning_rate": 3.6088583137982154e-05,
      "loss": 0.5995,
      "step": 12110
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8283449125046546,
      "learning_rate": 3.607830163170254e-05,
      "loss": 0.5865,
      "step": 12115
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8336540444103293,
      "learning_rate": 3.6068017793324385e-05,
      "loss": 0.6064,
      "step": 12120
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8838534548040841,
      "learning_rate": 3.605773162501255e-05,
      "loss": 0.6232,
      "step": 12125
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8218728945185004,
      "learning_rate": 3.6047443128932374e-05,
      "loss": 0.5761,
      "step": 12130
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9236400315218273,
      "learning_rate": 3.6037152307249706e-05,
      "loss": 0.5891,
      "step": 12135
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.881988398219993,
      "learning_rate": 3.6026859162130854e-05,
      "loss": 0.6041,
      "step": 12140
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8190175918841568,
      "learning_rate": 3.601656369574265e-05,
      "loss": 0.5823,
      "step": 12145
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9116519229688698,
      "learning_rate": 3.600626591025239e-05,
      "loss": 0.62,
      "step": 12150
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.9053885643530948,
      "learning_rate": 3.599596580782787e-05,
      "loss": 0.7042,
      "step": 12155
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.854121065594848,
      "learning_rate": 3.5985663390637354e-05,
      "loss": 0.5904,
      "step": 12160
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.939889025268169,
      "learning_rate": 3.5975358660849637e-05,
      "loss": 0.5553,
      "step": 12165
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9841472953902778,
      "learning_rate": 3.596505162063394e-05,
      "loss": 0.6373,
      "step": 12170
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0028632154873853,
      "learning_rate": 3.5954742272160026e-05,
      "loss": 0.651,
      "step": 12175
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.953966824591413,
      "learning_rate": 3.594443061759811e-05,
      "loss": 0.5832,
      "step": 12180
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8970291999675264,
      "learning_rate": 3.593411665911891e-05,
      "loss": 0.6377,
      "step": 12185
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2858605745665024,
      "learning_rate": 3.592380039889362e-05,
      "loss": 0.6246,
      "step": 12190
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7986784002870713,
      "learning_rate": 3.591348183909391e-05,
      "loss": 0.6582,
      "step": 12195
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9071050087626734,
      "learning_rate": 3.590316098189197e-05,
      "loss": 0.6737,
      "step": 12200
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9132510708771986,
      "learning_rate": 3.5892837829460426e-05,
      "loss": 0.5919,
      "step": 12205
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9433480879893589,
      "learning_rate": 3.588251238397242e-05,
      "loss": 0.5984,
      "step": 12210
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8425425497415544,
      "learning_rate": 3.587218464760157e-05,
      "loss": 0.674,
      "step": 12215
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8560379066177962,
      "learning_rate": 3.5861854622521956e-05,
      "loss": 0.6225,
      "step": 12220
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0216349670614215,
      "learning_rate": 3.585152231090817e-05,
      "loss": 0.5898,
      "step": 12225
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.941910058752599,
      "learning_rate": 3.584118771493527e-05,
      "loss": 0.6371,
      "step": 12230
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9450546824846788,
      "learning_rate": 3.583085083677879e-05,
      "loss": 0.67,
      "step": 12235
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9430983134358434,
      "learning_rate": 3.582051167861477e-05,
      "loss": 0.6077,
      "step": 12240
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2647311219912982,
      "learning_rate": 3.581017024261968e-05,
      "loss": 0.5574,
      "step": 12245
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8800855810701044,
      "learning_rate": 3.579982653097053e-05,
      "loss": 0.5983,
      "step": 12250
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8531335602475937,
      "learning_rate": 3.578948054584476e-05,
      "loss": 0.6076,
      "step": 12255
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9688059857971794,
      "learning_rate": 3.577913228942033e-05,
      "loss": 0.5814,
      "step": 12260
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8878522123624325,
      "learning_rate": 3.5768781763875634e-05,
      "loss": 0.6511,
      "step": 12265
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8859337872833457,
      "learning_rate": 3.575842897138957e-05,
      "loss": 0.6106,
      "step": 12270
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8392555265533592,
      "learning_rate": 3.5748073914141515e-05,
      "loss": 0.6318,
      "step": 12275
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0179826817227207,
      "learning_rate": 3.573771659431132e-05,
      "loss": 0.6587,
      "step": 12280
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8250855610755873,
      "learning_rate": 3.57273570140793e-05,
      "loss": 0.6277,
      "step": 12285
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.028761145323544,
      "learning_rate": 3.571699517562627e-05,
      "loss": 0.5805,
      "step": 12290
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8470798159598402,
      "learning_rate": 3.57066310811335e-05,
      "loss": 0.6357,
      "step": 12295
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9002010615128283,
      "learning_rate": 3.569626473278271e-05,
      "loss": 0.5889,
      "step": 12300
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9351117962530887,
      "learning_rate": 3.568589613275618e-05,
      "loss": 0.6457,
      "step": 12305
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9952238962551381,
      "learning_rate": 3.567552528323656e-05,
      "loss": 0.5789,
      "step": 12310
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8771019021548889,
      "learning_rate": 3.566515218640705e-05,
      "loss": 0.5995,
      "step": 12315
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8035809097945872,
      "learning_rate": 3.5654776844451296e-05,
      "loss": 0.5988,
      "step": 12320
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8893650027132582,
      "learning_rate": 3.56443992595534e-05,
      "loss": 0.5957,
      "step": 12325
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9225381879454705,
      "learning_rate": 3.5634019433897964e-05,
      "loss": 0.6473,
      "step": 12330
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9015631227736667,
      "learning_rate": 3.562363736967004e-05,
      "loss": 0.6037,
      "step": 12335
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8318607677704034,
      "learning_rate": 3.561325306905517e-05,
      "loss": 0.6522,
      "step": 12340
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8994217373668357,
      "learning_rate": 3.5602866534239365e-05,
      "loss": 0.5811,
      "step": 12345
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8591196579865831,
      "learning_rate": 3.559247776740908e-05,
      "loss": 0.5433,
      "step": 12350
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.865730815127164,
      "learning_rate": 3.558208677075127e-05,
      "loss": 0.5847,
      "step": 12355
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9072983135377769,
      "learning_rate": 3.5571693546453356e-05,
      "loss": 0.5503,
      "step": 12360
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8976836103305895,
      "learning_rate": 3.556129809670321e-05,
      "loss": 0.6447,
      "step": 12365
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9092058787096782,
      "learning_rate": 3.555090042368918e-05,
      "loss": 0.5973,
      "step": 12370
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8266958082415768,
      "learning_rate": 3.5540500529600096e-05,
      "loss": 0.6323,
      "step": 12375
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8145995325127587,
      "learning_rate": 3.553009841662524e-05,
      "loss": 0.599,
      "step": 12380
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.7600467175839406,
      "learning_rate": 3.551969408695436e-05,
      "loss": 0.6181,
      "step": 12385
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2119218629692157,
      "learning_rate": 3.550928754277769e-05,
      "loss": 0.6287,
      "step": 12390
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8219758616713628,
      "learning_rate": 3.5498878786285906e-05,
      "loss": 0.5444,
      "step": 12395
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9195114586262254,
      "learning_rate": 3.5488467819670165e-05,
      "loss": 0.5227,
      "step": 12400
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8354182979948097,
      "learning_rate": 3.5478054645122074e-05,
      "loss": 0.6294,
      "step": 12405
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.939150400634756,
      "learning_rate": 3.546763926483373e-05,
      "loss": 0.6228,
      "step": 12410
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8771575023674802,
      "learning_rate": 3.5457221680997665e-05,
      "loss": 0.6274,
      "step": 12415
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9647783644348853,
      "learning_rate": 3.5446801895806904e-05,
      "loss": 0.6751,
      "step": 12420
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.054634469694846,
      "learning_rate": 3.543637991145491e-05,
      "loss": 0.6449,
      "step": 12425
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8879257717479248,
      "learning_rate": 3.542595573013563e-05,
      "loss": 0.5864,
      "step": 12430
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8623663216663834,
      "learning_rate": 3.541552935404346e-05,
      "loss": 0.6031,
      "step": 12435
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0150487304074556,
      "learning_rate": 3.540510078537325e-05,
      "loss": 0.6144,
      "step": 12440
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9425391365070878,
      "learning_rate": 3.539467002632034e-05,
      "loss": 0.6682,
      "step": 12445
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8622312713916199,
      "learning_rate": 3.538423707908051e-05,
      "loss": 0.6847,
      "step": 12450
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8801524966858489,
      "learning_rate": 3.5373801945849985e-05,
      "loss": 0.6163,
      "step": 12455
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8299444575804841,
      "learning_rate": 3.53633646288255e-05,
      "loss": 0.5875,
      "step": 12460
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8759494556959604,
      "learning_rate": 3.53529251302042e-05,
      "loss": 0.5593,
      "step": 12465
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8985264861266203,
      "learning_rate": 3.534248345218371e-05,
      "loss": 0.6427,
      "step": 12470
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8519847751204643,
      "learning_rate": 3.533203959696212e-05,
      "loss": 0.7022,
      "step": 12475
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9825898764091011,
      "learning_rate": 3.532159356673797e-05,
      "loss": 0.5919,
      "step": 12480
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8653221102185527,
      "learning_rate": 3.531114536371026e-05,
      "loss": 0.6803,
      "step": 12485
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9239947369685201,
      "learning_rate": 3.530069499007843e-05,
      "loss": 0.6154,
      "step": 12490
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9062899279415221,
      "learning_rate": 3.529024244804242e-05,
      "loss": 0.65,
      "step": 12495
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9527406919374568,
      "learning_rate": 3.527978773980257e-05,
      "loss": 0.6002,
      "step": 12500
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0399162325622036,
      "learning_rate": 3.5269330867559734e-05,
      "loss": 0.672,
      "step": 12505
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8442645622694827,
      "learning_rate": 3.525887183351517e-05,
      "loss": 0.535,
      "step": 12510
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.938052602974706,
      "learning_rate": 3.524841063987063e-05,
      "loss": 0.5834,
      "step": 12515
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8828262617775489,
      "learning_rate": 3.52379472888283e-05,
      "loss": 0.5582,
      "step": 12520
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7867389542354907,
      "learning_rate": 3.522748178259082e-05,
      "loss": 0.6142,
      "step": 12525
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9845643637186133,
      "learning_rate": 3.5217014123361306e-05,
      "loss": 0.6257,
      "step": 12530
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9202685943685677,
      "learning_rate": 3.5206544313343274e-05,
      "loss": 0.5742,
      "step": 12535
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.039346776792003,
      "learning_rate": 3.519607235474077e-05,
      "loss": 0.6271,
      "step": 12540
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8564988677121582,
      "learning_rate": 3.518559824975822e-05,
      "loss": 0.6047,
      "step": 12545
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.015991476780951,
      "learning_rate": 3.517512200060055e-05,
      "loss": 0.7109,
      "step": 12550
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8922363318376536,
      "learning_rate": 3.5164643609473114e-05,
      "loss": 0.5472,
      "step": 12555
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8912317480108893,
      "learning_rate": 3.515416307858173e-05,
      "loss": 0.5998,
      "step": 12560
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0199078616697879,
      "learning_rate": 3.5143680410132654e-05,
      "loss": 0.6205,
      "step": 12565
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8246101578623768,
      "learning_rate": 3.513319560633259e-05,
      "loss": 0.6132,
      "step": 12570
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.916942633653224,
      "learning_rate": 3.5122708669388715e-05,
      "loss": 0.6402,
      "step": 12575
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.831173170164292,
      "learning_rate": 3.511221960150862e-05,
      "loss": 0.6278,
      "step": 12580
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0541321235758667,
      "learning_rate": 3.5101728404900376e-05,
      "loss": 0.6283,
      "step": 12585
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8331126187727365,
      "learning_rate": 3.50912350817725e-05,
      "loss": 0.5667,
      "step": 12590
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.071895814551917,
      "learning_rate": 3.508073963433392e-05,
      "loss": 0.6305,
      "step": 12595
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.901334249363068,
      "learning_rate": 3.507024206479406e-05,
      "loss": 0.6235,
      "step": 12600
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9740922386563966,
      "learning_rate": 3.505974237536276e-05,
      "loss": 0.608,
      "step": 12605
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8258422734556367,
      "learning_rate": 3.5049240568250314e-05,
      "loss": 0.5743,
      "step": 12610
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8115276989495022,
      "learning_rate": 3.503873664566746e-05,
      "loss": 0.5889,
      "step": 12615
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8906443008710788,
      "learning_rate": 3.502823060982539e-05,
      "loss": 0.6311,
      "step": 12620
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9575861203827718,
      "learning_rate": 3.501772246293573e-05,
      "loss": 0.6597,
      "step": 12625
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9253573690672728,
      "learning_rate": 3.500721220721056e-05,
      "loss": 0.6666,
      "step": 12630
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.956590163086714,
      "learning_rate": 3.499669984486239e-05,
      "loss": 0.6701,
      "step": 12635
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8584458801517105,
      "learning_rate": 3.49861853781042e-05,
      "loss": 0.7377,
      "step": 12640
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3546094304017708,
      "learning_rate": 3.4975668809149375e-05,
      "loss": 0.6653,
      "step": 12645
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9173080389352878,
      "learning_rate": 3.496515014021177e-05,
      "loss": 0.6029,
      "step": 12650
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7717861026546176,
      "learning_rate": 3.4954629373505675e-05,
      "loss": 0.5707,
      "step": 12655
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8941973636443095,
      "learning_rate": 3.494410651124583e-05,
      "loss": 0.6098,
      "step": 12660
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9301219672662374,
      "learning_rate": 3.4933581555647395e-05,
      "loss": 0.6382,
      "step": 12665
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9602475719939857,
      "learning_rate": 3.492305450892599e-05,
      "loss": 0.7317,
      "step": 12670
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8754327759681167,
      "learning_rate": 3.491252537329767e-05,
      "loss": 0.5679,
      "step": 12675
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8332519375768591,
      "learning_rate": 3.490199415097892e-05,
      "loss": 0.6359,
      "step": 12680
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0021545185332463,
      "learning_rate": 3.4891460844186686e-05,
      "loss": 0.6257,
      "step": 12685
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.6708521115211317,
      "learning_rate": 3.488092545513833e-05,
      "loss": 0.5801,
      "step": 12690
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9043971625533642,
      "learning_rate": 3.487038798605166e-05,
      "loss": 0.5934,
      "step": 12695
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9707006135387458,
      "learning_rate": 3.4859848439144935e-05,
      "loss": 0.6332,
      "step": 12700
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8671700202992485,
      "learning_rate": 3.4849306816636826e-05,
      "loss": 0.6898,
      "step": 12705
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.7868368116852693,
      "learning_rate": 3.4838763120746466e-05,
      "loss": 0.5487,
      "step": 12710
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9776706028230453,
      "learning_rate": 3.4828217353693406e-05,
      "loss": 0.6495,
      "step": 12715
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8105488864437738,
      "learning_rate": 3.4817669517697646e-05,
      "loss": 0.6229,
      "step": 12720
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8240676399476085,
      "learning_rate": 3.4807119614979604e-05,
      "loss": 0.5324,
      "step": 12725
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7860136215179412,
      "learning_rate": 3.479656764776017e-05,
      "loss": 0.6396,
      "step": 12730
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8547156450939054,
      "learning_rate": 3.4786013618260615e-05,
      "loss": 0.5825,
      "step": 12735
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8445157630854596,
      "learning_rate": 3.4775457528702694e-05,
      "loss": 0.6123,
      "step": 12740
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0213647765797331,
      "learning_rate": 3.476489938130855e-05,
      "loss": 0.61,
      "step": 12745
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8997373148241865,
      "learning_rate": 3.4754339178300815e-05,
      "loss": 0.5706,
      "step": 12750
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8349194461333793,
      "learning_rate": 3.4743776921902507e-05,
      "loss": 0.5866,
      "step": 12755
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0234969645557912,
      "learning_rate": 3.473321261433709e-05,
      "loss": 0.6119,
      "step": 12760
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0384233399407776,
      "learning_rate": 3.472264625782846e-05,
      "loss": 0.6104,
      "step": 12765
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9031259021096777,
      "learning_rate": 3.471207785460094e-05,
      "loss": 0.6553,
      "step": 12770
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9519468975360182,
      "learning_rate": 3.4701507406879316e-05,
      "loss": 0.6565,
      "step": 12775
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8750176568537622,
      "learning_rate": 3.4690934916888754e-05,
      "loss": 0.5794,
      "step": 12780
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9418170666749259,
      "learning_rate": 3.468036038685487e-05,
      "loss": 0.5943,
      "step": 12785
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9165266091499794,
      "learning_rate": 3.4669783819003744e-05,
      "loss": 0.6765,
      "step": 12790
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0415731301685915,
      "learning_rate": 3.465920521556181e-05,
      "loss": 0.5855,
      "step": 12795
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.918325395317344,
      "learning_rate": 3.4648624578756015e-05,
      "loss": 0.6199,
      "step": 12800
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8910389581495544,
      "learning_rate": 3.4638041910813676e-05,
      "loss": 0.6228,
      "step": 12805
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8550177516201493,
      "learning_rate": 3.462745721396256e-05,
      "loss": 0.5384,
      "step": 12810
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8784205415771655,
      "learning_rate": 3.461687049043084e-05,
      "loss": 0.8027,
      "step": 12815
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.873308321376839,
      "learning_rate": 3.460628174244717e-05,
      "loss": 0.5331,
      "step": 12820
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1242614668282056,
      "learning_rate": 3.459569097224054e-05,
      "loss": 0.6078,
      "step": 12825
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8942596638254138,
      "learning_rate": 3.4585098182040455e-05,
      "loss": 0.5706,
      "step": 12830
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8723261387454745,
      "learning_rate": 3.457450337407681e-05,
      "loss": 0.6101,
      "step": 12835
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8875896833951963,
      "learning_rate": 3.4563906550579906e-05,
      "loss": 0.6873,
      "step": 12840
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9253061229666167,
      "learning_rate": 3.4553307713780495e-05,
      "loss": 0.6052,
      "step": 12845
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9715378499755188,
      "learning_rate": 3.4542706865909734e-05,
      "loss": 0.6208,
      "step": 12850
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8408746971103221,
      "learning_rate": 3.4532104009199224e-05,
      "loss": 0.5636,
      "step": 12855
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9006259480128637,
      "learning_rate": 3.452149914588096e-05,
      "loss": 0.6393,
      "step": 12860
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8293184348962447,
      "learning_rate": 3.4510892278187406e-05,
      "loss": 0.676,
      "step": 12865
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8807055889725863,
      "learning_rate": 3.45002834083514e-05,
      "loss": 0.6504,
      "step": 12870
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0176297993998487,
      "learning_rate": 3.448967253860621e-05,
      "loss": 0.6295,
      "step": 12875
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8009997493787064,
      "learning_rate": 3.4479059671185554e-05,
      "loss": 0.6158,
      "step": 12880
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8560279805457361,
      "learning_rate": 3.446844480832354e-05,
      "loss": 0.66,
      "step": 12885
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.967252817282039,
      "learning_rate": 3.445782795225472e-05,
      "loss": 0.6953,
      "step": 12890
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1664043700480964,
      "learning_rate": 3.444720910521404e-05,
      "loss": 0.6085,
      "step": 12895
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9277619311992812,
      "learning_rate": 3.4436588269436887e-05,
      "loss": 0.6182,
      "step": 12900
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8382700027314575,
      "learning_rate": 3.442596544715906e-05,
      "loss": 0.5645,
      "step": 12905
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.7908477207082969,
      "learning_rate": 3.441534064061676e-05,
      "loss": 0.5883,
      "step": 12910
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8838387862921007,
      "learning_rate": 3.440471385204664e-05,
      "loss": 0.6224,
      "step": 12915
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1434419980146526,
      "learning_rate": 3.439408508368574e-05,
      "loss": 0.5898,
      "step": 12920
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0128506331169786,
      "learning_rate": 3.4383454337771526e-05,
      "loss": 0.6528,
      "step": 12925
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9618887083113228,
      "learning_rate": 3.437282161654188e-05,
      "loss": 0.6676,
      "step": 12930
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2541173464635784,
      "learning_rate": 3.43621869222351e-05,
      "loss": 0.6553,
      "step": 12935
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9366037361828637,
      "learning_rate": 3.435155025708992e-05,
      "loss": 0.6763,
      "step": 12940
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.943652775707637,
      "learning_rate": 3.4340911623345436e-05,
      "loss": 0.701,
      "step": 12945
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9108439836269759,
      "learning_rate": 3.433027102324122e-05,
      "loss": 0.6674,
      "step": 12950
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9898470852280427,
      "learning_rate": 3.4319628459017216e-05,
      "loss": 0.6382,
      "step": 12955
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9070685163297803,
      "learning_rate": 3.430898393291381e-05,
      "loss": 0.592,
      "step": 12960
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8759130030387231,
      "learning_rate": 3.4298337447171764e-05,
      "loss": 0.6295,
      "step": 12965
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7773532856823042,
      "learning_rate": 3.428768900403229e-05,
      "loss": 0.5401,
      "step": 12970
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8728781018623748,
      "learning_rate": 3.4277038605736986e-05,
      "loss": 0.7166,
      "step": 12975
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.813151309139467,
      "learning_rate": 3.426638625452789e-05,
      "loss": 0.5715,
      "step": 12980
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0215831652624239,
      "learning_rate": 3.425573195264742e-05,
      "loss": 0.555,
      "step": 12985
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8614008378157113,
      "learning_rate": 3.4245075702338415e-05,
      "loss": 0.695,
      "step": 12990
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.809144076015466,
      "learning_rate": 3.423441750584414e-05,
      "loss": 0.6254,
      "step": 12995
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8917639776273031,
      "learning_rate": 3.422375736540824e-05,
      "loss": 0.605,
      "step": 13000
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8995589927620352,
      "learning_rate": 3.4213095283274807e-05,
      "loss": 0.5897,
      "step": 13005
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8696231166601598,
      "learning_rate": 3.42024312616883e-05,
      "loss": 0.6283,
      "step": 13010
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0423269869481833,
      "learning_rate": 3.419176530289363e-05,
      "loss": 0.5981,
      "step": 13015
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8222009021178757,
      "learning_rate": 3.418109740913608e-05,
      "loss": 0.6892,
      "step": 13020
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8031873658234491,
      "learning_rate": 3.417042758266135e-05,
      "loss": 0.5313,
      "step": 13025
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7809903810806502,
      "learning_rate": 3.415975582571555e-05,
      "loss": 0.6371,
      "step": 13030
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8633127502310625,
      "learning_rate": 3.414908214054521e-05,
      "loss": 0.5565,
      "step": 13035
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9184847844091931,
      "learning_rate": 3.413840652939723e-05,
      "loss": 0.6381,
      "step": 13040
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1198374845791288,
      "learning_rate": 3.412772899451896e-05,
      "loss": 0.6213,
      "step": 13045
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8190319915825256,
      "learning_rate": 3.411704953815813e-05,
      "loss": 0.5723,
      "step": 13050
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.7995468855318312,
      "learning_rate": 3.410636816256286e-05,
      "loss": 0.625,
      "step": 13055
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9771296752971816,
      "learning_rate": 3.4095684869981715e-05,
      "loss": 0.5751,
      "step": 13060
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8837268572872942,
      "learning_rate": 3.408499966266362e-05,
      "loss": 0.6585,
      "step": 13065
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.84276977968608,
      "learning_rate": 3.4074312542857933e-05,
      "loss": 0.581,
      "step": 13070
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9201266801650666,
      "learning_rate": 3.40636235128144e-05,
      "loss": 0.5781,
      "step": 13075
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9520401434470717,
      "learning_rate": 3.405293257478319e-05,
      "loss": 0.6208,
      "step": 13080
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9102371922663296,
      "learning_rate": 3.4042239731014835e-05,
      "loss": 0.5991,
      "step": 13085
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9093704492946941,
      "learning_rate": 3.40315449837603e-05,
      "loss": 0.6248,
      "step": 13090
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8625920756576104,
      "learning_rate": 3.4020848335270944e-05,
      "loss": 0.5599,
      "step": 13095
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9254248842897794,
      "learning_rate": 3.401014978779852e-05,
      "loss": 0.6417,
      "step": 13100
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8380331488480008,
      "learning_rate": 3.399944934359518e-05,
      "loss": 0.6532,
      "step": 13105
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.946114144453383,
      "learning_rate": 3.3988747004913486e-05,
      "loss": 0.5843,
      "step": 13110
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9839002491888842,
      "learning_rate": 3.3978042774006395e-05,
      "loss": 0.594,
      "step": 13115
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9291678575353318,
      "learning_rate": 3.396733665312725e-05,
      "loss": 0.588,
      "step": 13120
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8915272873294703,
      "learning_rate": 3.3956628644529815e-05,
      "loss": 0.5793,
      "step": 13125
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8440532059398612,
      "learning_rate": 3.394591875046823e-05,
      "loss": 0.6231,
      "step": 13130
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.961022969741923,
      "learning_rate": 3.393520697319704e-05,
      "loss": 0.6165,
      "step": 13135
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.823843073127469,
      "learning_rate": 3.392449331497117e-05,
      "loss": 0.6481,
      "step": 13140
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0192337202739583,
      "learning_rate": 3.3913777778045994e-05,
      "loss": 0.6099,
      "step": 13145
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9025322981358163,
      "learning_rate": 3.3903060364677216e-05,
      "loss": 0.5629,
      "step": 13150
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8069845317325498,
      "learning_rate": 3.3892341077120974e-05,
      "loss": 0.5806,
      "step": 13155
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8124687021347363,
      "learning_rate": 3.388161991763379e-05,
      "loss": 0.6194,
      "step": 13160
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8875143177244339,
      "learning_rate": 3.387089688847258e-05,
      "loss": 0.6241,
      "step": 13165
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8359666089663408,
      "learning_rate": 3.386017199189465e-05,
      "loss": 0.6022,
      "step": 13170
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.858812621309702,
      "learning_rate": 3.384944523015771e-05,
      "loss": 0.5816,
      "step": 13175
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8896546849269568,
      "learning_rate": 3.383871660551985e-05,
      "loss": 0.5934,
      "step": 13180
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8647229392788754,
      "learning_rate": 3.3827986120239556e-05,
      "loss": 0.6686,
      "step": 13185
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.793345125802617,
      "learning_rate": 3.3817253776575714e-05,
      "loss": 0.5707,
      "step": 13190
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9480051849601898,
      "learning_rate": 3.38065195767876e-05,
      "loss": 0.6255,
      "step": 13195
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9191963606043647,
      "learning_rate": 3.379578352313485e-05,
      "loss": 0.59,
      "step": 13200
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9472888560047468,
      "learning_rate": 3.378504561787755e-05,
      "loss": 0.6342,
      "step": 13205
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8332083714559967,
      "learning_rate": 3.377430586327612e-05,
      "loss": 0.6058,
      "step": 13210
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.968968442777554,
      "learning_rate": 3.3763564261591385e-05,
      "loss": 0.6411,
      "step": 13215
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4511436808150782,
      "learning_rate": 3.375282081508459e-05,
      "loss": 0.5874,
      "step": 13220
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9868514432409948,
      "learning_rate": 3.3742075526017315e-05,
      "loss": 0.6434,
      "step": 13225
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2549177302774823,
      "learning_rate": 3.373132839665159e-05,
      "loss": 0.5901,
      "step": 13230
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.896072051615042,
      "learning_rate": 3.3720579429249755e-05,
      "loss": 0.5308,
      "step": 13235
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.986331797112231,
      "learning_rate": 3.3709828626074615e-05,
      "loss": 0.6494,
      "step": 13240
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.88956079434755,
      "learning_rate": 3.3699075989389305e-05,
      "loss": 0.6446,
      "step": 13245
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8432055006500154,
      "learning_rate": 3.368832152145738e-05,
      "loss": 0.6167,
      "step": 13250
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9007854163305742,
      "learning_rate": 3.367756522454277e-05,
      "loss": 0.6881,
      "step": 13255
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9950203619990622,
      "learning_rate": 3.3666807100909786e-05,
      "loss": 0.7618,
      "step": 13260
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8381708520680272,
      "learning_rate": 3.365604715282312e-05,
      "loss": 0.5727,
      "step": 13265
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9640117070742609,
      "learning_rate": 3.3645285382547855e-05,
      "loss": 0.5688,
      "step": 13270
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8211205577874293,
      "learning_rate": 3.363452179234946e-05,
      "loss": 0.5974,
      "step": 13275
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8357480242027399,
      "learning_rate": 3.362375638449378e-05,
      "loss": 0.5995,
      "step": 13280
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9715230295484729,
      "learning_rate": 3.361298916124704e-05,
      "loss": 0.5996,
      "step": 13285
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.7898587202706426,
      "learning_rate": 3.3602220124875875e-05,
      "loss": 0.6117,
      "step": 13290
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8818336958131766,
      "learning_rate": 3.359144927764727e-05,
      "loss": 0.6414,
      "step": 13295
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8275981280284921,
      "learning_rate": 3.358067662182859e-05,
      "loss": 0.6321,
      "step": 13300
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8827928716298319,
      "learning_rate": 3.3569902159687595e-05,
      "loss": 0.6043,
      "step": 13305
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9311754224703687,
      "learning_rate": 3.355912589349243e-05,
      "loss": 0.677,
      "step": 13310
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7913626804480359,
      "learning_rate": 3.35483478255116e-05,
      "loss": 0.5952,
      "step": 13315
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9396380034284258,
      "learning_rate": 3.353756795801402e-05,
      "loss": 0.5573,
      "step": 13320
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9239927191443369,
      "learning_rate": 3.3526786293268957e-05,
      "loss": 0.5941,
      "step": 13325
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9794809409295886,
      "learning_rate": 3.3516002833546056e-05,
      "loss": 0.6008,
      "step": 13330
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9225548682353489,
      "learning_rate": 3.3505217581115355e-05,
      "loss": 0.6832,
      "step": 13335
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9882205802588672,
      "learning_rate": 3.3494430538247253e-05,
      "loss": 0.6587,
      "step": 13340
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9719619059808527,
      "learning_rate": 3.3483641707212566e-05,
      "loss": 0.7756,
      "step": 13345
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1892106557209925,
      "learning_rate": 3.347285109028241e-05,
      "loss": 0.6402,
      "step": 13350
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9873130543692576,
      "learning_rate": 3.346205868972836e-05,
      "loss": 0.6866,
      "step": 13355
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0336494547972308,
      "learning_rate": 3.345126450782232e-05,
      "loss": 0.6036,
      "step": 13360
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8457080948817928,
      "learning_rate": 3.344046854683656e-05,
      "loss": 0.5957,
      "step": 13365
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1027643454746756,
      "learning_rate": 3.342967080904377e-05,
      "loss": 0.5772,
      "step": 13370
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8562088252487238,
      "learning_rate": 3.3418871296716977e-05,
      "loss": 0.6005,
      "step": 13375
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9466605641458102,
      "learning_rate": 3.340807001212958e-05,
      "loss": 0.6073,
      "step": 13380
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8965644460808967,
      "learning_rate": 3.339726695755537e-05,
      "loss": 0.6204,
      "step": 13385
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1096286163849778,
      "learning_rate": 3.338646213526852e-05,
      "loss": 0.635,
      "step": 13390
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8754853667874716,
      "learning_rate": 3.337565554754353e-05,
      "loss": 0.6358,
      "step": 13395
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9110482614717333,
      "learning_rate": 3.3364847196655305e-05,
      "loss": 0.6214,
      "step": 13400
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.7926845633276647,
      "learning_rate": 3.335403708487914e-05,
      "loss": 0.6243,
      "step": 13405
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9078940820337732,
      "learning_rate": 3.334322521449066e-05,
      "loss": 0.6026,
      "step": 13410
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9372460308302087,
      "learning_rate": 3.333241158776587e-05,
      "loss": 0.6094,
      "step": 13415
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9564483086346293,
      "learning_rate": 3.332159620698116e-05,
      "loss": 0.7157,
      "step": 13420
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9229741855777024,
      "learning_rate": 3.331077907441328e-05,
      "loss": 0.564,
      "step": 13425
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9769203270681543,
      "learning_rate": 3.3299960192339356e-05,
      "loss": 0.6285,
      "step": 13430
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.015604600610834,
      "learning_rate": 3.328913956303687e-05,
      "loss": 0.6421,
      "step": 13435
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9225261875128974,
      "learning_rate": 3.3278317188783684e-05,
      "loss": 0.6393,
      "step": 13440
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5001861035157056,
      "learning_rate": 3.3267493071858e-05,
      "loss": 0.6498,
      "step": 13445
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8612643013746522,
      "learning_rate": 3.325666721453843e-05,
      "loss": 0.6828,
      "step": 13450
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0309589730596793,
      "learning_rate": 3.3245839619103916e-05,
      "loss": 0.6204,
      "step": 13455
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8201016813229378,
      "learning_rate": 3.323501028783379e-05,
      "loss": 0.6237,
      "step": 13460
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.13957645634541,
      "learning_rate": 3.322417922300775e-05,
      "loss": 0.631,
      "step": 13465
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8453239478004159,
      "learning_rate": 3.321334642690582e-05,
      "loss": 0.6454,
      "step": 13470
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9482490037378013,
      "learning_rate": 3.320251190180845e-05,
      "loss": 0.6024,
      "step": 13475
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8549568743141279,
      "learning_rate": 3.319167564999639e-05,
      "loss": 0.6375,
      "step": 13480
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9089922574390278,
      "learning_rate": 3.318083767375081e-05,
      "loss": 0.6104,
      "step": 13485
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9671969586531175,
      "learning_rate": 3.3169997975353195e-05,
      "loss": 0.5837,
      "step": 13490
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9829591111528554,
      "learning_rate": 3.315915655708543e-05,
      "loss": 0.599,
      "step": 13495
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9878969304179483,
      "learning_rate": 3.314831342122974e-05,
      "loss": 0.6393,
      "step": 13500
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8643449190320925,
      "learning_rate": 3.313746857006873e-05,
      "loss": 0.63,
      "step": 13505
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9417188349940933,
      "learning_rate": 3.312662200588534e-05,
      "loss": 0.6288,
      "step": 13510
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.130525033006468,
      "learning_rate": 3.3115773730962896e-05,
      "loss": 0.5918,
      "step": 13515
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9404647424053684,
      "learning_rate": 3.310492374758507e-05,
      "loss": 0.6873,
      "step": 13520
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9975615755136265,
      "learning_rate": 3.3094072058035894e-05,
      "loss": 0.6591,
      "step": 13525
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8774444779428315,
      "learning_rate": 3.3083218664599755e-05,
      "loss": 0.6845,
      "step": 13530
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9092239097967811,
      "learning_rate": 3.307236356956144e-05,
      "loss": 0.62,
      "step": 13535
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8609363482888034,
      "learning_rate": 3.306150677520602e-05,
      "loss": 0.604,
      "step": 13540
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8140700714246218,
      "learning_rate": 3.3050648283818985e-05,
      "loss": 0.5633,
      "step": 13545
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0347541611333066,
      "learning_rate": 3.303978809768615e-05,
      "loss": 0.5718,
      "step": 13550
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0125114628229028,
      "learning_rate": 3.3028926219093714e-05,
      "loss": 0.578,
      "step": 13555
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8975432257995927,
      "learning_rate": 3.30180626503282e-05,
      "loss": 0.5827,
      "step": 13560
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.843827212748594,
      "learning_rate": 3.30071973936765e-05,
      "loss": 0.6177,
      "step": 13565
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.784180156204481,
      "learning_rate": 3.299633045142589e-05,
      "loss": 0.5826,
      "step": 13570
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9275241980706935,
      "learning_rate": 3.298546182586396e-05,
      "loss": 0.6339,
      "step": 13575
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9040153722117333,
      "learning_rate": 3.297459151927866e-05,
      "loss": 0.6973,
      "step": 13580
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9661519331134533,
      "learning_rate": 3.296371953395831e-05,
      "loss": 0.6597,
      "step": 13585
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.910274082636293,
      "learning_rate": 3.295284587219159e-05,
      "loss": 0.6241,
      "step": 13590
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8992370130049528,
      "learning_rate": 3.29419705362675e-05,
      "loss": 0.6826,
      "step": 13595
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.845200766365333,
      "learning_rate": 3.293109352847542e-05,
      "loss": 0.5987,
      "step": 13600
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0041001619269916,
      "learning_rate": 3.292021485110508e-05,
      "loss": 0.5825,
      "step": 13605
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8766907421245957,
      "learning_rate": 3.290933450644654e-05,
      "loss": 0.6311,
      "step": 13610
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8960473453380331,
      "learning_rate": 3.289845249679025e-05,
      "loss": 0.6706,
      "step": 13615
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9549203402504854,
      "learning_rate": 3.2887568824426965e-05,
      "loss": 0.6411,
      "step": 13620
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8567389877383453,
      "learning_rate": 3.287668349164783e-05,
      "loss": 0.5632,
      "step": 13625
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8035215975175422,
      "learning_rate": 3.28657965007443e-05,
      "loss": 0.6211,
      "step": 13630
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8128657085497275,
      "learning_rate": 3.285490785400822e-05,
      "loss": 0.6213,
      "step": 13635
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8270049212790569,
      "learning_rate": 3.284401755373176e-05,
      "loss": 0.5984,
      "step": 13640
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0343263940447744,
      "learning_rate": 3.283312560220743e-05,
      "loss": 0.6213,
      "step": 13645
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8168778107678732,
      "learning_rate": 3.2822232001728116e-05,
      "loss": 0.592,
      "step": 13650
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8367169041139833,
      "learning_rate": 3.2811336754587027e-05,
      "loss": 0.6354,
      "step": 13655
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.877453803181334,
      "learning_rate": 3.280043986307773e-05,
      "loss": 0.5926,
      "step": 13660
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9567992539893181,
      "learning_rate": 3.278954132949413e-05,
      "loss": 0.6709,
      "step": 13665
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9440890984372244,
      "learning_rate": 3.2778641156130483e-05,
      "loss": 0.6311,
      "step": 13670
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8550152848081506,
      "learning_rate": 3.27677393452814e-05,
      "loss": 0.5263,
      "step": 13675
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0679190497601727,
      "learning_rate": 3.275683589924181e-05,
      "loss": 0.6034,
      "step": 13680
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9056776230882122,
      "learning_rate": 3.274593082030702e-05,
      "loss": 0.5819,
      "step": 13685
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8613565987259736,
      "learning_rate": 3.273502411077265e-05,
      "loss": 0.6402,
      "step": 13690
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4131119352517771,
      "learning_rate": 3.272411577293468e-05,
      "loss": 0.5346,
      "step": 13695
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.841056332330446,
      "learning_rate": 3.271320580908943e-05,
      "loss": 0.5775,
      "step": 13700
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9991608484383396,
      "learning_rate": 3.270229422153356e-05,
      "loss": 0.6306,
      "step": 13705
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8252817791685058,
      "learning_rate": 3.269138101256408e-05,
      "loss": 0.602,
      "step": 13710
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8557066814007601,
      "learning_rate": 3.268046618447833e-05,
      "loss": 0.5983,
      "step": 13715
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.342185329347312,
      "learning_rate": 3.2669549739573994e-05,
      "loss": 0.6566,
      "step": 13720
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9293684467045353,
      "learning_rate": 3.26586316801491e-05,
      "loss": 0.551,
      "step": 13725
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9881476411545929,
      "learning_rate": 3.264771200850201e-05,
      "loss": 0.611,
      "step": 13730
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.892419704640757,
      "learning_rate": 3.2636790726931446e-05,
      "loss": 0.557,
      "step": 13735
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9388799630899802,
      "learning_rate": 3.262586783773642e-05,
      "loss": 0.6487,
      "step": 13740
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8817809748299225,
      "learning_rate": 3.261494334321635e-05,
      "loss": 0.6186,
      "step": 13745
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8421088804070335,
      "learning_rate": 3.260401724567094e-05,
      "loss": 0.63,
      "step": 13750
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.992382878836625,
      "learning_rate": 3.2593089547400234e-05,
      "loss": 0.6044,
      "step": 13755
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8855943216062995,
      "learning_rate": 3.2582160250704654e-05,
      "loss": 0.6103,
      "step": 13760
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8640879874274411,
      "learning_rate": 3.2571229357884915e-05,
      "loss": 0.6113,
      "step": 13765
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8737568180192352,
      "learning_rate": 3.256029687124209e-05,
      "loss": 0.6214,
      "step": 13770
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8365702078982379,
      "learning_rate": 3.254936279307757e-05,
      "loss": 0.6057,
      "step": 13775
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1079162458987608,
      "learning_rate": 3.253842712569312e-05,
      "loss": 0.6181,
      "step": 13780
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8169920397637487,
      "learning_rate": 3.252748987139078e-05,
      "loss": 0.6243,
      "step": 13785
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9998227717889913,
      "learning_rate": 3.251655103247298e-05,
      "loss": 0.6202,
      "step": 13790
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8792378919698017,
      "learning_rate": 3.250561061124245e-05,
      "loss": 0.642,
      "step": 13795
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8700829569105526,
      "learning_rate": 3.249466861000227e-05,
      "loss": 0.5613,
      "step": 13800
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8586101363117356,
      "learning_rate": 3.248372503105584e-05,
      "loss": 0.5551,
      "step": 13805
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8853985117030423,
      "learning_rate": 3.247277987670689e-05,
      "loss": 0.6033,
      "step": 13810
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8683083160803056,
      "learning_rate": 3.2461833149259516e-05,
      "loss": 0.5892,
      "step": 13815
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8949843862183975,
      "learning_rate": 3.2450884851018096e-05,
      "loss": 0.6292,
      "step": 13820
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9356885992030335,
      "learning_rate": 3.243993498428736e-05,
      "loss": 0.6513,
      "step": 13825
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8497374441324735,
      "learning_rate": 3.2428983551372374e-05,
      "loss": 0.6926,
      "step": 13830
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9157864165593556,
      "learning_rate": 3.241803055457855e-05,
      "loss": 0.6408,
      "step": 13835
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.001275078546633,
      "learning_rate": 3.240707599621158e-05,
      "loss": 0.5921,
      "step": 13840
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9215544155311559,
      "learning_rate": 3.2396119878577516e-05,
      "loss": 0.6493,
      "step": 13845
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.074733105041847,
      "learning_rate": 3.2385162203982765e-05,
      "loss": 0.6362,
      "step": 13850
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9492895870678364,
      "learning_rate": 3.237420297473399e-05,
      "loss": 0.5948,
      "step": 13855
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8169730446674744,
      "learning_rate": 3.236324219313826e-05,
      "loss": 0.6016,
      "step": 13860
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8708981236440343,
      "learning_rate": 3.2352279861502914e-05,
      "loss": 0.6146,
      "step": 13865
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8690444518168777,
      "learning_rate": 3.234131598213565e-05,
      "loss": 0.6295,
      "step": 13870
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8449880813128703,
      "learning_rate": 3.2330350557344455e-05,
      "loss": 0.6827,
      "step": 13875
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9335719888247228,
      "learning_rate": 3.231938358943769e-05,
      "loss": 0.6576,
      "step": 13880
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9016478062448239,
      "learning_rate": 3.2308415080724013e-05,
      "loss": 0.6144,
      "step": 13885
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9194336026418829,
      "learning_rate": 3.22974450335124e-05,
      "loss": 0.5881,
      "step": 13890
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.878014195494255,
      "learning_rate": 3.228647345011217e-05,
      "loss": 0.6076,
      "step": 13895
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8918343693158489,
      "learning_rate": 3.227550033283296e-05,
      "loss": 0.6161,
      "step": 13900
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.912385493336373,
      "learning_rate": 3.226452568398471e-05,
      "loss": 0.5138,
      "step": 13905
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8770431703359535,
      "learning_rate": 3.2253549505877715e-05,
      "loss": 0.6793,
      "step": 13910
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9705169824665427,
      "learning_rate": 3.2242571800822566e-05,
      "loss": 0.6704,
      "step": 13915
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9245329539519085,
      "learning_rate": 3.2231592571130196e-05,
      "loss": 0.6402,
      "step": 13920
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8747549649322257,
      "learning_rate": 3.2220611819111825e-05,
      "loss": 0.6515,
      "step": 13925
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.955157692671812,
      "learning_rate": 3.220962954707904e-05,
      "loss": 0.5855,
      "step": 13930
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8090221885265747,
      "learning_rate": 3.2198645757343715e-05,
      "loss": 0.6126,
      "step": 13935
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8299315608941946,
      "learning_rate": 3.218766045221805e-05,
      "loss": 0.6128,
      "step": 13940
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.873059829444481,
      "learning_rate": 3.217667363401456e-05,
      "loss": 0.6347,
      "step": 13945
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9092923994287389,
      "learning_rate": 3.216568530504611e-05,
      "loss": 0.6872,
      "step": 13950
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9338280320274359,
      "learning_rate": 3.215469546762584e-05,
      "loss": 0.6335,
      "step": 13955
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9477258003267339,
      "learning_rate": 3.214370412406723e-05,
      "loss": 0.6522,
      "step": 13960
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.7675122204643133,
      "learning_rate": 3.213271127668407e-05,
      "loss": 0.5975,
      "step": 13965
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9519957274619302,
      "learning_rate": 3.2121716927790476e-05,
      "loss": 0.5986,
      "step": 13970
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8993298996098514,
      "learning_rate": 3.211072107970086e-05,
      "loss": 0.6427,
      "step": 13975
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9692459849507432,
      "learning_rate": 3.209972373472998e-05,
      "loss": 0.615,
      "step": 13980
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9036038616248692,
      "learning_rate": 3.208872489519288e-05,
      "loss": 0.6085,
      "step": 13985
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9205927195955137,
      "learning_rate": 3.2077724563404946e-05,
      "loss": 0.6168,
      "step": 13990
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9840479366969124,
      "learning_rate": 3.2066722741681845e-05,
      "loss": 0.5396,
      "step": 13995
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8869098712612347,
      "learning_rate": 3.205571943233958e-05,
      "loss": 0.6423,
      "step": 14000
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8351403542328356,
      "learning_rate": 3.204471463769446e-05,
      "loss": 0.6563,
      "step": 14005
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8635962295641016,
      "learning_rate": 3.203370836006313e-05,
      "loss": 0.6148,
      "step": 14010
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8716443794972157,
      "learning_rate": 3.202270060176251e-05,
      "loss": 0.6199,
      "step": 14015
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7892797204684883,
      "learning_rate": 3.2011691365109844e-05,
      "loss": 0.5494,
      "step": 14020
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9138700219680662,
      "learning_rate": 3.200068065242271e-05,
      "loss": 0.6126,
      "step": 14025
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9477985254868797,
      "learning_rate": 3.198966846601896e-05,
      "loss": 0.6185,
      "step": 14030
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8768411577126559,
      "learning_rate": 3.197865480821678e-05,
      "loss": 0.5963,
      "step": 14035
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9538878594814884,
      "learning_rate": 3.196763968133466e-05,
      "loss": 0.5955,
      "step": 14040
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8485820899508701,
      "learning_rate": 3.195662308769142e-05,
      "loss": 0.5846,
      "step": 14045
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.7415826065087051,
      "learning_rate": 3.1945605029606134e-05,
      "loss": 0.6067,
      "step": 14050
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9631237404986648,
      "learning_rate": 3.193458550939823e-05,
      "loss": 0.6148,
      "step": 14055
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9579848931236165,
      "learning_rate": 3.192356452938745e-05,
      "loss": 0.6245,
      "step": 14060
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8494664317962217,
      "learning_rate": 3.19125420918938e-05,
      "loss": 0.6314,
      "step": 14065
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9093549111531063,
      "learning_rate": 3.190151819923765e-05,
      "loss": 0.6169,
      "step": 14070
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8360978546924288,
      "learning_rate": 3.1890492853739606e-05,
      "loss": 0.6328,
      "step": 14075
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9358108501322423,
      "learning_rate": 3.187946605772065e-05,
      "loss": 0.6187,
      "step": 14080
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8165968857856247,
      "learning_rate": 3.1868437813502026e-05,
      "loss": 0.651,
      "step": 14085
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8804855307176693,
      "learning_rate": 3.1857408123405293e-05,
      "loss": 0.6552,
      "step": 14090
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9219179133563361,
      "learning_rate": 3.184637698975232e-05,
      "loss": 0.6027,
      "step": 14095
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.934007970502787,
      "learning_rate": 3.1835344414865276e-05,
      "loss": 0.5353,
      "step": 14100
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.006284219952787,
      "learning_rate": 3.182431040106664e-05,
      "loss": 0.5708,
      "step": 14105
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.884477928120637,
      "learning_rate": 3.181327495067917e-05,
      "loss": 0.6436,
      "step": 14110
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9372121306421645,
      "learning_rate": 3.1802238066025956e-05,
      "loss": 0.6088,
      "step": 14115
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8290587024866012,
      "learning_rate": 3.1791199749430376e-05,
      "loss": 0.615,
      "step": 14120
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9088374548664306,
      "learning_rate": 3.178016000321612e-05,
      "loss": 0.628,
      "step": 14125
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8984696982000056,
      "learning_rate": 3.1769118829707156e-05,
      "loss": 0.5717,
      "step": 14130
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8520665306169647,
      "learning_rate": 3.175807623122777e-05,
      "loss": 0.6085,
      "step": 14135
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1054824102563734,
      "learning_rate": 3.1747032210102545e-05,
      "loss": 0.6174,
      "step": 14140
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8392065814313736,
      "learning_rate": 3.173598676865637e-05,
      "loss": 0.5699,
      "step": 14145
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.954578556851808,
      "learning_rate": 3.172493990921442e-05,
      "loss": 0.6186,
      "step": 14150
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8854617704570861,
      "learning_rate": 3.1713891634102176e-05,
      "loss": 0.6094,
      "step": 14155
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9433608289412669,
      "learning_rate": 3.170284194564542e-05,
      "loss": 0.6131,
      "step": 14160
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6056532075501486,
      "learning_rate": 3.1691790846170213e-05,
      "loss": 0.5993,
      "step": 14165
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0071999129562839,
      "learning_rate": 3.168073833800295e-05,
      "loss": 0.5685,
      "step": 14170
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8690295486427226,
      "learning_rate": 3.1669684423470275e-05,
      "loss": 0.5967,
      "step": 14175
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9742190459209951,
      "learning_rate": 3.165862910489917e-05,
      "loss": 0.6288,
      "step": 14180
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9068261467518752,
      "learning_rate": 3.164757238461689e-05,
      "loss": 0.6178,
      "step": 14185
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8753917139112746,
      "learning_rate": 3.1636514264950984e-05,
      "loss": 0.5892,
      "step": 14190
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8217207675729438,
      "learning_rate": 3.162545474822931e-05,
      "loss": 0.6106,
      "step": 14195
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9925949137916559,
      "learning_rate": 3.161439383678002e-05,
      "loss": 0.5905,
      "step": 14200
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8529060034107014,
      "learning_rate": 3.160333153293153e-05,
      "loss": 0.5647,
      "step": 14205
    },
    {
      "epoch": 1.66,
      "grad_norm": 127.03419464098543,
      "learning_rate": 3.15922678390126e-05,
      "loss": 0.7069,
      "step": 14210
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9821132542654768,
      "learning_rate": 3.158120275735221e-05,
      "loss": 0.6089,
      "step": 14215
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9239417418913255,
      "learning_rate": 3.157013629027972e-05,
      "loss": 0.6066,
      "step": 14220
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.934742165654382,
      "learning_rate": 3.155906844012472e-05,
      "loss": 0.6238,
      "step": 14225
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.810709193067882,
      "learning_rate": 3.15479992092171e-05,
      "loss": 0.6407,
      "step": 14230
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.897326263443791,
      "learning_rate": 3.153692859988706e-05,
      "loss": 0.5537,
      "step": 14235
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9700000544479225,
      "learning_rate": 3.152585661446508e-05,
      "loss": 0.6122,
      "step": 14240
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8670501614316243,
      "learning_rate": 3.151478325528192e-05,
      "loss": 0.6191,
      "step": 14245
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.7940522098542911,
      "learning_rate": 3.150370852466864e-05,
      "loss": 0.6351,
      "step": 14250
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8885844509213725,
      "learning_rate": 3.149263242495659e-05,
      "loss": 0.668,
      "step": 14255
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8031221834830546,
      "learning_rate": 3.148155495847742e-05,
      "loss": 0.5816,
      "step": 14260
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7719612473851699,
      "learning_rate": 3.147047612756302e-05,
      "loss": 0.5369,
      "step": 14265
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8490951241560902,
      "learning_rate": 3.1459395934545625e-05,
      "loss": 0.5952,
      "step": 14270
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9392766212076059,
      "learning_rate": 3.144831438175773e-05,
      "loss": 0.6449,
      "step": 14275
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8148533343490466,
      "learning_rate": 3.1437231471532104e-05,
      "loss": 0.6107,
      "step": 14280
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8364526118435871,
      "learning_rate": 3.142614720620182e-05,
      "loss": 0.6211,
      "step": 14285
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1214809028260668,
      "learning_rate": 3.141506158810024e-05,
      "loss": 0.5685,
      "step": 14290
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7782368941113667,
      "learning_rate": 3.1403974619561e-05,
      "loss": 0.6161,
      "step": 14295
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0546549739796423,
      "learning_rate": 3.1392886302918014e-05,
      "loss": 0.6173,
      "step": 14300
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9426567326441045,
      "learning_rate": 3.13817966405055e-05,
      "loss": 0.504,
      "step": 14305
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.7970881502192838,
      "learning_rate": 3.137070563465796e-05,
      "loss": 0.5922,
      "step": 14310
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8719743927258898,
      "learning_rate": 3.1359613287710135e-05,
      "loss": 0.5776,
      "step": 14315
    },
    {
      "epoch": 1.67,
      "grad_norm": 20.38565106905542,
      "learning_rate": 3.1348519601997095e-05,
      "loss": 0.6133,
      "step": 14320
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8797446176958362,
      "learning_rate": 3.133742457985418e-05,
      "loss": 0.5924,
      "step": 14325
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9586506553579648,
      "learning_rate": 3.132632822361701e-05,
      "loss": 0.6395,
      "step": 14330
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9052415787453856,
      "learning_rate": 3.1315230535621484e-05,
      "loss": 0.617,
      "step": 14335
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1641799510743467,
      "learning_rate": 3.1304131518203775e-05,
      "loss": 0.5867,
      "step": 14340
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9839868657007339,
      "learning_rate": 3.129303117370035e-05,
      "loss": 0.5656,
      "step": 14345
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.7945105450408615,
      "learning_rate": 3.128192950444794e-05,
      "loss": 0.5594,
      "step": 14350
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0125987081997354,
      "learning_rate": 3.127082651278357e-05,
      "loss": 0.5053,
      "step": 14355
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8633232910847957,
      "learning_rate": 3.1259722201044525e-05,
      "loss": 0.5713,
      "step": 14360
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9323865904612294,
      "learning_rate": 3.1248616571568395e-05,
      "loss": 0.6163,
      "step": 14365
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8918135421630677,
      "learning_rate": 3.123750962669301e-05,
      "loss": 0.6129,
      "step": 14370
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9172412216819962,
      "learning_rate": 3.122640136875653e-05,
      "loss": 0.6566,
      "step": 14375
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8561319207567425,
      "learning_rate": 3.121529180009733e-05,
      "loss": 0.6253,
      "step": 14380
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1558106866902784,
      "learning_rate": 3.120418092305411e-05,
      "loss": 0.6558,
      "step": 14385
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8920430044707267,
      "learning_rate": 3.119306873996581e-05,
      "loss": 0.5897,
      "step": 14390
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8333270672819031,
      "learning_rate": 3.118195525317166e-05,
      "loss": 0.5438,
      "step": 14395
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9584031754893765,
      "learning_rate": 3.117084046501119e-05,
      "loss": 0.6511,
      "step": 14400
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9936939750252267,
      "learning_rate": 3.1159724377824154e-05,
      "loss": 0.6499,
      "step": 14405
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8987091203563597,
      "learning_rate": 3.1148606993950614e-05,
      "loss": 0.5655,
      "step": 14410
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9136561564369415,
      "learning_rate": 3.113748831573089e-05,
      "loss": 0.5939,
      "step": 14415
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8973206809682199,
      "learning_rate": 3.11263683455056e-05,
      "loss": 0.5344,
      "step": 14420
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8867024741552035,
      "learning_rate": 3.1115247085615595e-05,
      "loss": 0.6523,
      "step": 14425
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.854185379361725,
      "learning_rate": 3.110412453840201e-05,
      "loss": 0.6022,
      "step": 14430
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7683978766451767,
      "learning_rate": 3.109300070620629e-05,
      "loss": 0.6511,
      "step": 14435
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8675161607514147,
      "learning_rate": 3.108187559137009e-05,
      "loss": 0.6076,
      "step": 14440
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7799821783691376,
      "learning_rate": 3.107074919623536e-05,
      "loss": 0.557,
      "step": 14445
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7907027951675817,
      "learning_rate": 3.105962152314435e-05,
      "loss": 0.6054,
      "step": 14450
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.02855111704502,
      "learning_rate": 3.104849257443953e-05,
      "loss": 0.6666,
      "step": 14455
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8173704925195387,
      "learning_rate": 3.1037362352463655e-05,
      "loss": 0.618,
      "step": 14460
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8963121320619791,
      "learning_rate": 3.102623085955977e-05,
      "loss": 0.7052,
      "step": 14465
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8587133538749553,
      "learning_rate": 3.101509809807117e-05,
      "loss": 0.6273,
      "step": 14470
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9192409451677703,
      "learning_rate": 3.10039640703414e-05,
      "loss": 0.6083,
      "step": 14475
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8778016404178711,
      "learning_rate": 3.0992828778714315e-05,
      "loss": 0.5685,
      "step": 14480
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.7979767553828838,
      "learning_rate": 3.0981692225533994e-05,
      "loss": 0.6405,
      "step": 14485
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8646106636423382,
      "learning_rate": 3.0970554413144805e-05,
      "loss": 0.6231,
      "step": 14490
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8900011275383782,
      "learning_rate": 3.095941534389136e-05,
      "loss": 0.6105,
      "step": 14495
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8290283065600704,
      "learning_rate": 3.0948275020118565e-05,
      "loss": 0.5956,
      "step": 14500
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8486155537079948,
      "learning_rate": 3.0937133444171585e-05,
      "loss": 0.6209,
      "step": 14505
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8209584881424392,
      "learning_rate": 3.0925990618395805e-05,
      "loss": 0.6764,
      "step": 14510
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.804981920476081,
      "learning_rate": 3.091484654513694e-05,
      "loss": 0.6399,
      "step": 14515
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9099640077272081,
      "learning_rate": 3.090370122674091e-05,
      "loss": 0.6057,
      "step": 14520
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2539892671479949,
      "learning_rate": 3.089255466555393e-05,
      "loss": 0.5685,
      "step": 14525
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8395571770173283,
      "learning_rate": 3.0881406863922474e-05,
      "loss": 0.568,
      "step": 14530
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8635041366718217,
      "learning_rate": 3.0870257824193263e-05,
      "loss": 0.5814,
      "step": 14535
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8986187790897435,
      "learning_rate": 3.085910754871329e-05,
      "loss": 0.6478,
      "step": 14540
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9041831731647924,
      "learning_rate": 3.0847956039829806e-05,
      "loss": 0.5643,
      "step": 14545
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8986964562669946,
      "learning_rate": 3.083680329989032e-05,
      "loss": 0.6201,
      "step": 14550
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7310458108687056,
      "learning_rate": 3.08256493312426e-05,
      "loss": 0.5765,
      "step": 14555
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8885560066656397,
      "learning_rate": 3.081449413623467e-05,
      "loss": 0.6367,
      "step": 14560
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2048526748579056,
      "learning_rate": 3.080333771721482e-05,
      "loss": 0.6058,
      "step": 14565
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.90271581557056,
      "learning_rate": 3.0792180076531594e-05,
      "loss": 0.6091,
      "step": 14570
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.811745238749785,
      "learning_rate": 3.078102121653378e-05,
      "loss": 0.6198,
      "step": 14575
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9053030017971876,
      "learning_rate": 3.076986113957044e-05,
      "loss": 0.5239,
      "step": 14580
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9801263041516508,
      "learning_rate": 3.075869984799091e-05,
      "loss": 0.572,
      "step": 14585
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.7736077071552256,
      "learning_rate": 3.074753734414473e-05,
      "loss": 0.5858,
      "step": 14590
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8937696389272364,
      "learning_rate": 3.073637363038173e-05,
      "loss": 0.5987,
      "step": 14595
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9102794810097287,
      "learning_rate": 3.0725208709051995e-05,
      "loss": 0.6691,
      "step": 14600
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9004238567524465,
      "learning_rate": 3.071404258250586e-05,
      "loss": 0.6766,
      "step": 14605
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8813464989942501,
      "learning_rate": 3.070287525309391e-05,
      "loss": 0.5912,
      "step": 14610
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0110013115712124,
      "learning_rate": 3.069170672316697e-05,
      "loss": 0.6944,
      "step": 14615
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9278201852419052,
      "learning_rate": 3.068053699507616e-05,
      "loss": 0.5681,
      "step": 14620
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.7648483584981083,
      "learning_rate": 3.066936607117279e-05,
      "loss": 0.5743,
      "step": 14625
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9574087744313496,
      "learning_rate": 3.0658193953808496e-05,
      "loss": 0.6013,
      "step": 14630
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8353107589488827,
      "learning_rate": 3.06470206453351e-05,
      "loss": 0.5471,
      "step": 14635
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8872461468913105,
      "learning_rate": 3.0635846148104695e-05,
      "loss": 0.6327,
      "step": 14640
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9061540783556211,
      "learning_rate": 3.062467046446965e-05,
      "loss": 0.5713,
      "step": 14645
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8305816397556461,
      "learning_rate": 3.061349359678255e-05,
      "loss": 0.631,
      "step": 14650
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8974785947761166,
      "learning_rate": 3.060231554739625e-05,
      "loss": 0.7026,
      "step": 14655
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8756909288269374,
      "learning_rate": 3.059113631866384e-05,
      "loss": 0.6555,
      "step": 14660
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.872034223424026,
      "learning_rate": 3.0579955912938676e-05,
      "loss": 0.5679,
      "step": 14665
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8355138767572048,
      "learning_rate": 3.056877433257434e-05,
      "loss": 0.587,
      "step": 14670
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9117033754783374,
      "learning_rate": 3.055759157992467e-05,
      "loss": 0.5837,
      "step": 14675
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8530265605140811,
      "learning_rate": 3.0546407657343756e-05,
      "loss": 0.6483,
      "step": 14680
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7763231446002526,
      "learning_rate": 3.053522256718594e-05,
      "loss": 0.6325,
      "step": 14685
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9020648029260706,
      "learning_rate": 3.0524036311805784e-05,
      "loss": 0.6497,
      "step": 14690
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.970187214484655,
      "learning_rate": 3.0512848893558127e-05,
      "loss": 0.5897,
      "step": 14695
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8484531605228595,
      "learning_rate": 3.050166031479803e-05,
      "loss": 0.5927,
      "step": 14700
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8161308322549383,
      "learning_rate": 3.04904705778808e-05,
      "loss": 0.5992,
      "step": 14705
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8543303417382723,
      "learning_rate": 3.047927968516201e-05,
      "loss": 0.5202,
      "step": 14710
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9863281093765126,
      "learning_rate": 3.046808763899745e-05,
      "loss": 0.5968,
      "step": 14715
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9075513079802398,
      "learning_rate": 3.0456894441743162e-05,
      "loss": 0.635,
      "step": 14720
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8781686159947228,
      "learning_rate": 3.0445700095755437e-05,
      "loss": 0.6409,
      "step": 14725
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7942874690493922,
      "learning_rate": 3.043450460339079e-05,
      "loss": 0.6029,
      "step": 14730
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9401660525442465,
      "learning_rate": 3.0423307967006016e-05,
      "loss": 0.6639,
      "step": 14735
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7951913814377338,
      "learning_rate": 3.0412110188958086e-05,
      "loss": 0.6135,
      "step": 14740
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.168511259090768,
      "learning_rate": 3.0400911271604278e-05,
      "loss": 0.5948,
      "step": 14745
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9279348721137287,
      "learning_rate": 3.0389711217302076e-05,
      "loss": 0.623,
      "step": 14750
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.01043980625006,
      "learning_rate": 3.0378510028409202e-05,
      "loss": 0.6818,
      "step": 14755
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8172186808017409,
      "learning_rate": 3.0367307707283626e-05,
      "loss": 0.5827,
      "step": 14760
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9067086546627713,
      "learning_rate": 3.0356104256283564e-05,
      "loss": 0.6682,
      "step": 14765
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8827044081760728,
      "learning_rate": 3.0344899677767445e-05,
      "loss": 0.5882,
      "step": 14770
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8591130134602417,
      "learning_rate": 3.0333693974093952e-05,
      "loss": 0.5362,
      "step": 14775
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8970185058699236,
      "learning_rate": 3.0322487147622014e-05,
      "loss": 0.5956,
      "step": 14780
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9620011834401389,
      "learning_rate": 3.031127920071078e-05,
      "loss": 0.6752,
      "step": 14785
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8208474687559303,
      "learning_rate": 3.030007013571963e-05,
      "loss": 0.6174,
      "step": 14790
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0667917692702664,
      "learning_rate": 3.028885995500821e-05,
      "loss": 0.6889,
      "step": 14795
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.807708090439241,
      "learning_rate": 3.0277648660936364e-05,
      "loss": 0.5148,
      "step": 14800
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0448599728959822,
      "learning_rate": 3.0266436255864195e-05,
      "loss": 0.62,
      "step": 14805
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7976554034356785,
      "learning_rate": 3.0255222742152018e-05,
      "loss": 0.6313,
      "step": 14810
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8699695422624385,
      "learning_rate": 3.0244008122160415e-05,
      "loss": 0.6253,
      "step": 14815
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.7994744383268445,
      "learning_rate": 3.023279239825017e-05,
      "loss": 0.6261,
      "step": 14820
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9674425525867826,
      "learning_rate": 3.0221575572782314e-05,
      "loss": 0.5977,
      "step": 14825
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9428129798667975,
      "learning_rate": 3.02103576481181e-05,
      "loss": 0.5971,
      "step": 14830
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8735977519404388,
      "learning_rate": 3.0199138626619027e-05,
      "loss": 0.6318,
      "step": 14835
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8862499366700649,
      "learning_rate": 3.018791851064681e-05,
      "loss": 0.6627,
      "step": 14840
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.8454794192143442,
      "learning_rate": 3.0176697302563405e-05,
      "loss": 0.6552,
      "step": 14845
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.93641355543239,
      "learning_rate": 3.0165475004730993e-05,
      "loss": 0.5273,
      "step": 14850
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9657752810686301,
      "learning_rate": 3.015425161951198e-05,
      "loss": 0.6083,
      "step": 14855
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8769095588710879,
      "learning_rate": 3.0143027149269014e-05,
      "loss": 0.6323,
      "step": 14860
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.022561444880132,
      "learning_rate": 3.0131801596364966e-05,
      "loss": 0.6103,
      "step": 14865
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.7843705192053287,
      "learning_rate": 3.012057496316292e-05,
      "loss": 0.6647,
      "step": 14870
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9139901418397574,
      "learning_rate": 3.0109347252026205e-05,
      "loss": 0.6511,
      "step": 14875
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.859650142540925,
      "learning_rate": 3.009811846531838e-05,
      "loss": 0.6503,
      "step": 14880
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0254925119533578,
      "learning_rate": 3.0086888605403208e-05,
      "loss": 0.6025,
      "step": 14885
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9402171720272638,
      "learning_rate": 3.0075657674644703e-05,
      "loss": 0.7411,
      "step": 14890
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.971803849491066,
      "learning_rate": 3.0064425675407092e-05,
      "loss": 0.5835,
      "step": 14895
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8461273985912147,
      "learning_rate": 3.005319261005483e-05,
      "loss": 0.8373,
      "step": 14900
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9259937114555273,
      "learning_rate": 3.004195848095258e-05,
      "loss": 0.6089,
      "step": 14905
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8604636731952273,
      "learning_rate": 3.0030723290465267e-05,
      "loss": 0.5828,
      "step": 14910
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9519233288910779,
      "learning_rate": 3.0019487040958004e-05,
      "loss": 0.6379,
      "step": 14915
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.833454807449613,
      "learning_rate": 3.0008249734796134e-05,
      "loss": 0.6281,
      "step": 14920
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9901037613201923,
      "learning_rate": 2.999701137434524e-05,
      "loss": 0.5581,
      "step": 14925
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.884230922647901,
      "learning_rate": 2.9985771961971104e-05,
      "loss": 0.6369,
      "step": 14930
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9078545673245122,
      "learning_rate": 2.9974531500039748e-05,
      "loss": 0.636,
      "step": 14935
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9149208943264309,
      "learning_rate": 2.99632899909174e-05,
      "loss": 0.5886,
      "step": 14940
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8108307065097886,
      "learning_rate": 2.995204743697052e-05,
      "loss": 0.6109,
      "step": 14945
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8663201113907295,
      "learning_rate": 2.9940803840565784e-05,
      "loss": 0.6138,
      "step": 14950
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8509709944392145,
      "learning_rate": 2.9929559204070078e-05,
      "loss": 0.5897,
      "step": 14955
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9060482229291699,
      "learning_rate": 2.9918313529850533e-05,
      "loss": 0.6479,
      "step": 14960
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8532734201153297,
      "learning_rate": 2.9907066820274466e-05,
      "loss": 0.6352,
      "step": 14965
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8456501479967143,
      "learning_rate": 2.9895819077709436e-05,
      "loss": 0.5928,
      "step": 14970
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9064687851107315,
      "learning_rate": 2.98845703045232e-05,
      "loss": 0.6689,
      "step": 14975
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8652678076261537,
      "learning_rate": 2.9873320503083756e-05,
      "loss": 0.5986,
      "step": 14980
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8401529460240302,
      "learning_rate": 2.9862069675759296e-05,
      "loss": 0.545,
      "step": 14985
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9102955944265461,
      "learning_rate": 2.9850817824918237e-05,
      "loss": 0.595,
      "step": 14990
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9749114740477626,
      "learning_rate": 2.9839564952929223e-05,
      "loss": 0.5803,
      "step": 14995
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8789190175440106,
      "learning_rate": 2.982831106216108e-05,
      "loss": 0.6571,
      "step": 15000
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8593762315789426,
      "learning_rate": 2.9817056154982888e-05,
      "loss": 0.6226,
      "step": 15005
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0075139965241253,
      "learning_rate": 2.980580023376392e-05,
      "loss": 0.5852,
      "step": 15010
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9118240855892054,
      "learning_rate": 2.9794543300873657e-05,
      "loss": 0.5793,
      "step": 15015
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9197131815881998,
      "learning_rate": 2.9783285358681806e-05,
      "loss": 0.5663,
      "step": 15020
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8890433750502471,
      "learning_rate": 2.9772026409558285e-05,
      "loss": 0.684,
      "step": 15025
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.4943136624401174,
      "learning_rate": 2.9760766455873222e-05,
      "loss": 0.6545,
      "step": 15030
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9147116041387249,
      "learning_rate": 2.9749505499996943e-05,
      "loss": 0.5876,
      "step": 15035
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7459538100279501,
      "learning_rate": 2.9738243544300015e-05,
      "loss": 0.5966,
      "step": 15040
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8956785128391742,
      "learning_rate": 2.972698059115318e-05,
      "loss": 0.5829,
      "step": 15045
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.7281108168341436,
      "learning_rate": 2.971571664292742e-05,
      "loss": 0.5968,
      "step": 15050
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9531359022710515,
      "learning_rate": 2.9704451701993906e-05,
      "loss": 0.5784,
      "step": 15055
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8752772059586545,
      "learning_rate": 2.9693185770724035e-05,
      "loss": 0.6822,
      "step": 15060
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.875999587143,
      "learning_rate": 2.9681918851489398e-05,
      "loss": 0.6772,
      "step": 15065
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.902186723360979,
      "learning_rate": 2.96706509466618e-05,
      "loss": 0.6529,
      "step": 15070
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9551614373463759,
      "learning_rate": 2.965938205861325e-05,
      "loss": 1.2168,
      "step": 15075
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9487659564948596,
      "learning_rate": 2.9648112189715977e-05,
      "loss": 0.569,
      "step": 15080
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0512244187765623,
      "learning_rate": 2.9636841342342392e-05,
      "loss": 0.6442,
      "step": 15085
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9570967230910677,
      "learning_rate": 2.9625569518865138e-05,
      "loss": 0.5953,
      "step": 15090
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0480209215855774,
      "learning_rate": 2.9614296721657042e-05,
      "loss": 0.564,
      "step": 15095
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9607319310527488,
      "learning_rate": 2.960302295309116e-05,
      "loss": 0.6626,
      "step": 15100
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9133064612445343,
      "learning_rate": 2.959174821554072e-05,
      "loss": 0.5827,
      "step": 15105
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1524396340438572,
      "learning_rate": 2.9580472511379197e-05,
      "loss": 0.6612,
      "step": 15110
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8550727310609498,
      "learning_rate": 2.9569195842980218e-05,
      "loss": 0.6398,
      "step": 15115
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8893716966803409,
      "learning_rate": 2.9557918212717657e-05,
      "loss": 0.5985,
      "step": 15120
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8542210506636788,
      "learning_rate": 2.9546639622965566e-05,
      "loss": 0.5318,
      "step": 15125
    },
    {
      "epoch": 1.77,
      "grad_norm": 95.57073773407024,
      "learning_rate": 2.9535360076098205e-05,
      "loss": 0.7631,
      "step": 15130
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9104315726838731,
      "learning_rate": 2.9524079574490047e-05,
      "loss": 0.625,
      "step": 15135
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9690479458585121,
      "learning_rate": 2.951279812051575e-05,
      "loss": 0.5394,
      "step": 15140
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9971123983145045,
      "learning_rate": 2.9501515716550178e-05,
      "loss": 0.5625,
      "step": 15145
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9058075794904135,
      "learning_rate": 2.949023236496839e-05,
      "loss": 0.615,
      "step": 15150
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.805836797371851,
      "learning_rate": 2.9478948068145655e-05,
      "loss": 0.6353,
      "step": 15155
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9032182189079547,
      "learning_rate": 2.946766282845744e-05,
      "loss": 0.5637,
      "step": 15160
    },
    {
      "epoch": 1.77,
      "grad_norm": 25.33904420349589,
      "learning_rate": 2.9456376648279393e-05,
      "loss": 0.6757,
      "step": 15165
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8391892317600869,
      "learning_rate": 2.9445089529987384e-05,
      "loss": 0.6036,
      "step": 15170
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8872383767708902,
      "learning_rate": 2.9433801475957475e-05,
      "loss": 0.7093,
      "step": 15175
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.821819208772899,
      "learning_rate": 2.94225124885659e-05,
      "loss": 0.6905,
      "step": 15180
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8132078498239131,
      "learning_rate": 2.9411222570189123e-05,
      "loss": 0.6638,
      "step": 15185
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.841537340818313,
      "learning_rate": 2.9399931723203796e-05,
      "loss": 0.5923,
      "step": 15190
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9559097040451198,
      "learning_rate": 2.9388639949986744e-05,
      "loss": 0.5672,
      "step": 15195
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9214939182203384,
      "learning_rate": 2.937734725291501e-05,
      "loss": 0.5893,
      "step": 15200
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.972882928842492,
      "learning_rate": 2.9366053634365835e-05,
      "loss": 0.6262,
      "step": 15205
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.7715616907361613,
      "learning_rate": 2.9354759096716634e-05,
      "loss": 0.58,
      "step": 15210
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8662036895757589,
      "learning_rate": 2.934346364234503e-05,
      "loss": 0.5611,
      "step": 15215
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8300598715474246,
      "learning_rate": 2.9332167273628823e-05,
      "loss": 0.6309,
      "step": 15220
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9357369214526228,
      "learning_rate": 2.932086999294604e-05,
      "loss": 0.6437,
      "step": 15225
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.815579467164919,
      "learning_rate": 2.9309571802674847e-05,
      "loss": 0.5935,
      "step": 15230
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9643868137658049,
      "learning_rate": 2.9298272705193652e-05,
      "loss": 0.6416,
      "step": 15235
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.85492742793463,
      "learning_rate": 2.9286972702881034e-05,
      "loss": 0.6264,
      "step": 15240
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9076639692940736,
      "learning_rate": 2.927567179811575e-05,
      "loss": 0.5829,
      "step": 15245
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8909180365387361,
      "learning_rate": 2.926436999327677e-05,
      "loss": 0.5887,
      "step": 15250
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8726015115225713,
      "learning_rate": 2.9253067290743235e-05,
      "loss": 0.6311,
      "step": 15255
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9277518081708441,
      "learning_rate": 2.9241763692894486e-05,
      "loss": 0.6007,
      "step": 15260
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9861246412172564,
      "learning_rate": 2.9230459202110046e-05,
      "loss": 0.5774,
      "step": 15265
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.886301681983085,
      "learning_rate": 2.9219153820769628e-05,
      "loss": 0.5534,
      "step": 15270
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9689578852269723,
      "learning_rate": 2.9207847551253142e-05,
      "loss": 0.5831,
      "step": 15275
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8994382022070937,
      "learning_rate": 2.919654039594066e-05,
      "loss": 0.5879,
      "step": 15280
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8963503428487793,
      "learning_rate": 2.9185232357212473e-05,
      "loss": 0.6201,
      "step": 15285
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8463400957560143,
      "learning_rate": 2.9173923437449036e-05,
      "loss": 0.6057,
      "step": 15290
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8430091402173308,
      "learning_rate": 2.9162613639030993e-05,
      "loss": 0.6042,
      "step": 15295
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.924570201297349,
      "learning_rate": 2.9151302964339172e-05,
      "loss": 0.6009,
      "step": 15300
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9357602735634031,
      "learning_rate": 2.9139991415754597e-05,
      "loss": 0.4952,
      "step": 15305
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8584845844509773,
      "learning_rate": 2.912867899565846e-05,
      "loss": 0.6335,
      "step": 15310
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9003463725769718,
      "learning_rate": 2.9117365706432144e-05,
      "loss": 0.5629,
      "step": 15315
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9045460629709449,
      "learning_rate": 2.9106051550457224e-05,
      "loss": 0.5758,
      "step": 15320
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9652131903777093,
      "learning_rate": 2.909473653011544e-05,
      "loss": 0.5894,
      "step": 15325
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8606927582561902,
      "learning_rate": 2.908342064778873e-05,
      "loss": 0.577,
      "step": 15330
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8914239692430463,
      "learning_rate": 2.9072103905859183e-05,
      "loss": 0.642,
      "step": 15335
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8290004897092969,
      "learning_rate": 2.9060786306709126e-05,
      "loss": 0.6131,
      "step": 15340
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8234281004776719,
      "learning_rate": 2.9049467852721007e-05,
      "loss": 0.637,
      "step": 15345
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9733989681776443,
      "learning_rate": 2.9038148546277488e-05,
      "loss": 0.5919,
      "step": 15350
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.7939712851349656,
      "learning_rate": 2.9026828389761406e-05,
      "loss": 0.5485,
      "step": 15355
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.8090157164040375,
      "learning_rate": 2.9015507385555763e-05,
      "loss": 0.6575,
      "step": 15360
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8220637441077542,
      "learning_rate": 2.9004185536043755e-05,
      "loss": 0.6048,
      "step": 15365
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8912668114056833,
      "learning_rate": 2.8992862843608746e-05,
      "loss": 0.5851,
      "step": 15370
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8414273747273939,
      "learning_rate": 2.898153931063428e-05,
      "loss": 0.6565,
      "step": 15375
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9075142797543228,
      "learning_rate": 2.8970214939504087e-05,
      "loss": 0.6455,
      "step": 15380
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5635072354235717,
      "learning_rate": 2.8958889732602064e-05,
      "loss": 0.638,
      "step": 15385
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.867668645195686,
      "learning_rate": 2.894756369231228e-05,
      "loss": 0.603,
      "step": 15390
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9022789682232436,
      "learning_rate": 2.8936236821018988e-05,
      "loss": 0.6011,
      "step": 15395
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9483143692107806,
      "learning_rate": 2.8924909121106612e-05,
      "loss": 0.601,
      "step": 15400
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8606154189427925,
      "learning_rate": 2.891358059495976e-05,
      "loss": 0.6186,
      "step": 15405
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8792379084804409,
      "learning_rate": 2.890225124496319e-05,
      "loss": 0.5889,
      "step": 15410
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.678016259254467,
      "learning_rate": 2.889092107350186e-05,
      "loss": 0.6146,
      "step": 15415
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9207736389941988,
      "learning_rate": 2.887959008296089e-05,
      "loss": 0.6457,
      "step": 15420
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9090070245499225,
      "learning_rate": 2.886825827572557e-05,
      "loss": 0.6205,
      "step": 15425
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.925285019670458,
      "learning_rate": 2.8856925654181356e-05,
      "loss": 0.6799,
      "step": 15430
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.02916783656782,
      "learning_rate": 2.8845592220713898e-05,
      "loss": 0.6014,
      "step": 15435
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8528647209907759,
      "learning_rate": 2.883425797770899e-05,
      "loss": 0.7026,
      "step": 15440
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8842330714333839,
      "learning_rate": 2.8822922927552616e-05,
      "loss": 0.6199,
      "step": 15445
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.786792938782078,
      "learning_rate": 2.881158707263092e-05,
      "loss": 0.5768,
      "step": 15450
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0406915896839435,
      "learning_rate": 2.880025041533022e-05,
      "loss": 0.7346,
      "step": 15455
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9759327538869409,
      "learning_rate": 2.8788912958037006e-05,
      "loss": 0.5937,
      "step": 15460
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9879829680698227,
      "learning_rate": 2.8777574703137915e-05,
      "loss": 0.6215,
      "step": 15465
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8361553037833519,
      "learning_rate": 2.876623565301979e-05,
      "loss": 0.5848,
      "step": 15470
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9417289077100779,
      "learning_rate": 2.8754895810069603e-05,
      "loss": 0.6114,
      "step": 15475
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7976990879482714,
      "learning_rate": 2.874355517667451e-05,
      "loss": 0.6002,
      "step": 15480
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.890926456435577,
      "learning_rate": 2.8732213755221855e-05,
      "loss": 0.6446,
      "step": 15485
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9451491166608306,
      "learning_rate": 2.8720871548099095e-05,
      "loss": 0.583,
      "step": 15490
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9099321630529936,
      "learning_rate": 2.8709528557693906e-05,
      "loss": 0.6244,
      "step": 15495
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8616978463790455,
      "learning_rate": 2.8698184786394094e-05,
      "loss": 0.644,
      "step": 15500
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8587283905001046,
      "learning_rate": 2.8686840236587646e-05,
      "loss": 0.5416,
      "step": 15505
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.946700712832544,
      "learning_rate": 2.8675494910662713e-05,
      "loss": 0.5252,
      "step": 15510
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1241769763091565,
      "learning_rate": 2.8664148811007596e-05,
      "loss": 0.6402,
      "step": 15515
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8107623762806399,
      "learning_rate": 2.8652801940010782e-05,
      "loss": 0.6053,
      "step": 15520
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.8811204274624614,
      "learning_rate": 2.8641454300060887e-05,
      "loss": 0.6016,
      "step": 15525
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.7812222171445361,
      "learning_rate": 2.863010589354672e-05,
      "loss": 0.6214,
      "step": 15530
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.802755397149214,
      "learning_rate": 2.8618756722857244e-05,
      "loss": 0.6961,
      "step": 15535
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0153862895503316,
      "learning_rate": 2.8607406790381564e-05,
      "loss": 0.6024,
      "step": 15540
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8864798088264312,
      "learning_rate": 2.8596056098508968e-05,
      "loss": 0.6065,
      "step": 15545
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8431454967113972,
      "learning_rate": 2.8584704649628895e-05,
      "loss": 0.5828,
      "step": 15550
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8992707917978078,
      "learning_rate": 2.857335244613095e-05,
      "loss": 0.5465,
      "step": 15555
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9566196774029382,
      "learning_rate": 2.856199949040488e-05,
      "loss": 0.639,
      "step": 15560
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8930180690240621,
      "learning_rate": 2.85506457848406e-05,
      "loss": 0.6598,
      "step": 15565
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9516695105041987,
      "learning_rate": 2.8539291331828198e-05,
      "loss": 0.6314,
      "step": 15570
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9142514895966403,
      "learning_rate": 2.8527936133757893e-05,
      "loss": 0.6409,
      "step": 15575
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2595048858553646,
      "learning_rate": 2.8516580193020075e-05,
      "loss": 0.6823,
      "step": 15580
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9038634882816586,
      "learning_rate": 2.8505223512005297e-05,
      "loss": 0.6373,
      "step": 15585
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.8395131364030178,
      "learning_rate": 2.8493866093104248e-05,
      "loss": 0.6452,
      "step": 15590
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9002844445332441,
      "learning_rate": 2.8482507938707794e-05,
      "loss": 0.6476,
      "step": 15595
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9488186000848398,
      "learning_rate": 2.8471149051206936e-05,
      "loss": 0.6223,
      "step": 15600
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9073022444458917,
      "learning_rate": 2.845978943299284e-05,
      "loss": 0.5655,
      "step": 15605
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9069415011986827,
      "learning_rate": 2.8448429086456825e-05,
      "loss": 0.5351,
      "step": 15610
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.873628949718507,
      "learning_rate": 2.8437068013990377e-05,
      "loss": 0.6071,
      "step": 15615
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9604049892919877,
      "learning_rate": 2.84257062179851e-05,
      "loss": 0.6448,
      "step": 15620
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8308200298302506,
      "learning_rate": 2.8414343700832778e-05,
      "loss": 0.5532,
      "step": 15625
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9027708754985672,
      "learning_rate": 2.8402980464925344e-05,
      "loss": 0.6373,
      "step": 15630
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8919052142974081,
      "learning_rate": 2.839161651265489e-05,
      "loss": 0.5416,
      "step": 15635
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8321245606009727,
      "learning_rate": 2.8380251846413613e-05,
      "loss": 0.6215,
      "step": 15640
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8564624880366666,
      "learning_rate": 2.836888646859392e-05,
      "loss": 0.5784,
      "step": 15645
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0011875896902096,
      "learning_rate": 2.835752038158835e-05,
      "loss": 0.6294,
      "step": 15650
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8151643297421427,
      "learning_rate": 2.8346153587789553e-05,
      "loss": 0.6102,
      "step": 15655
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.865754298984823,
      "learning_rate": 2.833478608959038e-05,
      "loss": 0.6642,
      "step": 15660
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.854560387439589,
      "learning_rate": 2.832341788938381e-05,
      "loss": 0.5829,
      "step": 15665
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8804218472430021,
      "learning_rate": 2.8312048989562955e-05,
      "loss": 0.6074,
      "step": 15670
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.8545618799163368,
      "learning_rate": 2.8300679392521096e-05,
      "loss": 0.5895,
      "step": 15675
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9652038609049712,
      "learning_rate": 2.828930910065165e-05,
      "loss": 0.6483,
      "step": 15680
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.785036131334889,
      "learning_rate": 2.8277938116348184e-05,
      "loss": 0.5727,
      "step": 15685
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.7781872319644279,
      "learning_rate": 2.8266566442004406e-05,
      "loss": 0.5165,
      "step": 15690
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9088117541912111,
      "learning_rate": 2.825519408001418e-05,
      "loss": 0.6232,
      "step": 15695
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9001654774268263,
      "learning_rate": 2.82438210327715e-05,
      "loss": 0.5938,
      "step": 15700
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9664343477830007,
      "learning_rate": 2.8232447302670513e-05,
      "loss": 0.6753,
      "step": 15705
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8339831249463702,
      "learning_rate": 2.8221072892105505e-05,
      "loss": 0.6622,
      "step": 15710
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0084503753988086,
      "learning_rate": 2.820969780347092e-05,
      "loss": 0.578,
      "step": 15715
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.003284467156913,
      "learning_rate": 2.8198322039161322e-05,
      "loss": 0.6289,
      "step": 15720
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8883027528047214,
      "learning_rate": 2.8186945601571428e-05,
      "loss": 0.6119,
      "step": 15725
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9284869809696922,
      "learning_rate": 2.8175568493096115e-05,
      "loss": 0.6524,
      "step": 15730
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9161117199431074,
      "learning_rate": 2.816419071613036e-05,
      "loss": 0.672,
      "step": 15735
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8726369520586122,
      "learning_rate": 2.815281227306932e-05,
      "loss": 0.5792,
      "step": 15740
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9594375541580534,
      "learning_rate": 2.8141433166308263e-05,
      "loss": 0.5308,
      "step": 15745
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9782893364223625,
      "learning_rate": 2.813005339824263e-05,
      "loss": 0.6294,
      "step": 15750
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9043232377410992,
      "learning_rate": 2.8118672971267963e-05,
      "loss": 0.6249,
      "step": 15755
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8909021373088903,
      "learning_rate": 2.810729188777996e-05,
      "loss": 0.6219,
      "step": 15760
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8638747358728026,
      "learning_rate": 2.809591015017448e-05,
      "loss": 0.6145,
      "step": 15765
    },
    {
      "epoch": 1.84,
      "grad_norm": 2.2259314996671353,
      "learning_rate": 2.808452776084748e-05,
      "loss": 0.5233,
      "step": 15770
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.00790397103229,
      "learning_rate": 2.807314472219507e-05,
      "loss": 0.5996,
      "step": 15775
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.7995895086216097,
      "learning_rate": 2.8061761036613514e-05,
      "loss": 0.5967,
      "step": 15780
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8298635074958586,
      "learning_rate": 2.8050376706499183e-05,
      "loss": 0.6003,
      "step": 15785
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.8977249628470072,
      "learning_rate": 2.80389917342486e-05,
      "loss": 0.6264,
      "step": 15790
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0262435288309792,
      "learning_rate": 2.802760612225843e-05,
      "loss": 0.6015,
      "step": 15795
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7437627141196989,
      "learning_rate": 2.8016219872925453e-05,
      "loss": 0.5647,
      "step": 15800
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8625797903378296,
      "learning_rate": 2.8004832988646594e-05,
      "loss": 0.5896,
      "step": 15805
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8563876446849167,
      "learning_rate": 2.7993445471818926e-05,
      "loss": 0.5638,
      "step": 15810
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9110849079384351,
      "learning_rate": 2.798205732483963e-05,
      "loss": 0.5721,
      "step": 15815
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8706606739055289,
      "learning_rate": 2.7970668550106017e-05,
      "loss": 0.6834,
      "step": 15820
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8706579403217899,
      "learning_rate": 2.7959279150015567e-05,
      "loss": 0.5853,
      "step": 15825
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.7752437116512222,
      "learning_rate": 2.794788912696586e-05,
      "loss": 0.5763,
      "step": 15830
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1087155870782206,
      "learning_rate": 2.7936498483354605e-05,
      "loss": 0.5552,
      "step": 15835
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8706300922872348,
      "learning_rate": 2.792510722157966e-05,
      "loss": 0.5767,
      "step": 15840
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8492687112401328,
      "learning_rate": 2.791371534403901e-05,
      "loss": 0.6507,
      "step": 15845
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9121793610697639,
      "learning_rate": 2.7902322853130757e-05,
      "loss": 0.5322,
      "step": 15850
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9189718096351744,
      "learning_rate": 2.7890929751253142e-05,
      "loss": 0.6173,
      "step": 15855
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8821655189226002,
      "learning_rate": 2.7879536040804533e-05,
      "loss": 0.6,
      "step": 15860
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0864360265978323,
      "learning_rate": 2.7868141724183423e-05,
      "loss": 0.5779,
      "step": 15865
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9574193815415722,
      "learning_rate": 2.7856746803788435e-05,
      "loss": 0.6123,
      "step": 15870
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.8910269556806888,
      "learning_rate": 2.7845351282018324e-05,
      "loss": 0.5723,
      "step": 15875
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9529094205532579,
      "learning_rate": 2.7833955161271968e-05,
      "loss": 0.6605,
      "step": 15880
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8758479065335699,
      "learning_rate": 2.7822558443948354e-05,
      "loss": 0.6688,
      "step": 15885
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9096254482813643,
      "learning_rate": 2.7811161132446624e-05,
      "loss": 0.6436,
      "step": 15890
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8443034074249516,
      "learning_rate": 2.7799763229166038e-05,
      "loss": 0.5973,
      "step": 15895
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8205982941896848,
      "learning_rate": 2.778836473650595e-05,
      "loss": 0.5606,
      "step": 15900
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9547737518306739,
      "learning_rate": 2.7776965656865888e-05,
      "loss": 0.5735,
      "step": 15905
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1558645525388844,
      "learning_rate": 2.7765565992645465e-05,
      "loss": 0.6548,
      "step": 15910
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8363521802205757,
      "learning_rate": 2.7754165746244433e-05,
      "loss": 0.6069,
      "step": 15915
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.7795771399962282,
      "learning_rate": 2.7742764920062657e-05,
      "loss": 0.5617,
      "step": 15920
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8557510305810598,
      "learning_rate": 2.7731363516500138e-05,
      "loss": 0.642,
      "step": 15925
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8543614234786983,
      "learning_rate": 2.771996153795699e-05,
      "loss": 0.5914,
      "step": 15930
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.876360395363473,
      "learning_rate": 2.7708558986833437e-05,
      "loss": 0.572,
      "step": 15935
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.891020462313241,
      "learning_rate": 2.7697155865529856e-05,
      "loss": 0.662,
      "step": 15940
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9042436366135987,
      "learning_rate": 2.768575217644671e-05,
      "loss": 0.6147,
      "step": 15945
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9571926136739703,
      "learning_rate": 2.76743479219846e-05,
      "loss": 0.625,
      "step": 15950
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8429153692879391,
      "learning_rate": 2.7662943104544232e-05,
      "loss": 0.6348,
      "step": 15955
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.8835499717886393,
      "learning_rate": 2.765153772652645e-05,
      "loss": 0.5903,
      "step": 15960
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9325576787556494,
      "learning_rate": 2.7640131790332207e-05,
      "loss": 0.6216,
      "step": 15965
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8862124946222595,
      "learning_rate": 2.7628725298362555e-05,
      "loss": 0.5942,
      "step": 15970
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8818170829648544,
      "learning_rate": 2.7617318253018702e-05,
      "loss": 0.5894,
      "step": 15975
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1057024370006303,
      "learning_rate": 2.760591065670194e-05,
      "loss": 0.6122,
      "step": 15980
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.886506665632321,
      "learning_rate": 2.7594502511813685e-05,
      "loss": 0.6534,
      "step": 15985
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8152950589213906,
      "learning_rate": 2.7583093820755472e-05,
      "loss": 0.6085,
      "step": 15990
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.945828116042296,
      "learning_rate": 2.7571684585928958e-05,
      "loss": 0.5736,
      "step": 15995
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.098688746584009,
      "learning_rate": 2.7560274809735898e-05,
      "loss": 0.6811,
      "step": 16000
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8332466340426692,
      "learning_rate": 2.754886449457817e-05,
      "loss": 0.6163,
      "step": 16005
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.7996120088231948,
      "learning_rate": 2.7537453642857773e-05,
      "loss": 0.5831,
      "step": 16010
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8819697375827106,
      "learning_rate": 2.7526042256976798e-05,
      "loss": 0.6018,
      "step": 16015
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8417995282798617,
      "learning_rate": 2.7514630339337473e-05,
      "loss": 0.5892,
      "step": 16020
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8875331153906554,
      "learning_rate": 2.7503217892342116e-05,
      "loss": 0.5747,
      "step": 16025
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8057349187041387,
      "learning_rate": 2.749180491839317e-05,
      "loss": 0.5617,
      "step": 16030
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8020825363525532,
      "learning_rate": 2.7480391419893193e-05,
      "loss": 0.5508,
      "step": 16035
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9500815451598762,
      "learning_rate": 2.7468977399244837e-05,
      "loss": 0.5503,
      "step": 16040
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.8965576506239832,
      "learning_rate": 2.745756285885088e-05,
      "loss": 0.5632,
      "step": 16045
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.819190099085965,
      "learning_rate": 2.744614780111419e-05,
      "loss": 0.5793,
      "step": 16050
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8656872801588841,
      "learning_rate": 2.743473222843777e-05,
      "loss": 0.6403,
      "step": 16055
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9275576130610358,
      "learning_rate": 2.7423316143224716e-05,
      "loss": 0.7171,
      "step": 16060
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8611193076102043,
      "learning_rate": 2.7411899547878222e-05,
      "loss": 0.6863,
      "step": 16065
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9699922676618187,
      "learning_rate": 2.7400482444801612e-05,
      "loss": 0.6487,
      "step": 16070
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9271339272446701,
      "learning_rate": 2.7389064836398303e-05,
      "loss": 0.6285,
      "step": 16075
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9044821352559754,
      "learning_rate": 2.7377646725071827e-05,
      "loss": 0.5975,
      "step": 16080
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8673870922871352,
      "learning_rate": 2.73662281132258e-05,
      "loss": 0.51,
      "step": 16085
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.8562655756504537,
      "learning_rate": 2.735480900326397e-05,
      "loss": 0.5777,
      "step": 16090
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0837876640952775,
      "learning_rate": 2.7343389397590184e-05,
      "loss": 0.5815,
      "step": 16095
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9740839357332823,
      "learning_rate": 2.7331969298608378e-05,
      "loss": 0.6277,
      "step": 16100
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9668206650183705,
      "learning_rate": 2.732054870872261e-05,
      "loss": 0.5675,
      "step": 16105
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9321700443566489,
      "learning_rate": 2.7309127630337033e-05,
      "loss": 0.6736,
      "step": 16110
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.850400796749632,
      "learning_rate": 2.7297706065855898e-05,
      "loss": 0.59,
      "step": 16115
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.88337331368329,
      "learning_rate": 2.7286284017683567e-05,
      "loss": 0.6245,
      "step": 16120
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9228023934065946,
      "learning_rate": 2.727486148822451e-05,
      "loss": 0.6126,
      "step": 16125
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9297848327807844,
      "learning_rate": 2.7263438479883275e-05,
      "loss": 0.6041,
      "step": 16130
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8897542897390996,
      "learning_rate": 2.7252014995064522e-05,
      "loss": 0.6164,
      "step": 16135
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8354749432178362,
      "learning_rate": 2.724059103617304e-05,
      "loss": 0.6245,
      "step": 16140
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9522027141721513,
      "learning_rate": 2.7229166605613664e-05,
      "loss": 0.6494,
      "step": 16145
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8991223295241942,
      "learning_rate": 2.7217741705791373e-05,
      "loss": 0.6171,
      "step": 16150
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8626663455817438,
      "learning_rate": 2.720631633911122e-05,
      "loss": 0.5996,
      "step": 16155
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.89915405906176,
      "learning_rate": 2.7194890507978366e-05,
      "loss": 0.5941,
      "step": 16160
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9764651439984724,
      "learning_rate": 2.7183464214798065e-05,
      "loss": 0.5903,
      "step": 16165
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8043282982413243,
      "learning_rate": 2.717203746197568e-05,
      "loss": 0.5275,
      "step": 16170
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8424033337242354,
      "learning_rate": 2.7160610251916658e-05,
      "loss": 0.6002,
      "step": 16175
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8392593205409246,
      "learning_rate": 2.714918258702654e-05,
      "loss": 0.5489,
      "step": 16180
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8555816970175268,
      "learning_rate": 2.7137754469710974e-05,
      "loss": 0.5691,
      "step": 16185
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9432108375139995,
      "learning_rate": 2.712632590237571e-05,
      "loss": 0.6494,
      "step": 16190
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9629151410031676,
      "learning_rate": 2.7114896887426565e-05,
      "loss": 0.6422,
      "step": 16195
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.8760125629876795,
      "learning_rate": 2.7103467427269462e-05,
      "loss": 0.6639,
      "step": 16200
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0141312487457848,
      "learning_rate": 2.709203752431045e-05,
      "loss": 0.559,
      "step": 16205
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9715762700049224,
      "learning_rate": 2.7080607180955615e-05,
      "loss": 0.6759,
      "step": 16210
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9885706554451825,
      "learning_rate": 2.7069176399611167e-05,
      "loss": 0.636,
      "step": 16215
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9214957989058412,
      "learning_rate": 2.7057745182683424e-05,
      "loss": 0.6248,
      "step": 16220
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8412140512772429,
      "learning_rate": 2.7046313532578767e-05,
      "loss": 0.6241,
      "step": 16225
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8698811885760802,
      "learning_rate": 2.703488145170367e-05,
      "loss": 0.5535,
      "step": 16230
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8648144351536741,
      "learning_rate": 2.7023448942464715e-05,
      "loss": 0.6362,
      "step": 16235
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8774876929432863,
      "learning_rate": 2.7012016007268558e-05,
      "loss": 0.5914,
      "step": 16240
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.887143315732413,
      "learning_rate": 2.700058264852196e-05,
      "loss": 0.6272,
      "step": 16245
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.935063005758012,
      "learning_rate": 2.6989148868631752e-05,
      "loss": 0.629,
      "step": 16250
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0821960174018377,
      "learning_rate": 2.6977714670004883e-05,
      "loss": 0.6524,
      "step": 16255
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8373458304438235,
      "learning_rate": 2.6966280055048348e-05,
      "loss": 0.6123,
      "step": 16260
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9265005171783958,
      "learning_rate": 2.6954845026169274e-05,
      "loss": 0.669,
      "step": 16265
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9390398529339612,
      "learning_rate": 2.6943409585774842e-05,
      "loss": 0.6138,
      "step": 16270
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8827613329609988,
      "learning_rate": 2.6931973736272332e-05,
      "loss": 0.6328,
      "step": 16275
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9182019884321903,
      "learning_rate": 2.6920537480069115e-05,
      "loss": 0.6652,
      "step": 16280
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9856380537138498,
      "learning_rate": 2.690910081957264e-05,
      "loss": 0.6202,
      "step": 16285
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8983374203247493,
      "learning_rate": 2.689766375719045e-05,
      "loss": 0.5754,
      "step": 16290
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8701257426198524,
      "learning_rate": 2.688622629533015e-05,
      "loss": 0.6083,
      "step": 16295
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.8944257789411685,
      "learning_rate": 2.6874788436399468e-05,
      "loss": 0.6473,
      "step": 16300
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.7986303831329686,
      "learning_rate": 2.686335018280618e-05,
      "loss": 0.6428,
      "step": 16305
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8097106579939339,
      "learning_rate": 2.6851911536958153e-05,
      "loss": 0.6428,
      "step": 16310
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.830689177166533,
      "learning_rate": 2.6840472501263353e-05,
      "loss": 0.6745,
      "step": 16315
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8298203205638941,
      "learning_rate": 2.6829033078129818e-05,
      "loss": 0.5736,
      "step": 16320
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9171280769701649,
      "learning_rate": 2.6817593269965656e-05,
      "loss": 0.6021,
      "step": 16325
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8516406706788282,
      "learning_rate": 2.680615307917907e-05,
      "loss": 0.6042,
      "step": 16330
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9653169827711248,
      "learning_rate": 2.6794712508178344e-05,
      "loss": 0.6341,
      "step": 16335
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.056215567765017,
      "learning_rate": 2.6783271559371836e-05,
      "loss": 0.5396,
      "step": 16340
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8384006411542563,
      "learning_rate": 2.6771830235167977e-05,
      "loss": 0.607,
      "step": 16345
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3336453249967208,
      "learning_rate": 2.6760388537975307e-05,
      "loss": 0.6252,
      "step": 16350
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8887856382899293,
      "learning_rate": 2.6748946470202402e-05,
      "loss": 0.6445,
      "step": 16355
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8663139256509812,
      "learning_rate": 2.673750403425795e-05,
      "loss": 0.5686,
      "step": 16360
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8246043116283,
      "learning_rate": 2.672606123255069e-05,
      "loss": 0.5149,
      "step": 16365
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8532783920532206,
      "learning_rate": 2.671461806748947e-05,
      "loss": 0.6382,
      "step": 16370
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9817389534989502,
      "learning_rate": 2.6703174541483177e-05,
      "loss": 0.6358,
      "step": 16375
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8444300865838213,
      "learning_rate": 2.6691730656940806e-05,
      "loss": 0.5604,
      "step": 16380
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9282467998731498,
      "learning_rate": 2.6680286416271422e-05,
      "loss": 0.5599,
      "step": 16385
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.8245567772369279,
      "learning_rate": 2.6668841821884126e-05,
      "loss": 0.5736,
      "step": 16390
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9371113506022917,
      "learning_rate": 2.6657396876188162e-05,
      "loss": 0.613,
      "step": 16395
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8300184279600211,
      "learning_rate": 2.6645951581592786e-05,
      "loss": 0.5858,
      "step": 16400
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9271838086884295,
      "learning_rate": 2.663450594050737e-05,
      "loss": 0.5869,
      "step": 16405
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9884078459708783,
      "learning_rate": 2.6623059955341323e-05,
      "loss": 0.6117,
      "step": 16410
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9399718188479353,
      "learning_rate": 2.6611613628504156e-05,
      "loss": 0.6711,
      "step": 16415
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8235430722960222,
      "learning_rate": 2.660016696240544e-05,
      "loss": 0.6408,
      "step": 16420
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8540579428831879,
      "learning_rate": 2.658871995945481e-05,
      "loss": 0.6056,
      "step": 16425
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.846066542908944,
      "learning_rate": 2.6577272622061993e-05,
      "loss": 0.5666,
      "step": 16430
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8734639685391002,
      "learning_rate": 2.656582495263677e-05,
      "loss": 0.6051,
      "step": 16435
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0118201204888633,
      "learning_rate": 2.6554376953588993e-05,
      "loss": 0.6603,
      "step": 16440
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.948139380146183,
      "learning_rate": 2.654292862732858e-05,
      "loss": 0.5653,
      "step": 16445
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.046031784317703,
      "learning_rate": 2.6531479976265534e-05,
      "loss": 0.6146,
      "step": 16450
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7937036579389388,
      "learning_rate": 2.6520031002809908e-05,
      "loss": 0.5019,
      "step": 16455
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8335148953635896,
      "learning_rate": 2.6508581709371828e-05,
      "loss": 0.5851,
      "step": 16460
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7570566263529973,
      "learning_rate": 2.6497132098361506e-05,
      "loss": 0.5716,
      "step": 16465
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9469409523832679,
      "learning_rate": 2.648568217218919e-05,
      "loss": 0.6053,
      "step": 16470
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.8349934403837173,
      "learning_rate": 2.6474231933265214e-05,
      "loss": 0.5199,
      "step": 16475
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9583480579930664,
      "learning_rate": 2.646278138399997e-05,
      "loss": 0.6359,
      "step": 16480
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9462513146533498,
      "learning_rate": 2.645133052680393e-05,
      "loss": 0.7017,
      "step": 16485
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8753439140524353,
      "learning_rate": 2.6439879364087612e-05,
      "loss": 0.6299,
      "step": 16490
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8437288100362328,
      "learning_rate": 2.6428427898261595e-05,
      "loss": 0.5901,
      "step": 16495
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9124240236022785,
      "learning_rate": 2.6416976131736553e-05,
      "loss": 0.6147,
      "step": 16500
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0390665817586586,
      "learning_rate": 2.6405524066923188e-05,
      "loss": 0.6224,
      "step": 16505
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0416717994843472,
      "learning_rate": 2.6394071706232286e-05,
      "loss": 0.6315,
      "step": 16510
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.7920638661302623,
      "learning_rate": 2.6382619052074687e-05,
      "loss": 0.5539,
      "step": 16515
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9084293560134566,
      "learning_rate": 2.637116610686129e-05,
      "loss": 0.5834,
      "step": 16520
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8316448292525853,
      "learning_rate": 2.6359712873003066e-05,
      "loss": 0.6082,
      "step": 16525
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9500532616079839,
      "learning_rate": 2.6348259352911047e-05,
      "loss": 0.5725,
      "step": 16530
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8984691047263826,
      "learning_rate": 2.6336805548996306e-05,
      "loss": 0.6271,
      "step": 16535
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8282147108429584,
      "learning_rate": 2.632535146366999e-05,
      "loss": 0.5914,
      "step": 16540
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8506475092981782,
      "learning_rate": 2.631389709934331e-05,
      "loss": 0.5297,
      "step": 16545
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8163470172035223,
      "learning_rate": 2.6302442458427535e-05,
      "loss": 0.6161,
      "step": 16550
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8843141338832471,
      "learning_rate": 2.6290987543333967e-05,
      "loss": 0.6149,
      "step": 16555
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.8110236544239569,
      "learning_rate": 2.6279532356474008e-05,
      "loss": 0.5899,
      "step": 16560
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.89710013290311,
      "learning_rate": 2.6268076900259083e-05,
      "loss": 0.6152,
      "step": 16565
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8453458929522863,
      "learning_rate": 2.6256621177100687e-05,
      "loss": 0.5538,
      "step": 16570
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9995764988508382,
      "learning_rate": 2.624516518941037e-05,
      "loss": 0.5506,
      "step": 16575
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7985082591789547,
      "learning_rate": 2.6233708939599738e-05,
      "loss": 0.6672,
      "step": 16580
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.7951410022637205,
      "learning_rate": 2.622225243008045e-05,
      "loss": 0.5468,
      "step": 16585
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.925844545735007,
      "learning_rate": 2.621079566326422e-05,
      "loss": 0.5619,
      "step": 16590
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8973457390987843,
      "learning_rate": 2.6199338641562825e-05,
      "loss": 0.5496,
      "step": 16595
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9555595312502254,
      "learning_rate": 2.6187881367388078e-05,
      "loss": 0.572,
      "step": 16600
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.858256456733164,
      "learning_rate": 2.6176423843151865e-05,
      "loss": 0.5845,
      "step": 16605
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0504556295738423,
      "learning_rate": 2.6164966071266104e-05,
      "loss": 0.6216,
      "step": 16610
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0106378262720044,
      "learning_rate": 2.6153508054142795e-05,
      "loss": 0.561,
      "step": 16615
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8001979265884642,
      "learning_rate": 2.6142049794193947e-05,
      "loss": 0.5893,
      "step": 16620
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8868633271089033,
      "learning_rate": 2.6130591293831652e-05,
      "loss": 0.6672,
      "step": 16625
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8552408210797487,
      "learning_rate": 2.6119132555468055e-05,
      "loss": 0.6252,
      "step": 16630
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0055482549177566,
      "learning_rate": 2.610767358151533e-05,
      "loss": 0.5638,
      "step": 16635
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8464189486452749,
      "learning_rate": 2.6096214374385707e-05,
      "loss": 0.6157,
      "step": 16640
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.8884619050402282,
      "learning_rate": 2.6084754936491484e-05,
      "loss": 0.5963,
      "step": 16645
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8305666401989756,
      "learning_rate": 2.6073295270244984e-05,
      "loss": 0.6318,
      "step": 16650
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8718791739176128,
      "learning_rate": 2.606183537805858e-05,
      "loss": 0.5727,
      "step": 16655
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8404766048331345,
      "learning_rate": 2.6050375262344707e-05,
      "loss": 0.5444,
      "step": 16660
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.804592121241699,
      "learning_rate": 2.6038914925515846e-05,
      "loss": 0.5891,
      "step": 16665
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8035335923587422,
      "learning_rate": 2.6027454369984498e-05,
      "loss": 0.5794,
      "step": 16670
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9228027019338096,
      "learning_rate": 2.6015993598163257e-05,
      "loss": 0.5544,
      "step": 16675
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8365424636355522,
      "learning_rate": 2.6004532612464715e-05,
      "loss": 0.6053,
      "step": 16680
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8551052746039735,
      "learning_rate": 2.5993071415301536e-05,
      "loss": 0.5737,
      "step": 16685
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9170677809172477,
      "learning_rate": 2.5981610009086417e-05,
      "loss": 0.5848,
      "step": 16690
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8602008361428335,
      "learning_rate": 2.5970148396232115e-05,
      "loss": 0.651,
      "step": 16695
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9653960906626363,
      "learning_rate": 2.5958686579151413e-05,
      "loss": 0.5969,
      "step": 16700
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8434143686757325,
      "learning_rate": 2.594722456025714e-05,
      "loss": 0.6052,
      "step": 16705
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.8422190248696347,
      "learning_rate": 2.593576234196218e-05,
      "loss": 0.5969,
      "step": 16710
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9962820126693613,
      "learning_rate": 2.592429992667944e-05,
      "loss": 0.6164,
      "step": 16715
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.7985716172985903,
      "learning_rate": 2.5912837316821888e-05,
      "loss": 0.5559,
      "step": 16720
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.817605352178141,
      "learning_rate": 2.5901374514802513e-05,
      "loss": 0.5903,
      "step": 16725
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9248492532734429,
      "learning_rate": 2.5889911523034365e-05,
      "loss": 0.6226,
      "step": 16730
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2289085615072273,
      "learning_rate": 2.5878448343930516e-05,
      "loss": 0.609,
      "step": 16735
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7378848106477746,
      "learning_rate": 2.586698497990409e-05,
      "loss": 0.563,
      "step": 16740
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0005531804580519,
      "learning_rate": 2.585552143336825e-05,
      "loss": 0.5924,
      "step": 16745
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.768968006608705,
      "learning_rate": 2.584405770673618e-05,
      "loss": 0.6359,
      "step": 16750
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8862142709284415,
      "learning_rate": 2.583259380242113e-05,
      "loss": 0.5704,
      "step": 16755
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9621465603691413,
      "learning_rate": 2.5821129722836357e-05,
      "loss": 0.6572,
      "step": 16760
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9006019843685333,
      "learning_rate": 2.5809665470395172e-05,
      "loss": 0.6479,
      "step": 16765
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9092680938267377,
      "learning_rate": 2.5798201047510928e-05,
      "loss": 0.5876,
      "step": 16770
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9391622878245258,
      "learning_rate": 2.5786736456597e-05,
      "loss": 0.5706,
      "step": 16775
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9431639716438284,
      "learning_rate": 2.5775271700066816e-05,
      "loss": 0.6805,
      "step": 16780
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9030138272808294,
      "learning_rate": 2.576380678033381e-05,
      "loss": 0.6415,
      "step": 16785
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.7992925270642898,
      "learning_rate": 2.575234169981148e-05,
      "loss": 0.5662,
      "step": 16790
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.120085046764829,
      "learning_rate": 2.5740876460913345e-05,
      "loss": 0.6798,
      "step": 16795
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9397953257168703,
      "learning_rate": 2.572941106605294e-05,
      "loss": 0.5723,
      "step": 16800
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8598085551464412,
      "learning_rate": 2.5717945517643877e-05,
      "loss": 0.6011,
      "step": 16805
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.868106234903999,
      "learning_rate": 2.570647981809976e-05,
      "loss": 0.6699,
      "step": 16810
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.8118746870352042,
      "learning_rate": 2.5695013969834242e-05,
      "loss": 0.5899,
      "step": 16815
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8880877589467654,
      "learning_rate": 2.5683547975260997e-05,
      "loss": 0.6217,
      "step": 16820
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.915873503949781,
      "learning_rate": 2.5672081836793748e-05,
      "loss": 0.5892,
      "step": 16825
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.856335943551882,
      "learning_rate": 2.5660615556846236e-05,
      "loss": 0.5516,
      "step": 16830
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8286314251356766,
      "learning_rate": 2.564914913783222e-05,
      "loss": 0.519,
      "step": 16835
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7498305594885852,
      "learning_rate": 2.563768258216552e-05,
      "loss": 0.5765,
      "step": 16840
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8991917670335136,
      "learning_rate": 2.5626215892259953e-05,
      "loss": 0.6489,
      "step": 16845
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8736694743046873,
      "learning_rate": 2.5614749070529382e-05,
      "loss": 0.5791,
      "step": 16850
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.904213413843636,
      "learning_rate": 2.5603282119387685e-05,
      "loss": 0.5843,
      "step": 16855
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7930718957465704,
      "learning_rate": 2.5591815041248796e-05,
      "loss": 0.6205,
      "step": 16860
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8974345724699936,
      "learning_rate": 2.558034783852663e-05,
      "loss": 0.5887,
      "step": 16865
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9841381922253502,
      "learning_rate": 2.5568880513635162e-05,
      "loss": 0.6442,
      "step": 16870
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9042253499119333,
      "learning_rate": 2.55574130689884e-05,
      "loss": 0.655,
      "step": 16875
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.7442416514466498,
      "learning_rate": 2.554594550700034e-05,
      "loss": 0.5884,
      "step": 16880
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2975522908602373,
      "learning_rate": 2.5534477830085036e-05,
      "loss": 0.5685,
      "step": 16885
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.8543090316384669,
      "learning_rate": 2.552301004065654e-05,
      "loss": 0.6303,
      "step": 16890
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9595412894206504,
      "learning_rate": 2.551154214112897e-05,
      "loss": 0.5736,
      "step": 16895
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9529149076817772,
      "learning_rate": 2.5500074133916406e-05,
      "loss": 0.6283,
      "step": 16900
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8358242303560207,
      "learning_rate": 2.5488606021433005e-05,
      "loss": 0.6515,
      "step": 16905
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8982360038691455,
      "learning_rate": 2.5477137806092923e-05,
      "loss": 0.5438,
      "step": 16910
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7951213507262371,
      "learning_rate": 2.546566949031032e-05,
      "loss": 0.5642,
      "step": 16915
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8782406187411251,
      "learning_rate": 2.5454201076499422e-05,
      "loss": 0.6379,
      "step": 16920
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9153368453589131,
      "learning_rate": 2.544273256707444e-05,
      "loss": 0.6252,
      "step": 16925
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8148545242138204,
      "learning_rate": 2.5431263964449614e-05,
      "loss": 0.563,
      "step": 16930
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7197345301449499,
      "learning_rate": 2.5419795271039203e-05,
      "loss": 0.6045,
      "step": 16935
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8393789454018323,
      "learning_rate": 2.5408326489257496e-05,
      "loss": 0.6014,
      "step": 16940
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7903400215076338,
      "learning_rate": 2.5396857621518783e-05,
      "loss": 0.6736,
      "step": 16945
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8782324369881968,
      "learning_rate": 2.5385388670237374e-05,
      "loss": 0.6072,
      "step": 16950
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7465079656224758,
      "learning_rate": 2.5373919637827627e-05,
      "loss": 0.5816,
      "step": 16955
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8933501622674589,
      "learning_rate": 2.5362450526703874e-05,
      "loss": 0.6506,
      "step": 16960
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.7706031489737,
      "learning_rate": 2.535098133928049e-05,
      "loss": 0.6377,
      "step": 16965
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.852046615275629,
      "learning_rate": 2.533951207797185e-05,
      "loss": 0.5703,
      "step": 16970
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8156620801240052,
      "learning_rate": 2.532804274519237e-05,
      "loss": 0.6108,
      "step": 16975
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0297750440049749,
      "learning_rate": 2.5316573343356454e-05,
      "loss": 0.6874,
      "step": 16980
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.8602869864432551,
      "learning_rate": 2.5305103874878533e-05,
      "loss": 0.6578,
      "step": 16985
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0371792763361753,
      "learning_rate": 2.529363434217305e-05,
      "loss": 0.6033,
      "step": 16990
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9238445910411287,
      "learning_rate": 2.5282164747654463e-05,
      "loss": 0.6279,
      "step": 16995
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9401134678293528,
      "learning_rate": 2.5270695093737235e-05,
      "loss": 0.6026,
      "step": 17000
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9249105829062576,
      "learning_rate": 2.5259225382835862e-05,
      "loss": 0.5666,
      "step": 17005
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0329416712213435,
      "learning_rate": 2.5247755617364828e-05,
      "loss": 0.5789,
      "step": 17010
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8996205439380367,
      "learning_rate": 2.523628579973864e-05,
      "loss": 0.6134,
      "step": 17015
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.996982043749138,
      "learning_rate": 2.5224815932371815e-05,
      "loss": 0.5876,
      "step": 17020
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9602185979829699,
      "learning_rate": 2.521334601767889e-05,
      "loss": 0.5928,
      "step": 17025
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9207581679596757,
      "learning_rate": 2.5201876058074386e-05,
      "loss": 0.6419,
      "step": 17030
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9976838175883397,
      "learning_rate": 2.5190406055972854e-05,
      "loss": 0.5844,
      "step": 17035
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.908672282345896,
      "learning_rate": 2.5178936013788862e-05,
      "loss": 0.6398,
      "step": 17040
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9513677094943923,
      "learning_rate": 2.5167465933936956e-05,
      "loss": 0.6254,
      "step": 17045
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8867481863888266,
      "learning_rate": 2.5155995818831725e-05,
      "loss": 0.6178,
      "step": 17050
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9079319261230471,
      "learning_rate": 2.514452567088773e-05,
      "loss": 0.6834,
      "step": 17055
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9006898502150343,
      "learning_rate": 2.5133055492519574e-05,
      "loss": 0.623,
      "step": 17060
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.8621736262313544,
      "learning_rate": 2.512158528614184e-05,
      "loss": 0.5758,
      "step": 17065
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.9459559248324776,
      "learning_rate": 2.5110115054169136e-05,
      "loss": 0.6371,
      "step": 17070
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0604462835247377,
      "learning_rate": 2.5098644799016053e-05,
      "loss": 0.578,
      "step": 17075
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9134108857072495,
      "learning_rate": 2.5087174523097206e-05,
      "loss": 0.5858,
      "step": 17080
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0205067462570563,
      "learning_rate": 2.5075704228827213e-05,
      "loss": 0.5686,
      "step": 17085
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9497294364292642,
      "learning_rate": 2.506423391862069e-05,
      "loss": 0.6471,
      "step": 17090
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8423440191690329,
      "learning_rate": 2.505276359489225e-05,
      "loss": 0.5846,
      "step": 17095
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.904922017254225,
      "learning_rate": 2.5041293260056514e-05,
      "loss": 0.5598,
      "step": 17100
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9168715776753865,
      "learning_rate": 2.5029822916528124e-05,
      "loss": 0.6542,
      "step": 17105
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8420788907513983,
      "learning_rate": 2.501835256672169e-05,
      "loss": 0.5946,
      "step": 17110
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8939470086470097,
      "learning_rate": 2.5006882213051853e-05,
      "loss": 0.6605,
      "step": 17115
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8333788677494608,
      "learning_rate": 2.4995411857933236e-05,
      "loss": 0.5478,
      "step": 17120
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9603733468073307,
      "learning_rate": 2.4983941503780474e-05,
      "loss": 0.3922,
      "step": 17125
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8859297788558389,
      "learning_rate": 2.497247115300819e-05,
      "loss": 0.3093,
      "step": 17130
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0192122631091693,
      "learning_rate": 2.4961000808031028e-05,
      "loss": 0.3381,
      "step": 17135
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9176969557641856,
      "learning_rate": 2.4949530471263593e-05,
      "loss": 0.3637,
      "step": 17140
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0327964684954671,
      "learning_rate": 2.4938060145120523e-05,
      "loss": 0.3189,
      "step": 17145
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9816091433877076,
      "learning_rate": 2.4926589832016438e-05,
      "loss": 0.3442,
      "step": 17150
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9140664639954762,
      "learning_rate": 2.4915119534365968e-05,
      "loss": 0.3332,
      "step": 17155
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9379566439341127,
      "learning_rate": 2.4903649254583732e-05,
      "loss": 0.3005,
      "step": 17160
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.9916249328406576,
      "learning_rate": 2.4892178995084327e-05,
      "loss": 0.3448,
      "step": 17165
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8194002451321004,
      "learning_rate": 2.4880708758282376e-05,
      "loss": 0.2776,
      "step": 17170
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8248128527151911,
      "learning_rate": 2.4869238546592478e-05,
      "loss": 0.3432,
      "step": 17175
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0940180123900358,
      "learning_rate": 2.485776836242923e-05,
      "loss": 0.3571,
      "step": 17180
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8077867248896436,
      "learning_rate": 2.484629820820725e-05,
      "loss": 0.3184,
      "step": 17185
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8799185092935281,
      "learning_rate": 2.48348280863411e-05,
      "loss": 0.3134,
      "step": 17190
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8492333707808328,
      "learning_rate": 2.482335799924536e-05,
      "loss": 0.2756,
      "step": 17195
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.9653531459324595,
      "learning_rate": 2.481188794933462e-05,
      "loss": 0.3286,
      "step": 17200
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.9910112047087751,
      "learning_rate": 2.4800417939023436e-05,
      "loss": 0.3609,
      "step": 17205
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.060884259039683,
      "learning_rate": 2.4788947970726362e-05,
      "loss": 0.3572,
      "step": 17210
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8864800880756659,
      "learning_rate": 2.477747804685795e-05,
      "loss": 0.2775,
      "step": 17215
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.949222088740071,
      "learning_rate": 2.476600816983274e-05,
      "loss": 0.2796,
      "step": 17220
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.9699748257748605,
      "learning_rate": 2.4754538342065262e-05,
      "loss": 0.3025,
      "step": 17225
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.8788443815813005,
      "learning_rate": 2.4743068565970044e-05,
      "loss": 0.3457,
      "step": 17230
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1169666507318705,
      "learning_rate": 2.4731598843961572e-05,
      "loss": 0.3445,
      "step": 17235
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.917127682219297,
      "learning_rate": 2.472012917845436e-05,
      "loss": 0.3034,
      "step": 17240
    },
    {
      "epoch": 2.01,
      "grad_norm": 0.953704266404627,
      "learning_rate": 2.4708659571862887e-05,
      "loss": 0.2491,
      "step": 17245
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0856311027224432,
      "learning_rate": 2.4697190026601623e-05,
      "loss": 0.3461,
      "step": 17250
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8834116242457338,
      "learning_rate": 2.4685720545085042e-05,
      "loss": 0.3397,
      "step": 17255
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9189022126420227,
      "learning_rate": 2.4674251129727566e-05,
      "loss": 0.309,
      "step": 17260
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0504422218122913,
      "learning_rate": 2.4662781782943638e-05,
      "loss": 0.336,
      "step": 17265
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8912952056533301,
      "learning_rate": 2.4651312507147684e-05,
      "loss": 0.315,
      "step": 17270
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.878345353339824,
      "learning_rate": 2.4639843304754104e-05,
      "loss": 0.3055,
      "step": 17275
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9684809126947463,
      "learning_rate": 2.4628374178177272e-05,
      "loss": 0.343,
      "step": 17280
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9746897312903375,
      "learning_rate": 2.4616905129831572e-05,
      "loss": 0.2882,
      "step": 17285
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8404782475832167,
      "learning_rate": 2.4605436162131347e-05,
      "loss": 0.2955,
      "step": 17290
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8506945892178218,
      "learning_rate": 2.4593967277490946e-05,
      "loss": 0.4619,
      "step": 17295
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9541116771277358,
      "learning_rate": 2.4582498478324698e-05,
      "loss": 0.3051,
      "step": 17300
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8236177620344743,
      "learning_rate": 2.457102976704689e-05,
      "loss": 0.3451,
      "step": 17305
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9213990587629239,
      "learning_rate": 2.45595611460718e-05,
      "loss": 0.3607,
      "step": 17310
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.8723357222497063,
      "learning_rate": 2.454809261781371e-05,
      "loss": 0.2929,
      "step": 17315
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9480733391434952,
      "learning_rate": 2.4536624184686847e-05,
      "loss": 0.3125,
      "step": 17320
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5675813618845456,
      "learning_rate": 2.4525155849105465e-05,
      "loss": 0.3024,
      "step": 17325
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9627573940108838,
      "learning_rate": 2.4513687613483734e-05,
      "loss": 0.3457,
      "step": 17330
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0243274837634995,
      "learning_rate": 2.4502219480235862e-05,
      "loss": 0.3384,
      "step": 17335
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8480022057754414,
      "learning_rate": 2.4490751451776002e-05,
      "loss": 0.3139,
      "step": 17340
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9903741283030525,
      "learning_rate": 2.4479283530518302e-05,
      "loss": 0.2594,
      "step": 17345
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0199372197175123,
      "learning_rate": 2.4467815718876863e-05,
      "loss": 0.3362,
      "step": 17350
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8618474272851386,
      "learning_rate": 2.4456348019265794e-05,
      "loss": 0.2858,
      "step": 17355
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9735975817160154,
      "learning_rate": 2.444488043409916e-05,
      "loss": 0.3503,
      "step": 17360
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8898212656121688,
      "learning_rate": 2.4433412965791014e-05,
      "loss": 0.3476,
      "step": 17365
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.020482655923461,
      "learning_rate": 2.442194561675538e-05,
      "loss": 0.3517,
      "step": 17370
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8976920607162705,
      "learning_rate": 2.4410478389406236e-05,
      "loss": 0.3012,
      "step": 17375
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.7375671295520114,
      "learning_rate": 2.4399011286157564e-05,
      "loss": 0.2654,
      "step": 17380
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8872065117202679,
      "learning_rate": 2.438754430942332e-05,
      "loss": 0.2868,
      "step": 17385
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9658662036254859,
      "learning_rate": 2.4376077461617407e-05,
      "loss": 0.2665,
      "step": 17390
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9271367226076082,
      "learning_rate": 2.4364610745153736e-05,
      "loss": 0.2678,
      "step": 17395
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.8490318071965303,
      "learning_rate": 2.4353144162446153e-05,
      "loss": 0.2665,
      "step": 17400
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.858582147563123,
      "learning_rate": 2.4341677715908494e-05,
      "loss": 0.3059,
      "step": 17405
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0230979388841956,
      "learning_rate": 2.4330211407954574e-05,
      "loss": 0.3552,
      "step": 17410
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.988131308725531,
      "learning_rate": 2.4318745240998178e-05,
      "loss": 0.3,
      "step": 17415
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0460092365578237,
      "learning_rate": 2.430727921745304e-05,
      "loss": 0.3038,
      "step": 17420
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8903441366020896,
      "learning_rate": 2.4295813339732875e-05,
      "loss": 0.3011,
      "step": 17425
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9269108066605277,
      "learning_rate": 2.4284347610251384e-05,
      "loss": 0.303,
      "step": 17430
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8970700240366455,
      "learning_rate": 2.4272882031422215e-05,
      "loss": 0.324,
      "step": 17435
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8670441823301787,
      "learning_rate": 2.4261416605659005e-05,
      "loss": 0.3332,
      "step": 17440
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.947934789143799,
      "learning_rate": 2.4249951335375322e-05,
      "loss": 0.3078,
      "step": 17445
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.7845898370122005,
      "learning_rate": 2.423848622298475e-05,
      "loss": 0.3079,
      "step": 17450
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9041180105863424,
      "learning_rate": 2.4227021270900796e-05,
      "loss": 0.3561,
      "step": 17455
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9969265199985096,
      "learning_rate": 2.4215556481536962e-05,
      "loss": 0.3632,
      "step": 17460
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9843266685052504,
      "learning_rate": 2.4204091857306715e-05,
      "loss": 0.3265,
      "step": 17465
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0200931355251979,
      "learning_rate": 2.4192627400623467e-05,
      "loss": 0.2828,
      "step": 17470
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9250327050669283,
      "learning_rate": 2.41811631139006e-05,
      "loss": 0.3227,
      "step": 17475
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.967646323264111,
      "learning_rate": 2.4169698999551483e-05,
      "loss": 0.3231,
      "step": 17480
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8174524454148893,
      "learning_rate": 2.4158235059989426e-05,
      "loss": 0.2788,
      "step": 17485
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.8596024109022202,
      "learning_rate": 2.4146771297627697e-05,
      "loss": 0.3339,
      "step": 17490
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.789800962791306,
      "learning_rate": 2.4135307714879546e-05,
      "loss": 0.4363,
      "step": 17495
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.9778444661411706,
      "learning_rate": 2.4123844314158186e-05,
      "loss": 0.3007,
      "step": 17500
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8572374376686605,
      "learning_rate": 2.4112381097876772e-05,
      "loss": 0.2994,
      "step": 17505
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9458634889415379,
      "learning_rate": 2.4100918068448442e-05,
      "loss": 0.305,
      "step": 17510
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2084205887146289,
      "learning_rate": 2.408945522828627e-05,
      "loss": 0.2995,
      "step": 17515
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8589052641322348,
      "learning_rate": 2.4077992579803306e-05,
      "loss": 0.2545,
      "step": 17520
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0193983154870474,
      "learning_rate": 2.4066530125412565e-05,
      "loss": 0.2631,
      "step": 17525
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8330716395593406,
      "learning_rate": 2.4055067867527014e-05,
      "loss": 0.3248,
      "step": 17530
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8001572156382677,
      "learning_rate": 2.404360580855958e-05,
      "loss": 0.3142,
      "step": 17535
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9512966692156783,
      "learning_rate": 2.403214395092313e-05,
      "loss": 0.2876,
      "step": 17540
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8008460898710926,
      "learning_rate": 2.4020682297030524e-05,
      "loss": 0.3022,
      "step": 17545
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0463965729666374,
      "learning_rate": 2.4009220849294544e-05,
      "loss": 0.317,
      "step": 17550
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.9652821189327888,
      "learning_rate": 2.3997759610127966e-05,
      "loss": 0.3073,
      "step": 17555
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8204171538106345,
      "learning_rate": 2.3986298581943482e-05,
      "loss": 0.294,
      "step": 17560
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8837246568376214,
      "learning_rate": 2.3974837767153767e-05,
      "loss": 0.3237,
      "step": 17565
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0141254398224684,
      "learning_rate": 2.3963377168171435e-05,
      "loss": 0.2863,
      "step": 17570
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0463994619453385,
      "learning_rate": 2.3951916787409065e-05,
      "loss": 0.3563,
      "step": 17575
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.91538276156957,
      "learning_rate": 2.3940456627279202e-05,
      "loss": 0.2804,
      "step": 17580
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.8199897389856885,
      "learning_rate": 2.3928996690194314e-05,
      "loss": 0.3123,
      "step": 17585
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9019888526692975,
      "learning_rate": 2.3917536978566835e-05,
      "loss": 0.323,
      "step": 17590
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9566856084089564,
      "learning_rate": 2.390607749480917e-05,
      "loss": 0.3045,
      "step": 17595
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.933611932594224,
      "learning_rate": 2.3894618241333638e-05,
      "loss": 0.3521,
      "step": 17600
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.7277770884507857,
      "learning_rate": 2.388315922055256e-05,
      "loss": 0.2754,
      "step": 17605
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.739979901483226,
      "learning_rate": 2.3871700434878154e-05,
      "loss": 0.2781,
      "step": 17610
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9622292982299171,
      "learning_rate": 2.3860241886722632e-05,
      "loss": 0.3037,
      "step": 17615
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8778439527538725,
      "learning_rate": 2.3848783578498124e-05,
      "loss": 0.2683,
      "step": 17620
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9834064417215478,
      "learning_rate": 2.383732551261674e-05,
      "loss": 0.2865,
      "step": 17625
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8127924211541777,
      "learning_rate": 2.3825867691490513e-05,
      "loss": 0.295,
      "step": 17630
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0956783553251446,
      "learning_rate": 2.3814410117531433e-05,
      "loss": 0.3838,
      "step": 17635
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0508621690876265,
      "learning_rate": 2.3802952793151438e-05,
      "loss": 0.3265,
      "step": 17640
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9314797582825766,
      "learning_rate": 2.3791495720762426e-05,
      "loss": 0.2913,
      "step": 17645
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0664927404042088,
      "learning_rate": 2.3780038902776226e-05,
      "loss": 0.322,
      "step": 17650
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.990039974345198,
      "learning_rate": 2.376858234160461e-05,
      "loss": 0.3088,
      "step": 17655
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.9445338526591907,
      "learning_rate": 2.3757126039659314e-05,
      "loss": 0.3368,
      "step": 17660
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.8944551492459483,
      "learning_rate": 2.3745669999352e-05,
      "loss": 0.3041,
      "step": 17665
    },
    {
      "epoch": 2.06,
      "grad_norm": 0.7361238350213304,
      "learning_rate": 2.3734214223094288e-05,
      "loss": 0.2971,
      "step": 17670
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9845168746919798,
      "learning_rate": 2.3722758713297756e-05,
      "loss": 0.2899,
      "step": 17675
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9908751426023606,
      "learning_rate": 2.3711303472373882e-05,
      "loss": 0.3371,
      "step": 17680
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0223196264062282,
      "learning_rate": 2.3699848502734126e-05,
      "loss": 0.3635,
      "step": 17685
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8192788140094526,
      "learning_rate": 2.3688393806789876e-05,
      "loss": 0.3069,
      "step": 17690
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9177860908789507,
      "learning_rate": 2.367693938695248e-05,
      "loss": 0.3248,
      "step": 17695
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9126884935041745,
      "learning_rate": 2.36654852456332e-05,
      "loss": 0.2971,
      "step": 17700
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8346638059648912,
      "learning_rate": 2.3654031385243248e-05,
      "loss": 0.4914,
      "step": 17705
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.7579956220568959,
      "learning_rate": 2.3642577808193793e-05,
      "loss": 0.2594,
      "step": 17710
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9826270141906627,
      "learning_rate": 2.3631124516895924e-05,
      "loss": 0.3118,
      "step": 17715
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0299733042950374,
      "learning_rate": 2.361967151376069e-05,
      "loss": 0.3255,
      "step": 17720
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.070308244402529,
      "learning_rate": 2.3608218801199054e-05,
      "loss": 0.3384,
      "step": 17725
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8309321333151022,
      "learning_rate": 2.3596766381621944e-05,
      "loss": 0.336,
      "step": 17730
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0075054481170602,
      "learning_rate": 2.3585314257440202e-05,
      "loss": 0.3402,
      "step": 17735
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8817988330814701,
      "learning_rate": 2.3573862431064636e-05,
      "loss": 0.271,
      "step": 17740
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.9206174617142147,
      "learning_rate": 2.3562410904905973e-05,
      "loss": 0.3056,
      "step": 17745
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.7865613476564897,
      "learning_rate": 2.3550959681374858e-05,
      "loss": 0.256,
      "step": 17750
    },
    {
      "epoch": 2.07,
      "grad_norm": 0.8522575872914615,
      "learning_rate": 2.353950876288191e-05,
      "loss": 0.349,
      "step": 17755
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.7993515698440024,
      "learning_rate": 2.3528058151837674e-05,
      "loss": 0.3321,
      "step": 17760
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9547298977592046,
      "learning_rate": 2.3516607850652604e-05,
      "loss": 0.2937,
      "step": 17765
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9414497661479948,
      "learning_rate": 2.3505157861737137e-05,
      "loss": 0.3422,
      "step": 17770
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.860544991148803,
      "learning_rate": 2.3493708187501592e-05,
      "loss": 0.3225,
      "step": 17775
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.7513022922227058,
      "learning_rate": 2.3482258830356243e-05,
      "loss": 0.3321,
      "step": 17780
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.7909143198209917,
      "learning_rate": 2.3470809792711312e-05,
      "loss": 0.258,
      "step": 17785
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1973799650308607,
      "learning_rate": 2.345936107697695e-05,
      "loss": 0.3477,
      "step": 17790
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.7830698848979976,
      "learning_rate": 2.3447912685563208e-05,
      "loss": 0.327,
      "step": 17795
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8341165224803347,
      "learning_rate": 2.3436464620880104e-05,
      "loss": 0.2748,
      "step": 17800
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.949280199241921,
      "learning_rate": 2.342501688533757e-05,
      "loss": 0.3201,
      "step": 17805
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8998789013407463,
      "learning_rate": 2.3413569481345487e-05,
      "loss": 0.3298,
      "step": 17810
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9546963615119591,
      "learning_rate": 2.3402122411313654e-05,
      "loss": 0.2971,
      "step": 17815
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1113313676138408,
      "learning_rate": 2.3390675677651777e-05,
      "loss": 0.3424,
      "step": 17820
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.97477551525791,
      "learning_rate": 2.3379229282769536e-05,
      "loss": 0.3156,
      "step": 17825
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.9457693468387023,
      "learning_rate": 2.33677832290765e-05,
      "loss": 0.2806,
      "step": 17830
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.825824585582643,
      "learning_rate": 2.3356337518982193e-05,
      "loss": 0.2881,
      "step": 17835
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8299506431675037,
      "learning_rate": 2.3344892154896063e-05,
      "loss": 0.3074,
      "step": 17840
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.8890205053188829,
      "learning_rate": 2.333344713922747e-05,
      "loss": 0.373,
      "step": 17845
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.7929268480744854,
      "learning_rate": 2.3322002474385708e-05,
      "loss": 0.2844,
      "step": 17850
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8673610739122468,
      "learning_rate": 2.3310558162780002e-05,
      "loss": 0.3196,
      "step": 17855
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.6794440522669515,
      "learning_rate": 2.329911420681951e-05,
      "loss": 0.3136,
      "step": 17860
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8922434111330548,
      "learning_rate": 2.328767060891328e-05,
      "loss": 0.3092,
      "step": 17865
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9063661460113771,
      "learning_rate": 2.327622737147032e-05,
      "loss": 0.3014,
      "step": 17870
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2526418271553343,
      "learning_rate": 2.3264784496899567e-05,
      "loss": 0.3411,
      "step": 17875
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0397573969835696,
      "learning_rate": 2.3253341987609844e-05,
      "loss": 0.3094,
      "step": 17880
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0375345942566607,
      "learning_rate": 2.324189984600994e-05,
      "loss": 0.3142,
      "step": 17885
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9354433773364085,
      "learning_rate": 2.323045807450853e-05,
      "loss": 0.3201,
      "step": 17890
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0440716763013076,
      "learning_rate": 2.3219016675514226e-05,
      "loss": 0.3009,
      "step": 17895
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8227562198699478,
      "learning_rate": 2.3207575651435567e-05,
      "loss": 0.2786,
      "step": 17900
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.8470677011857068,
      "learning_rate": 2.3196135004681015e-05,
      "loss": 0.3022,
      "step": 17905
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0751179292977704,
      "learning_rate": 2.3184694737658942e-05,
      "loss": 0.3175,
      "step": 17910
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.117780919164468,
      "learning_rate": 2.3173254852777635e-05,
      "loss": 0.3277,
      "step": 17915
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9172643757132686,
      "learning_rate": 2.3161815352445314e-05,
      "loss": 0.311,
      "step": 17920
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.944647364858098,
      "learning_rate": 2.3150376239070122e-05,
      "loss": 0.3401,
      "step": 17925
    },
    {
      "epoch": 2.09,
      "grad_norm": 0.9739125334211022,
      "learning_rate": 2.3138937515060113e-05,
      "loss": 0.3075,
      "step": 17930
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.7872825078347776,
      "learning_rate": 2.3127499182823238e-05,
      "loss": 0.3423,
      "step": 17935
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8438321710149619,
      "learning_rate": 2.311606124476741e-05,
      "loss": 0.3619,
      "step": 17940
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9577101179855763,
      "learning_rate": 2.3104623703300416e-05,
      "loss": 0.3245,
      "step": 17945
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8867860968444758,
      "learning_rate": 2.3093186560829984e-05,
      "loss": 0.3493,
      "step": 17950
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.7615531967554799,
      "learning_rate": 2.308174981976377e-05,
      "loss": 0.3213,
      "step": 17955
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8700634825323005,
      "learning_rate": 2.30703134825093e-05,
      "loss": 0.2801,
      "step": 17960
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1121489729524707,
      "learning_rate": 2.3058877551474056e-05,
      "loss": 0.3429,
      "step": 17965
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.015920965890387,
      "learning_rate": 2.3047442029065424e-05,
      "loss": 0.2998,
      "step": 17970
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9317820690993767,
      "learning_rate": 2.303600691769069e-05,
      "loss": 0.3194,
      "step": 17975
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9308927772521824,
      "learning_rate": 2.3024572219757087e-05,
      "loss": 0.3121,
      "step": 17980
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8715719095838016,
      "learning_rate": 2.3013137937671713e-05,
      "loss": 0.2792,
      "step": 17985
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.6480854795825243,
      "learning_rate": 2.3001704073841616e-05,
      "loss": 0.2478,
      "step": 17990
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8524422127329091,
      "learning_rate": 2.2990270630673743e-05,
      "loss": 0.3308,
      "step": 17995
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.8209643624248277,
      "learning_rate": 2.2978837610574964e-05,
      "loss": 0.2808,
      "step": 18000
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0015748101581854,
      "learning_rate": 2.2967405015952033e-05,
      "loss": 0.2942,
      "step": 18005
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.9178753838370047,
      "learning_rate": 2.2955972849211636e-05,
      "loss": 0.3218,
      "step": 18010
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.880660627565955,
      "learning_rate": 2.294454111276036e-05,
      "loss": 0.2732,
      "step": 18015
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9259840538083625,
      "learning_rate": 2.293310980900472e-05,
      "loss": 0.3087,
      "step": 18020
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.003457708044618,
      "learning_rate": 2.2921678940351117e-05,
      "loss": 0.3571,
      "step": 18025
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0543284321793844,
      "learning_rate": 2.2910248509205858e-05,
      "loss": 0.2857,
      "step": 18030
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.829469513748819,
      "learning_rate": 2.289881851797518e-05,
      "loss": 0.3608,
      "step": 18035
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9760987994680816,
      "learning_rate": 2.2887388969065215e-05,
      "loss": 0.3032,
      "step": 18040
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9954090334760022,
      "learning_rate": 2.2875959864882002e-05,
      "loss": 0.2673,
      "step": 18045
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9537178743048064,
      "learning_rate": 2.286453120783149e-05,
      "loss": 0.2574,
      "step": 18050
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0105368157624715,
      "learning_rate": 2.285310300031953e-05,
      "loss": 0.3446,
      "step": 18055
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9028534388400108,
      "learning_rate": 2.2841675244751864e-05,
      "loss": 0.2917,
      "step": 18060
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1006166058146305,
      "learning_rate": 2.2830247943534173e-05,
      "loss": 0.3239,
      "step": 18065
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8314069673933822,
      "learning_rate": 2.2818821099072026e-05,
      "loss": 0.3083,
      "step": 18070
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8732818465923707,
      "learning_rate": 2.280739471377088e-05,
      "loss": 0.2459,
      "step": 18075
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9263975825910838,
      "learning_rate": 2.279596879003611e-05,
      "loss": 0.2691,
      "step": 18080
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8929642614822747,
      "learning_rate": 2.2784543330273e-05,
      "loss": 0.3194,
      "step": 18085
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.9267118318702681,
      "learning_rate": 2.2773118336886724e-05,
      "loss": 0.3024,
      "step": 18090
    },
    {
      "epoch": 2.11,
      "grad_norm": 0.8654807620027696,
      "learning_rate": 2.2761693812282374e-05,
      "loss": 0.275,
      "step": 18095
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1141103936568186,
      "learning_rate": 2.2750269758864908e-05,
      "loss": 0.2969,
      "step": 18100
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9331348976820988,
      "learning_rate": 2.273884617903923e-05,
      "loss": 0.357,
      "step": 18105
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8641249789300807,
      "learning_rate": 2.2727423075210114e-05,
      "loss": 0.3441,
      "step": 18110
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0425715696452151,
      "learning_rate": 2.271600044978225e-05,
      "loss": 0.3697,
      "step": 18115
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8700057932394241,
      "learning_rate": 2.2704578305160224e-05,
      "loss": 0.2618,
      "step": 18120
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8765256429701753,
      "learning_rate": 2.2693156643748494e-05,
      "loss": 0.2987,
      "step": 18125
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8818888911087939,
      "learning_rate": 2.2681735467951455e-05,
      "loss": 0.3049,
      "step": 18130
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9990160192901573,
      "learning_rate": 2.267031478017339e-05,
      "loss": 0.315,
      "step": 18135
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9316250696122101,
      "learning_rate": 2.2658894582818473e-05,
      "loss": 0.3028,
      "step": 18140
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9582781812524059,
      "learning_rate": 2.264747487829076e-05,
      "loss": 0.2614,
      "step": 18145
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.8207848899410607,
      "learning_rate": 2.2636055668994223e-05,
      "loss": 0.2931,
      "step": 18150
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9568272704311052,
      "learning_rate": 2.262463695733274e-05,
      "loss": 0.3393,
      "step": 18155
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0587143343536545,
      "learning_rate": 2.261321874571005e-05,
      "loss": 0.2952,
      "step": 18160
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9705452970853029,
      "learning_rate": 2.2601801036529828e-05,
      "loss": 0.3477,
      "step": 18165
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9392396884471369,
      "learning_rate": 2.2590383832195607e-05,
      "loss": 0.3223,
      "step": 18170
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9669233535316395,
      "learning_rate": 2.2578967135110817e-05,
      "loss": 0.3629,
      "step": 18175
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9447899098335893,
      "learning_rate": 2.2567550947678812e-05,
      "loss": 0.301,
      "step": 18180
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.9715206016803206,
      "learning_rate": 2.2556135272302815e-05,
      "loss": 0.3335,
      "step": 18185
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8898709982860302,
      "learning_rate": 2.2544720111385946e-05,
      "loss": 0.3133,
      "step": 18190
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.0451956513969018,
      "learning_rate": 2.2533305467331205e-05,
      "loss": 0.3485,
      "step": 18195
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.7884296102556158,
      "learning_rate": 2.2521891342541504e-05,
      "loss": 0.3105,
      "step": 18200
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9206978124265363,
      "learning_rate": 2.2510477739419628e-05,
      "loss": 0.3507,
      "step": 18205
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9290153798588782,
      "learning_rate": 2.2499064660368276e-05,
      "loss": 0.333,
      "step": 18210
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9108268388135041,
      "learning_rate": 2.248765210779e-05,
      "loss": 0.2871,
      "step": 18215
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8729223519674997,
      "learning_rate": 2.247624008408728e-05,
      "loss": 0.2964,
      "step": 18220
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.870354251454955,
      "learning_rate": 2.246482859166245e-05,
      "loss": 0.2991,
      "step": 18225
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.078234666437474,
      "learning_rate": 2.2453417632917764e-05,
      "loss": 0.3253,
      "step": 18230
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.145530000137898,
      "learning_rate": 2.2442007210255345e-05,
      "loss": 0.327,
      "step": 18235
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8315235260130985,
      "learning_rate": 2.2430597326077195e-05,
      "loss": 0.3206,
      "step": 18240
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.8934828764310307,
      "learning_rate": 2.241918798278522e-05,
      "loss": 0.3156,
      "step": 18245
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.9226141806632508,
      "learning_rate": 2.2407779182781214e-05,
      "loss": 0.3273,
      "step": 18250
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.6704774582282024,
      "learning_rate": 2.239637092846684e-05,
      "loss": 0.2567,
      "step": 18255
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.153973848334559,
      "learning_rate": 2.2384963222243667e-05,
      "loss": 0.3028,
      "step": 18260
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.749385826493055,
      "learning_rate": 2.2373556066513118e-05,
      "loss": 0.2319,
      "step": 18265
    },
    {
      "epoch": 2.13,
      "grad_norm": 0.935174964858879,
      "learning_rate": 2.2362149463676536e-05,
      "loss": 0.2981,
      "step": 18270
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9733176057387588,
      "learning_rate": 2.2350743416135114e-05,
      "loss": 0.3924,
      "step": 18275
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8524327605791269,
      "learning_rate": 2.2339337926289965e-05,
      "loss": 0.349,
      "step": 18280
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8025760346566144,
      "learning_rate": 2.232793299654205e-05,
      "loss": 0.298,
      "step": 18285
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0213137232814853,
      "learning_rate": 2.231652862929222e-05,
      "loss": 0.3126,
      "step": 18290
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8798503065456813,
      "learning_rate": 2.2305124826941227e-05,
      "loss": 0.3112,
      "step": 18295
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8838245903461434,
      "learning_rate": 2.2293721591889687e-05,
      "loss": 0.2912,
      "step": 18300
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9019523235492393,
      "learning_rate": 2.228231892653811e-05,
      "loss": 0.3018,
      "step": 18305
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8957271836172304,
      "learning_rate": 2.2270916833286854e-05,
      "loss": 0.3394,
      "step": 18310
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8526951451315468,
      "learning_rate": 2.22595153145362e-05,
      "loss": 0.2867,
      "step": 18315
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0688657507311814,
      "learning_rate": 2.2248114372686267e-05,
      "loss": 0.3185,
      "step": 18320
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9883563035033653,
      "learning_rate": 2.2236714010137085e-05,
      "loss": 0.3437,
      "step": 18325
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0003111808483434,
      "learning_rate": 2.2225314229288563e-05,
      "loss": 0.3597,
      "step": 18330
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.9178490395686039,
      "learning_rate": 2.2213915032540457e-05,
      "loss": 0.2907,
      "step": 18335
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.7771970475602619,
      "learning_rate": 2.220251642229241e-05,
      "loss": 0.2581,
      "step": 18340
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1254601013105048,
      "learning_rate": 2.2191118400943968e-05,
      "loss": 0.284,
      "step": 18345
    },
    {
      "epoch": 2.14,
      "grad_norm": 0.8648829345674997,
      "learning_rate": 2.217972097089453e-05,
      "loss": 0.3172,
      "step": 18350
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0761924695941634,
      "learning_rate": 2.216832413454336e-05,
      "loss": 0.2933,
      "step": 18355
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.010747136086127,
      "learning_rate": 2.215692789428962e-05,
      "loss": 0.3212,
      "step": 18360
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8228467981530794,
      "learning_rate": 2.214553225253234e-05,
      "loss": 0.3518,
      "step": 18365
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9307398095949934,
      "learning_rate": 2.2134137211670423e-05,
      "loss": 0.3428,
      "step": 18370
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8408019872828656,
      "learning_rate": 2.212274277410265e-05,
      "loss": 0.3365,
      "step": 18375
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1117320976727552,
      "learning_rate": 2.2111348942227654e-05,
      "loss": 0.3178,
      "step": 18380
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.897556555638027,
      "learning_rate": 2.209995571844397e-05,
      "loss": 0.3081,
      "step": 18385
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8564390697391289,
      "learning_rate": 2.208856310514997e-05,
      "loss": 0.3313,
      "step": 18390
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9180043817157929,
      "learning_rate": 2.2077171104743952e-05,
      "loss": 0.3905,
      "step": 18395
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8611305380974957,
      "learning_rate": 2.206577971962403e-05,
      "loss": 0.3271,
      "step": 18400
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0356129986553817,
      "learning_rate": 2.2054388952188205e-05,
      "loss": 0.3212,
      "step": 18405
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0433556352168651,
      "learning_rate": 2.2042998804834362e-05,
      "loss": 0.3268,
      "step": 18410
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8957551962509526,
      "learning_rate": 2.2031609279960248e-05,
      "loss": 0.3076,
      "step": 18415
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9469706031649314,
      "learning_rate": 2.202022037996348e-05,
      "loss": 0.3488,
      "step": 18420
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.069691306867801,
      "learning_rate": 2.2008832107241523e-05,
      "loss": 0.2991,
      "step": 18425
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.8680007043591736,
      "learning_rate": 2.1997444464191745e-05,
      "loss": 0.3185,
      "step": 18430
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.9524391239303843,
      "learning_rate": 2.1986057453211357e-05,
      "loss": 0.3189,
      "step": 18435
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1634891773684417,
      "learning_rate": 2.1974671076697443e-05,
      "loss": 0.3099,
      "step": 18440
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.112284340124006,
      "learning_rate": 2.196328533704697e-05,
      "loss": 0.3404,
      "step": 18445
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.7276850845865299,
      "learning_rate": 2.1951900236656732e-05,
      "loss": 0.3269,
      "step": 18450
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0083017044753924,
      "learning_rate": 2.1940515777923423e-05,
      "loss": 0.327,
      "step": 18455
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8643501837061109,
      "learning_rate": 2.1929131963243595e-05,
      "loss": 0.3166,
      "step": 18460
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.008850448068876,
      "learning_rate": 2.1917748795013652e-05,
      "loss": 0.3069,
      "step": 18465
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.936178350374074,
      "learning_rate": 2.190636627562988e-05,
      "loss": 0.2842,
      "step": 18470
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0486685031021898,
      "learning_rate": 2.1894984407488407e-05,
      "loss": 0.3338,
      "step": 18475
    },
    {
      "epoch": 2.16,
      "grad_norm": 3.3593178941955615,
      "learning_rate": 2.1883603192985242e-05,
      "loss": 0.3639,
      "step": 18480
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8611971490316048,
      "learning_rate": 2.1872222634516247e-05,
      "loss": 0.3481,
      "step": 18485
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.973783465887101,
      "learning_rate": 2.1860842734477157e-05,
      "loss": 0.3446,
      "step": 18490
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9059663310557655,
      "learning_rate": 2.1849463495263546e-05,
      "loss": 0.3494,
      "step": 18495
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.7233760658678366,
      "learning_rate": 2.1838084919270878e-05,
      "loss": 0.3107,
      "step": 18500
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9666266908941069,
      "learning_rate": 2.1826707008894447e-05,
      "loss": 0.2824,
      "step": 18505
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2927638142707638,
      "learning_rate": 2.181532976652944e-05,
      "loss": 0.26,
      "step": 18510
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.112396121685976,
      "learning_rate": 2.1803953194570876e-05,
      "loss": 0.3153,
      "step": 18515
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.9197697793642021,
      "learning_rate": 2.1792577295413636e-05,
      "loss": 0.3381,
      "step": 18520
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8808141042044956,
      "learning_rate": 2.1781202071452473e-05,
      "loss": 0.3361,
      "step": 18525
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.8501400473105637,
      "learning_rate": 2.1769827525081992e-05,
      "loss": 0.3362,
      "step": 18530
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0049353574789925,
      "learning_rate": 2.1758453658696652e-05,
      "loss": 0.378,
      "step": 18535
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9190250097282301,
      "learning_rate": 2.1747080474690778e-05,
      "loss": 0.2885,
      "step": 18540
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.017124404057533,
      "learning_rate": 2.1735707975458536e-05,
      "loss": 0.3525,
      "step": 18545
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.85261659111868,
      "learning_rate": 2.1724336163393953e-05,
      "loss": 0.3307,
      "step": 18550
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8491688998189677,
      "learning_rate": 2.171296504089092e-05,
      "loss": 0.3086,
      "step": 18555
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8852672571434125,
      "learning_rate": 2.1701594610343186e-05,
      "loss": 0.2915,
      "step": 18560
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8785399334865858,
      "learning_rate": 2.1690224874144334e-05,
      "loss": 0.3112,
      "step": 18565
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9189243416216638,
      "learning_rate": 2.1678855834687808e-05,
      "loss": 0.3608,
      "step": 18570
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8851670260650392,
      "learning_rate": 2.1667487494366924e-05,
      "loss": 0.3172,
      "step": 18575
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9444106717230838,
      "learning_rate": 2.1656119855574823e-05,
      "loss": 0.2757,
      "step": 18580
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0490894692485158,
      "learning_rate": 2.1644752920704534e-05,
      "loss": 0.301,
      "step": 18585
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9044117705653503,
      "learning_rate": 2.1633386692148887e-05,
      "loss": 0.271,
      "step": 18590
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0132114129186218,
      "learning_rate": 2.1622021172300615e-05,
      "loss": 0.3379,
      "step": 18595
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8262285607743437,
      "learning_rate": 2.1610656363552263e-05,
      "loss": 0.2998,
      "step": 18600
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.886276912329097,
      "learning_rate": 2.159929226829625e-05,
      "loss": 0.3211,
      "step": 18605
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.9520381706715765,
      "learning_rate": 2.1587928888924846e-05,
      "loss": 0.3424,
      "step": 18610
    },
    {
      "epoch": 2.17,
      "grad_norm": 0.8813676335843862,
      "learning_rate": 2.1576566227830148e-05,
      "loss": 0.3187,
      "step": 18615
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8842688610429247,
      "learning_rate": 2.1565204287404112e-05,
      "loss": 0.2931,
      "step": 18620
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8635441066615837,
      "learning_rate": 2.155384307003856e-05,
      "loss": 0.309,
      "step": 18625
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1814391282253924,
      "learning_rate": 2.1542482578125143e-05,
      "loss": 0.3379,
      "step": 18630
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.98510832562715,
      "learning_rate": 2.1531122814055354e-05,
      "loss": 0.3187,
      "step": 18635
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8463320101883766,
      "learning_rate": 2.151976378022055e-05,
      "loss": 0.3154,
      "step": 18640
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9945810369310236,
      "learning_rate": 2.1508405479011924e-05,
      "loss": 0.2917,
      "step": 18645
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.9354568415984246,
      "learning_rate": 2.149704791282052e-05,
      "loss": 0.3189,
      "step": 18650
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.079737917001233,
      "learning_rate": 2.1485691084037227e-05,
      "loss": 0.2811,
      "step": 18655
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.934457210198989,
      "learning_rate": 2.1474334995052773e-05,
      "loss": 0.3033,
      "step": 18660
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.983614608893491,
      "learning_rate": 2.1462979648257723e-05,
      "loss": 0.3364,
      "step": 18665
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8757070321472686,
      "learning_rate": 2.145162504604251e-05,
      "loss": 0.321,
      "step": 18670
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.7639473189930548,
      "learning_rate": 2.1440271190797403e-05,
      "loss": 0.2831,
      "step": 18675
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.0222393347235514,
      "learning_rate": 2.1428918084912502e-05,
      "loss": 0.3378,
      "step": 18680
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.7759554606294536,
      "learning_rate": 2.141756573077774e-05,
      "loss": 0.2837,
      "step": 18685
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2215849766691051,
      "learning_rate": 2.140621413078292e-05,
      "loss": 0.3186,
      "step": 18690
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1188905490077006,
      "learning_rate": 2.1394863287317665e-05,
      "loss": 0.3247,
      "step": 18695
    },
    {
      "epoch": 2.18,
      "grad_norm": 0.8869815403849061,
      "learning_rate": 2.1383513202771463e-05,
      "loss": 0.3212,
      "step": 18700
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.0014657774041709,
      "learning_rate": 2.1372163879533605e-05,
      "loss": 0.3036,
      "step": 18705
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8399256289840166,
      "learning_rate": 2.1360815319993254e-05,
      "loss": 0.2979,
      "step": 18710
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9587365053171065,
      "learning_rate": 2.1349467526539396e-05,
      "loss": 0.3174,
      "step": 18715
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9773621666350173,
      "learning_rate": 2.1338120501560862e-05,
      "loss": 0.3673,
      "step": 18720
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9699215562631516,
      "learning_rate": 2.132677424744633e-05,
      "loss": 0.3199,
      "step": 18725
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1659867432406463,
      "learning_rate": 2.131542876658429e-05,
      "loss": 0.2846,
      "step": 18730
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2586784513783118,
      "learning_rate": 2.1304084061363086e-05,
      "loss": 0.2963,
      "step": 18735
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9883578232800853,
      "learning_rate": 2.1292740134170906e-05,
      "loss": 0.2978,
      "step": 18740
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8348168014760463,
      "learning_rate": 2.1281396987395758e-05,
      "loss": 0.3214,
      "step": 18745
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9235468088555026,
      "learning_rate": 2.1270054623425505e-05,
      "loss": 0.2982,
      "step": 18750
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.874670486891471,
      "learning_rate": 2.1258713044647814e-05,
      "loss": 0.275,
      "step": 18755
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.0817822379405924,
      "learning_rate": 2.124737225345022e-05,
      "loss": 0.2546,
      "step": 18760
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8760625455528133,
      "learning_rate": 2.123603225222007e-05,
      "loss": 0.3082,
      "step": 18765
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8080183160380161,
      "learning_rate": 2.1224693043344578e-05,
      "loss": 0.3184,
      "step": 18770
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.0758553886981022,
      "learning_rate": 2.121335462921073e-05,
      "loss": 0.2751,
      "step": 18775
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.8908569631670024,
      "learning_rate": 2.1202017012205397e-05,
      "loss": 0.3136,
      "step": 18780
    },
    {
      "epoch": 2.19,
      "grad_norm": 0.9068004507143597,
      "learning_rate": 2.1190680194715265e-05,
      "loss": 0.3649,
      "step": 18785
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8845124608014854,
      "learning_rate": 2.117934417912686e-05,
      "loss": 0.2974,
      "step": 18790
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.035245114722013,
      "learning_rate": 2.1168008967826534e-05,
      "loss": 0.3506,
      "step": 18795
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8648860971381006,
      "learning_rate": 2.1156674563200446e-05,
      "loss": 0.3232,
      "step": 18800
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8722927111964861,
      "learning_rate": 2.1145340967634628e-05,
      "loss": 0.3233,
      "step": 18805
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1044457948745663,
      "learning_rate": 2.1134008183514906e-05,
      "loss": 0.2843,
      "step": 18810
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8937792684462321,
      "learning_rate": 2.112267621322696e-05,
      "loss": 0.3102,
      "step": 18815
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.855978502051007,
      "learning_rate": 2.1111345059156295e-05,
      "loss": 0.2817,
      "step": 18820
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9737418379836537,
      "learning_rate": 2.1100014723688226e-05,
      "loss": 0.3009,
      "step": 18825
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8629444786204769,
      "learning_rate": 2.1088685209207898e-05,
      "loss": 0.2945,
      "step": 18830
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4690368546840122,
      "learning_rate": 2.107735651810031e-05,
      "loss": 0.4054,
      "step": 18835
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8851498431032873,
      "learning_rate": 2.1066028652750272e-05,
      "loss": 0.324,
      "step": 18840
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8824187781320911,
      "learning_rate": 2.1054701615542407e-05,
      "loss": 0.311,
      "step": 18845
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8342603278944576,
      "learning_rate": 2.104337540886117e-05,
      "loss": 0.3074,
      "step": 18850
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.07601561771372,
      "learning_rate": 2.1032050035090865e-05,
      "loss": 0.337,
      "step": 18855
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8537132451686101,
      "learning_rate": 2.1020725496615585e-05,
      "loss": 0.2634,
      "step": 18860
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.8515139235900822,
      "learning_rate": 2.1009401795819277e-05,
      "loss": 0.2713,
      "step": 18865
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.9345747288714233,
      "learning_rate": 2.0998078935085686e-05,
      "loss": 0.3411,
      "step": 18870
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9128851451492647,
      "learning_rate": 2.0986756916798403e-05,
      "loss": 0.3187,
      "step": 18875
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9787631223096794,
      "learning_rate": 2.0975435743340822e-05,
      "loss": 0.31,
      "step": 18880
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.0338863888769156,
      "learning_rate": 2.096411541709618e-05,
      "loss": 0.2932,
      "step": 18885
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9340066213116687,
      "learning_rate": 2.0952795940447522e-05,
      "loss": 0.3045,
      "step": 18890
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8842521690733544,
      "learning_rate": 2.09414773157777e-05,
      "loss": 0.2818,
      "step": 18895
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9653041989853287,
      "learning_rate": 2.093015954546942e-05,
      "loss": 0.2922,
      "step": 18900
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9981339006893417,
      "learning_rate": 2.0918842631905195e-05,
      "loss": 0.3978,
      "step": 18905
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9597187235787491,
      "learning_rate": 2.0907526577467345e-05,
      "loss": 0.2983,
      "step": 18910
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.964709160695622,
      "learning_rate": 2.089621138453801e-05,
      "loss": 0.3161,
      "step": 18915
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1290528259316996,
      "learning_rate": 2.0884897055499173e-05,
      "loss": 0.3303,
      "step": 18920
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8965214514996026,
      "learning_rate": 2.0873583592732602e-05,
      "loss": 0.2818,
      "step": 18925
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9468307837590071,
      "learning_rate": 2.0862270998619908e-05,
      "loss": 0.3145,
      "step": 18930
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9813663049968601,
      "learning_rate": 2.0850959275542527e-05,
      "loss": 0.3266,
      "step": 18935
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9818995648649408,
      "learning_rate": 2.0839648425881668e-05,
      "loss": 0.3798,
      "step": 18940
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2844426306133414,
      "learning_rate": 2.0828338452018396e-05,
      "loss": 0.3421,
      "step": 18945
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.8618822871814008,
      "learning_rate": 2.0817029356333574e-05,
      "loss": 0.3069,
      "step": 18950
    },
    {
      "epoch": 2.21,
      "grad_norm": 0.9598016210865685,
      "learning_rate": 2.0805721141207894e-05,
      "loss": 0.2491,
      "step": 18955
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8545341745506732,
      "learning_rate": 2.0794413809021853e-05,
      "loss": 0.2721,
      "step": 18960
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9669650056308706,
      "learning_rate": 2.078310736215575e-05,
      "loss": 0.3128,
      "step": 18965
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9885806359417516,
      "learning_rate": 2.0771801802989724e-05,
      "loss": 0.2915,
      "step": 18970
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.0368117798149952,
      "learning_rate": 2.07604971339037e-05,
      "loss": 0.3407,
      "step": 18975
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9066354449033249,
      "learning_rate": 2.0749193357277452e-05,
      "loss": 0.3776,
      "step": 18980
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.992363809705979,
      "learning_rate": 2.0737890475490516e-05,
      "loss": 0.343,
      "step": 18985
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.019680567735476,
      "learning_rate": 2.0726588490922288e-05,
      "loss": 0.3354,
      "step": 18990
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.8962852655172553,
      "learning_rate": 2.071528740595194e-05,
      "loss": 0.3085,
      "step": 18995
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9242524408274967,
      "learning_rate": 2.070398722295848e-05,
      "loss": 0.2694,
      "step": 19000
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.0338168520135511,
      "learning_rate": 2.0692687944320716e-05,
      "loss": 0.3348,
      "step": 19005
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.9408727091633743,
      "learning_rate": 2.0681389572417244e-05,
      "loss": 0.3465,
      "step": 19010
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.0213095798049854,
      "learning_rate": 2.067009210962651e-05,
      "loss": 0.2908,
      "step": 19015
    },
    {
      "epoch": 2.22,
      "grad_norm": 0.99087844897229,
      "learning_rate": 2.0658795558326743e-05,
      "loss": 0.2452,
      "step": 19020
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.0056084115234374,
      "learning_rate": 2.064749992089598e-05,
      "loss": 0.2724,
      "step": 19025
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.0505525269988698,
      "learning_rate": 2.0636205199712088e-05,
      "loss": 0.3437,
      "step": 19030
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.0061621842171282,
      "learning_rate": 2.062491139715271e-05,
      "loss": 0.328,
      "step": 19035
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.097324635588219,
      "learning_rate": 2.0613618515595302e-05,
      "loss": 0.3263,
      "step": 19040
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9248932166460767,
      "learning_rate": 2.060232655741714e-05,
      "loss": 0.287,
      "step": 19045
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2223474490151356,
      "learning_rate": 2.059103552499532e-05,
      "loss": 0.3625,
      "step": 19050
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9914663311262445,
      "learning_rate": 2.0579745420706698e-05,
      "loss": 0.337,
      "step": 19055
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0915530577272503,
      "learning_rate": 2.0568456246927957e-05,
      "loss": 0.3098,
      "step": 19060
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0191701121067873,
      "learning_rate": 2.0557168006035604e-05,
      "loss": 0.3384,
      "step": 19065
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0424222717286833,
      "learning_rate": 2.0545880700405924e-05,
      "loss": 0.276,
      "step": 19070
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.88243624258224,
      "learning_rate": 2.0534594332415022e-05,
      "loss": 0.3213,
      "step": 19075
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9466989063921294,
      "learning_rate": 2.0523308904438775e-05,
      "loss": 0.2971,
      "step": 19080
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8594658974255374,
      "learning_rate": 2.0512024418852902e-05,
      "loss": 0.2777,
      "step": 19085
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0495013322238314,
      "learning_rate": 2.0500740878032894e-05,
      "loss": 0.3512,
      "step": 19090
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9080577979051785,
      "learning_rate": 2.048945828435406e-05,
      "loss": 0.331,
      "step": 19095
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8811052628893118,
      "learning_rate": 2.0478176640191517e-05,
      "loss": 0.2599,
      "step": 19100
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.0627314793894536,
      "learning_rate": 2.046689594792015e-05,
      "loss": 0.3088,
      "step": 19105
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3021815605428533,
      "learning_rate": 2.0455616209914668e-05,
      "loss": 0.336,
      "step": 19110
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.9436131936328238,
      "learning_rate": 2.044433742854958e-05,
      "loss": 0.3173,
      "step": 19115
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.8156132958112147,
      "learning_rate": 2.0433059606199187e-05,
      "loss": 0.3251,
      "step": 19120
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1041518148813803,
      "learning_rate": 2.0421782745237574e-05,
      "loss": 0.312,
      "step": 19125
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8949361717279436,
      "learning_rate": 2.0410506848038653e-05,
      "loss": 0.3462,
      "step": 19130
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9772182318208312,
      "learning_rate": 2.0399231916976118e-05,
      "loss": 0.3461,
      "step": 19135
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9071619935038618,
      "learning_rate": 2.038795795442345e-05,
      "loss": 0.3412,
      "step": 19140
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8126851347200398,
      "learning_rate": 2.037668496275396e-05,
      "loss": 0.2758,
      "step": 19145
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8636289614269506,
      "learning_rate": 2.0365412944340705e-05,
      "loss": 0.3192,
      "step": 19150
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.0520663722279355,
      "learning_rate": 2.0354141901556564e-05,
      "loss": 0.2935,
      "step": 19155
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8565307240481904,
      "learning_rate": 2.034287183677422e-05,
      "loss": 0.3315,
      "step": 19160
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9643274687575706,
      "learning_rate": 2.0331602752366145e-05,
      "loss": 0.2867,
      "step": 19165
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1266583053840489,
      "learning_rate": 2.0320334650704594e-05,
      "loss": 0.35,
      "step": 19170
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9972158709128763,
      "learning_rate": 2.0309067534161607e-05,
      "loss": 0.3005,
      "step": 19175
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.962264043008785,
      "learning_rate": 2.0297801405109045e-05,
      "loss": 0.3173,
      "step": 19180
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8794429143721346,
      "learning_rate": 2.0286536265918545e-05,
      "loss": 0.3136,
      "step": 19185
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8807659876711478,
      "learning_rate": 2.027527211896153e-05,
      "loss": 0.3009,
      "step": 19190
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.0474060369169909,
      "learning_rate": 2.0264008966609235e-05,
      "loss": 0.3033,
      "step": 19195
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.9802655467410194,
      "learning_rate": 2.025274681123266e-05,
      "loss": 0.2801,
      "step": 19200
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.8487097465263694,
      "learning_rate": 2.0241485655202595e-05,
      "loss": 0.2609,
      "step": 19205
    },
    {
      "epoch": 2.24,
      "grad_norm": 85.35932775830166,
      "learning_rate": 2.023022550088965e-05,
      "loss": 0.5221,
      "step": 19210
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8750779760450615,
      "learning_rate": 2.021896635066421e-05,
      "loss": 0.3384,
      "step": 19215
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.0004307742160832,
      "learning_rate": 2.0207708206896432e-05,
      "loss": 0.2956,
      "step": 19220
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8804472066511255,
      "learning_rate": 2.0196451071956267e-05,
      "loss": 0.2984,
      "step": 19225
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8400089124620665,
      "learning_rate": 2.018519494821347e-05,
      "loss": 0.2818,
      "step": 19230
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9278831009910146,
      "learning_rate": 2.017393983803757e-05,
      "loss": 0.3385,
      "step": 19235
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8982325636562568,
      "learning_rate": 2.0162685743797895e-05,
      "loss": 0.2913,
      "step": 19240
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.243173813072387,
      "learning_rate": 2.0151432667863533e-05,
      "loss": 0.3507,
      "step": 19245
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8428910331745797,
      "learning_rate": 2.014018061260339e-05,
      "loss": 0.2919,
      "step": 19250
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.0254555271890058,
      "learning_rate": 2.0128929580386124e-05,
      "loss": 0.3088,
      "step": 19255
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.0208843920868895,
      "learning_rate": 2.011767957358021e-05,
      "loss": 0.2895,
      "step": 19260
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.856232668905099,
      "learning_rate": 2.01064305945539e-05,
      "loss": 0.3078,
      "step": 19265
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9353336958152558,
      "learning_rate": 2.0095182645675192e-05,
      "loss": 0.4259,
      "step": 19270
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.0303151759896785,
      "learning_rate": 2.0083935729311917e-05,
      "loss": 0.3164,
      "step": 19275
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9514570444388372,
      "learning_rate": 2.007268984783167e-05,
      "loss": 0.302,
      "step": 19280
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.8084884408234481,
      "learning_rate": 2.0061445003601835e-05,
      "loss": 0.3265,
      "step": 19285
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9002291361118132,
      "learning_rate": 2.0050201198989543e-05,
      "loss": 0.2801,
      "step": 19290
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.092576421543211,
      "learning_rate": 2.0038958436361746e-05,
      "loss": 0.3198,
      "step": 19295
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.9471667202195733,
      "learning_rate": 2.002771671808517e-05,
      "loss": 0.3527,
      "step": 19300
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9247703460511966,
      "learning_rate": 2.0016476046526305e-05,
      "loss": 0.2766,
      "step": 19305
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9019095311137985,
      "learning_rate": 2.0005236424051444e-05,
      "loss": 0.309,
      "step": 19310
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8688007671477888,
      "learning_rate": 1.999399785302663e-05,
      "loss": 0.2816,
      "step": 19315
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8514171760719479,
      "learning_rate": 1.9982760335817703e-05,
      "loss": 0.3171,
      "step": 19320
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9165854555566252,
      "learning_rate": 1.9971523874790278e-05,
      "loss": 0.3974,
      "step": 19325
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.0556633992010753,
      "learning_rate": 1.9960288472309756e-05,
      "loss": 0.2982,
      "step": 19330
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8758221087250375,
      "learning_rate": 1.9949054130741307e-05,
      "loss": 0.2897,
      "step": 19335
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.0588190353192888,
      "learning_rate": 1.993782085244986e-05,
      "loss": 0.3258,
      "step": 19340
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.0266069382429226,
      "learning_rate": 1.9926588639800154e-05,
      "loss": 0.3194,
      "step": 19345
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.082861249746999,
      "learning_rate": 1.991535749515668e-05,
      "loss": 0.3441,
      "step": 19350
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8490691732837352,
      "learning_rate": 1.990412742088372e-05,
      "loss": 0.344,
      "step": 19355
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.03528974974153,
      "learning_rate": 1.9892898419345312e-05,
      "loss": 0.328,
      "step": 19360
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8596815033383882,
      "learning_rate": 1.988167049290529e-05,
      "loss": 0.3007,
      "step": 19365
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.9555655197639726,
      "learning_rate": 1.9870443643927233e-05,
      "loss": 0.3161,
      "step": 19370
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1913956772242842,
      "learning_rate": 1.9859217874774528e-05,
      "loss": 0.3618,
      "step": 19375
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8704937448994219,
      "learning_rate": 1.9847993187810316e-05,
      "loss": 0.2792,
      "step": 19380
    },
    {
      "epoch": 2.26,
      "grad_norm": 0.8960534221328859,
      "learning_rate": 1.9836769585397492e-05,
      "loss": 0.3149,
      "step": 19385
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.8544079730529959,
      "learning_rate": 1.982554706989875e-05,
      "loss": 0.2909,
      "step": 19390
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.7973887460843961,
      "learning_rate": 1.981432564367657e-05,
      "loss": 0.3271,
      "step": 19395
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.878333975664547,
      "learning_rate": 1.9803105309093145e-05,
      "loss": 0.2919,
      "step": 19400
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1221867278169215,
      "learning_rate": 1.9791886068510503e-05,
      "loss": 0.3255,
      "step": 19405
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9822828299529107,
      "learning_rate": 1.978066792429039e-05,
      "loss": 0.3309,
      "step": 19410
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9738605039542356,
      "learning_rate": 1.9769450878794356e-05,
      "loss": 0.3105,
      "step": 19415
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9346212942476916,
      "learning_rate": 1.9758234934383694e-05,
      "loss": 0.3294,
      "step": 19420
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.8317241589287779,
      "learning_rate": 1.9747020093419495e-05,
      "loss": 0.2977,
      "step": 19425
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9284375730977665,
      "learning_rate": 1.9735806358262588e-05,
      "loss": 0.2905,
      "step": 19430
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9742098549584641,
      "learning_rate": 1.9724593731273577e-05,
      "loss": 0.3668,
      "step": 19435
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.050345894412471,
      "learning_rate": 1.971338221481285e-05,
      "loss": 0.2973,
      "step": 19440
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.8305778127957896,
      "learning_rate": 1.9702171811240543e-05,
      "loss": 0.2391,
      "step": 19445
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.8919978112261086,
      "learning_rate": 1.9690962522916574e-05,
      "loss": 0.3464,
      "step": 19450
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.9360553419905666,
      "learning_rate": 1.967975435220059e-05,
      "loss": 0.3149,
      "step": 19455
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.0851103178228876,
      "learning_rate": 1.9668547301452055e-05,
      "loss": 0.3382,
      "step": 19460
    },
    {
      "epoch": 2.27,
      "grad_norm": 0.8640620889860231,
      "learning_rate": 1.9657341373030153e-05,
      "loss": 0.2878,
      "step": 19465
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.0539855484445155,
      "learning_rate": 1.9646136569293852e-05,
      "loss": 0.2915,
      "step": 19470
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.884082682077803,
      "learning_rate": 1.96349328926019e-05,
      "loss": 0.2854,
      "step": 19475
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.7959663538683446,
      "learning_rate": 1.9623730345312763e-05,
      "loss": 0.3067,
      "step": 19480
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8980690506314459,
      "learning_rate": 1.96125289297847e-05,
      "loss": 0.3378,
      "step": 19485
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.069052419234371,
      "learning_rate": 1.960132864837574e-05,
      "loss": 0.3395,
      "step": 19490
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.7989861579540782,
      "learning_rate": 1.959012950344365e-05,
      "loss": 0.2348,
      "step": 19495
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9175251853040606,
      "learning_rate": 1.957893149734596e-05,
      "loss": 0.3043,
      "step": 19500
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8844698023071313,
      "learning_rate": 1.9567734632439974e-05,
      "loss": 0.255,
      "step": 19505
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.967259586317296,
      "learning_rate": 1.9556538911082755e-05,
      "loss": 0.3042,
      "step": 19510
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9339863995744397,
      "learning_rate": 1.9545344335631114e-05,
      "loss": 0.3416,
      "step": 19515
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8276360288505716,
      "learning_rate": 1.9534150908441633e-05,
      "loss": 0.2616,
      "step": 19520
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9926911385271557,
      "learning_rate": 1.9522958631870632e-05,
      "loss": 0.3257,
      "step": 19525
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9526508330226306,
      "learning_rate": 1.9511767508274214e-05,
      "loss": 0.2965,
      "step": 19530
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2362134828162996,
      "learning_rate": 1.950057754000822e-05,
      "loss": 0.3141,
      "step": 19535
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.857833732272555,
      "learning_rate": 1.9489388729428264e-05,
      "loss": 0.3232,
      "step": 19540
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.9793739174366688,
      "learning_rate": 1.9478201078889713e-05,
      "loss": 0.3348,
      "step": 19545
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.7979001385092815,
      "learning_rate": 1.9467014590747662e-05,
      "loss": 0.3015,
      "step": 19550
    },
    {
      "epoch": 2.28,
      "grad_norm": 0.8553307227067145,
      "learning_rate": 1.9455829267357e-05,
      "loss": 0.2654,
      "step": 19555
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1191995705594253,
      "learning_rate": 1.9444645111072352e-05,
      "loss": 0.369,
      "step": 19560
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9193459249953835,
      "learning_rate": 1.9433462124248108e-05,
      "loss": 0.3283,
      "step": 19565
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.836878614391158,
      "learning_rate": 1.9422280309238384e-05,
      "loss": 0.3285,
      "step": 19570
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1016396311345666,
      "learning_rate": 1.9411099668397085e-05,
      "loss": 0.3217,
      "step": 19575
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0582528640245685,
      "learning_rate": 1.939992020407785e-05,
      "loss": 0.3323,
      "step": 19580
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9747675360002149,
      "learning_rate": 1.9388741918634067e-05,
      "loss": 0.3395,
      "step": 19585
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.0523604250854264,
      "learning_rate": 1.93775648144189e-05,
      "loss": 0.3129,
      "step": 19590
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.8831891685961533,
      "learning_rate": 1.9366388893785228e-05,
      "loss": 0.3044,
      "step": 19595
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.983763036952166,
      "learning_rate": 1.9355214159085702e-05,
      "loss": 0.3027,
      "step": 19600
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.8033630850470433,
      "learning_rate": 1.934404061267273e-05,
      "loss": 0.2847,
      "step": 19605
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9486630028808671,
      "learning_rate": 1.9332868256898447e-05,
      "loss": 0.2703,
      "step": 19610
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9180339767802568,
      "learning_rate": 1.932169709411477e-05,
      "loss": 0.3165,
      "step": 19615
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9960696458027846,
      "learning_rate": 1.931052712667332e-05,
      "loss": 0.2882,
      "step": 19620
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9388677311214171,
      "learning_rate": 1.9299358356925517e-05,
      "loss": 0.317,
      "step": 19625
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.8693859957602147,
      "learning_rate": 1.928819078722249e-05,
      "loss": 0.3413,
      "step": 19630
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9481992395450557,
      "learning_rate": 1.927702441991514e-05,
      "loss": 0.3446,
      "step": 19635
    },
    {
      "epoch": 2.29,
      "grad_norm": 0.9603315436364415,
      "learning_rate": 1.9265859257354084e-05,
      "loss": 0.3323,
      "step": 19640
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.0232614614090032,
      "learning_rate": 1.925469530188973e-05,
      "loss": 0.3099,
      "step": 19645
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.871830739955941,
      "learning_rate": 1.924353255587219e-05,
      "loss": 0.3054,
      "step": 19650
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.8832566520955866,
      "learning_rate": 1.9232371021651348e-05,
      "loss": 0.322,
      "step": 19655
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9049802543319408,
      "learning_rate": 1.922121070157683e-05,
      "loss": 0.3193,
      "step": 19660
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9497730070594884,
      "learning_rate": 1.921005159799798e-05,
      "loss": 0.359,
      "step": 19665
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.7390709369195414,
      "learning_rate": 1.9198893713263915e-05,
      "loss": 0.3002,
      "step": 19670
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9233609742716121,
      "learning_rate": 1.9187737049723498e-05,
      "loss": 0.3276,
      "step": 19675
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9162390278085042,
      "learning_rate": 1.9176581609725307e-05,
      "loss": 0.2535,
      "step": 19680
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9043474635288291,
      "learning_rate": 1.9165427395617704e-05,
      "loss": 0.3366,
      "step": 19685
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.041571717313108,
      "learning_rate": 1.9154274409748742e-05,
      "loss": 0.3126,
      "step": 19690
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.0242304000187896,
      "learning_rate": 1.914312265446624e-05,
      "loss": 0.2696,
      "step": 19695
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9500210492750076,
      "learning_rate": 1.9131972132117776e-05,
      "loss": 0.3057,
      "step": 19700
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9595355388664257,
      "learning_rate": 1.912082284505065e-05,
      "loss": 0.3239,
      "step": 19705
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9843058650941644,
      "learning_rate": 1.9109674795611898e-05,
      "loss": 0.295,
      "step": 19710
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9497760161615868,
      "learning_rate": 1.9098527986148294e-05,
      "loss": 0.3141,
      "step": 19715
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9319852658392956,
      "learning_rate": 1.908738241900638e-05,
      "loss": 0.3126,
      "step": 19720
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.9517394251860755,
      "learning_rate": 1.9076238096532387e-05,
      "loss": 0.3052,
      "step": 19725
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8776288579568556,
      "learning_rate": 1.9065095021072338e-05,
      "loss": 0.3147,
      "step": 19730
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9956225821780308,
      "learning_rate": 1.9053953194971947e-05,
      "loss": 0.275,
      "step": 19735
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9430343748420174,
      "learning_rate": 1.9042812620576693e-05,
      "loss": 0.3057,
      "step": 19740
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.92481447819531,
      "learning_rate": 1.9031673300231782e-05,
      "loss": 0.3181,
      "step": 19745
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8562383790311145,
      "learning_rate": 1.902053523628216e-05,
      "loss": 0.3391,
      "step": 19750
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8178204357095703,
      "learning_rate": 1.900939843107251e-05,
      "loss": 0.2893,
      "step": 19755
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.0002941132167138,
      "learning_rate": 1.899826288694724e-05,
      "loss": 0.3398,
      "step": 19760
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9376293134034597,
      "learning_rate": 1.8987128606250496e-05,
      "loss": 0.3366,
      "step": 19765
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9937934518182798,
      "learning_rate": 1.8975995591326168e-05,
      "loss": 0.3415,
      "step": 19770
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9731590454269231,
      "learning_rate": 1.8964863844517873e-05,
      "loss": 0.3192,
      "step": 19775
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.9642478291439845,
      "learning_rate": 1.895373336816895e-05,
      "loss": 0.3168,
      "step": 19780
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.0842544573081947,
      "learning_rate": 1.8942604164622484e-05,
      "loss": 0.2827,
      "step": 19785
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8549012298612148,
      "learning_rate": 1.8931476236221298e-05,
      "loss": 0.3364,
      "step": 19790
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1556458673996979,
      "learning_rate": 1.8920349585307925e-05,
      "loss": 0.2749,
      "step": 19795
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.7874059307206615,
      "learning_rate": 1.8909224214224662e-05,
      "loss": 0.3078,
      "step": 19800
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.862194654250148,
      "learning_rate": 1.8898100125313494e-05,
      "loss": 0.2583,
      "step": 19805
    },
    {
      "epoch": 2.31,
      "grad_norm": 0.8802550167148678,
      "learning_rate": 1.8886977320916163e-05,
      "loss": 0.3166,
      "step": 19810
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.807661318717498,
      "learning_rate": 1.887585580337414e-05,
      "loss": 0.2529,
      "step": 19815
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9838482239265266,
      "learning_rate": 1.8864735575028623e-05,
      "loss": 0.3496,
      "step": 19820
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9378745616535644,
      "learning_rate": 1.885361663822054e-05,
      "loss": 0.2921,
      "step": 19825
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9683712557290653,
      "learning_rate": 1.884249899529053e-05,
      "loss": 0.3071,
      "step": 19830
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8797188305993103,
      "learning_rate": 1.883138264857898e-05,
      "loss": 0.3515,
      "step": 19835
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.923157112158475,
      "learning_rate": 1.8820267600425996e-05,
      "loss": 0.3242,
      "step": 19840
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.0043636550225985,
      "learning_rate": 1.8809153853171426e-05,
      "loss": 0.3603,
      "step": 19845
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.884430837449619,
      "learning_rate": 1.8798041409154807e-05,
      "loss": 0.3015,
      "step": 19850
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8884575530989697,
      "learning_rate": 1.878693027071544e-05,
      "loss": 0.2682,
      "step": 19855
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8892073683106652,
      "learning_rate": 1.8775820440192332e-05,
      "loss": 0.2965,
      "step": 19860
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9757747820687703,
      "learning_rate": 1.8764711919924215e-05,
      "loss": 0.347,
      "step": 19865
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1857940083995246,
      "learning_rate": 1.8753604712249565e-05,
      "loss": 0.3276,
      "step": 19870
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.780213776198026,
      "learning_rate": 1.874249881950655e-05,
      "loss": 0.2825,
      "step": 19875
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.8247014195197875,
      "learning_rate": 1.873139424403307e-05,
      "loss": 0.3045,
      "step": 19880
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9065511886311877,
      "learning_rate": 1.8720290988166773e-05,
      "loss": 0.3149,
      "step": 19885
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9543313082655942,
      "learning_rate": 1.8709189054244996e-05,
      "loss": 0.3119,
      "step": 19890
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.9439106028514123,
      "learning_rate": 1.8698088444604833e-05,
      "loss": 0.3012,
      "step": 19895
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2122520388016078,
      "learning_rate": 1.8686989161583053e-05,
      "loss": 0.3253,
      "step": 19900
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9243484416547644,
      "learning_rate": 1.8675891207516185e-05,
      "loss": 0.3067,
      "step": 19905
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.8879016294564582,
      "learning_rate": 1.866479458474046e-05,
      "loss": 0.3367,
      "step": 19910
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9467859236463569,
      "learning_rate": 1.8653699295591843e-05,
      "loss": 0.3168,
      "step": 19915
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.8253396517583389,
      "learning_rate": 1.8642605342405998e-05,
      "loss": 0.3012,
      "step": 19920
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.897623523671182,
      "learning_rate": 1.863151272751831e-05,
      "loss": 0.2899,
      "step": 19925
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9663078010236645,
      "learning_rate": 1.8620421453263915e-05,
      "loss": 0.3093,
      "step": 19930
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1058123989036994,
      "learning_rate": 1.8609331521977623e-05,
      "loss": 0.2874,
      "step": 19935
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.979926423581943,
      "learning_rate": 1.8598242935994e-05,
      "loss": 0.3074,
      "step": 19940
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9119635970877846,
      "learning_rate": 1.8587155697647283e-05,
      "loss": 0.2864,
      "step": 19945
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.0677179479240284,
      "learning_rate": 1.8576069809271475e-05,
      "loss": 0.3143,
      "step": 19950
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.0704260422345384,
      "learning_rate": 1.8564985273200253e-05,
      "loss": 0.2875,
      "step": 19955
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.0827876965857768,
      "learning_rate": 1.855390209176704e-05,
      "loss": 0.3595,
      "step": 19960
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7358631958751243,
      "learning_rate": 1.8542820267304966e-05,
      "loss": 0.2646,
      "step": 19965
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9378841438888215,
      "learning_rate": 1.853173980214686e-05,
      "loss": 0.3298,
      "step": 19970
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9520672581606249,
      "learning_rate": 1.8520660698625275e-05,
      "loss": 0.3034,
      "step": 19975
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9944129246574176,
      "learning_rate": 1.8509582959072486e-05,
      "loss": 0.3381,
      "step": 19980
    },
    {
      "epoch": 2.33,
      "grad_norm": 0.9070642683200693,
      "learning_rate": 1.849850658582048e-05,
      "loss": 0.2888,
      "step": 19985
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.0002980590119825,
      "learning_rate": 1.8487431581200933e-05,
      "loss": 0.3067,
      "step": 19990
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9572080732898964,
      "learning_rate": 1.8476357947545252e-05,
      "loss": 0.2876,
      "step": 19995
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.17871576249345,
      "learning_rate": 1.8465285687184562e-05,
      "loss": 0.3532,
      "step": 20000
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.0592207583832656,
      "learning_rate": 1.845421480244969e-05,
      "loss": 0.352,
      "step": 20005
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9706162294261611,
      "learning_rate": 1.8443145295671165e-05,
      "loss": 0.3119,
      "step": 20010
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.806236961695897,
      "learning_rate": 1.8432077169179235e-05,
      "loss": 0.2668,
      "step": 20015
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8750205284897073,
      "learning_rate": 1.8421010425303865e-05,
      "loss": 0.3402,
      "step": 20020
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9711022131903482,
      "learning_rate": 1.8409945066374706e-05,
      "loss": 0.3262,
      "step": 20025
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8227846744115173,
      "learning_rate": 1.839888109472115e-05,
      "loss": 0.2799,
      "step": 20030
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9430191483338972,
      "learning_rate": 1.838781851267227e-05,
      "loss": 0.3193,
      "step": 20035
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8561815774260501,
      "learning_rate": 1.8376757322556847e-05,
      "loss": 0.2757,
      "step": 20040
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1202063880113722,
      "learning_rate": 1.8365697526703384e-05,
      "loss": 0.3899,
      "step": 20045
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.9679879927630963,
      "learning_rate": 1.8354639127440097e-05,
      "loss": 0.337,
      "step": 20050
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8964955923152649,
      "learning_rate": 1.8343582127094884e-05,
      "loss": 0.3162,
      "step": 20055
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8628627585749146,
      "learning_rate": 1.8332526527995347e-05,
      "loss": 0.3038,
      "step": 20060
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.7741338316261027,
      "learning_rate": 1.8321472332468825e-05,
      "loss": 0.306,
      "step": 20065
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.8581152877946427,
      "learning_rate": 1.8310419542842327e-05,
      "loss": 0.351,
      "step": 20070
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.259252654960738,
      "learning_rate": 1.8299368161442588e-05,
      "loss": 0.3686,
      "step": 20075
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9003352820308149,
      "learning_rate": 1.828831819059605e-05,
      "loss": 0.2956,
      "step": 20080
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.898095328432729,
      "learning_rate": 1.827726963262883e-05,
      "loss": 0.2842,
      "step": 20085
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8104156384289827,
      "learning_rate": 1.8266222489866775e-05,
      "loss": 0.319,
      "step": 20090
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9225468555985851,
      "learning_rate": 1.8255176764635414e-05,
      "loss": 0.2555,
      "step": 20095
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9545276851350571,
      "learning_rate": 1.824413245926e-05,
      "loss": 0.2826,
      "step": 20100
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9444292032530351,
      "learning_rate": 1.823308957606548e-05,
      "loss": 0.2617,
      "step": 20105
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9305023289632784,
      "learning_rate": 1.8222048117376477e-05,
      "loss": 0.3235,
      "step": 20110
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.925720308203352,
      "learning_rate": 1.821100808551735e-05,
      "loss": 0.3112,
      "step": 20115
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.0307969690814232,
      "learning_rate": 1.8199969482812124e-05,
      "loss": 0.3259,
      "step": 20120
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.8948437195312565,
      "learning_rate": 1.8188932311584565e-05,
      "loss": 0.2991,
      "step": 20125
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9424291443738005,
      "learning_rate": 1.817789657415809e-05,
      "loss": 0.296,
      "step": 20130
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9862261356785185,
      "learning_rate": 1.816686227285585e-05,
      "loss": 0.3486,
      "step": 20135
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.759140054054554,
      "learning_rate": 1.815582941000068e-05,
      "loss": 0.2994,
      "step": 20140
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9512712707924539,
      "learning_rate": 1.814479798791511e-05,
      "loss": 0.2484,
      "step": 20145
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9188747932844091,
      "learning_rate": 1.813376800892138e-05,
      "loss": 0.3367,
      "step": 20150
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.9277092399760558,
      "learning_rate": 1.8122739475341394e-05,
      "loss": 0.3215,
      "step": 20155
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9746324153220186,
      "learning_rate": 1.811171238949679e-05,
      "loss": 0.36,
      "step": 20160
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9308516314055012,
      "learning_rate": 1.8100686753708883e-05,
      "loss": 0.2864,
      "step": 20165
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8672692702521405,
      "learning_rate": 1.808966257029868e-05,
      "loss": 0.3292,
      "step": 20170
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.0259830739178795,
      "learning_rate": 1.80786398415869e-05,
      "loss": 0.3078,
      "step": 20175
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8356647581383919,
      "learning_rate": 1.8067618569893934e-05,
      "loss": 0.3183,
      "step": 20180
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9135833041312057,
      "learning_rate": 1.8056598757539868e-05,
      "loss": 0.3449,
      "step": 20185
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.036834507167456,
      "learning_rate": 1.8045580406844498e-05,
      "loss": 0.3243,
      "step": 20190
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9362926626029245,
      "learning_rate": 1.8034563520127307e-05,
      "loss": 0.3352,
      "step": 20195
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8872973538161422,
      "learning_rate": 1.802354809970745e-05,
      "loss": 0.3348,
      "step": 20200
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.928004858577566,
      "learning_rate": 1.801253414790379e-05,
      "loss": 0.2836,
      "step": 20205
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8778173207054601,
      "learning_rate": 1.800152166703489e-05,
      "loss": 0.2704,
      "step": 20210
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9205221847500501,
      "learning_rate": 1.7990510659418992e-05,
      "loss": 0.2771,
      "step": 20215
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9048422029957066,
      "learning_rate": 1.7979501127374025e-05,
      "loss": 0.3259,
      "step": 20220
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.771507383270431,
      "learning_rate": 1.7968493073217603e-05,
      "loss": 0.312,
      "step": 20225
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.9524902863447129,
      "learning_rate": 1.7957486499267044e-05,
      "loss": 0.3293,
      "step": 20230
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.0341427448944533,
      "learning_rate": 1.794648140783935e-05,
      "loss": 0.3011,
      "step": 20235
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.8530488169788732,
      "learning_rate": 1.7935477801251198e-05,
      "loss": 0.2806,
      "step": 20240
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8850242125129613,
      "learning_rate": 1.792447568181898e-05,
      "loss": 0.2993,
      "step": 20245
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.1175102612321604,
      "learning_rate": 1.7913475051858744e-05,
      "loss": 0.3698,
      "step": 20250
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.930189418267071,
      "learning_rate": 1.7902475913686233e-05,
      "loss": 0.3069,
      "step": 20255
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.043525356659347,
      "learning_rate": 1.7891478269616895e-05,
      "loss": 0.3052,
      "step": 20260
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.0077921462624668,
      "learning_rate": 1.7880482121965846e-05,
      "loss": 0.3247,
      "step": 20265
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.99922877145686,
      "learning_rate": 1.786948747304788e-05,
      "loss": 0.2995,
      "step": 20270
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9261530706677262,
      "learning_rate": 1.785849432517749e-05,
      "loss": 0.3458,
      "step": 20275
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.928385961082353,
      "learning_rate": 1.7847502680668853e-05,
      "loss": 0.2871,
      "step": 20280
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9707647471317726,
      "learning_rate": 1.783651254183582e-05,
      "loss": 0.3758,
      "step": 20285
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9544664370722635,
      "learning_rate": 1.7825523910991948e-05,
      "loss": 0.3155,
      "step": 20290
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.0412653462424266,
      "learning_rate": 1.7814536790450437e-05,
      "loss": 0.3145,
      "step": 20295
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.8931060674317187,
      "learning_rate": 1.780355118252419e-05,
      "loss": 0.3357,
      "step": 20300
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9414105831261608,
      "learning_rate": 1.779256708952581e-05,
      "loss": 0.3354,
      "step": 20305
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.875494951333337,
      "learning_rate": 1.7781584513767552e-05,
      "loss": 0.2973,
      "step": 20310
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.901421317662818,
      "learning_rate": 1.7770603457561374e-05,
      "loss": 0.285,
      "step": 20315
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.0382276614992962,
      "learning_rate": 1.7759623923218887e-05,
      "loss": 0.274,
      "step": 20320
    },
    {
      "epoch": 2.37,
      "grad_norm": 0.9712509780882456,
      "learning_rate": 1.7748645913051408e-05,
      "loss": 0.3247,
      "step": 20325
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9923133228320421,
      "learning_rate": 1.7737669429369925e-05,
      "loss": 0.3054,
      "step": 20330
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.0475887270092925,
      "learning_rate": 1.77266944744851e-05,
      "loss": 0.334,
      "step": 20335
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9223358575970703,
      "learning_rate": 1.771572105070727e-05,
      "loss": 0.3115,
      "step": 20340
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9961148002246296,
      "learning_rate": 1.7704749160346465e-05,
      "loss": 0.3091,
      "step": 20345
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.1185843876837405,
      "learning_rate": 1.7693778805712376e-05,
      "loss": 0.3235,
      "step": 20350
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9868665116080396,
      "learning_rate": 1.7682809989114373e-05,
      "loss": 0.3396,
      "step": 20355
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9719937218883374,
      "learning_rate": 1.767184271286153e-05,
      "loss": 0.2979,
      "step": 20360
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.8678706609228284,
      "learning_rate": 1.7660876979262552e-05,
      "loss": 0.31,
      "step": 20365
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9225186965328284,
      "learning_rate": 1.7649912790625843e-05,
      "loss": 0.2641,
      "step": 20370
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.0249936988213915,
      "learning_rate": 1.7638950149259485e-05,
      "loss": 0.336,
      "step": 20375
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.7492566501317238,
      "learning_rate": 1.7627989057471227e-05,
      "loss": 0.2871,
      "step": 20380
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9753784196827595,
      "learning_rate": 1.7617029517568502e-05,
      "loss": 0.3094,
      "step": 20385
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9774791800120136,
      "learning_rate": 1.7606071531858386e-05,
      "loss": 0.304,
      "step": 20390
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.854841108571231,
      "learning_rate": 1.7595115102647672e-05,
      "loss": 0.3253,
      "step": 20395
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.973166471733524,
      "learning_rate": 1.7584160232242784e-05,
      "loss": 0.3098,
      "step": 20400
    },
    {
      "epoch": 2.38,
      "grad_norm": 0.9790106172918394,
      "learning_rate": 1.7573206922949864e-05,
      "loss": 0.3505,
      "step": 20405
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.0426348851970602,
      "learning_rate": 1.7562255177074677e-05,
      "loss": 0.305,
      "step": 20410
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9537398421609526,
      "learning_rate": 1.7551304996922675e-05,
      "loss": 0.3112,
      "step": 20415
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4043081186463895,
      "learning_rate": 1.7540356384798996e-05,
      "loss": 0.3744,
      "step": 20420
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8768677469798624,
      "learning_rate": 1.7529409343008445e-05,
      "loss": 0.3183,
      "step": 20425
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9103740863512297,
      "learning_rate": 1.7518463873855486e-05,
      "loss": 0.3253,
      "step": 20430
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9925106643874901,
      "learning_rate": 1.750751997964424e-05,
      "loss": 0.3322,
      "step": 20435
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9560424441306414,
      "learning_rate": 1.749657766267852e-05,
      "loss": 0.3543,
      "step": 20440
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8437797542296586,
      "learning_rate": 1.7485636925261806e-05,
      "loss": 0.3068,
      "step": 20445
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.09168729894445,
      "learning_rate": 1.7474697769697222e-05,
      "loss": 0.3427,
      "step": 20450
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8503990414237921,
      "learning_rate": 1.7463760198287602e-05,
      "loss": 0.3394,
      "step": 20455
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8572862986765358,
      "learning_rate": 1.745282421333539e-05,
      "loss": 0.2855,
      "step": 20460
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.951134552703443,
      "learning_rate": 1.7441889817142737e-05,
      "loss": 0.327,
      "step": 20465
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9700613902614589,
      "learning_rate": 1.7430957012011444e-05,
      "loss": 0.3073,
      "step": 20470
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9610966526333646,
      "learning_rate": 1.7420025800243e-05,
      "loss": 0.288,
      "step": 20475
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9680948412201371,
      "learning_rate": 1.7409096184138517e-05,
      "loss": 0.3166,
      "step": 20480
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.8133368103559565,
      "learning_rate": 1.7398168165998798e-05,
      "loss": 0.322,
      "step": 20485
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.9689577117767062,
      "learning_rate": 1.738724174812431e-05,
      "loss": 0.3244,
      "step": 20490
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.0866970303455001,
      "learning_rate": 1.7376316932815174e-05,
      "loss": 0.3001,
      "step": 20495
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8133781800096883,
      "learning_rate": 1.7365393722371194e-05,
      "loss": 0.3048,
      "step": 20500
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8121343138556024,
      "learning_rate": 1.735447211909179e-05,
      "loss": 0.325,
      "step": 20505
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9108360070294981,
      "learning_rate": 1.73435521252761e-05,
      "loss": 0.3447,
      "step": 20510
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9053299857111836,
      "learning_rate": 1.7332633743222887e-05,
      "loss": 0.2491,
      "step": 20515
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9232985741818027,
      "learning_rate": 1.732171697523059e-05,
      "loss": 0.2873,
      "step": 20520
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.0195852979904876,
      "learning_rate": 1.73108018235973e-05,
      "loss": 0.2974,
      "step": 20525
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.087397788177389,
      "learning_rate": 1.7299888290620762e-05,
      "loss": 0.3484,
      "step": 20530
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9307975027803674,
      "learning_rate": 1.72889763785984e-05,
      "loss": 0.305,
      "step": 20535
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.1617616823165626,
      "learning_rate": 1.7278066089827282e-05,
      "loss": 0.3136,
      "step": 20540
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3359904433851528,
      "learning_rate": 1.7267157426604148e-05,
      "loss": 0.3886,
      "step": 20545
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.0804653042748416,
      "learning_rate": 1.725625039122536e-05,
      "loss": 0.313,
      "step": 20550
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9179991876521328,
      "learning_rate": 1.7245344985986987e-05,
      "loss": 0.3091,
      "step": 20555
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9345260839164207,
      "learning_rate": 1.7234441213184726e-05,
      "loss": 0.2925,
      "step": 20560
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9544697820682926,
      "learning_rate": 1.722353907511393e-05,
      "loss": 0.2916,
      "step": 20565
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.136804425881183,
      "learning_rate": 1.7212638574069623e-05,
      "loss": 0.3242,
      "step": 20570
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.9068613494483474,
      "learning_rate": 1.7201739712346464e-05,
      "loss": 0.2891,
      "step": 20575
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.8867231952479596,
      "learning_rate": 1.719084249223878e-05,
      "loss": 0.3496,
      "step": 20580
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9129609215482772,
      "learning_rate": 1.7179946916040553e-05,
      "loss": 0.2766,
      "step": 20585
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.7625365671053158,
      "learning_rate": 1.7169052986045413e-05,
      "loss": 0.2599,
      "step": 20590
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8844413078907383,
      "learning_rate": 1.7158160704546656e-05,
      "loss": 0.2618,
      "step": 20595
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8351459548422423,
      "learning_rate": 1.7147270073837203e-05,
      "loss": 0.2811,
      "step": 20600
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9604317844847353,
      "learning_rate": 1.7136381096209664e-05,
      "loss": 0.346,
      "step": 20605
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.020036289302891,
      "learning_rate": 1.7125493773956265e-05,
      "loss": 0.2953,
      "step": 20610
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9252697031509793,
      "learning_rate": 1.7114608109368924e-05,
      "loss": 0.3301,
      "step": 20615
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8607711306172855,
      "learning_rate": 1.7103724104739167e-05,
      "loss": 0.3075,
      "step": 20620
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.933881258289097,
      "learning_rate": 1.7092841762358203e-05,
      "loss": 0.2776,
      "step": 20625
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9013882227731405,
      "learning_rate": 1.7081961084516866e-05,
      "loss": 0.2868,
      "step": 20630
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9918354175891291,
      "learning_rate": 1.707108207350567e-05,
      "loss": 0.3237,
      "step": 20635
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.019497834107873,
      "learning_rate": 1.7060204731614755e-05,
      "loss": 0.3267,
      "step": 20640
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9166277589657522,
      "learning_rate": 1.7049329061133907e-05,
      "loss": 0.3176,
      "step": 20645
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8899707757904882,
      "learning_rate": 1.7038455064352572e-05,
      "loss": 0.3125,
      "step": 20650
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.8585113793155914,
      "learning_rate": 1.7027582743559843e-05,
      "loss": 0.3147,
      "step": 20655
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9388739924671904,
      "learning_rate": 1.7016712101044456e-05,
      "loss": 0.3305,
      "step": 20660
    },
    {
      "epoch": 2.41,
      "grad_norm": 0.9882207033822796,
      "learning_rate": 1.7005843139094807e-05,
      "loss": 0.3064,
      "step": 20665
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.0144948178408904,
      "learning_rate": 1.69949758599989e-05,
      "loss": 0.3115,
      "step": 20670
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8036509267305074,
      "learning_rate": 1.698411026604444e-05,
      "loss": 0.2843,
      "step": 20675
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8517237684574767,
      "learning_rate": 1.6973246359518724e-05,
      "loss": 0.294,
      "step": 20680
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9234673618868314,
      "learning_rate": 1.6962384142708736e-05,
      "loss": 0.3341,
      "step": 20685
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8854195602713585,
      "learning_rate": 1.6951523617901084e-05,
      "loss": 0.3327,
      "step": 20690
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8830809344350237,
      "learning_rate": 1.6940664787382007e-05,
      "loss": 0.3364,
      "step": 20695
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9080807803233512,
      "learning_rate": 1.6929807653437412e-05,
      "loss": 0.3209,
      "step": 20700
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.934693526725521,
      "learning_rate": 1.6918952218352846e-05,
      "loss": 0.3175,
      "step": 20705
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9434819447256206,
      "learning_rate": 1.6908098484413493e-05,
      "loss": 0.2822,
      "step": 20710
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.8651975521694826,
      "learning_rate": 1.689724645390416e-05,
      "loss": 0.337,
      "step": 20715
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.0821995155315314,
      "learning_rate": 1.688639612910933e-05,
      "loss": 0.285,
      "step": 20720
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9046286616752537,
      "learning_rate": 1.68755475123131e-05,
      "loss": 0.3192,
      "step": 20725
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9587134528252275,
      "learning_rate": 1.686470060579922e-05,
      "loss": 0.3072,
      "step": 20730
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9697384738696976,
      "learning_rate": 1.6853855411851087e-05,
      "loss": 0.3531,
      "step": 20735
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.890594451563599,
      "learning_rate": 1.6843011932751717e-05,
      "loss": 0.3155,
      "step": 20740
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9752737066820358,
      "learning_rate": 1.6832170170783776e-05,
      "loss": 0.3066,
      "step": 20745
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9038019834976873,
      "learning_rate": 1.6821330128229573e-05,
      "loss": 0.2846,
      "step": 20750
    },
    {
      "epoch": 2.42,
      "grad_norm": 0.9688024861410557,
      "learning_rate": 1.681049180737105e-05,
      "loss": 0.2701,
      "step": 20755
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.8357071365403306,
      "learning_rate": 1.679965521048979e-05,
      "loss": 0.2982,
      "step": 20760
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9845445436958881,
      "learning_rate": 1.6788820339866995e-05,
      "loss": 0.3069,
      "step": 20765
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.822498152565524,
      "learning_rate": 1.677798719778354e-05,
      "loss": 0.2701,
      "step": 20770
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.881184037213802,
      "learning_rate": 1.6767155786519895e-05,
      "loss": 0.3361,
      "step": 20775
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.0960580204089654,
      "learning_rate": 1.675632610835621e-05,
      "loss": 0.3533,
      "step": 20780
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.881113603312044,
      "learning_rate": 1.674549816557222e-05,
      "loss": 0.2994,
      "step": 20785
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.8609811422833646,
      "learning_rate": 1.6734671960447333e-05,
      "loss": 0.3276,
      "step": 20790
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.853432243819674,
      "learning_rate": 1.6723847495260573e-05,
      "loss": 0.3173,
      "step": 20795
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.8940328699733874,
      "learning_rate": 1.6713024772290614e-05,
      "loss": 0.3047,
      "step": 20800
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.0294205464621227,
      "learning_rate": 1.670220379381575e-05,
      "loss": 0.3541,
      "step": 20805
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9774537385969103,
      "learning_rate": 1.66913845621139e-05,
      "loss": 0.3288,
      "step": 20810
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9703909878845425,
      "learning_rate": 1.6680567079462625e-05,
      "loss": 0.3471,
      "step": 20815
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.1556860907985436,
      "learning_rate": 1.6669751348139134e-05,
      "loss": 0.3527,
      "step": 20820
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9276049684644704,
      "learning_rate": 1.6658937370420236e-05,
      "loss": 0.3755,
      "step": 20825
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.8648750472734364,
      "learning_rate": 1.6648125148582406e-05,
      "loss": 0.3026,
      "step": 20830
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.8573433900894533,
      "learning_rate": 1.6637314684901713e-05,
      "loss": 0.3244,
      "step": 20835
    },
    {
      "epoch": 2.43,
      "grad_norm": 0.9443938285571626,
      "learning_rate": 1.6626505981653867e-05,
      "loss": 0.3236,
      "step": 20840
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9657790650448367,
      "learning_rate": 1.6615699041114225e-05,
      "loss": 0.3456,
      "step": 20845
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9753422846569205,
      "learning_rate": 1.6604893865557773e-05,
      "loss": 0.3278,
      "step": 20850
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2491490489624901,
      "learning_rate": 1.6594090457259092e-05,
      "loss": 0.3047,
      "step": 20855
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9474631513214848,
      "learning_rate": 1.658328881849242e-05,
      "loss": 0.3028,
      "step": 20860
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.8847222692210336,
      "learning_rate": 1.6572488951531617e-05,
      "loss": 0.2928,
      "step": 20865
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.7130706051034976,
      "learning_rate": 1.6561690858650164e-05,
      "loss": 0.2608,
      "step": 20870
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.061097368136985,
      "learning_rate": 1.6550894542121186e-05,
      "loss": 0.3109,
      "step": 20875
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.0742504512238875,
      "learning_rate": 1.6540100004217402e-05,
      "loss": 0.3064,
      "step": 20880
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9336147598275403,
      "learning_rate": 1.6529307247211187e-05,
      "loss": 0.2522,
      "step": 20885
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9896140210309206,
      "learning_rate": 1.6518516273374522e-05,
      "loss": 0.3639,
      "step": 20890
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.8502843148498878,
      "learning_rate": 1.6507727084979026e-05,
      "loss": 0.3006,
      "step": 20895
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.0080279193669577,
      "learning_rate": 1.649693968429594e-05,
      "loss": 0.3241,
      "step": 20900
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6558955517805702,
      "learning_rate": 1.6486154073596117e-05,
      "loss": 0.3661,
      "step": 20905
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.048229699332212,
      "learning_rate": 1.6475370255150037e-05,
      "loss": 0.2927,
      "step": 20910
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.9849249594950062,
      "learning_rate": 1.646458823122782e-05,
      "loss": 0.3282,
      "step": 20915
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.7842460666991516,
      "learning_rate": 1.6453808004099192e-05,
      "loss": 0.3099,
      "step": 20920
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.7856172373810463,
      "learning_rate": 1.644302957603349e-05,
      "loss": 0.2576,
      "step": 20925
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9245349575648175,
      "learning_rate": 1.6432252949299698e-05,
      "loss": 0.3118,
      "step": 20930
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.8560391212522578,
      "learning_rate": 1.6421478126166406e-05,
      "loss": 0.2792,
      "step": 20935
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.049366249690819,
      "learning_rate": 1.641070510890183e-05,
      "loss": 0.3484,
      "step": 20940
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.8710677198901201,
      "learning_rate": 1.6399933899773805e-05,
      "loss": 0.2494,
      "step": 20945
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9297356372925153,
      "learning_rate": 1.6389164501049774e-05,
      "loss": 0.3285,
      "step": 20950
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9408150799930833,
      "learning_rate": 1.6378396914996812e-05,
      "loss": 0.3072,
      "step": 20955
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9654945901133778,
      "learning_rate": 1.6367631143881608e-05,
      "loss": 0.2907,
      "step": 20960
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.916107020986903,
      "learning_rate": 1.6356867189970475e-05,
      "loss": 0.3095,
      "step": 20965
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.8770344997279952,
      "learning_rate": 1.634610505552934e-05,
      "loss": 0.3104,
      "step": 20970
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.0910499935207085,
      "learning_rate": 1.6335344742823737e-05,
      "loss": 0.3505,
      "step": 20975
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.8453726345725098,
      "learning_rate": 1.6324586254118823e-05,
      "loss": 0.264,
      "step": 20980
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9731022217076088,
      "learning_rate": 1.6313829591679377e-05,
      "loss": 0.3277,
      "step": 20985
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.8964074944333037,
      "learning_rate": 1.63030747577698e-05,
      "loss": 0.3091,
      "step": 20990
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.9797503908378691,
      "learning_rate": 1.629232175465407e-05,
      "loss": 0.3036,
      "step": 20995
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.1043463387354728,
      "learning_rate": 1.6281570584595833e-05,
      "loss": 0.3116,
      "step": 21000
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.8566110963898634,
      "learning_rate": 1.6270821249858305e-05,
      "loss": 0.341,
      "step": 21005
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.8645923223264744,
      "learning_rate": 1.6260073752704343e-05,
      "loss": 0.2899,
      "step": 21010
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9319372269066225,
      "learning_rate": 1.6249328095396415e-05,
      "loss": 0.3885,
      "step": 21015
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.8348021237681238,
      "learning_rate": 1.6238584280196578e-05,
      "loss": 0.3243,
      "step": 21020
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2796334645012506,
      "learning_rate": 1.6227842309366525e-05,
      "loss": 0.3697,
      "step": 21025
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.8753334106917359,
      "learning_rate": 1.6217102185167555e-05,
      "loss": 0.3334,
      "step": 21030
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.0364566555468364,
      "learning_rate": 1.6206363909860573e-05,
      "loss": 0.3071,
      "step": 21035
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9702234932821351,
      "learning_rate": 1.619562748570611e-05,
      "loss": 0.3092,
      "step": 21040
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9433252560829067,
      "learning_rate": 1.618489291496427e-05,
      "loss": 0.2943,
      "step": 21045
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.8497851155983013,
      "learning_rate": 1.617416019989482e-05,
      "loss": 0.3227,
      "step": 21050
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.7951703507666193,
      "learning_rate": 1.6163429342757098e-05,
      "loss": 0.2961,
      "step": 21055
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9139584647556377,
      "learning_rate": 1.6152700345810063e-05,
      "loss": 0.3552,
      "step": 21060
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.998207735343552,
      "learning_rate": 1.614197321131228e-05,
      "loss": 0.3026,
      "step": 21065
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.894406830485271,
      "learning_rate": 1.6131247941521916e-05,
      "loss": 0.312,
      "step": 21070
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.007765064582181,
      "learning_rate": 1.6120524538696757e-05,
      "loss": 0.3256,
      "step": 21075
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.931676126096992,
      "learning_rate": 1.6109803005094203e-05,
      "loss": 0.3272,
      "step": 21080
    },
    {
      "epoch": 2.46,
      "grad_norm": 0.9227146381814008,
      "learning_rate": 1.6099083342971243e-05,
      "loss": 0.3193,
      "step": 21085
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3632590369105373,
      "learning_rate": 1.608836555458447e-05,
      "loss": 0.3132,
      "step": 21090
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.228437500653841,
      "learning_rate": 1.6077649642190098e-05,
      "loss": 0.2778,
      "step": 21095
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.255178901707754,
      "learning_rate": 1.6066935608043932e-05,
      "loss": 0.3171,
      "step": 21100
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.094881540919093,
      "learning_rate": 1.6056223454401396e-05,
      "loss": 0.3329,
      "step": 21105
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8917136376999013,
      "learning_rate": 1.6045513183517518e-05,
      "loss": 0.3105,
      "step": 21110
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8138921819723233,
      "learning_rate": 1.6034804797646903e-05,
      "loss": 0.327,
      "step": 21115
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9964847694209485,
      "learning_rate": 1.602409829904379e-05,
      "loss": 0.3354,
      "step": 21120
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8926927176072407,
      "learning_rate": 1.6013393689962e-05,
      "loss": 0.2865,
      "step": 21125
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8478548005242057,
      "learning_rate": 1.600269097265499e-05,
      "loss": 0.2932,
      "step": 21130
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.004624752413318,
      "learning_rate": 1.5991990149375763e-05,
      "loss": 0.2294,
      "step": 21135
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9891482440374775,
      "learning_rate": 1.5981291222376965e-05,
      "loss": 0.3007,
      "step": 21140
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9190798932379494,
      "learning_rate": 1.5970594193910842e-05,
      "loss": 0.304,
      "step": 21145
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.058838624775201,
      "learning_rate": 1.5959899066229218e-05,
      "loss": 0.3432,
      "step": 21150
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.8837833311307045,
      "learning_rate": 1.594920584158354e-05,
      "loss": 0.3194,
      "step": 21155
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9768447308248345,
      "learning_rate": 1.5938514522224836e-05,
      "loss": 0.2978,
      "step": 21160
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9130036586167442,
      "learning_rate": 1.5927825110403747e-05,
      "loss": 0.2978,
      "step": 21165
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.0443342817975974,
      "learning_rate": 1.5917137608370496e-05,
      "loss": 0.3234,
      "step": 21170
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9358383767714993,
      "learning_rate": 1.5906452018374926e-05,
      "loss": 0.3153,
      "step": 21175
    },
    {
      "epoch": 2.47,
      "grad_norm": 0.9779753493468136,
      "learning_rate": 1.5895768342666464e-05,
      "loss": 0.3321,
      "step": 21180
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9531950154804638,
      "learning_rate": 1.5885086583494126e-05,
      "loss": 0.3092,
      "step": 21185
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9815328359454177,
      "learning_rate": 1.5874406743106536e-05,
      "loss": 0.3882,
      "step": 21190
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0536605715420615,
      "learning_rate": 1.5863728823751923e-05,
      "loss": 0.2998,
      "step": 21195
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9306718431105918,
      "learning_rate": 1.5853052827678096e-05,
      "loss": 0.2839,
      "step": 21200
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8420643194261997,
      "learning_rate": 1.5842378757132457e-05,
      "loss": 0.3093,
      "step": 21205
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.011613633405876,
      "learning_rate": 1.583170661436202e-05,
      "loss": 0.2913,
      "step": 21210
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9154677363296995,
      "learning_rate": 1.5821036401613372e-05,
      "loss": 0.3225,
      "step": 21215
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.0061547615463968,
      "learning_rate": 1.5810368121132707e-05,
      "loss": 0.3198,
      "step": 21220
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8333003271567757,
      "learning_rate": 1.5799701775165827e-05,
      "loss": 0.357,
      "step": 21225
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8620641799749553,
      "learning_rate": 1.5789037365958086e-05,
      "loss": 0.2813,
      "step": 21230
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9178676620739203,
      "learning_rate": 1.5778374895754462e-05,
      "loss": 0.3116,
      "step": 21235
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9689875611703933,
      "learning_rate": 1.576771436679952e-05,
      "loss": 0.3332,
      "step": 21240
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9341003327637706,
      "learning_rate": 1.5757055781337413e-05,
      "loss": 0.3163,
      "step": 21245
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.9037989852739657,
      "learning_rate": 1.574639914161189e-05,
      "loss": 0.3519,
      "step": 21250
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8911504656732542,
      "learning_rate": 1.5735744449866265e-05,
      "loss": 0.3826,
      "step": 21255
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8536462281017833,
      "learning_rate": 1.5725091708343488e-05,
      "loss": 0.3162,
      "step": 21260
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.8938295547254937,
      "learning_rate": 1.5714440919286046e-05,
      "loss": 0.3062,
      "step": 21265
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9444821201941503,
      "learning_rate": 1.570379208493607e-05,
      "loss": 0.2876,
      "step": 21270
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8476946156698107,
      "learning_rate": 1.5693145207535232e-05,
      "loss": 0.2514,
      "step": 21275
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.1291391211192547,
      "learning_rate": 1.5682500289324812e-05,
      "loss": 0.2881,
      "step": 21280
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9809673886703361,
      "learning_rate": 1.5671857332545685e-05,
      "loss": 0.3014,
      "step": 21285
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8915515000075215,
      "learning_rate": 1.56612163394383e-05,
      "loss": 0.3185,
      "step": 21290
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9313460242350698,
      "learning_rate": 1.565057731224271e-05,
      "loss": 0.3252,
      "step": 21295
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.7874113080840245,
      "learning_rate": 1.5639940253198516e-05,
      "loss": 0.296,
      "step": 21300
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8563869005385844,
      "learning_rate": 1.5629305164544946e-05,
      "loss": 0.2947,
      "step": 21305
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8183212889462009,
      "learning_rate": 1.5618672048520795e-05,
      "loss": 0.4039,
      "step": 21310
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9447373465993163,
      "learning_rate": 1.5608040907364448e-05,
      "loss": 0.3016,
      "step": 21315
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.217664155674163,
      "learning_rate": 1.559741174331388e-05,
      "loss": 0.3035,
      "step": 21320
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9640455232239884,
      "learning_rate": 1.5586784558606622e-05,
      "loss": 0.347,
      "step": 21325
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.995528036862888,
      "learning_rate": 1.5576159355479812e-05,
      "loss": 0.2933,
      "step": 21330
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8930067045422301,
      "learning_rate": 1.556553613617018e-05,
      "loss": 0.2754,
      "step": 21335
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.9209789298288272,
      "learning_rate": 1.5554914902914015e-05,
      "loss": 0.3289,
      "step": 21340
    },
    {
      "epoch": 2.49,
      "grad_norm": 0.8976493568693282,
      "learning_rate": 1.5544295657947203e-05,
      "loss": 0.2918,
      "step": 21345
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.0422036247979658,
      "learning_rate": 1.5533678403505195e-05,
      "loss": 0.3133,
      "step": 21350
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0630563397839716,
      "learning_rate": 1.5523063141823042e-05,
      "loss": 0.3355,
      "step": 21355
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.841155753226513,
      "learning_rate": 1.551244987513537e-05,
      "loss": 0.2827,
      "step": 21360
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8899685364821089,
      "learning_rate": 1.5501838605676393e-05,
      "loss": 0.2858,
      "step": 21365
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9183498263743104,
      "learning_rate": 1.5491229335679866e-05,
      "loss": 0.2877,
      "step": 21370
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8818492997732088,
      "learning_rate": 1.5480622067379176e-05,
      "loss": 0.2936,
      "step": 21375
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.098851486905534,
      "learning_rate": 1.5470016803007244e-05,
      "loss": 0.2858,
      "step": 21380
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.001947808749853,
      "learning_rate": 1.5459413544796608e-05,
      "loss": 0.3084,
      "step": 21385
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8044809446530183,
      "learning_rate": 1.5448812294979366e-05,
      "loss": 0.3297,
      "step": 21390
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.1354642131570167,
      "learning_rate": 1.5438213055787175e-05,
      "loss": 0.3572,
      "step": 21395
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9555562354743699,
      "learning_rate": 1.5427615829451293e-05,
      "loss": 0.3258,
      "step": 21400
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9041572980961791,
      "learning_rate": 1.5417020618202554e-05,
      "loss": 0.2744,
      "step": 21405
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.92828666985902,
      "learning_rate": 1.540642742427136e-05,
      "loss": 0.3579,
      "step": 21410
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9714841916203468,
      "learning_rate": 1.5395836249887673e-05,
      "loss": 0.2795,
      "step": 21415
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9729645569409078,
      "learning_rate": 1.538524709728106e-05,
      "loss": 0.2857,
      "step": 21420
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.9648888031954621,
      "learning_rate": 1.5374659968680656e-05,
      "loss": 0.3358,
      "step": 21425
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0237146928994092,
      "learning_rate": 1.5364074866315144e-05,
      "loss": 0.3014,
      "step": 21430
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.0018081448364091,
      "learning_rate": 1.5353491792412817e-05,
      "loss": 0.3353,
      "step": 21435
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.8784137250141396,
      "learning_rate": 1.5342910749201512e-05,
      "loss": 0.3098,
      "step": 21440
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8431315728926133,
      "learning_rate": 1.5332331738908643e-05,
      "loss": 0.2958,
      "step": 21445
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8431432780814623,
      "learning_rate": 1.5321754763761213e-05,
      "loss": 0.2835,
      "step": 21450
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9303355836360612,
      "learning_rate": 1.5311179825985787e-05,
      "loss": 0.3325,
      "step": 21455
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.0197415995910337,
      "learning_rate": 1.5300606927808506e-05,
      "loss": 0.2789,
      "step": 21460
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8085242134872769,
      "learning_rate": 1.5290036071455055e-05,
      "loss": 0.3242,
      "step": 21465
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8451584232445656,
      "learning_rate": 1.5279467259150724e-05,
      "loss": 0.2614,
      "step": 21470
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2059538937613414,
      "learning_rate": 1.5268900493120358e-05,
      "loss": 0.2983,
      "step": 21475
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.1868231319031395,
      "learning_rate": 1.5258335775588378e-05,
      "loss": 0.3298,
      "step": 21480
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9366140089236894,
      "learning_rate": 1.524777310877875e-05,
      "loss": 0.3165,
      "step": 21485
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8517976116392272,
      "learning_rate": 1.5237212494915042e-05,
      "loss": 0.328,
      "step": 21490
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8999219767268843,
      "learning_rate": 1.5226653936220364e-05,
      "loss": 0.3095,
      "step": 21495
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9716401143297234,
      "learning_rate": 1.5216097434917407e-05,
      "loss": 0.3172,
      "step": 21500
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9967004859855729,
      "learning_rate": 1.5205542993228434e-05,
      "loss": 0.3873,
      "step": 21505
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8556761493451421,
      "learning_rate": 1.5194990613375253e-05,
      "loss": 0.2538,
      "step": 21510
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.8934369242786062,
      "learning_rate": 1.518444029757925e-05,
      "loss": 0.2548,
      "step": 21515
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.7622862399215884,
      "learning_rate": 1.5173892048061383e-05,
      "loss": 0.2745,
      "step": 21520
    },
    {
      "epoch": 2.51,
      "grad_norm": 0.9950398043817247,
      "learning_rate": 1.5163345867042167e-05,
      "loss": 0.3396,
      "step": 21525
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9985447911690505,
      "learning_rate": 1.5152801756741691e-05,
      "loss": 0.3238,
      "step": 21530
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9748434500915857,
      "learning_rate": 1.5142259719379587e-05,
      "loss": 0.3188,
      "step": 21535
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0306792962321856,
      "learning_rate": 1.5131719757175076e-05,
      "loss": 0.3002,
      "step": 21540
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0281294746838727,
      "learning_rate": 1.5121181872346918e-05,
      "loss": 0.282,
      "step": 21545
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9616497316132993,
      "learning_rate": 1.5110646067113466e-05,
      "loss": 0.3332,
      "step": 21550
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.8279731514348087,
      "learning_rate": 1.5100112343692604e-05,
      "loss": 0.3131,
      "step": 21555
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.45107633871601,
      "learning_rate": 1.5089580704301787e-05,
      "loss": 0.308,
      "step": 21560
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9087351292104445,
      "learning_rate": 1.5079051151158044e-05,
      "loss": 0.2575,
      "step": 21565
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.8484494290989507,
      "learning_rate": 1.5068523686477958e-05,
      "loss": 0.3036,
      "step": 21570
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.899554809311599,
      "learning_rate": 1.5057998312477678e-05,
      "loss": 0.3071,
      "step": 21575
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.8961586307425095,
      "learning_rate": 1.5047475031372878e-05,
      "loss": 0.308,
      "step": 21580
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9209952729958419,
      "learning_rate": 1.5036953845378838e-05,
      "loss": 0.2782,
      "step": 21585
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0869939772725805,
      "learning_rate": 1.5026434756710384e-05,
      "loss": 0.3754,
      "step": 21590
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9656953909379147,
      "learning_rate": 1.5015917767581878e-05,
      "loss": 0.3399,
      "step": 21595
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9976930024795567,
      "learning_rate": 1.5005402880207273e-05,
      "loss": 0.2385,
      "step": 21600
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.8720208845712702,
      "learning_rate": 1.4994890096800052e-05,
      "loss": 0.3139,
      "step": 21605
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.9752751614726635,
      "learning_rate": 1.498437941957326e-05,
      "loss": 0.3337,
      "step": 21610
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.1258240022550678,
      "learning_rate": 1.4973870850739516e-05,
      "loss": 0.347,
      "step": 21615
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0253648923424215,
      "learning_rate": 1.496336439251099e-05,
      "loss": 0.3298,
      "step": 21620
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9862069058778715,
      "learning_rate": 1.4952860047099387e-05,
      "loss": 0.2947,
      "step": 21625
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8725384443656609,
      "learning_rate": 1.4942357816715981e-05,
      "loss": 0.2954,
      "step": 21630
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4489002970792106,
      "learning_rate": 1.4931857703571616e-05,
      "loss": 0.34,
      "step": 21635
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9150676855976392,
      "learning_rate": 1.4921359709876658e-05,
      "loss": 0.2853,
      "step": 21640
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8415966628341504,
      "learning_rate": 1.4910863837841068e-05,
      "loss": 0.3725,
      "step": 21645
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9018462955910234,
      "learning_rate": 1.4900370089674313e-05,
      "loss": 0.306,
      "step": 21650
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.865625382019111,
      "learning_rate": 1.4889878467585449e-05,
      "loss": 0.2874,
      "step": 21655
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8711490168610587,
      "learning_rate": 1.4879388973783065e-05,
      "loss": 0.3086,
      "step": 21660
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8603117155347523,
      "learning_rate": 1.4868901610475322e-05,
      "loss": 0.33,
      "step": 21665
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9271282845041959,
      "learning_rate": 1.4858416379869917e-05,
      "loss": 0.3376,
      "step": 21670
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8589838925297786,
      "learning_rate": 1.4847933284174086e-05,
      "loss": 0.27,
      "step": 21675
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9557897100628673,
      "learning_rate": 1.4837452325594642e-05,
      "loss": 0.3324,
      "step": 21680
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.0404946909050719,
      "learning_rate": 1.4826973506337937e-05,
      "loss": 0.338,
      "step": 21685
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.9056108852906005,
      "learning_rate": 1.4816496828609878e-05,
      "loss": 0.3039,
      "step": 21690
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.8622947791761709,
      "learning_rate": 1.4806022294615896e-05,
      "loss": 0.3446,
      "step": 21695
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.0779904081132172,
      "learning_rate": 1.4795549906561005e-05,
      "loss": 0.3913,
      "step": 21700
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.8984380911638086,
      "learning_rate": 1.4785079666649754e-05,
      "loss": 0.3157,
      "step": 21705
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.0543290010377133,
      "learning_rate": 1.4774611577086226e-05,
      "loss": 0.3345,
      "step": 21710
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9341774053294449,
      "learning_rate": 1.4764145640074083e-05,
      "loss": 0.3099,
      "step": 21715
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9704998914701681,
      "learning_rate": 1.4753681857816499e-05,
      "loss": 0.3215,
      "step": 21720
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9637876248542727,
      "learning_rate": 1.4743220232516203e-05,
      "loss": 0.2813,
      "step": 21725
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9637558940049057,
      "learning_rate": 1.4732760766375491e-05,
      "loss": 0.3042,
      "step": 21730
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.8846325978183377,
      "learning_rate": 1.472230346159619e-05,
      "loss": 0.2785,
      "step": 21735
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9453883799357359,
      "learning_rate": 1.4711848320379673e-05,
      "loss": 0.331,
      "step": 21740
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.004157022319753,
      "learning_rate": 1.4701395344926843e-05,
      "loss": 0.2834,
      "step": 21745
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9172977530267649,
      "learning_rate": 1.4690944537438179e-05,
      "loss": 0.2968,
      "step": 21750
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.8867169088187941,
      "learning_rate": 1.4680495900113667e-05,
      "loss": 0.3188,
      "step": 21755
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.0909762921342887,
      "learning_rate": 1.467004943515287e-05,
      "loss": 0.3144,
      "step": 21760
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.8780589425033255,
      "learning_rate": 1.4659605144754868e-05,
      "loss": 0.3529,
      "step": 21765
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9217936574749216,
      "learning_rate": 1.4649163031118302e-05,
      "loss": 0.3199,
      "step": 21770
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9703598843043927,
      "learning_rate": 1.4638723096441337e-05,
      "loss": 0.3449,
      "step": 21775
    },
    {
      "epoch": 2.54,
      "grad_norm": 0.9290620496605048,
      "learning_rate": 1.46282853429217e-05,
      "loss": 0.2795,
      "step": 21780
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9201341096369683,
      "learning_rate": 1.4617849772756642e-05,
      "loss": 0.3291,
      "step": 21785
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9117022379286077,
      "learning_rate": 1.4607416388142952e-05,
      "loss": 0.3016,
      "step": 21790
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.204010209721409,
      "learning_rate": 1.4596985191276973e-05,
      "loss": 0.3136,
      "step": 21795
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.8260889622901302,
      "learning_rate": 1.4586556184354583e-05,
      "loss": 0.301,
      "step": 21800
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9335207222344308,
      "learning_rate": 1.4576129369571207e-05,
      "loss": 0.2673,
      "step": 21805
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9442132899585794,
      "learning_rate": 1.4565704749121784e-05,
      "loss": 0.2842,
      "step": 21810
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.858087870080883,
      "learning_rate": 1.4555282325200802e-05,
      "loss": 0.3121,
      "step": 21815
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.8908228275451331,
      "learning_rate": 1.4544862100002298e-05,
      "loss": 0.3074,
      "step": 21820
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.0055442115253612,
      "learning_rate": 1.4534444075719839e-05,
      "loss": 0.3184,
      "step": 21825
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9926824519971305,
      "learning_rate": 1.452402825454654e-05,
      "loss": 0.301,
      "step": 21830
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.8926720005926836,
      "learning_rate": 1.4513614638675017e-05,
      "loss": 0.3001,
      "step": 21835
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.8576112056427009,
      "learning_rate": 1.4503203230297462e-05,
      "loss": 0.3164,
      "step": 21840
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2227561495462187,
      "learning_rate": 1.4492794031605577e-05,
      "loss": 0.3562,
      "step": 21845
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.8762471013867944,
      "learning_rate": 1.4482387044790607e-05,
      "loss": 0.3234,
      "step": 21850
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.1274170474211664,
      "learning_rate": 1.4471982272043344e-05,
      "loss": 0.349,
      "step": 21855
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9190755223805522,
      "learning_rate": 1.4461579715554086e-05,
      "loss": 0.3633,
      "step": 21860
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.9431066321046293,
      "learning_rate": 1.4451179377512685e-05,
      "loss": 0.3077,
      "step": 21865
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9766465029319518,
      "learning_rate": 1.4440781260108521e-05,
      "loss": 0.3168,
      "step": 21870
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.980798157247656,
      "learning_rate": 1.4430385365530525e-05,
      "loss": 0.3461,
      "step": 21875
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9537709701778024,
      "learning_rate": 1.441999169596712e-05,
      "loss": 0.3398,
      "step": 21880
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0998677352492243,
      "learning_rate": 1.4409600253606279e-05,
      "loss": 0.2806,
      "step": 21885
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0949361133837958,
      "learning_rate": 1.4399211040635519e-05,
      "loss": 0.2775,
      "step": 21890
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9590641548640746,
      "learning_rate": 1.4388824059241877e-05,
      "loss": 0.2966,
      "step": 21895
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.0177205487811838,
      "learning_rate": 1.437843931161193e-05,
      "loss": 0.2879,
      "step": 21900
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9492964748487662,
      "learning_rate": 1.4368056799931761e-05,
      "loss": 0.2671,
      "step": 21905
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.8265834262847414,
      "learning_rate": 1.4357676526387011e-05,
      "loss": 0.3135,
      "step": 21910
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.7950916529130719,
      "learning_rate": 1.4347298493162823e-05,
      "loss": 0.3439,
      "step": 21915
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.8420633802320864,
      "learning_rate": 1.4336922702443886e-05,
      "loss": 0.356,
      "step": 21920
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.8702770802355497,
      "learning_rate": 1.4326549156414425e-05,
      "loss": 0.3261,
      "step": 21925
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.9036489919601084,
      "learning_rate": 1.4316177857258162e-05,
      "loss": 0.2752,
      "step": 21930
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.8985751600578105,
      "learning_rate": 1.4305808807158371e-05,
      "loss": 0.3287,
      "step": 21935
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.8925960711606314,
      "learning_rate": 1.4295442008297855e-05,
      "loss": 0.3474,
      "step": 21940
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.8487347851618743,
      "learning_rate": 1.428507746285892e-05,
      "loss": 0.3373,
      "step": 21945
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.8411604135176117,
      "learning_rate": 1.4274715173023424e-05,
      "loss": 0.2984,
      "step": 21950
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.8968618547731546,
      "learning_rate": 1.426435514097272e-05,
      "loss": 0.2991,
      "step": 21955
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9819905059089123,
      "learning_rate": 1.4253997368887717e-05,
      "loss": 0.3157,
      "step": 21960
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9888725866185373,
      "learning_rate": 1.4243641858948828e-05,
      "loss": 0.3046,
      "step": 21965
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.926040934764479,
      "learning_rate": 1.423328861333601e-05,
      "loss": 0.2801,
      "step": 21970
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.0373093783367293,
      "learning_rate": 1.4222937634228723e-05,
      "loss": 0.3542,
      "step": 21975
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.7467546127030646,
      "learning_rate": 1.4212588923805939e-05,
      "loss": 0.3083,
      "step": 21980
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9288160630406694,
      "learning_rate": 1.4202242484246187e-05,
      "loss": 0.2745,
      "step": 21985
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.0316764878541007,
      "learning_rate": 1.4191898317727493e-05,
      "loss": 0.3658,
      "step": 21990
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.7459196824500384,
      "learning_rate": 1.418155642642743e-05,
      "loss": 0.3142,
      "step": 21995
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9044159052033965,
      "learning_rate": 1.4171216812523053e-05,
      "loss": 0.2959,
      "step": 22000
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9539881445206426,
      "learning_rate": 1.4160879478190974e-05,
      "loss": 0.3374,
      "step": 22005
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9373581899158129,
      "learning_rate": 1.4150544425607298e-05,
      "loss": 0.3233,
      "step": 22010
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.8690571068705678,
      "learning_rate": 1.4140211656947664e-05,
      "loss": 0.287,
      "step": 22015
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.9674882461493011,
      "learning_rate": 1.4129881174387245e-05,
      "loss": 0.3475,
      "step": 22020
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.8364071193852061,
      "learning_rate": 1.4119552980100686e-05,
      "loss": 0.2963,
      "step": 22025
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2259107520284953,
      "learning_rate": 1.4109227076262204e-05,
      "loss": 0.3318,
      "step": 22030
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.8894802608889425,
      "learning_rate": 1.409890346504551e-05,
      "loss": 0.3229,
      "step": 22035
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0790495184870377,
      "learning_rate": 1.4088582148623821e-05,
      "loss": 0.3631,
      "step": 22040
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8795077450911526,
      "learning_rate": 1.4078263129169881e-05,
      "loss": 0.3219,
      "step": 22045
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8788546073802213,
      "learning_rate": 1.4067946408855953e-05,
      "loss": 0.3304,
      "step": 22050
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2314295934309314,
      "learning_rate": 1.405763198985382e-05,
      "loss": 0.3338,
      "step": 22055
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.940266360497376,
      "learning_rate": 1.4047319874334775e-05,
      "loss": 0.3321,
      "step": 22060
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.950986158184618,
      "learning_rate": 1.403701006446963e-05,
      "loss": 0.2729,
      "step": 22065
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8544533092020391,
      "learning_rate": 1.4026702562428706e-05,
      "loss": 0.3221,
      "step": 22070
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9276787665533494,
      "learning_rate": 1.4016397370381828e-05,
      "loss": 0.3282,
      "step": 22075
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9299458928615699,
      "learning_rate": 1.4006094490498357e-05,
      "loss": 0.3091,
      "step": 22080
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.0240973407710294,
      "learning_rate": 1.3995793924947154e-05,
      "loss": 0.2674,
      "step": 22085
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9795739138584069,
      "learning_rate": 1.3985495675896606e-05,
      "loss": 0.3,
      "step": 22090
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9756987880235342,
      "learning_rate": 1.3975199745514587e-05,
      "loss": 0.3562,
      "step": 22095
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9266357004842157,
      "learning_rate": 1.3964906135968502e-05,
      "loss": 0.3221,
      "step": 22100
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9880875179234331,
      "learning_rate": 1.3954614849425278e-05,
      "loss": 0.2753,
      "step": 22105
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9715887432818779,
      "learning_rate": 1.3944325888051332e-05,
      "loss": 0.341,
      "step": 22110
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9319262744596718,
      "learning_rate": 1.3934039254012581e-05,
      "loss": 0.2815,
      "step": 22115
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.9355321231137924,
      "learning_rate": 1.3923754949474482e-05,
      "loss": 0.2836,
      "step": 22120
    },
    {
      "epoch": 2.58,
      "grad_norm": 0.8493254490597868,
      "learning_rate": 1.391347297660199e-05,
      "loss": 0.3268,
      "step": 22125
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9593297755050466,
      "learning_rate": 1.3903193337559567e-05,
      "loss": 0.3339,
      "step": 22130
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9337554509211765,
      "learning_rate": 1.3892916034511194e-05,
      "loss": 0.3134,
      "step": 22135
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.0314420833235727,
      "learning_rate": 1.3882641069620339e-05,
      "loss": 0.3131,
      "step": 22140
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8467455956369683,
      "learning_rate": 1.3872368445049988e-05,
      "loss": 0.3113,
      "step": 22145
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9612854707358428,
      "learning_rate": 1.3862098162962634e-05,
      "loss": 0.361,
      "step": 22150
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.037164286377405,
      "learning_rate": 1.385183022552029e-05,
      "loss": 0.3351,
      "step": 22155
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.8871833782297859,
      "learning_rate": 1.3841564634884468e-05,
      "loss": 0.2969,
      "step": 22160
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.138127781359419,
      "learning_rate": 1.3831301393216162e-05,
      "loss": 0.327,
      "step": 22165
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.0491251261188774,
      "learning_rate": 1.3821040502675917e-05,
      "loss": 0.2956,
      "step": 22170
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.910138684742929,
      "learning_rate": 1.3810781965423733e-05,
      "loss": 0.2876,
      "step": 22175
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.916874112830565,
      "learning_rate": 1.3800525783619153e-05,
      "loss": 0.3252,
      "step": 22180
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9871472402334044,
      "learning_rate": 1.3790271959421219e-05,
      "loss": 0.2914,
      "step": 22185
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9667719324447465,
      "learning_rate": 1.3780020494988446e-05,
      "loss": 0.3316,
      "step": 22190
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9021139700593398,
      "learning_rate": 1.376977139247889e-05,
      "loss": 0.2942,
      "step": 22195
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9812378086512034,
      "learning_rate": 1.3759524654050099e-05,
      "loss": 0.3077,
      "step": 22200
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.0369509005832003,
      "learning_rate": 1.3749280281859115e-05,
      "loss": 0.3216,
      "step": 22205
    },
    {
      "epoch": 2.59,
      "grad_norm": 0.9880425173466699,
      "learning_rate": 1.3739038278062469e-05,
      "loss": 0.2905,
      "step": 22210
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.0030333932721018,
      "learning_rate": 1.3728798644816226e-05,
      "loss": 0.2905,
      "step": 22215
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9145963169819733,
      "learning_rate": 1.371856138427593e-05,
      "loss": 0.3243,
      "step": 22220
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9094212256059082,
      "learning_rate": 1.3708326498596635e-05,
      "loss": 0.3262,
      "step": 22225
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7211366808369744,
      "learning_rate": 1.3698093989932904e-05,
      "loss": 0.2928,
      "step": 22230
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9290274270175071,
      "learning_rate": 1.368786386043877e-05,
      "loss": 0.3348,
      "step": 22235
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.8849402483176418,
      "learning_rate": 1.3677636112267783e-05,
      "loss": 0.3255,
      "step": 22240
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.8907632563006805,
      "learning_rate": 1.3667410747572993e-05,
      "loss": 0.3148,
      "step": 22245
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2672466276436878,
      "learning_rate": 1.3657187768506947e-05,
      "loss": 0.3489,
      "step": 22250
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9445377996551351,
      "learning_rate": 1.3646967177221703e-05,
      "loss": 0.3111,
      "step": 22255
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9370018840528048,
      "learning_rate": 1.363674897586878e-05,
      "loss": 0.3457,
      "step": 22260
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.8729678721048351,
      "learning_rate": 1.3626533166599234e-05,
      "loss": 0.2879,
      "step": 22265
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9129632297947332,
      "learning_rate": 1.3616319751563583e-05,
      "loss": 0.345,
      "step": 22270
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9279111883711738,
      "learning_rate": 1.3606108732911882e-05,
      "loss": 0.3475,
      "step": 22275
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.9085686116761412,
      "learning_rate": 1.3595900112793628e-05,
      "loss": 0.326,
      "step": 22280
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.8487967564167486,
      "learning_rate": 1.3585693893357859e-05,
      "loss": 0.3286,
      "step": 22285
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.8462189507118796,
      "learning_rate": 1.357549007675309e-05,
      "loss": 0.2757,
      "step": 22290
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.7874921812807293,
      "learning_rate": 1.356528866512734e-05,
      "loss": 0.3138,
      "step": 22295
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.8288795215023179,
      "learning_rate": 1.35550896606281e-05,
      "loss": 0.332,
      "step": 22300
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9836124785910392,
      "learning_rate": 1.3544893065402367e-05,
      "loss": 0.3086,
      "step": 22305
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.0011073229720042,
      "learning_rate": 1.3534698881596634e-05,
      "loss": 0.2994,
      "step": 22310
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.1202544185226706,
      "learning_rate": 1.3524507111356882e-05,
      "loss": 0.2973,
      "step": 22315
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9927803989554137,
      "learning_rate": 1.3514317756828587e-05,
      "loss": 0.2989,
      "step": 22320
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.8851821084364495,
      "learning_rate": 1.350413082015673e-05,
      "loss": 0.3116,
      "step": 22325
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9357914636791029,
      "learning_rate": 1.349394630348574e-05,
      "loss": 0.3163,
      "step": 22330
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9357242386931341,
      "learning_rate": 1.348376420895959e-05,
      "loss": 0.3144,
      "step": 22335
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.055088450514771,
      "learning_rate": 1.3473584538721695e-05,
      "loss": 0.3318,
      "step": 22340
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.8079073834693146,
      "learning_rate": 1.3463407294915003e-05,
      "loss": 0.2924,
      "step": 22345
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9241905447655252,
      "learning_rate": 1.345323247968191e-05,
      "loss": 0.3139,
      "step": 22350
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.200051772921454,
      "learning_rate": 1.3443060095164334e-05,
      "loss": 0.2796,
      "step": 22355
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.94352949674795,
      "learning_rate": 1.3432890143503662e-05,
      "loss": 0.3038,
      "step": 22360
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.8614922011973818,
      "learning_rate": 1.3422722626840791e-05,
      "loss": 0.2795,
      "step": 22365
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.9335999127995972,
      "learning_rate": 1.3412557547316078e-05,
      "loss": 0.282,
      "step": 22370
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.1108151826143688,
      "learning_rate": 1.3402394907069371e-05,
      "loss": 0.2669,
      "step": 22375
    },
    {
      "epoch": 2.61,
      "grad_norm": 0.8977587632381109,
      "learning_rate": 1.3392234708240014e-05,
      "loss": 0.2529,
      "step": 22380
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9005057039331144,
      "learning_rate": 1.338207695296685e-05,
      "loss": 0.3057,
      "step": 22385
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.7995389290467421,
      "learning_rate": 1.3371921643388177e-05,
      "loss": 0.3341,
      "step": 22390
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.8662523631006348,
      "learning_rate": 1.3361768781641815e-05,
      "loss": 0.2716,
      "step": 22395
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9834719270831563,
      "learning_rate": 1.3351618369865033e-05,
      "loss": 0.3302,
      "step": 22400
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9287366305473104,
      "learning_rate": 1.3341470410194589e-05,
      "loss": 0.2872,
      "step": 22405
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9187165831535372,
      "learning_rate": 1.3331324904766745e-05,
      "loss": 0.3323,
      "step": 22410
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0300046433533967,
      "learning_rate": 1.332118185571725e-05,
      "loss": 0.2982,
      "step": 22415
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0061265493377358,
      "learning_rate": 1.3311041265181295e-05,
      "loss": 0.3156,
      "step": 22420
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9755586053544507,
      "learning_rate": 1.33009031352936e-05,
      "loss": 0.3387,
      "step": 22425
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9528265438692267,
      "learning_rate": 1.3290767468188347e-05,
      "loss": 0.2738,
      "step": 22430
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.9918111684227027,
      "learning_rate": 1.328063426599919e-05,
      "loss": 0.3154,
      "step": 22435
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.0567538575954991,
      "learning_rate": 1.3270503530859285e-05,
      "loss": 0.3355,
      "step": 22440
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.994060291909992,
      "learning_rate": 1.3260375264901246e-05,
      "loss": 0.302,
      "step": 22445
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3885612477963847,
      "learning_rate": 1.3250249470257181e-05,
      "loss": 0.3221,
      "step": 22450
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3426906968977006,
      "learning_rate": 1.3240126149058685e-05,
      "loss": 0.3113,
      "step": 22455
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.902323349376137,
      "learning_rate": 1.3230005303436827e-05,
      "loss": 0.3226,
      "step": 22460
    },
    {
      "epoch": 2.62,
      "grad_norm": 0.8948102539833652,
      "learning_rate": 1.321988693552214e-05,
      "loss": 0.3024,
      "step": 22465
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8720083584700585,
      "learning_rate": 1.320977104744464e-05,
      "loss": 0.3368,
      "step": 22470
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8176955567209032,
      "learning_rate": 1.3199657641333834e-05,
      "loss": 0.3016,
      "step": 22475
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9915903249964074,
      "learning_rate": 1.3189546719318707e-05,
      "loss": 0.2988,
      "step": 22480
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9763553630051854,
      "learning_rate": 1.3179438283527718e-05,
      "loss": 0.3508,
      "step": 22485
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9728259857812372,
      "learning_rate": 1.3169332336088777e-05,
      "loss": 0.3427,
      "step": 22490
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9537774092069546,
      "learning_rate": 1.3159228879129315e-05,
      "loss": 0.3316,
      "step": 22495
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9063923724301263,
      "learning_rate": 1.3149127914776196e-05,
      "loss": 0.3076,
      "step": 22500
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8530138379409117,
      "learning_rate": 1.3139029445155787e-05,
      "loss": 0.2639,
      "step": 22505
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8501351601698401,
      "learning_rate": 1.3128933472393933e-05,
      "loss": 0.2753,
      "step": 22510
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8754644724040899,
      "learning_rate": 1.3118839998615923e-05,
      "loss": 0.2801,
      "step": 22515
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0782942094429984,
      "learning_rate": 1.3108749025946548e-05,
      "loss": 0.3156,
      "step": 22520
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.977207959497981,
      "learning_rate": 1.309866055651007e-05,
      "loss": 0.3556,
      "step": 22525
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9827450199941207,
      "learning_rate": 1.3088574592430203e-05,
      "loss": 0.2921,
      "step": 22530
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.9917675416297038,
      "learning_rate": 1.3078491135830162e-05,
      "loss": 0.3382,
      "step": 22535
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0068197277109445,
      "learning_rate": 1.3068410188832608e-05,
      "loss": 0.3286,
      "step": 22540
    },
    {
      "epoch": 2.63,
      "grad_norm": 0.8704928072284427,
      "learning_rate": 1.3058331753559688e-05,
      "loss": 0.3166,
      "step": 22545
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.0238253220711975,
      "learning_rate": 1.3048255832133021e-05,
      "loss": 0.3175,
      "step": 22550
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9657460608900847,
      "learning_rate": 1.3038182426673708e-05,
      "loss": 0.3299,
      "step": 22555
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.91784610433293,
      "learning_rate": 1.3028111539302284e-05,
      "loss": 0.3142,
      "step": 22560
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9897676336839573,
      "learning_rate": 1.3018043172138795e-05,
      "loss": 0.3354,
      "step": 22565
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9026601028583013,
      "learning_rate": 1.3007977327302718e-05,
      "loss": 0.3476,
      "step": 22570
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.1044215284632963,
      "learning_rate": 1.2997914006913025e-05,
      "loss": 0.3696,
      "step": 22575
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8579009026236615,
      "learning_rate": 1.298785321308817e-05,
      "loss": 0.274,
      "step": 22580
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9641122322953601,
      "learning_rate": 1.2977794947946027e-05,
      "loss": 0.288,
      "step": 22585
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8277263305868358,
      "learning_rate": 1.296773921360398e-05,
      "loss": 0.2994,
      "step": 22590
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.1993870908859585,
      "learning_rate": 1.2957686012178877e-05,
      "loss": 0.2986,
      "step": 22595
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.0301932819527575,
      "learning_rate": 1.2947635345786998e-05,
      "loss": 0.2961,
      "step": 22600
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9601215288273234,
      "learning_rate": 1.293758721654414e-05,
      "loss": 0.296,
      "step": 22605
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.1635434021530018,
      "learning_rate": 1.292754162656551e-05,
      "loss": 0.2782,
      "step": 22610
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8829817937768688,
      "learning_rate": 1.291749857796583e-05,
      "loss": 0.3161,
      "step": 22615
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.1350742722734048,
      "learning_rate": 1.290745807285926e-05,
      "loss": 0.3428,
      "step": 22620
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.9924071622341578,
      "learning_rate": 1.2897420113359445e-05,
      "loss": 0.3302,
      "step": 22625
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.008556091355803,
      "learning_rate": 1.288738470157947e-05,
      "loss": 0.3827,
      "step": 22630
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.8654524650235262,
      "learning_rate": 1.2877351839631884e-05,
      "loss": 0.3023,
      "step": 22635
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8770868541537176,
      "learning_rate": 1.286732152962872e-05,
      "loss": 0.3003,
      "step": 22640
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9385612017530551,
      "learning_rate": 1.2857293773681462e-05,
      "loss": 0.293,
      "step": 22645
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.0058701076610725,
      "learning_rate": 1.2847268573901067e-05,
      "loss": 0.3275,
      "step": 22650
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.1351550407047633,
      "learning_rate": 1.2837245932397928e-05,
      "loss": 0.2854,
      "step": 22655
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8812545562750342,
      "learning_rate": 1.2827225851281935e-05,
      "loss": 0.3026,
      "step": 22660
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9334285091298161,
      "learning_rate": 1.2817208332662398e-05,
      "loss": 0.3087,
      "step": 22665
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8975354225181005,
      "learning_rate": 1.280719337864812e-05,
      "loss": 0.3279,
      "step": 22670
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8866553139546327,
      "learning_rate": 1.2797180991347365e-05,
      "loss": 0.3334,
      "step": 22675
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2602813394651082,
      "learning_rate": 1.2787171172867826e-05,
      "loss": 0.2196,
      "step": 22680
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8192613215676635,
      "learning_rate": 1.2777163925316682e-05,
      "loss": 0.2976,
      "step": 22685
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.0084069949842545,
      "learning_rate": 1.2767159250800575e-05,
      "loss": 0.3489,
      "step": 22690
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2190776303772917,
      "learning_rate": 1.2757157151425585e-05,
      "loss": 0.359,
      "step": 22695
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9386472738505037,
      "learning_rate": 1.2747157629297246e-05,
      "loss": 0.3185,
      "step": 22700
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.9288912079926299,
      "learning_rate": 1.2737160686520578e-05,
      "loss": 0.3335,
      "step": 22705
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.0490074872445914,
      "learning_rate": 1.2727166325200029e-05,
      "loss": 0.2788,
      "step": 22710
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.956614048346336,
      "learning_rate": 1.2717174547439531e-05,
      "loss": 0.323,
      "step": 22715
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.8576283768806858,
      "learning_rate": 1.2707185355342463e-05,
      "loss": 0.3005,
      "step": 22720
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.0152148478707892,
      "learning_rate": 1.2697198751011641e-05,
      "loss": 0.3248,
      "step": 22725
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.069992861691313,
      "learning_rate": 1.2687214736549347e-05,
      "loss": 0.314,
      "step": 22730
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9161206915867551,
      "learning_rate": 1.2677233314057325e-05,
      "loss": 0.3414,
      "step": 22735
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8880615504729791,
      "learning_rate": 1.2667254485636768e-05,
      "loss": 0.3206,
      "step": 22740
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.0020043974106367,
      "learning_rate": 1.2657278253388339e-05,
      "loss": 0.3089,
      "step": 22745
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.041960170687636,
      "learning_rate": 1.2647304619412115e-05,
      "loss": 0.387,
      "step": 22750
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9159069360908418,
      "learning_rate": 1.263733358580767e-05,
      "loss": 0.2891,
      "step": 22755
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.0165755903192042,
      "learning_rate": 1.2627365154673996e-05,
      "loss": 0.2849,
      "step": 22760
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8613151940899465,
      "learning_rate": 1.2617399328109569e-05,
      "loss": 0.2735,
      "step": 22765
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9403422949616127,
      "learning_rate": 1.2607436108212278e-05,
      "loss": 0.34,
      "step": 22770
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9925258914315268,
      "learning_rate": 1.25974754970795e-05,
      "loss": 0.3683,
      "step": 22775
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.0820569641316524,
      "learning_rate": 1.2587517496808043e-05,
      "loss": 0.3219,
      "step": 22780
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9830223344350332,
      "learning_rate": 1.2577562109494176e-05,
      "loss": 0.2828,
      "step": 22785
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8993939195910761,
      "learning_rate": 1.2567609337233616e-05,
      "loss": 0.2778,
      "step": 22790
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.8928369391557056,
      "learning_rate": 1.2557659182121521e-05,
      "loss": 0.3312,
      "step": 22795
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9871003396337353,
      "learning_rate": 1.2547711646252488e-05,
      "loss": 0.2489,
      "step": 22800
    },
    {
      "epoch": 2.66,
      "grad_norm": 0.9881226477898498,
      "learning_rate": 1.253776673172059e-05,
      "loss": 0.3138,
      "step": 22805
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9751302778861034,
      "learning_rate": 1.252782444061934e-05,
      "loss": 0.2824,
      "step": 22810
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.8227718195045954,
      "learning_rate": 1.25178847750417e-05,
      "loss": 0.3184,
      "step": 22815
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9784556220522435,
      "learning_rate": 1.250794773708005e-05,
      "loss": 0.2949,
      "step": 22820
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.8974417702940808,
      "learning_rate": 1.2498013328826266e-05,
      "loss": 0.3271,
      "step": 22825
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9215540502532096,
      "learning_rate": 1.2488081552371625e-05,
      "loss": 0.3078,
      "step": 22830
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9041214809423772,
      "learning_rate": 1.2478152409806887e-05,
      "loss": 0.3268,
      "step": 22835
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0712076396392887,
      "learning_rate": 1.2468225903222222e-05,
      "loss": 0.3354,
      "step": 22840
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.7961951267326319,
      "learning_rate": 1.2458302034707273e-05,
      "loss": 0.2792,
      "step": 22845
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9036218869960682,
      "learning_rate": 1.244838080635112e-05,
      "loss": 0.2816,
      "step": 22850
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9248018611702412,
      "learning_rate": 1.2438462220242294e-05,
      "loss": 0.2807,
      "step": 22855
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.8496000027457463,
      "learning_rate": 1.2428546278468753e-05,
      "loss": 0.327,
      "step": 22860
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0322936271653458,
      "learning_rate": 1.2418632983117895e-05,
      "loss": 0.3154,
      "step": 22865
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.0752425689502767,
      "learning_rate": 1.2408722336276582e-05,
      "loss": 0.3314,
      "step": 22870
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9154246028702314,
      "learning_rate": 1.2398814340031111e-05,
      "loss": 0.2862,
      "step": 22875
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.8775205498737138,
      "learning_rate": 1.238890899646722e-05,
      "loss": 0.3083,
      "step": 22880
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9402785406703711,
      "learning_rate": 1.237900630767009e-05,
      "loss": 0.2985,
      "step": 22885
    },
    {
      "epoch": 2.67,
      "grad_norm": 0.9048407571226267,
      "learning_rate": 1.2369106275724338e-05,
      "loss": 0.2926,
      "step": 22890
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.106763611185044,
      "learning_rate": 1.2359208902714015e-05,
      "loss": 0.3305,
      "step": 22895
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8170659558529766,
      "learning_rate": 1.2349314190722628e-05,
      "loss": 0.2429,
      "step": 22900
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9095155861723698,
      "learning_rate": 1.2339422141833127e-05,
      "loss": 0.3375,
      "step": 22905
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9850619858537725,
      "learning_rate": 1.2329532758127876e-05,
      "loss": 0.305,
      "step": 22910
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9414565556280247,
      "learning_rate": 1.2319646041688699e-05,
      "loss": 0.3297,
      "step": 22915
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9239046040538416,
      "learning_rate": 1.230976199459687e-05,
      "loss": 0.3156,
      "step": 22920
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.881956410083199,
      "learning_rate": 1.2299880618933054e-05,
      "loss": 0.2611,
      "step": 22925
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9519661667281967,
      "learning_rate": 1.2290001916777417e-05,
      "loss": 0.3146,
      "step": 22930
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9628756422793493,
      "learning_rate": 1.22801258902095e-05,
      "loss": 0.3105,
      "step": 22935
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.14828743599253,
      "learning_rate": 1.2270252541308321e-05,
      "loss": 0.2648,
      "step": 22940
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8678421024765048,
      "learning_rate": 1.2260381872152326e-05,
      "loss": 0.3029,
      "step": 22945
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9573770585311471,
      "learning_rate": 1.2250513884819403e-05,
      "loss": 0.3216,
      "step": 22950
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9844073693907955,
      "learning_rate": 1.224064858138686e-05,
      "loss": 0.3145,
      "step": 22955
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0625608962780768,
      "learning_rate": 1.2230785963931433e-05,
      "loss": 0.3331,
      "step": 22960
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.9365157892447141,
      "learning_rate": 1.2220926034529315e-05,
      "loss": 0.297,
      "step": 22965
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8385356007864888,
      "learning_rate": 1.2211068795256129e-05,
      "loss": 0.2705,
      "step": 22970
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.8311017205813936,
      "learning_rate": 1.2201214248186931e-05,
      "loss": 0.2961,
      "step": 22975
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.0809393138060064,
      "learning_rate": 1.2191362395396192e-05,
      "loss": 0.3141,
      "step": 22980
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9303754817227148,
      "learning_rate": 1.2181513238957848e-05,
      "loss": 0.3245,
      "step": 22985
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.920008901625138,
      "learning_rate": 1.2171666780945232e-05,
      "loss": 0.2564,
      "step": 22990
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.0481922293324,
      "learning_rate": 1.2161823023431133e-05,
      "loss": 0.3241,
      "step": 22995
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.8904455686199985,
      "learning_rate": 1.2151981968487775e-05,
      "loss": 0.2721,
      "step": 23000
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9426783715872329,
      "learning_rate": 1.2142143618186786e-05,
      "loss": 0.2923,
      "step": 23005
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.00934636607528,
      "learning_rate": 1.213230797459925e-05,
      "loss": 0.308,
      "step": 23010
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9328454098011226,
      "learning_rate": 1.2122475039795672e-05,
      "loss": 0.3043,
      "step": 23015
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9282216598172034,
      "learning_rate": 1.2112644815845999e-05,
      "loss": 0.2925,
      "step": 23020
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.8891395519608644,
      "learning_rate": 1.2102817304819589e-05,
      "loss": 0.3171,
      "step": 23025
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.0581740837403106,
      "learning_rate": 1.2092992508785223e-05,
      "loss": 0.3388,
      "step": 23030
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9850361554562754,
      "learning_rate": 1.208317042981113e-05,
      "loss": 0.2748,
      "step": 23035
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9352614588879432,
      "learning_rate": 1.207335106996497e-05,
      "loss": 0.8495,
      "step": 23040
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9244927194442741,
      "learning_rate": 1.2063534431313819e-05,
      "loss": 0.2738,
      "step": 23045
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.0403690905006984,
      "learning_rate": 1.2053720515924174e-05,
      "loss": 0.3679,
      "step": 23050
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.9372004716169946,
      "learning_rate": 1.2043909325861979e-05,
      "loss": 0.4102,
      "step": 23055
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.8419795794585395,
      "learning_rate": 1.2034100863192576e-05,
      "loss": 0.2657,
      "step": 23060
    },
    {
      "epoch": 2.69,
      "grad_norm": 0.8940436812343538,
      "learning_rate": 1.2024295129980756e-05,
      "loss": 0.2525,
      "step": 23065
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9682780694913872,
      "learning_rate": 1.2014492128290744e-05,
      "loss": 0.3663,
      "step": 23070
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9289494309975811,
      "learning_rate": 1.2004691860186148e-05,
      "loss": 0.2787,
      "step": 23075
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8755275997308232,
      "learning_rate": 1.1994894327730042e-05,
      "loss": 0.2513,
      "step": 23080
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.036489585957659,
      "learning_rate": 1.1985099532984917e-05,
      "loss": 0.3112,
      "step": 23085
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9363246013821599,
      "learning_rate": 1.197530747801266e-05,
      "loss": 0.3112,
      "step": 23090
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9965596474671534,
      "learning_rate": 1.196551816487462e-05,
      "loss": 0.3401,
      "step": 23095
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8801080314559394,
      "learning_rate": 1.1955731595631533e-05,
      "loss": 0.3098,
      "step": 23100
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.7800694148980456,
      "learning_rate": 1.1945947772343585e-05,
      "loss": 0.3022,
      "step": 23105
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.0400503352529242,
      "learning_rate": 1.193616669707037e-05,
      "loss": 0.3802,
      "step": 23110
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8515452029840317,
      "learning_rate": 1.1926388371870913e-05,
      "loss": 0.3027,
      "step": 23115
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8999896412599249,
      "learning_rate": 1.1916612798803655e-05,
      "loss": 0.332,
      "step": 23120
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.47350156682016,
      "learning_rate": 1.1906839979926438e-05,
      "loss": 0.317,
      "step": 23125
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.9591569056072188,
      "learning_rate": 1.1897069917296555e-05,
      "loss": 0.2909,
      "step": 23130
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8642377846151067,
      "learning_rate": 1.1887302612970704e-05,
      "loss": 0.2794,
      "step": 23135
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.1177579897166943,
      "learning_rate": 1.187753806900502e-05,
      "loss": 0.3409,
      "step": 23140
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8855697457554093,
      "learning_rate": 1.1867776287455021e-05,
      "loss": 0.2737,
      "step": 23145
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.8677257562352807,
      "learning_rate": 1.1858017270375679e-05,
      "loss": 0.6087,
      "step": 23150
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.8720100939063528,
      "learning_rate": 1.1848261019821355e-05,
      "loss": 0.2826,
      "step": 23155
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0674767780478314,
      "learning_rate": 1.183850753784585e-05,
      "loss": 0.3158,
      "step": 23160
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9337730019343822,
      "learning_rate": 1.1828756826502382e-05,
      "loss": 0.3378,
      "step": 23165
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.0981605244517039,
      "learning_rate": 1.181900888784356e-05,
      "loss": 0.3013,
      "step": 23170
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.1357751781323333,
      "learning_rate": 1.1809263723921438e-05,
      "loss": 0.3377,
      "step": 23175
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9350243569688483,
      "learning_rate": 1.1799521336787483e-05,
      "loss": 0.3216,
      "step": 23180
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.7818555007876645,
      "learning_rate": 1.178978172849256e-05,
      "loss": 0.2675,
      "step": 23185
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9635830398252786,
      "learning_rate": 1.178004490108695e-05,
      "loss": 0.296,
      "step": 23190
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9675959787159427,
      "learning_rate": 1.1770310856620368e-05,
      "loss": 0.3001,
      "step": 23195
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9657993732349076,
      "learning_rate": 1.1760579597141927e-05,
      "loss": 0.2896,
      "step": 23200
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.9554835133283328,
      "learning_rate": 1.1750851124700166e-05,
      "loss": 0.3138,
      "step": 23205
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.8974886250895949,
      "learning_rate": 1.1741125441343034e-05,
      "loss": 0.3078,
      "step": 23210
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.969468758922195,
      "learning_rate": 1.1731402549117885e-05,
      "loss": 0.2867,
      "step": 23215
    },
    {
      "epoch": 2.71,
      "grad_norm": 92.14569291240348,
      "learning_rate": 1.1721682450071476e-05,
      "loss": 0.3363,
      "step": 23220
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.1018434913945603,
      "learning_rate": 1.1711965146250001e-05,
      "loss": 0.3024,
      "step": 23225
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.8225866825145253,
      "learning_rate": 1.1702250639699053e-05,
      "loss": 0.2941,
      "step": 23230
    },
    {
      "epoch": 2.71,
      "grad_norm": 0.8124022526661175,
      "learning_rate": 1.1692538932463648e-05,
      "loss": 0.2454,
      "step": 23235
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9436323813880034,
      "learning_rate": 1.1682830026588181e-05,
      "loss": 0.2853,
      "step": 23240
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9318640069811314,
      "learning_rate": 1.167312392411649e-05,
      "loss": 0.407,
      "step": 23245
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.0172614894097036,
      "learning_rate": 1.1663420627091822e-05,
      "loss": 0.2768,
      "step": 23250
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.097797890608603,
      "learning_rate": 1.1653720137556814e-05,
      "loss": 0.3432,
      "step": 23255
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9135469122716455,
      "learning_rate": 1.1644022457553506e-05,
      "loss": 0.309,
      "step": 23260
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.0021623665089956,
      "learning_rate": 1.1634327589123373e-05,
      "loss": 0.3867,
      "step": 23265
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.8967069354500952,
      "learning_rate": 1.1624635534307288e-05,
      "loss": 0.2773,
      "step": 23270
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.7630547449795746,
      "learning_rate": 1.1614946295145529e-05,
      "loss": 0.2874,
      "step": 23275
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.8757736405807965,
      "learning_rate": 1.160525987367779e-05,
      "loss": 0.3174,
      "step": 23280
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.7304558363140028,
      "learning_rate": 1.1595576271943157e-05,
      "loss": 0.2611,
      "step": 23285
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9214452185153639,
      "learning_rate": 1.158589549198012e-05,
      "loss": 0.285,
      "step": 23290
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9717582566065859,
      "learning_rate": 1.1576217535826594e-05,
      "loss": 0.2624,
      "step": 23295
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9928811557681508,
      "learning_rate": 1.156654240551989e-05,
      "loss": 0.3046,
      "step": 23300
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.966256974647924,
      "learning_rate": 1.1556870103096731e-05,
      "loss": 0.3205,
      "step": 23305
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9360870584149181,
      "learning_rate": 1.1547200630593224e-05,
      "loss": 0.3169,
      "step": 23310
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.9291298702386991,
      "learning_rate": 1.153753399004491e-05,
      "loss": 0.2691,
      "step": 23315
    },
    {
      "epoch": 2.72,
      "grad_norm": 0.8677978034457244,
      "learning_rate": 1.15278701834867e-05,
      "loss": 0.3149,
      "step": 23320
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9178350772908358,
      "learning_rate": 1.1518209212952943e-05,
      "loss": 0.333,
      "step": 23325
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8582811803734776,
      "learning_rate": 1.1508551080477362e-05,
      "loss": 0.2709,
      "step": 23330
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0674644550282384,
      "learning_rate": 1.1498895788093097e-05,
      "loss": 0.2977,
      "step": 23335
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8162154138435762,
      "learning_rate": 1.1489243337832695e-05,
      "loss": 0.302,
      "step": 23340
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0016102446118273,
      "learning_rate": 1.14795937317281e-05,
      "loss": 0.3436,
      "step": 23345
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8282365695876247,
      "learning_rate": 1.1469946971810652e-05,
      "loss": 0.3037,
      "step": 23350
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0606606541997659,
      "learning_rate": 1.1460303060111083e-05,
      "loss": 0.2995,
      "step": 23355
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0060869548756155,
      "learning_rate": 1.1450661998659548e-05,
      "loss": 0.2688,
      "step": 23360
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8571141652928844,
      "learning_rate": 1.1441023789485589e-05,
      "loss": 0.2797,
      "step": 23365
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.932008462361764,
      "learning_rate": 1.143138843461815e-05,
      "loss": 0.2731,
      "step": 23370
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8478332005755546,
      "learning_rate": 1.1421755936085587e-05,
      "loss": 0.2918,
      "step": 23375
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8891455074469505,
      "learning_rate": 1.1412126295915632e-05,
      "loss": 0.3147,
      "step": 23380
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.8864718154402954,
      "learning_rate": 1.1402499516135412e-05,
      "loss": 0.2855,
      "step": 23385
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.9832832938254445,
      "learning_rate": 1.1392875598771477e-05,
      "loss": 0.3033,
      "step": 23390
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0168586417932541,
      "learning_rate": 1.138325454584977e-05,
      "loss": 0.3291,
      "step": 23395
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.0824853837646562,
      "learning_rate": 1.137363635939561e-05,
      "loss": 0.295,
      "step": 23400
    },
    {
      "epoch": 2.73,
      "grad_norm": 0.7687830909532848,
      "learning_rate": 1.1364021041433725e-05,
      "loss": 0.2661,
      "step": 23405
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.983860772846003,
      "learning_rate": 1.135440859398826e-05,
      "loss": 0.2712,
      "step": 23410
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9377613189200062,
      "learning_rate": 1.1344799019082713e-05,
      "loss": 0.3005,
      "step": 23415
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9877198462161915,
      "learning_rate": 1.133519231874002e-05,
      "loss": 0.3292,
      "step": 23420
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9978212050040649,
      "learning_rate": 1.132558849498247e-05,
      "loss": 0.307,
      "step": 23425
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.005941789063786,
      "learning_rate": 1.1315987549831782e-05,
      "loss": 0.3214,
      "step": 23430
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.1876485332302549,
      "learning_rate": 1.1306389485309052e-05,
      "loss": 0.2902,
      "step": 23435
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.0957393105014936,
      "learning_rate": 1.1296794303434785e-05,
      "loss": 0.2857,
      "step": 23440
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8808072967665173,
      "learning_rate": 1.1287202006228858e-05,
      "loss": 0.3033,
      "step": 23445
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.0399370331087272,
      "learning_rate": 1.1277612595710538e-05,
      "loss": 0.3018,
      "step": 23450
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9565855267443694,
      "learning_rate": 1.1268026073898508e-05,
      "loss": 0.3007,
      "step": 23455
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9695752108566751,
      "learning_rate": 1.1258442442810833e-05,
      "loss": 0.3246,
      "step": 23460
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8858621291842715,
      "learning_rate": 1.1248861704464971e-05,
      "loss": 0.3326,
      "step": 23465
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9662613576891774,
      "learning_rate": 1.1239283860877756e-05,
      "loss": 0.3875,
      "step": 23470
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.9665851921586993,
      "learning_rate": 1.122970891406543e-05,
      "loss": 0.3164,
      "step": 23475
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8920046177154548,
      "learning_rate": 1.1220136866043629e-05,
      "loss": 0.3219,
      "step": 23480
    },
    {
      "epoch": 2.74,
      "grad_norm": 0.8301284085478287,
      "learning_rate": 1.121056771882735e-05,
      "loss": 0.2779,
      "step": 23485
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.0708710290620955,
      "learning_rate": 1.1201001474431022e-05,
      "loss": 0.3004,
      "step": 23490
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9160839116538275,
      "learning_rate": 1.1191438134868414e-05,
      "loss": 0.2972,
      "step": 23495
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8986245165879151,
      "learning_rate": 1.1181877702152725e-05,
      "loss": 0.3049,
      "step": 23500
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8913374620198132,
      "learning_rate": 1.1172320178296522e-05,
      "loss": 0.3168,
      "step": 23505
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8395060466096411,
      "learning_rate": 1.1162765565311775e-05,
      "loss": 0.2649,
      "step": 23510
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9027930251491513,
      "learning_rate": 1.1153213865209821e-05,
      "loss": 0.3267,
      "step": 23515
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8577130979311428,
      "learning_rate": 1.1143665080001382e-05,
      "loss": 0.2876,
      "step": 23520
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.069821600651597,
      "learning_rate": 1.1134119211696589e-05,
      "loss": 0.351,
      "step": 23525
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.8846252990133306,
      "learning_rate": 1.1124576262304944e-05,
      "loss": 0.2868,
      "step": 23530
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9389204534337867,
      "learning_rate": 1.1115036233835349e-05,
      "loss": 0.323,
      "step": 23535
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.905324547311633,
      "learning_rate": 1.1105499128296062e-05,
      "loss": 0.2857,
      "step": 23540
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.0434526042134222,
      "learning_rate": 1.1095964947694761e-05,
      "loss": 0.2864,
      "step": 23545
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.1948811192937119,
      "learning_rate": 1.1086433694038476e-05,
      "loss": 0.3767,
      "step": 23550
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9855581532827002,
      "learning_rate": 1.107690536933364e-05,
      "loss": 0.2921,
      "step": 23555
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9798418335090091,
      "learning_rate": 1.1067379975586077e-05,
      "loss": 0.2997,
      "step": 23560
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.903539605706739,
      "learning_rate": 1.1057857514800959e-05,
      "loss": 0.3094,
      "step": 23565
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9374344894487049,
      "learning_rate": 1.1048337988982883e-05,
      "loss": 0.3198,
      "step": 23570
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.0006717022993645,
      "learning_rate": 1.1038821400135807e-05,
      "loss": 0.2578,
      "step": 23575
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.9603612301139439,
      "learning_rate": 1.102930775026306e-05,
      "loss": 0.3162,
      "step": 23580
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.8797527273472097,
      "learning_rate": 1.1019797041367386e-05,
      "loss": 0.3177,
      "step": 23585
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9466543099457245,
      "learning_rate": 1.1010289275450864e-05,
      "loss": 0.2613,
      "step": 23590
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0794091578765985,
      "learning_rate": 1.1000784454514992e-05,
      "loss": 0.3242,
      "step": 23595
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0583407393853788,
      "learning_rate": 1.0991282580560633e-05,
      "loss": 0.3046,
      "step": 23600
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9470761468887988,
      "learning_rate": 1.0981783655588038e-05,
      "loss": 0.3098,
      "step": 23605
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0091157325675195,
      "learning_rate": 1.0972287681596824e-05,
      "loss": 0.2901,
      "step": 23610
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9282651434372654,
      "learning_rate": 1.0962794660585984e-05,
      "loss": 0.2749,
      "step": 23615
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.069238462542684,
      "learning_rate": 1.0953304594553906e-05,
      "loss": 0.2848,
      "step": 23620
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9208592207905248,
      "learning_rate": 1.094381748549835e-05,
      "loss": 0.3125,
      "step": 23625
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9759889117580864,
      "learning_rate": 1.0934333335416458e-05,
      "loss": 0.321,
      "step": 23630
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.8520053159356505,
      "learning_rate": 1.092485214630473e-05,
      "loss": 0.3363,
      "step": 23635
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0416556282099745,
      "learning_rate": 1.0915373920159067e-05,
      "loss": 0.3111,
      "step": 23640
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0005953944544534,
      "learning_rate": 1.0905898658974719e-05,
      "loss": 0.3018,
      "step": 23645
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9879947021385429,
      "learning_rate": 1.089642636474634e-05,
      "loss": 0.2662,
      "step": 23650
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.9098697306046369,
      "learning_rate": 1.0886957039467954e-05,
      "loss": 0.2973,
      "step": 23655
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.1403501140169232,
      "learning_rate": 1.0877490685132935e-05,
      "loss": 0.2864,
      "step": 23660
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.0968395990976476,
      "learning_rate": 1.0868027303734057e-05,
      "loss": 0.298,
      "step": 23665
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.8532057197963661,
      "learning_rate": 1.0858566897263475e-05,
      "loss": 0.3434,
      "step": 23670
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9417573683852283,
      "learning_rate": 1.0849109467712681e-05,
      "loss": 0.3072,
      "step": 23675
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.8550553077331028,
      "learning_rate": 1.0839655017072583e-05,
      "loss": 0.2988,
      "step": 23680
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9972584699976879,
      "learning_rate": 1.0830203547333422e-05,
      "loss": 0.2996,
      "step": 23685
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0017408881083572,
      "learning_rate": 1.0820755060484844e-05,
      "loss": 0.3076,
      "step": 23690
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9377687191213354,
      "learning_rate": 1.081130955851585e-05,
      "loss": 0.3408,
      "step": 23695
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.8865272909002163,
      "learning_rate": 1.0801867043414828e-05,
      "loss": 0.2973,
      "step": 23700
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9055830495038591,
      "learning_rate": 1.0792427517169512e-05,
      "loss": 0.3146,
      "step": 23705
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.2083013782768683,
      "learning_rate": 1.0782990981767033e-05,
      "loss": 0.3924,
      "step": 23710
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0168900333033841,
      "learning_rate": 1.0773557439193865e-05,
      "loss": 0.3041,
      "step": 23715
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9154854672836121,
      "learning_rate": 1.0764126891435877e-05,
      "loss": 0.2904,
      "step": 23720
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9268696915625491,
      "learning_rate": 1.0754699340478311e-05,
      "loss": 0.2855,
      "step": 23725
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.9365293492548428,
      "learning_rate": 1.074527478830574e-05,
      "loss": 0.2945,
      "step": 23730
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0241618296670367,
      "learning_rate": 1.073585323690214e-05,
      "loss": 0.3134,
      "step": 23735
    },
    {
      "epoch": 2.77,
      "grad_norm": 0.8317413520640192,
      "learning_rate": 1.0726434688250864e-05,
      "loss": 0.3482,
      "step": 23740
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.0491610838366687,
      "learning_rate": 1.071701914433459e-05,
      "loss": 0.2888,
      "step": 23745
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.1826379442010508,
      "learning_rate": 1.0707606607135407e-05,
      "loss": 0.2835,
      "step": 23750
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.0022085838235593,
      "learning_rate": 1.0698197078634733e-05,
      "loss": 0.3022,
      "step": 23755
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.1295260043233306,
      "learning_rate": 1.0688790560813388e-05,
      "loss": 0.2994,
      "step": 23760
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8590583871989996,
      "learning_rate": 1.0679387055651538e-05,
      "loss": 0.3298,
      "step": 23765
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.7896318522487846,
      "learning_rate": 1.0669986565128732e-05,
      "loss": 0.274,
      "step": 23770
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9470255666600508,
      "learning_rate": 1.0660589091223855e-05,
      "loss": 0.2834,
      "step": 23775
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.806691952222429,
      "learning_rate": 1.0651194635915172e-05,
      "loss": 0.2869,
      "step": 23780
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9920244881950887,
      "learning_rate": 1.064180320118032e-05,
      "loss": 0.3048,
      "step": 23785
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9811043245735639,
      "learning_rate": 1.0632414788996294e-05,
      "loss": 0.3593,
      "step": 23790
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.981571241604834,
      "learning_rate": 1.0623029401339466e-05,
      "loss": 0.2827,
      "step": 23795
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.908638091882898,
      "learning_rate": 1.0613647040185535e-05,
      "loss": 0.2473,
      "step": 23800
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8804751881187239,
      "learning_rate": 1.0604267707509608e-05,
      "loss": 0.2593,
      "step": 23805
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9621528614821562,
      "learning_rate": 1.0594891405286112e-05,
      "loss": 0.2765,
      "step": 23810
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8760314269025647,
      "learning_rate": 1.058551813548887e-05,
      "loss": 0.3164,
      "step": 23815
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.0026624709941225,
      "learning_rate": 1.0576147900091058e-05,
      "loss": 0.2776,
      "step": 23820
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8610740877250057,
      "learning_rate": 1.0566780701065193e-05,
      "loss": 0.2928,
      "step": 23825
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.9211160043873191,
      "learning_rate": 1.055741654038318e-05,
      "loss": 0.2913,
      "step": 23830
    },
    {
      "epoch": 2.78,
      "grad_norm": 0.8762788904048749,
      "learning_rate": 1.0548055420016276e-05,
      "loss": 0.3495,
      "step": 23835
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9220130035727188,
      "learning_rate": 1.0538697341935091e-05,
      "loss": 0.2629,
      "step": 23840
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8588414662232006,
      "learning_rate": 1.0529342308109586e-05,
      "loss": 0.3281,
      "step": 23845
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9718662344894322,
      "learning_rate": 1.0519990320509104e-05,
      "loss": 0.2874,
      "step": 23850
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9142675264104025,
      "learning_rate": 1.0510641381102337e-05,
      "loss": 0.2962,
      "step": 23855
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8873390971475601,
      "learning_rate": 1.0501295491857335e-05,
      "loss": 0.3124,
      "step": 23860
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.073030921493654,
      "learning_rate": 1.0491952654741514e-05,
      "loss": 0.2933,
      "step": 23865
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0891083550397802,
      "learning_rate": 1.0482612871721628e-05,
      "loss": 0.3159,
      "step": 23870
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9627610015886772,
      "learning_rate": 1.0473276144763793e-05,
      "loss": 0.3195,
      "step": 23875
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9846751290510684,
      "learning_rate": 1.0463942475833496e-05,
      "loss": 0.296,
      "step": 23880
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0213771648613001,
      "learning_rate": 1.0454611866895572e-05,
      "loss": 0.3272,
      "step": 23885
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.878120819137557,
      "learning_rate": 1.044528431991422e-05,
      "loss": 0.323,
      "step": 23890
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.0376013751257847,
      "learning_rate": 1.0435959836852967e-05,
      "loss": 0.3301,
      "step": 23895
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.6877638430729498,
      "learning_rate": 1.0426638419674739e-05,
      "loss": 0.2612,
      "step": 23900
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.014633177373045,
      "learning_rate": 1.0417320070341765e-05,
      "loss": 0.3062,
      "step": 23905
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8993083758190988,
      "learning_rate": 1.040800479081568e-05,
      "loss": 0.3274,
      "step": 23910
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.9128952190570753,
      "learning_rate": 1.0398692583057429e-05,
      "loss": 0.3708,
      "step": 23915
    },
    {
      "epoch": 2.79,
      "grad_norm": 0.8357886771472559,
      "learning_rate": 1.0389383449027335e-05,
      "loss": 0.3347,
      "step": 23920
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3085852767781703,
      "learning_rate": 1.038007739068507e-05,
      "loss": 0.2539,
      "step": 23925
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.868166987701541,
      "learning_rate": 1.0370774409989655e-05,
      "loss": 0.3505,
      "step": 23930
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.8720868415654188,
      "learning_rate": 1.036147450889948e-05,
      "loss": 0.3045,
      "step": 23935
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9692423511286279,
      "learning_rate": 1.0352177689372256e-05,
      "loss": 0.3012,
      "step": 23940
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.1488121193790917,
      "learning_rate": 1.0342883953365051e-05,
      "loss": 0.3919,
      "step": 23945
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9253449880918742,
      "learning_rate": 1.033359330283431e-05,
      "loss": 0.2679,
      "step": 23950
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.8402356983312381,
      "learning_rate": 1.0324305739735807e-05,
      "loss": 0.2856,
      "step": 23955
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.0368873730232566,
      "learning_rate": 1.0315021266024683e-05,
      "loss": 0.3022,
      "step": 23960
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.1496190175764356,
      "learning_rate": 1.0305739883655395e-05,
      "loss": 0.3141,
      "step": 23965
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.8834647969007838,
      "learning_rate": 1.0296461594581791e-05,
      "loss": 0.2887,
      "step": 23970
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9370827068333535,
      "learning_rate": 1.0287186400757032e-05,
      "loss": 0.2785,
      "step": 23975
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.0303960318240577,
      "learning_rate": 1.0277914304133661e-05,
      "loss": 0.3355,
      "step": 23980
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9915618216413891,
      "learning_rate": 1.0268645306663532e-05,
      "loss": 0.2747,
      "step": 23985
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3257351002283646,
      "learning_rate": 1.0259379410297875e-05,
      "loss": 0.3253,
      "step": 23990
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.8704719409336908,
      "learning_rate": 1.025011661698726e-05,
      "loss": 0.2971,
      "step": 23995
    },
    {
      "epoch": 2.8,
      "grad_norm": 16.970800750921004,
      "learning_rate": 1.0240856928681611e-05,
      "loss": 0.4302,
      "step": 24000
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.9618174684253242,
      "learning_rate": 1.0231600347330179e-05,
      "loss": 0.3103,
      "step": 24005
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.93887331601894,
      "learning_rate": 1.022234687488156e-05,
      "loss": 0.2672,
      "step": 24010
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9079812314562753,
      "learning_rate": 1.021309651328372e-05,
      "loss": 0.2884,
      "step": 24015
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.811229992269716,
      "learning_rate": 1.0203849264483955e-05,
      "loss": 0.3001,
      "step": 24020
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.8988081282422115,
      "learning_rate": 1.019460513042891e-05,
      "loss": 0.3309,
      "step": 24025
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.8694364127794912,
      "learning_rate": 1.0185364113064577e-05,
      "loss": 0.2498,
      "step": 24030
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.064006727748688,
      "learning_rate": 1.0176126214336282e-05,
      "loss": 0.2985,
      "step": 24035
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9328393222840907,
      "learning_rate": 1.016689143618869e-05,
      "loss": 0.3444,
      "step": 24040
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9201184699376819,
      "learning_rate": 1.0157659780565823e-05,
      "loss": 0.3114,
      "step": 24045
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.831694478570635,
      "learning_rate": 1.0148431249411056e-05,
      "loss": 0.2788,
      "step": 24050
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.0546481758656725,
      "learning_rate": 1.013920584466707e-05,
      "loss": 0.3586,
      "step": 24055
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9665089853766448,
      "learning_rate": 1.0129983568275919e-05,
      "loss": 0.2626,
      "step": 24060
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9440834510312459,
      "learning_rate": 1.0120764422179e-05,
      "loss": 0.3136,
      "step": 24065
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.894586536893201,
      "learning_rate": 1.0111548408317017e-05,
      "loss": 0.2692,
      "step": 24070
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.095624332163829,
      "learning_rate": 1.0102335528630061e-05,
      "loss": 0.2727,
      "step": 24075
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9391056339302459,
      "learning_rate": 1.0093125785057517e-05,
      "loss": 0.2903,
      "step": 24080
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.066638189111323,
      "learning_rate": 1.0083919179538146e-05,
      "loss": 0.3365,
      "step": 24085
    },
    {
      "epoch": 2.81,
      "grad_norm": 0.9294220742852979,
      "learning_rate": 1.0074715714010031e-05,
      "loss": 0.3135,
      "step": 24090
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9222853252288646,
      "learning_rate": 1.0065515390410607e-05,
      "loss": 0.2619,
      "step": 24095
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9088450074899105,
      "learning_rate": 1.0056318210676633e-05,
      "loss": 0.2775,
      "step": 24100
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9113917583432781,
      "learning_rate": 1.0047124176744198e-05,
      "loss": 0.3094,
      "step": 24105
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9123154466025912,
      "learning_rate": 1.0037933290548757e-05,
      "loss": 0.3239,
      "step": 24110
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.8819793087263148,
      "learning_rate": 1.0028745554025082e-05,
      "loss": 0.3033,
      "step": 24115
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.04487368529544,
      "learning_rate": 1.0019560969107302e-05,
      "loss": 0.3469,
      "step": 24120
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9133161019310184,
      "learning_rate": 1.0010379537728845e-05,
      "loss": 0.2764,
      "step": 24125
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9286874522791484,
      "learning_rate": 1.000120126182252e-05,
      "loss": 0.3253,
      "step": 24130
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9415924465774013,
      "learning_rate": 9.992026143320434e-06,
      "loss": 0.3031,
      "step": 24135
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.8500447394001744,
      "learning_rate": 9.982854184154053e-06,
      "loss": 0.3126,
      "step": 24140
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.0655944706528835,
      "learning_rate": 9.973685386254174e-06,
      "loss": 0.2806,
      "step": 24145
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9758574861517874,
      "learning_rate": 9.964519751550916e-06,
      "loss": 0.2787,
      "step": 24150
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.8918688811302502,
      "learning_rate": 9.955357281973743e-06,
      "loss": 0.2993,
      "step": 24155
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.888293877938757,
      "learning_rate": 9.946197979451454e-06,
      "loss": 0.2925,
      "step": 24160
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.843654989719251,
      "learning_rate": 9.937041845912188e-06,
      "loss": 0.3359,
      "step": 24165
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.1291216942967641,
      "learning_rate": 9.927888883283395e-06,
      "loss": 0.298,
      "step": 24170
    },
    {
      "epoch": 2.82,
      "grad_norm": 0.9005704624947608,
      "learning_rate": 9.918739093491863e-06,
      "loss": 0.3058,
      "step": 24175
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8744905349602763,
      "learning_rate": 9.909592478463725e-06,
      "loss": 0.2782,
      "step": 24180
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9765592927756971,
      "learning_rate": 9.900449040124444e-06,
      "loss": 0.3324,
      "step": 24185
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8608439155318706,
      "learning_rate": 9.891308780398813e-06,
      "loss": 0.2496,
      "step": 24190
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9443155856131693,
      "learning_rate": 9.88217170121094e-06,
      "loss": 0.3242,
      "step": 24195
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9430139713644371,
      "learning_rate": 9.87303780448429e-06,
      "loss": 0.2959,
      "step": 24200
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.964402836844983,
      "learning_rate": 9.863907092141624e-06,
      "loss": 0.2831,
      "step": 24205
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9884365774866519,
      "learning_rate": 9.854779566105068e-06,
      "loss": 0.3134,
      "step": 24210
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.0053095844951974,
      "learning_rate": 9.845655228296067e-06,
      "loss": 0.3005,
      "step": 24215
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.948097139787577,
      "learning_rate": 9.836534080635371e-06,
      "loss": 0.3142,
      "step": 24220
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.892488647252319,
      "learning_rate": 9.827416125043085e-06,
      "loss": 0.3265,
      "step": 24225
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.893053947844135,
      "learning_rate": 9.818301363438647e-06,
      "loss": 0.3282,
      "step": 24230
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9258964399894913,
      "learning_rate": 9.809189797740792e-06,
      "loss": 0.296,
      "step": 24235
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8800263191295711,
      "learning_rate": 9.800081429867614e-06,
      "loss": 0.3035,
      "step": 24240
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.1475290789393873,
      "learning_rate": 9.790976261736504e-06,
      "loss": 0.3253,
      "step": 24245
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.8461825763490542,
      "learning_rate": 9.781874295264204e-06,
      "loss": 0.2804,
      "step": 24250
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.9579451385308144,
      "learning_rate": 9.772775532366774e-06,
      "loss": 0.3204,
      "step": 24255
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.893914078122257,
      "learning_rate": 9.763679974959608e-06,
      "loss": 0.3487,
      "step": 24260
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.8925869635932685,
      "learning_rate": 9.754587624957406e-06,
      "loss": 0.315,
      "step": 24265
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.8943505956822083,
      "learning_rate": 9.745498484274192e-06,
      "loss": 0.3058,
      "step": 24270
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9838052552361592,
      "learning_rate": 9.73641255482334e-06,
      "loss": 0.3012,
      "step": 24275
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9122863458162808,
      "learning_rate": 9.727329838517529e-06,
      "loss": 0.335,
      "step": 24280
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9941288036306175,
      "learning_rate": 9.718250337268778e-06,
      "loss": 0.2565,
      "step": 24285
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.944530994635024,
      "learning_rate": 9.709174052988393e-06,
      "loss": 0.3127,
      "step": 24290
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.8257866544916291,
      "learning_rate": 9.700100987587052e-06,
      "loss": 0.2639,
      "step": 24295
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9174961357745193,
      "learning_rate": 9.691031142974707e-06,
      "loss": 0.3362,
      "step": 24300
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9575370788479027,
      "learning_rate": 9.681964521060669e-06,
      "loss": 0.2745,
      "step": 24305
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.8954608120326626,
      "learning_rate": 9.672901123753561e-06,
      "loss": 0.3191,
      "step": 24310
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.96033961140247,
      "learning_rate": 9.663840952961312e-06,
      "loss": 0.2855,
      "step": 24315
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9656117809561124,
      "learning_rate": 9.654784010591184e-06,
      "loss": 0.2707,
      "step": 24320
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9336011373036461,
      "learning_rate": 9.645730298549771e-06,
      "loss": 0.2401,
      "step": 24325
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9060333854482696,
      "learning_rate": 9.636679818742963e-06,
      "loss": 0.3044,
      "step": 24330
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.0578848328033112,
      "learning_rate": 9.627632573075977e-06,
      "loss": 0.3122,
      "step": 24335
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.8086007794739632,
      "learning_rate": 9.618588563453357e-06,
      "loss": 0.3296,
      "step": 24340
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.1310610018705838,
      "learning_rate": 9.609547791778964e-06,
      "loss": 0.3143,
      "step": 24345
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.9452698888828003,
      "learning_rate": 9.600510259955975e-06,
      "loss": 0.3578,
      "step": 24350
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.998464299457308,
      "learning_rate": 9.591475969886893e-06,
      "loss": 0.3508,
      "step": 24355
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0542119406339037,
      "learning_rate": 9.58244492347352e-06,
      "loss": 0.3273,
      "step": 24360
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.8454439583324652,
      "learning_rate": 9.573417122616981e-06,
      "loss": 0.2897,
      "step": 24365
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0502961480556605,
      "learning_rate": 9.56439256921773e-06,
      "loss": 0.3027,
      "step": 24370
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.548830058672662,
      "learning_rate": 9.555371265175533e-06,
      "loss": 0.3567,
      "step": 24375
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9410034588960128,
      "learning_rate": 9.546353212389472e-06,
      "loss": 0.3176,
      "step": 24380
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.8484823982551155,
      "learning_rate": 9.537338412757927e-06,
      "loss": 0.3084,
      "step": 24385
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9862240061867517,
      "learning_rate": 9.528326868178616e-06,
      "loss": 0.3441,
      "step": 24390
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.0163199165368089,
      "learning_rate": 9.519318580548572e-06,
      "loss": 0.2759,
      "step": 24395
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.8309326565634324,
      "learning_rate": 9.510313551764128e-06,
      "loss": 0.2636,
      "step": 24400
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9829718270033582,
      "learning_rate": 9.501311783720923e-06,
      "loss": 0.2574,
      "step": 24405
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.8814662162771878,
      "learning_rate": 9.492313278313936e-06,
      "loss": 0.2722,
      "step": 24410
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9291428404439623,
      "learning_rate": 9.483318037437447e-06,
      "loss": 0.2412,
      "step": 24415
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.8990467108792146,
      "learning_rate": 9.474326062985044e-06,
      "loss": 0.2678,
      "step": 24420
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.8854606173102194,
      "learning_rate": 9.465337356849646e-06,
      "loss": 0.2733,
      "step": 24425
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.9367297271360102,
      "learning_rate": 9.456351920923459e-06,
      "loss": 0.3471,
      "step": 24430
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.1542124346690832,
      "learning_rate": 9.447369757098002e-06,
      "loss": 0.3439,
      "step": 24435
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.8136588297985716,
      "learning_rate": 9.438390867264121e-06,
      "loss": 0.3132,
      "step": 24440
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9276599989330045,
      "learning_rate": 9.429415253311973e-06,
      "loss": 0.3015,
      "step": 24445
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9264355796206835,
      "learning_rate": 9.420442917131023e-06,
      "loss": 0.3267,
      "step": 24450
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9778534344340757,
      "learning_rate": 9.411473860610024e-06,
      "loss": 0.352,
      "step": 24455
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9749908719170531,
      "learning_rate": 9.402508085637074e-06,
      "loss": 0.3242,
      "step": 24460
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.0216443893015401,
      "learning_rate": 9.393545594099545e-06,
      "loss": 0.2599,
      "step": 24465
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.073380628884854,
      "learning_rate": 9.384586387884154e-06,
      "loss": 0.3333,
      "step": 24470
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.004031629808907,
      "learning_rate": 9.37563046887689e-06,
      "loss": 0.2896,
      "step": 24475
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.049920041637581,
      "learning_rate": 9.366677838963078e-06,
      "loss": 0.2889,
      "step": 24480
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.8857842529992311,
      "learning_rate": 9.357728500027335e-06,
      "loss": 0.2873,
      "step": 24485
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.1858521971637972,
      "learning_rate": 9.348782453953605e-06,
      "loss": 0.2461,
      "step": 24490
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.300941750191885,
      "learning_rate": 9.339839702625117e-06,
      "loss": 0.3624,
      "step": 24495
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9108355469535715,
      "learning_rate": 9.330900247924396e-06,
      "loss": 0.3194,
      "step": 24500
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.0251410109567916,
      "learning_rate": 9.321964091733309e-06,
      "loss": 0.3045,
      "step": 24505
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.873628049375326,
      "learning_rate": 9.313031235933006e-06,
      "loss": 0.2968,
      "step": 24510
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.9425558829599604,
      "learning_rate": 9.304101682403948e-06,
      "loss": 0.2931,
      "step": 24515
    },
    {
      "epoch": 2.86,
      "grad_norm": 0.8856538997826833,
      "learning_rate": 9.295175433025908e-06,
      "loss": 0.3086,
      "step": 24520
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.951910918674955,
      "learning_rate": 9.286252489677944e-06,
      "loss": 0.3129,
      "step": 24525
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.895392300327379,
      "learning_rate": 9.277332854238424e-06,
      "loss": 0.2881,
      "step": 24530
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.921277631091775,
      "learning_rate": 9.26841652858503e-06,
      "loss": 0.339,
      "step": 24535
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9034321573935192,
      "learning_rate": 9.259503514594755e-06,
      "loss": 0.3136,
      "step": 24540
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.0666000335582098,
      "learning_rate": 9.250593814143862e-06,
      "loss": 0.3038,
      "step": 24545
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9381343135718994,
      "learning_rate": 9.241687429107942e-06,
      "loss": 0.3199,
      "step": 24550
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9415290930121821,
      "learning_rate": 9.2327843613619e-06,
      "loss": 0.2776,
      "step": 24555
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8276495863971298,
      "learning_rate": 9.223884612779895e-06,
      "loss": 0.3221,
      "step": 24560
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9501980539342294,
      "learning_rate": 9.214988185235443e-06,
      "loss": 0.2676,
      "step": 24565
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.015005000412691,
      "learning_rate": 9.206095080601319e-06,
      "loss": 0.3148,
      "step": 24570
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8803260434759336,
      "learning_rate": 9.197205300749616e-06,
      "loss": 0.273,
      "step": 24575
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9683203169687521,
      "learning_rate": 9.188318847551734e-06,
      "loss": 0.2614,
      "step": 24580
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.0098958851834654,
      "learning_rate": 9.179435722878366e-06,
      "loss": 0.288,
      "step": 24585
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.9664564904216176,
      "learning_rate": 9.1705559285995e-06,
      "loss": 0.313,
      "step": 24590
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.847107745954699,
      "learning_rate": 9.161679466584414e-06,
      "loss": 0.2922,
      "step": 24595
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8335466475318957,
      "learning_rate": 9.152806338701702e-06,
      "loss": 0.2948,
      "step": 24600
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.8642420220658538,
      "learning_rate": 9.143936546819257e-06,
      "loss": 0.3179,
      "step": 24605
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.857499274055507,
      "learning_rate": 9.135070092804268e-06,
      "loss": 0.289,
      "step": 24610
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8468770120674177,
      "learning_rate": 9.126206978523202e-06,
      "loss": 0.2744,
      "step": 24615
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.1222186531946703,
      "learning_rate": 9.117347205841842e-06,
      "loss": 0.3313,
      "step": 24620
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.0387109032163027,
      "learning_rate": 9.108490776625272e-06,
      "loss": 0.33,
      "step": 24625
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.052204403017723,
      "learning_rate": 9.099637692737847e-06,
      "loss": 0.3172,
      "step": 24630
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9204754344447226,
      "learning_rate": 9.090787956043254e-06,
      "loss": 0.2953,
      "step": 24635
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9777513645130192,
      "learning_rate": 9.081941568404431e-06,
      "loss": 0.3142,
      "step": 24640
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8169307014216313,
      "learning_rate": 9.073098531683652e-06,
      "loss": 0.2662,
      "step": 24645
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9508802286425354,
      "learning_rate": 9.064258847742465e-06,
      "loss": 0.3173,
      "step": 24650
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9882138493539279,
      "learning_rate": 9.055422518441723e-06,
      "loss": 0.3045,
      "step": 24655
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.0014893014288486,
      "learning_rate": 9.04658954564156e-06,
      "loss": 0.3147,
      "step": 24660
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.8920019982657743,
      "learning_rate": 9.037759931201401e-06,
      "loss": 0.2803,
      "step": 24665
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9169261574206906,
      "learning_rate": 9.028933676979975e-06,
      "loss": 0.3046,
      "step": 24670
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9619180282976926,
      "learning_rate": 9.020110784835309e-06,
      "loss": 0.3194,
      "step": 24675
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.0412445137432709,
      "learning_rate": 9.011291256624722e-06,
      "loss": 0.3042,
      "step": 24680
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9475581134203468,
      "learning_rate": 9.002475094204793e-06,
      "loss": 0.3198,
      "step": 24685
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.9363921415624746,
      "learning_rate": 8.993662299431438e-06,
      "loss": 0.3045,
      "step": 24690
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9427388755947488,
      "learning_rate": 8.984852874159828e-06,
      "loss": 0.3174,
      "step": 24695
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9072574323751448,
      "learning_rate": 8.976046820244444e-06,
      "loss": 0.3243,
      "step": 24700
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.1344893055326293,
      "learning_rate": 8.967244139539064e-06,
      "loss": 0.3097,
      "step": 24705
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9083667456910753,
      "learning_rate": 8.958444833896722e-06,
      "loss": 0.329,
      "step": 24710
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.0562803420074898,
      "learning_rate": 8.94964890516978e-06,
      "loss": 0.2935,
      "step": 24715
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.012198085776275,
      "learning_rate": 8.940856355209879e-06,
      "loss": 0.3238,
      "step": 24720
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9521868464235179,
      "learning_rate": 8.932067185867923e-06,
      "loss": 0.3065,
      "step": 24725
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.049505578202553,
      "learning_rate": 8.923281398994142e-06,
      "loss": 0.3147,
      "step": 24730
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9006298487894506,
      "learning_rate": 8.914498996438023e-06,
      "loss": 0.3121,
      "step": 24735
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9531838808650365,
      "learning_rate": 8.905719980048357e-06,
      "loss": 0.3185,
      "step": 24740
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.8874956465339281,
      "learning_rate": 8.896944351673223e-06,
      "loss": 0.3331,
      "step": 24745
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9455213941980077,
      "learning_rate": 8.888172113159989e-06,
      "loss": 0.3334,
      "step": 24750
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.9489368419238016,
      "learning_rate": 8.879403266355295e-06,
      "loss": 0.2781,
      "step": 24755
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.884408675047062,
      "learning_rate": 8.870637813105067e-06,
      "loss": 0.3338,
      "step": 24760
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.7607836167263174,
      "learning_rate": 8.861875755254536e-06,
      "loss": 0.2626,
      "step": 24765
    },
    {
      "epoch": 2.89,
      "grad_norm": 0.8606595833042103,
      "learning_rate": 8.8531170946482e-06,
      "loss": 0.3219,
      "step": 24770
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.0075728878169883,
      "learning_rate": 8.844361833129863e-06,
      "loss": 0.2816,
      "step": 24775
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0042239632571994,
      "learning_rate": 8.835609972542583e-06,
      "loss": 0.2671,
      "step": 24780
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9931029966640522,
      "learning_rate": 8.826861514728729e-06,
      "loss": 0.2486,
      "step": 24785
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9048359932877287,
      "learning_rate": 8.818116461529932e-06,
      "loss": 0.2602,
      "step": 24790
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9576046174799073,
      "learning_rate": 8.809374814787124e-06,
      "loss": 0.3446,
      "step": 24795
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9434411265982885,
      "learning_rate": 8.800636576340521e-06,
      "loss": 0.3335,
      "step": 24800
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.8994936482141401,
      "learning_rate": 8.791901748029596e-06,
      "loss": 0.2641,
      "step": 24805
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.045448196175751,
      "learning_rate": 8.783170331693132e-06,
      "loss": 0.337,
      "step": 24810
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9759772445120798,
      "learning_rate": 8.774442329169194e-06,
      "loss": 0.3026,
      "step": 24815
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9642300159610239,
      "learning_rate": 8.765717742295104e-06,
      "loss": 0.2677,
      "step": 24820
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9209249174773894,
      "learning_rate": 8.756996572907475e-06,
      "loss": 0.2849,
      "step": 24825
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0521239083363665,
      "learning_rate": 8.748278822842212e-06,
      "loss": 0.3004,
      "step": 24830
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.1630370278283324,
      "learning_rate": 8.73956449393449e-06,
      "loss": 0.3435,
      "step": 24835
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.8847554266545826,
      "learning_rate": 8.730853588018772e-06,
      "loss": 0.3388,
      "step": 24840
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.9065749128496903,
      "learning_rate": 8.722146106928797e-06,
      "loss": 0.2851,
      "step": 24845
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.056609312345644,
      "learning_rate": 8.713442052497569e-06,
      "loss": 0.2904,
      "step": 24850
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.0676033295107987,
      "learning_rate": 8.704741426557397e-06,
      "loss": 0.3076,
      "step": 24855
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.8007208284557547,
      "learning_rate": 8.69604423093984e-06,
      "loss": 0.2564,
      "step": 24860
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.8789258588724301,
      "learning_rate": 8.687350467475753e-06,
      "loss": 0.338,
      "step": 24865
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9267851243235534,
      "learning_rate": 8.678660137995277e-06,
      "loss": 0.3078,
      "step": 24870
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6057379325578072,
      "learning_rate": 8.669973244327795e-06,
      "loss": 0.3259,
      "step": 24875
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9370285568380858,
      "learning_rate": 8.661289788302001e-06,
      "loss": 0.295,
      "step": 24880
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.8671113771512887,
      "learning_rate": 8.652609771745862e-06,
      "loss": 0.3139,
      "step": 24885
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9005948265731976,
      "learning_rate": 8.643933196486606e-06,
      "loss": 0.2872,
      "step": 24890
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9835958307674249,
      "learning_rate": 8.63526006435073e-06,
      "loss": 0.3104,
      "step": 24895
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9932819649744413,
      "learning_rate": 8.62659037716403e-06,
      "loss": 0.2988,
      "step": 24900
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.8756829091910165,
      "learning_rate": 8.617924136751565e-06,
      "loss": 0.271,
      "step": 24905
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9841391651223605,
      "learning_rate": 8.609261344937672e-06,
      "loss": 0.345,
      "step": 24910
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9218290008804002,
      "learning_rate": 8.600602003545965e-06,
      "loss": 0.3074,
      "step": 24915
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.0064452276414906,
      "learning_rate": 8.591946114399318e-06,
      "loss": 0.2912,
      "step": 24920
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9811380341919945,
      "learning_rate": 8.583293679319879e-06,
      "loss": 0.3443,
      "step": 24925
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.8557771398955576,
      "learning_rate": 8.574644700129087e-06,
      "loss": 0.2676,
      "step": 24930
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.091720722784849,
      "learning_rate": 8.565999178647635e-06,
      "loss": 0.2928,
      "step": 24935
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.0502109626464502,
      "learning_rate": 8.557357116695513e-06,
      "loss": 0.2733,
      "step": 24940
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9918073036821434,
      "learning_rate": 8.548718516091941e-06,
      "loss": 0.3308,
      "step": 24945
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.0696325538985048,
      "learning_rate": 8.540083378655459e-06,
      "loss": 0.3054,
      "step": 24950
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9801403717362307,
      "learning_rate": 8.531451706203834e-06,
      "loss": 0.285,
      "step": 24955
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9264429454334463,
      "learning_rate": 8.522823500554136e-06,
      "loss": 0.2651,
      "step": 24960
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9832433898782803,
      "learning_rate": 8.514198763522685e-06,
      "loss": 0.3028,
      "step": 24965
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9220175379039554,
      "learning_rate": 8.505577496925076e-06,
      "loss": 0.3085,
      "step": 24970
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9703688560850237,
      "learning_rate": 8.496959702576187e-06,
      "loss": 0.2635,
      "step": 24975
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9698795623140767,
      "learning_rate": 8.488345382290153e-06,
      "loss": 0.2775,
      "step": 24980
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.002558600075366,
      "learning_rate": 8.479734537880376e-06,
      "loss": 0.2977,
      "step": 24985
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9086077302224278,
      "learning_rate": 8.47112717115952e-06,
      "loss": 0.2845,
      "step": 24990
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9543291941718527,
      "learning_rate": 8.46252328393953e-06,
      "loss": 0.2794,
      "step": 24995
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9165797481909358,
      "learning_rate": 8.45392287803162e-06,
      "loss": 0.2867,
      "step": 25000
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9548700714444911,
      "learning_rate": 8.445325955246263e-06,
      "loss": 0.3252,
      "step": 25005
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9443692603011027,
      "learning_rate": 8.436732517393209e-06,
      "loss": 0.3121,
      "step": 25010
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.029144686390705,
      "learning_rate": 8.428142566281457e-06,
      "loss": 0.3065,
      "step": 25015
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.111544576800941,
      "learning_rate": 8.419556103719279e-06,
      "loss": 0.2902,
      "step": 25020
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.9725286974420745,
      "learning_rate": 8.410973131514219e-06,
      "loss": 0.3221,
      "step": 25025
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.8972052985056075,
      "learning_rate": 8.40239365147309e-06,
      "loss": 0.3449,
      "step": 25030
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.1226855913493445,
      "learning_rate": 8.393817665401948e-06,
      "loss": 0.3322,
      "step": 25035
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.070646222368591,
      "learning_rate": 8.385245175106137e-06,
      "loss": 0.3184,
      "step": 25040
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.077157184125533,
      "learning_rate": 8.376676182390264e-06,
      "loss": 0.3291,
      "step": 25045
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9450512489402122,
      "learning_rate": 8.368110689058173e-06,
      "loss": 0.2896,
      "step": 25050
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.8858532441525451,
      "learning_rate": 8.359548696913005e-06,
      "loss": 0.3069,
      "step": 25055
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.8031783898826441,
      "learning_rate": 8.350990207757139e-06,
      "loss": 0.2939,
      "step": 25060
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.8884756988335079,
      "learning_rate": 8.342435223392232e-06,
      "loss": 0.2905,
      "step": 25065
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9668259502950691,
      "learning_rate": 8.333883745619195e-06,
      "loss": 0.3151,
      "step": 25070
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9268049264399959,
      "learning_rate": 8.325335776238205e-06,
      "loss": 0.3137,
      "step": 25075
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0110813515722838,
      "learning_rate": 8.316791317048711e-06,
      "loss": 0.3085,
      "step": 25080
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.1284587518167715,
      "learning_rate": 8.308250369849399e-06,
      "loss": 0.3173,
      "step": 25085
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.8554973317775217,
      "learning_rate": 8.299712936438219e-06,
      "loss": 0.2982,
      "step": 25090
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.8601800805064899,
      "learning_rate": 8.291179018612403e-06,
      "loss": 0.2881,
      "step": 25095
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.0123035345647424,
      "learning_rate": 8.282648618168423e-06,
      "loss": 0.328,
      "step": 25100
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.8490959403173556,
      "learning_rate": 8.274121736902032e-06,
      "loss": 0.3249,
      "step": 25105
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9562105723442911,
      "learning_rate": 8.265598376608211e-06,
      "loss": 0.2747,
      "step": 25110
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5227998599867598,
      "learning_rate": 8.257078539081228e-06,
      "loss": 0.3261,
      "step": 25115
    },
    {
      "epoch": 2.93,
      "grad_norm": 0.9725892292393016,
      "learning_rate": 8.24856222611459e-06,
      "loss": 0.2482,
      "step": 25120
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.7829322934508481,
      "learning_rate": 8.240049439501076e-06,
      "loss": 0.2532,
      "step": 25125
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.0435525207066698,
      "learning_rate": 8.23154018103271e-06,
      "loss": 0.3127,
      "step": 25130
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9068990636688932,
      "learning_rate": 8.223034452500783e-06,
      "loss": 0.3466,
      "step": 25135
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9018229599526959,
      "learning_rate": 8.21453225569584e-06,
      "loss": 0.2839,
      "step": 25140
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.2041744910979784,
      "learning_rate": 8.206033592407692e-06,
      "loss": 0.2774,
      "step": 25145
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.7509338575046008,
      "learning_rate": 8.197538464425388e-06,
      "loss": 0.329,
      "step": 25150
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9817463508991848,
      "learning_rate": 8.189046873537237e-06,
      "loss": 0.317,
      "step": 25155
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.863682598043019,
      "learning_rate": 8.180558821530809e-06,
      "loss": 0.2984,
      "step": 25160
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9799862878965853,
      "learning_rate": 8.172074310192932e-06,
      "loss": 0.2505,
      "step": 25165
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.934625093552519,
      "learning_rate": 8.163593341309683e-06,
      "loss": 0.3764,
      "step": 25170
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.2261049081646873,
      "learning_rate": 8.155115916666406e-06,
      "loss": 0.3337,
      "step": 25175
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.1071113235931016,
      "learning_rate": 8.146642038047676e-06,
      "loss": 0.3307,
      "step": 25180
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9405602993412395,
      "learning_rate": 8.138171707237327e-06,
      "loss": 0.3154,
      "step": 25185
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.895897066437274,
      "learning_rate": 8.129704926018458e-06,
      "loss": 0.2916,
      "step": 25190
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.9193982511243515,
      "learning_rate": 8.121241696173427e-06,
      "loss": 0.2758,
      "step": 25195
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.8691080857432063,
      "learning_rate": 8.112782019483813e-06,
      "loss": 0.3183,
      "step": 25200
    },
    {
      "epoch": 2.94,
      "grad_norm": 0.8533154443780022,
      "learning_rate": 8.104325897730478e-06,
      "loss": 0.3113,
      "step": 25205
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9698906157091091,
      "learning_rate": 8.09587333269353e-06,
      "loss": 0.3093,
      "step": 25210
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9345616452731546,
      "learning_rate": 8.087424326152307e-06,
      "loss": 0.304,
      "step": 25215
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.948408823700671,
      "learning_rate": 8.078978879885427e-06,
      "loss": 0.2897,
      "step": 25220
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0329578825174837,
      "learning_rate": 8.070536995670733e-06,
      "loss": 0.2946,
      "step": 25225
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9031067363888123,
      "learning_rate": 8.062098675285336e-06,
      "loss": 0.2691,
      "step": 25230
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5348889285378948,
      "learning_rate": 8.05366392050559e-06,
      "loss": 0.3223,
      "step": 25235
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0753213492704967,
      "learning_rate": 8.04523273310711e-06,
      "loss": 0.3055,
      "step": 25240
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.941987526197896,
      "learning_rate": 8.036805114864736e-06,
      "loss": 0.2797,
      "step": 25245
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9025154619081572,
      "learning_rate": 8.02838106755257e-06,
      "loss": 0.3306,
      "step": 25250
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9794634482544485,
      "learning_rate": 8.019960592943964e-06,
      "loss": 0.2781,
      "step": 25255
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9164668418013237,
      "learning_rate": 8.011543692811518e-06,
      "loss": 0.2989,
      "step": 25260
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.056848921657759,
      "learning_rate": 8.003130368927084e-06,
      "loss": 0.3397,
      "step": 25265
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.8543138805055576,
      "learning_rate": 7.994720623061742e-06,
      "loss": 0.3336,
      "step": 25270
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.866552380301082,
      "learning_rate": 7.986314456985844e-06,
      "loss": 0.3498,
      "step": 25275
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.8957177408395095,
      "learning_rate": 7.97791187246896e-06,
      "loss": 0.3068,
      "step": 25280
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9403020253776941,
      "learning_rate": 7.969512871279936e-06,
      "loss": 0.2793,
      "step": 25285
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.9637639700025175,
      "learning_rate": 7.96111745518685e-06,
      "loss": 0.3043,
      "step": 25290
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9485857804683302,
      "learning_rate": 7.952725625957017e-06,
      "loss": 0.278,
      "step": 25295
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.910895684876032,
      "learning_rate": 7.944337385357003e-06,
      "loss": 0.2929,
      "step": 25300
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.258688954227687,
      "learning_rate": 7.93595273515263e-06,
      "loss": 0.2719,
      "step": 25305
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.0280424248620488,
      "learning_rate": 7.927571677108959e-06,
      "loss": 0.3092,
      "step": 25310
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.8918193652901086,
      "learning_rate": 7.91919421299028e-06,
      "loss": 0.3171,
      "step": 25315
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9830915256539035,
      "learning_rate": 7.910820344560133e-06,
      "loss": 0.3601,
      "step": 25320
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9556672171145505,
      "learning_rate": 7.902450073581311e-06,
      "loss": 0.3134,
      "step": 25325
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.954231053543131,
      "learning_rate": 7.894083401815842e-06,
      "loss": 0.3439,
      "step": 25330
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.0639659962908548,
      "learning_rate": 7.88572033102501e-06,
      "loss": 0.3284,
      "step": 25335
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9731087610442579,
      "learning_rate": 7.87736086296931e-06,
      "loss": 0.2972,
      "step": 25340
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.0207396307907552,
      "learning_rate": 7.869004999408514e-06,
      "loss": 0.3531,
      "step": 25345
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.8670107072061473,
      "learning_rate": 7.860652742101604e-06,
      "loss": 0.2896,
      "step": 25350
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9303874843214798,
      "learning_rate": 7.852304092806825e-06,
      "loss": 0.2729,
      "step": 25355
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9174337877363012,
      "learning_rate": 7.843959053281663e-06,
      "loss": 0.302,
      "step": 25360
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.8070936212042978,
      "learning_rate": 7.835617625282823e-06,
      "loss": 0.3262,
      "step": 25365
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9095039615039022,
      "learning_rate": 7.827279810566266e-06,
      "loss": 0.3218,
      "step": 25370
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.9894243417439872,
      "learning_rate": 7.818945610887199e-06,
      "loss": 0.3248,
      "step": 25375
    },
    {
      "epoch": 2.97,
      "grad_norm": 51.67964912522946,
      "learning_rate": 7.810615028000045e-06,
      "loss": 0.3222,
      "step": 25380
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.0860220630808157,
      "learning_rate": 7.802288063658492e-06,
      "loss": 0.321,
      "step": 25385
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9146857739040198,
      "learning_rate": 7.793964719615438e-06,
      "loss": 0.3003,
      "step": 25390
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.8986280899323342,
      "learning_rate": 7.785644997623043e-06,
      "loss": 0.302,
      "step": 25395
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.8016509258170548,
      "learning_rate": 7.777328899432695e-06,
      "loss": 0.2454,
      "step": 25400
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.933538865578769,
      "learning_rate": 7.769016426795028e-06,
      "loss": 0.2979,
      "step": 25405
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9563963607873767,
      "learning_rate": 7.760707581459894e-06,
      "loss": 0.2882,
      "step": 25410
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.320537311757936,
      "learning_rate": 7.75240236517639e-06,
      "loss": 0.2718,
      "step": 25415
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.8874759288697339,
      "learning_rate": 7.744100779692853e-06,
      "loss": 0.3079,
      "step": 25420
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9549955471653746,
      "learning_rate": 7.735802826756856e-06,
      "loss": 0.3094,
      "step": 25425
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9110601784095469,
      "learning_rate": 7.727508508115214e-06,
      "loss": 0.319,
      "step": 25430
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9028471801879644,
      "learning_rate": 7.71921782551395e-06,
      "loss": 0.2791,
      "step": 25435
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9734956441893885,
      "learning_rate": 7.710930780698356e-06,
      "loss": 0.3037,
      "step": 25440
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9617939557045352,
      "learning_rate": 7.702647375412927e-06,
      "loss": 0.3197,
      "step": 25445
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.9257312035679582,
      "learning_rate": 7.694367611401416e-06,
      "loss": 0.2735,
      "step": 25450
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.914998502597473,
      "learning_rate": 7.686091490406802e-06,
      "loss": 0.3045,
      "step": 25455
    },
    {
      "epoch": 2.97,
      "grad_norm": 0.989967742539965,
      "learning_rate": 7.677819014171286e-06,
      "loss": 0.2977,
      "step": 25460
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9827260944602146,
      "learning_rate": 7.669550184436313e-06,
      "loss": 0.272,
      "step": 25465
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9007182908011555,
      "learning_rate": 7.661285002942572e-06,
      "loss": 0.305,
      "step": 25470
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.8976312410532772,
      "learning_rate": 7.65302347142996e-06,
      "loss": 0.2709,
      "step": 25475
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.1873652781078856,
      "learning_rate": 7.644765591637607e-06,
      "loss": 0.3129,
      "step": 25480
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.8611721232147667,
      "learning_rate": 7.63651136530389e-06,
      "loss": 0.253,
      "step": 25485
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9757808787533969,
      "learning_rate": 7.628260794166412e-06,
      "loss": 0.3511,
      "step": 25490
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8856369973388363,
      "learning_rate": 7.620013879962004e-06,
      "loss": 0.3206,
      "step": 25495
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0326700916031482,
      "learning_rate": 7.611770624426737e-06,
      "loss": 0.28,
      "step": 25500
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.8523192368453756,
      "learning_rate": 7.603531029295893e-06,
      "loss": 0.2911,
      "step": 25505
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9816393367238673,
      "learning_rate": 7.5952950963039865e-06,
      "loss": 0.2839,
      "step": 25510
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9831184260138018,
      "learning_rate": 7.5870628271847765e-06,
      "loss": 0.3339,
      "step": 25515
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.8898209461236166,
      "learning_rate": 7.578834223671238e-06,
      "loss": 0.2744,
      "step": 25520
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.970413732195831,
      "learning_rate": 7.570609287495592e-06,
      "loss": 0.2895,
      "step": 25525
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9230828492351917,
      "learning_rate": 7.562388020389252e-06,
      "loss": 0.2742,
      "step": 25530
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.0587845345117999,
      "learning_rate": 7.554170424082893e-06,
      "loss": 0.3381,
      "step": 25535
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9560823806000868,
      "learning_rate": 7.545956500306409e-06,
      "loss": 0.2926,
      "step": 25540
    },
    {
      "epoch": 2.98,
      "grad_norm": 0.9977047617581358,
      "learning_rate": 7.537746250788913e-06,
      "loss": 0.2996,
      "step": 25545
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.8579617788187831,
      "learning_rate": 7.5295396772587376e-06,
      "loss": 0.3436,
      "step": 25550
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9515549313546731,
      "learning_rate": 7.521336781443464e-06,
      "loss": 0.277,
      "step": 25555
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9289571590436669,
      "learning_rate": 7.5131375650698835e-06,
      "loss": 0.3019,
      "step": 25560
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0714930756891181,
      "learning_rate": 7.5049420298640175e-06,
      "loss": 0.2972,
      "step": 25565
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0679839569446519,
      "learning_rate": 7.496750177551126e-06,
      "loss": 0.2693,
      "step": 25570
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.7943835500224294,
      "learning_rate": 7.488562009855663e-06,
      "loss": 0.328,
      "step": 25575
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.8810813658279644,
      "learning_rate": 7.480377528501323e-06,
      "loss": 0.287,
      "step": 25580
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9873446119957718,
      "learning_rate": 7.472196735211029e-06,
      "loss": 0.3034,
      "step": 25585
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.1409084100736497,
      "learning_rate": 7.4640196317069225e-06,
      "loss": 0.32,
      "step": 25590
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9064147784656964,
      "learning_rate": 7.455846219710383e-06,
      "loss": 0.3348,
      "step": 25595
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.8767706195156121,
      "learning_rate": 7.4476765009419775e-06,
      "loss": 0.3092,
      "step": 25600
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.007118586183292,
      "learning_rate": 7.439510477121536e-06,
      "loss": 0.3057,
      "step": 25605
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.0466972292582857,
      "learning_rate": 7.431348149968076e-06,
      "loss": 0.3463,
      "step": 25610
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9055093635662006,
      "learning_rate": 7.423189521199867e-06,
      "loss": 0.3069,
      "step": 25615
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.8815073452368876,
      "learning_rate": 7.415034592534373e-06,
      "loss": 0.2835,
      "step": 25620
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.9796541721478743,
      "learning_rate": 7.4068833656883e-06,
      "loss": 0.334,
      "step": 25625
    },
    {
      "epoch": 2.99,
      "grad_norm": 0.939487768224349,
      "learning_rate": 7.3987358423775634e-06,
      "loss": 0.2473,
      "step": 25630
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.0164116096828941,
      "learning_rate": 7.390592024317311e-06,
      "loss": 0.2894,
      "step": 25635
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9088123317314842,
      "learning_rate": 7.382451913221894e-06,
      "loss": 0.2943,
      "step": 25640
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9855048029428187,
      "learning_rate": 7.374315510804886e-06,
      "loss": 0.2654,
      "step": 25645
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8004988873093533,
      "learning_rate": 7.366182818779088e-06,
      "loss": 0.2896,
      "step": 25650
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9663828356806119,
      "learning_rate": 7.358053838856521e-06,
      "loss": 0.2732,
      "step": 25655
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9749653199155557,
      "learning_rate": 7.3499285727484184e-06,
      "loss": 0.3009,
      "step": 25660
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.9486304246992414,
      "learning_rate": 7.341807022165245e-06,
      "loss": 0.2978,
      "step": 25665
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8988789510433882,
      "learning_rate": 7.333689188816658e-06,
      "loss": 0.2961,
      "step": 25670
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.8739588118307898,
      "learning_rate": 7.325575074411545e-06,
      "loss": 0.3114,
      "step": 25675
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6666462630897451,
      "learning_rate": 7.317464680658015e-06,
      "loss": 0.1854,
      "step": 25680
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.5792332061408854,
      "learning_rate": 7.3093580092634004e-06,
      "loss": 0.128,
      "step": 25685
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.5800027182233656,
      "learning_rate": 7.301255061934226e-06,
      "loss": 0.1165,
      "step": 25690
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.590470133985006,
      "learning_rate": 7.2931558403762535e-06,
      "loss": 0.1099,
      "step": 25695
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6843244980344247,
      "learning_rate": 7.2850603462944616e-06,
      "loss": 0.1189,
      "step": 25700
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6698734669524816,
      "learning_rate": 7.27696858139302e-06,
      "loss": 0.1001,
      "step": 25705
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.82789249783902,
      "learning_rate": 7.268880547375351e-06,
      "loss": 0.1393,
      "step": 25710
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.7624113790559901,
      "learning_rate": 7.2607962459440485e-06,
      "loss": 0.1061,
      "step": 25715
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7789374637318257,
      "learning_rate": 7.2527156788009556e-06,
      "loss": 0.143,
      "step": 25720
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.6067680235070263,
      "learning_rate": 7.244638847647112e-06,
      "loss": 0.0931,
      "step": 25725
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7527601937177296,
      "learning_rate": 7.236565754182783e-06,
      "loss": 0.1137,
      "step": 25730
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7040234194527107,
      "learning_rate": 7.228496400107434e-06,
      "loss": 0.1121,
      "step": 25735
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.6230299647399241,
      "learning_rate": 7.220430787119742e-06,
      "loss": 0.1043,
      "step": 25740
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8032028417597421,
      "learning_rate": 7.212368916917609e-06,
      "loss": 0.1133,
      "step": 25745
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7110139189186594,
      "learning_rate": 7.204310791198141e-06,
      "loss": 0.1294,
      "step": 25750
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.6508090702228945,
      "learning_rate": 7.196256411657667e-06,
      "loss": 0.1155,
      "step": 25755
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7348275068098351,
      "learning_rate": 7.188205779991705e-06,
      "loss": 0.1276,
      "step": 25760
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.8073421940574722,
      "learning_rate": 7.1801588978950036e-06,
      "loss": 0.1155,
      "step": 25765
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7497618927437621,
      "learning_rate": 7.172115767061524e-06,
      "loss": 0.0996,
      "step": 25770
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7059470049647905,
      "learning_rate": 7.164076389184413e-06,
      "loss": 0.1235,
      "step": 25775
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.9038235295073999,
      "learning_rate": 7.156040765956057e-06,
      "loss": 0.0955,
      "step": 25780
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.6955374781103616,
      "learning_rate": 7.148008899068029e-06,
      "loss": 0.0978,
      "step": 25785
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7957312573531673,
      "learning_rate": 7.139980790211123e-06,
      "loss": 0.1081,
      "step": 25790
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7466190161407813,
      "learning_rate": 7.1319564410753465e-06,
      "loss": 0.2046,
      "step": 25795
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.6904227609716052,
      "learning_rate": 7.123935853349911e-06,
      "loss": 0.0943,
      "step": 25800
    },
    {
      "epoch": 3.01,
      "grad_norm": 1.0197191763284226,
      "learning_rate": 7.115919028723228e-06,
      "loss": 0.116,
      "step": 25805
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.8762461118157909,
      "learning_rate": 7.1079059688829164e-06,
      "loss": 0.1139,
      "step": 25810
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.8896254201440681,
      "learning_rate": 7.099896675515819e-06,
      "loss": 0.1199,
      "step": 25815
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.6035578091906693,
      "learning_rate": 7.091891150307972e-06,
      "loss": 0.0891,
      "step": 25820
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.61801022746837,
      "learning_rate": 7.083889394944629e-06,
      "loss": 0.1069,
      "step": 25825
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.6852288254533534,
      "learning_rate": 7.0758914111102335e-06,
      "loss": 0.1203,
      "step": 25830
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.698277667934384,
      "learning_rate": 7.067897200488457e-06,
      "loss": 0.1002,
      "step": 25835
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.7585407421949465,
      "learning_rate": 7.059906764762148e-06,
      "loss": 0.0979,
      "step": 25840
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.7002183196178202,
      "learning_rate": 7.051920105613388e-06,
      "loss": 0.1006,
      "step": 25845
    },
    {
      "epoch": 3.02,
      "grad_norm": 1.0840698577683,
      "learning_rate": 7.043937224723457e-06,
      "loss": 0.1257,
      "step": 25850
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.8124452490345206,
      "learning_rate": 7.035958123772821e-06,
      "loss": 0.1269,
      "step": 25855
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.7366707511174803,
      "learning_rate": 7.027982804441171e-06,
      "loss": 0.1141,
      "step": 25860
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.6306136980050606,
      "learning_rate": 7.020011268407406e-06,
      "loss": 0.107,
      "step": 25865
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.6173998104994289,
      "learning_rate": 7.012043517349598e-06,
      "loss": 0.0934,
      "step": 25870
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.7062152809136335,
      "learning_rate": 7.004079552945062e-06,
      "loss": 0.1198,
      "step": 25875
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.659383273148436,
      "learning_rate": 6.9961193768702745e-06,
      "loss": 0.1191,
      "step": 25880
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.682355103218542,
      "learning_rate": 6.98816299080095e-06,
      "loss": 0.0819,
      "step": 25885
    },
    {
      "epoch": 3.02,
      "grad_norm": 0.6453553331248967,
      "learning_rate": 6.980210396411987e-06,
      "loss": 0.1212,
      "step": 25890
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.7318894613121328,
      "learning_rate": 6.9722615953774996e-06,
      "loss": 0.125,
      "step": 25895
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.7625115410481241,
      "learning_rate": 6.9643165893707865e-06,
      "loss": 0.1159,
      "step": 25900
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.6356848020563247,
      "learning_rate": 6.956375380064345e-06,
      "loss": 0.0985,
      "step": 25905
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.7048958348134202,
      "learning_rate": 6.948437969129892e-06,
      "loss": 0.1266,
      "step": 25910
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.5732019712783071,
      "learning_rate": 6.940504358238337e-06,
      "loss": 0.1083,
      "step": 25915
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.6783886104219875,
      "learning_rate": 6.932574549059789e-06,
      "loss": 0.1216,
      "step": 25920
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.659643945342329,
      "learning_rate": 6.924648543263551e-06,
      "loss": 0.1338,
      "step": 25925
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.7032557462918197,
      "learning_rate": 6.916726342518137e-06,
      "loss": 0.1154,
      "step": 25930
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.5495182729776876,
      "learning_rate": 6.908807948491241e-06,
      "loss": 0.1047,
      "step": 25935
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.5985940204560098,
      "learning_rate": 6.900893362849778e-06,
      "loss": 0.099,
      "step": 25940
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.720134198341354,
      "learning_rate": 6.892982587259852e-06,
      "loss": 0.1105,
      "step": 25945
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.7530826576061642,
      "learning_rate": 6.885075623386758e-06,
      "loss": 0.1093,
      "step": 25950
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.7880013058248595,
      "learning_rate": 6.877172472894996e-06,
      "loss": 0.1138,
      "step": 25955
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.6986562461350952,
      "learning_rate": 6.869273137448271e-06,
      "loss": 0.1155,
      "step": 25960
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.8489108713953327,
      "learning_rate": 6.861377618709466e-06,
      "loss": 0.125,
      "step": 25965
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.6991534866152611,
      "learning_rate": 6.853485918340669e-06,
      "loss": 0.1211,
      "step": 25970
    },
    {
      "epoch": 3.03,
      "grad_norm": 0.6760740065478332,
      "learning_rate": 6.845598038003168e-06,
      "loss": 0.1316,
      "step": 25975
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.5424280130641633,
      "learning_rate": 6.837713979357446e-06,
      "loss": 0.0901,
      "step": 25980
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6310513669801356,
      "learning_rate": 6.82983374406318e-06,
      "loss": 0.1274,
      "step": 25985
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6472820295166074,
      "learning_rate": 6.821957333779249e-06,
      "loss": 0.1243,
      "step": 25990
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.7395819056318942,
      "learning_rate": 6.814084750163705e-06,
      "loss": 0.1116,
      "step": 25995
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.8112187535618616,
      "learning_rate": 6.806215994873827e-06,
      "loss": 0.106,
      "step": 26000
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.7712546623283802,
      "learning_rate": 6.79835106956605e-06,
      "loss": 0.1054,
      "step": 26005
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6585336437449546,
      "learning_rate": 6.790489975896033e-06,
      "loss": 0.103,
      "step": 26010
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.5712067309109954,
      "learning_rate": 6.78263271551863e-06,
      "loss": 0.1163,
      "step": 26015
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6953360387018156,
      "learning_rate": 6.774779290087857e-06,
      "loss": 0.1126,
      "step": 26020
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6022364608823308,
      "learning_rate": 6.76692970125695e-06,
      "loss": 0.0821,
      "step": 26025
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6276423851743334,
      "learning_rate": 6.7590839506783385e-06,
      "loss": 0.0966,
      "step": 26030
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.7558626237999606,
      "learning_rate": 6.751242040003628e-06,
      "loss": 0.1168,
      "step": 26035
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6427284842209443,
      "learning_rate": 6.743403970883616e-06,
      "loss": 0.0957,
      "step": 26040
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6563102159911584,
      "learning_rate": 6.735569744968303e-06,
      "loss": 0.1188,
      "step": 26045
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.8075500551366093,
      "learning_rate": 6.727739363906876e-06,
      "loss": 0.1106,
      "step": 26050
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.7586503712845535,
      "learning_rate": 6.719912829347716e-06,
      "loss": 0.1291,
      "step": 26055
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.6329569960480992,
      "learning_rate": 6.712090142938396e-06,
      "loss": 0.1174,
      "step": 26060
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.9599491186613833,
      "learning_rate": 6.704271306325663e-06,
      "loss": 0.1111,
      "step": 26065
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.5789830062540582,
      "learning_rate": 6.696456321155459e-06,
      "loss": 0.1105,
      "step": 26070
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.6688107528963738,
      "learning_rate": 6.688645189072929e-06,
      "loss": 0.1107,
      "step": 26075
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.638058702836562,
      "learning_rate": 6.6808379117223925e-06,
      "loss": 0.11,
      "step": 26080
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.6873110907466587,
      "learning_rate": 6.673034490747379e-06,
      "loss": 0.0971,
      "step": 26085
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.8192393902004805,
      "learning_rate": 6.66523492779057e-06,
      "loss": 0.1141,
      "step": 26090
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.7217670932676725,
      "learning_rate": 6.657439224493872e-06,
      "loss": 0.0989,
      "step": 26095
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.628813249978428,
      "learning_rate": 6.649647382498345e-06,
      "loss": 0.0972,
      "step": 26100
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.7656671164556205,
      "learning_rate": 6.64185940344427e-06,
      "loss": 0.1156,
      "step": 26105
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.6608351175345727,
      "learning_rate": 6.634075288971084e-06,
      "loss": 0.0999,
      "step": 26110
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.6887990486861082,
      "learning_rate": 6.626295040717431e-06,
      "loss": 0.0992,
      "step": 26115
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.6267976043256926,
      "learning_rate": 6.618518660321135e-06,
      "loss": 0.1098,
      "step": 26120
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.5821058528785312,
      "learning_rate": 6.610746149419214e-06,
      "loss": 0.0863,
      "step": 26125
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.6636853664323545,
      "learning_rate": 6.602977509647854e-06,
      "loss": 0.0909,
      "step": 26130
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.7223944871968184,
      "learning_rate": 6.5952127426424286e-06,
      "loss": 0.1053,
      "step": 26135
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.6640745565999715,
      "learning_rate": 6.587451850037507e-06,
      "loss": 0.0902,
      "step": 26140
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.6031629585219299,
      "learning_rate": 6.579694833466843e-06,
      "loss": 0.1099,
      "step": 26145
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.6204414585622152,
      "learning_rate": 6.571941694563369e-06,
      "loss": 0.1052,
      "step": 26150
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7827397136665145,
      "learning_rate": 6.564192434959205e-06,
      "loss": 0.1257,
      "step": 26155
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7955731670164823,
      "learning_rate": 6.55644705628565e-06,
      "loss": 0.0989,
      "step": 26160
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7194559130686795,
      "learning_rate": 6.548705560173174e-06,
      "loss": 0.1093,
      "step": 26165
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7427976229280938,
      "learning_rate": 6.540967948251456e-06,
      "loss": 0.1082,
      "step": 26170
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.8049930168062688,
      "learning_rate": 6.533234222149348e-06,
      "loss": 0.1005,
      "step": 26175
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7869616922831549,
      "learning_rate": 6.525504383494865e-06,
      "loss": 0.0946,
      "step": 26180
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7245779254144922,
      "learning_rate": 6.517778433915231e-06,
      "loss": 0.0875,
      "step": 26185
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7195324852015156,
      "learning_rate": 6.510056375036841e-06,
      "loss": 0.0974,
      "step": 26190
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.6596443541433088,
      "learning_rate": 6.502338208485259e-06,
      "loss": 0.0969,
      "step": 26195
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.6714694117818513,
      "learning_rate": 6.494623935885255e-06,
      "loss": 0.112,
      "step": 26200
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.6951530175989664,
      "learning_rate": 6.486913558860744e-06,
      "loss": 0.1081,
      "step": 26205
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.918517832724242,
      "learning_rate": 6.479207079034855e-06,
      "loss": 0.1826,
      "step": 26210
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7481462899702817,
      "learning_rate": 6.471504498029876e-06,
      "loss": 0.1378,
      "step": 26215
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7781801200232565,
      "learning_rate": 6.463805817467286e-06,
      "loss": 0.1028,
      "step": 26220
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.6305727540011729,
      "learning_rate": 6.4561110389677465e-06,
      "loss": 0.1043,
      "step": 26225
    },
    {
      "epoch": 3.06,
      "grad_norm": 0.7414077335804938,
      "learning_rate": 6.448420164151076e-06,
      "loss": 0.1062,
      "step": 26230
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.779076066844843,
      "learning_rate": 6.440733194636281e-06,
      "loss": 0.1178,
      "step": 26235
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.5959009473978787,
      "learning_rate": 6.433050132041554e-06,
      "loss": 0.1063,
      "step": 26240
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.7411986353085473,
      "learning_rate": 6.425370977984271e-06,
      "loss": 0.0967,
      "step": 26245
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.78144453500627,
      "learning_rate": 6.4176957340809556e-06,
      "loss": 0.114,
      "step": 26250
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.6617298682824334,
      "learning_rate": 6.410024401947337e-06,
      "loss": 0.0826,
      "step": 26255
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.7841786828968641,
      "learning_rate": 6.402356983198319e-06,
      "loss": 0.1096,
      "step": 26260
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.6780428882784549,
      "learning_rate": 6.394693479447958e-06,
      "loss": 0.118,
      "step": 26265
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.6340536895837628,
      "learning_rate": 6.387033892309513e-06,
      "loss": 0.0876,
      "step": 26270
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.584421856402789,
      "learning_rate": 6.379378223395396e-06,
      "loss": 0.0953,
      "step": 26275
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.6735257359327141,
      "learning_rate": 6.3717264743172134e-06,
      "loss": 0.0973,
      "step": 26280
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.6344762737763165,
      "learning_rate": 6.36407864668574e-06,
      "loss": 0.093,
      "step": 26285
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.6012282402288402,
      "learning_rate": 6.356434742110926e-06,
      "loss": 0.1055,
      "step": 26290
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.6984866310526668,
      "learning_rate": 6.348794762201887e-06,
      "loss": 0.1159,
      "step": 26295
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.5663254297759245,
      "learning_rate": 6.341158708566916e-06,
      "loss": 0.0892,
      "step": 26300
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.7066668094270536,
      "learning_rate": 6.333526582813487e-06,
      "loss": 0.1051,
      "step": 26305
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.7820246602332909,
      "learning_rate": 6.32589838654824e-06,
      "loss": 0.0796,
      "step": 26310
    },
    {
      "epoch": 3.07,
      "grad_norm": 0.7779764016634094,
      "learning_rate": 6.318274121376999e-06,
      "loss": 0.1333,
      "step": 26315
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.7856533319848613,
      "learning_rate": 6.310653788904741e-06,
      "loss": 0.1228,
      "step": 26320
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.6777718896520394,
      "learning_rate": 6.303037390735634e-06,
      "loss": 0.1007,
      "step": 26325
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.6724219744148024,
      "learning_rate": 6.295424928472998e-06,
      "loss": 0.1039,
      "step": 26330
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.5944257338427684,
      "learning_rate": 6.287816403719346e-06,
      "loss": 0.1141,
      "step": 26335
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.7244686837654882,
      "learning_rate": 6.2802118180763565e-06,
      "loss": 0.111,
      "step": 26340
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.7376764639433125,
      "learning_rate": 6.27261117314486e-06,
      "loss": 0.1163,
      "step": 26345
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.6425688286861146,
      "learning_rate": 6.265014470524877e-06,
      "loss": 0.1198,
      "step": 26350
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.7828557117103871,
      "learning_rate": 6.257421711815606e-06,
      "loss": 0.1126,
      "step": 26355
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.7648900480180448,
      "learning_rate": 6.249832898615382e-06,
      "loss": 0.1046,
      "step": 26360
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.6229532976416061,
      "learning_rate": 6.242248032521747e-06,
      "loss": 0.1104,
      "step": 26365
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.8228338217744782,
      "learning_rate": 6.234667115131382e-06,
      "loss": 0.1291,
      "step": 26370
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.648720665762594,
      "learning_rate": 6.227090148040151e-06,
      "loss": 0.1043,
      "step": 26375
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.6109161005446335,
      "learning_rate": 6.219517132843092e-06,
      "loss": 0.1329,
      "step": 26380
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.7832889781037061,
      "learning_rate": 6.211948071134405e-06,
      "loss": 0.0952,
      "step": 26385
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.6406279830428531,
      "learning_rate": 6.204382964507455e-06,
      "loss": 0.1041,
      "step": 26390
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.6500546169680852,
      "learning_rate": 6.196821814554765e-06,
      "loss": 0.101,
      "step": 26395
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.6953458208336976,
      "learning_rate": 6.1892646228680436e-06,
      "loss": 0.0916,
      "step": 26400
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7197682220683861,
      "learning_rate": 6.181711391038161e-06,
      "loss": 0.1235,
      "step": 26405
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7472525102366198,
      "learning_rate": 6.174162120655158e-06,
      "loss": 0.1221,
      "step": 26410
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7543055334480568,
      "learning_rate": 6.166616813308221e-06,
      "loss": 0.1254,
      "step": 26415
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.778530498108572,
      "learning_rate": 6.1590754705857286e-06,
      "loss": 0.1029,
      "step": 26420
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7668497198221197,
      "learning_rate": 6.151538094075202e-06,
      "loss": 0.1249,
      "step": 26425
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.68060243855134,
      "learning_rate": 6.1440046853633415e-06,
      "loss": 0.1124,
      "step": 26430
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.8991230533280472,
      "learning_rate": 6.1364752460360165e-06,
      "loss": 0.1351,
      "step": 26435
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7823041953362571,
      "learning_rate": 6.128949777678239e-06,
      "loss": 0.118,
      "step": 26440
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.6946285467655421,
      "learning_rate": 6.121428281874206e-06,
      "loss": 0.1049,
      "step": 26445
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7437546497377329,
      "learning_rate": 6.1139107602072724e-06,
      "loss": 0.1092,
      "step": 26450
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.6219489966003742,
      "learning_rate": 6.1063972142599594e-06,
      "loss": 0.1205,
      "step": 26455
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7207845145834177,
      "learning_rate": 6.098887645613943e-06,
      "loss": 0.1259,
      "step": 26460
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.5872623675405833,
      "learning_rate": 6.0913820558500565e-06,
      "loss": 0.0922,
      "step": 26465
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.6650476410227526,
      "learning_rate": 6.083880446548315e-06,
      "loss": 0.0989,
      "step": 26470
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.6471236933645951,
      "learning_rate": 6.0763828192878815e-06,
      "loss": 0.0925,
      "step": 26475
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.6463297381498803,
      "learning_rate": 6.068889175647099e-06,
      "loss": 0.1048,
      "step": 26480
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.8025955386858256,
      "learning_rate": 6.061399517203437e-06,
      "loss": 0.0917,
      "step": 26485
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.7742384055578915,
      "learning_rate": 6.0539138455335654e-06,
      "loss": 0.1196,
      "step": 26490
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.7724864261958942,
      "learning_rate": 6.046432162213281e-06,
      "loss": 0.1106,
      "step": 26495
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.6099700977648117,
      "learning_rate": 6.038954468817562e-06,
      "loss": 0.1056,
      "step": 26500
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.9475686445717705,
      "learning_rate": 6.03148076692055e-06,
      "loss": 0.1262,
      "step": 26505
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.666432364610513,
      "learning_rate": 6.0240110580955245e-06,
      "loss": 0.1014,
      "step": 26510
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.7301857662848863,
      "learning_rate": 6.016545343914942e-06,
      "loss": 0.0851,
      "step": 26515
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.6728897174201404,
      "learning_rate": 6.0090836259504235e-06,
      "loss": 0.0995,
      "step": 26520
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.6330719176786265,
      "learning_rate": 6.001625905772728e-06,
      "loss": 0.1067,
      "step": 26525
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.6607892889945931,
      "learning_rate": 5.99417218495178e-06,
      "loss": 0.0975,
      "step": 26530
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.6760077819146996,
      "learning_rate": 5.986722465056671e-06,
      "loss": 0.1105,
      "step": 26535
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.5823624055652249,
      "learning_rate": 5.979276747655646e-06,
      "loss": 0.0791,
      "step": 26540
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.7732439781230626,
      "learning_rate": 5.971835034316106e-06,
      "loss": 0.0891,
      "step": 26545
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.8518185811550463,
      "learning_rate": 5.9643973266046145e-06,
      "loss": 0.0991,
      "step": 26550
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.889079466197909,
      "learning_rate": 5.956963626086887e-06,
      "loss": 0.1106,
      "step": 26555
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.7358821376254031,
      "learning_rate": 5.949533934327781e-06,
      "loss": 0.087,
      "step": 26560
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.6852286418982134,
      "learning_rate": 5.942108252891332e-06,
      "loss": 0.1004,
      "step": 26565
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.7288681796790348,
      "learning_rate": 5.934686583340726e-06,
      "loss": 0.1144,
      "step": 26570
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.7534896634630351,
      "learning_rate": 5.927268927238311e-06,
      "loss": 0.13,
      "step": 26575
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.6724731061119408,
      "learning_rate": 5.919855286145565e-06,
      "loss": 0.1146,
      "step": 26580
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.5433158235176958,
      "learning_rate": 5.91244566162315e-06,
      "loss": 0.0996,
      "step": 26585
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.7941017448479711,
      "learning_rate": 5.905040055230857e-06,
      "loss": 0.109,
      "step": 26590
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.7311546127556436,
      "learning_rate": 5.897638468527653e-06,
      "loss": 0.1388,
      "step": 26595
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.7554797565473562,
      "learning_rate": 5.89024090307165e-06,
      "loss": 0.1292,
      "step": 26600
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.6747894223440208,
      "learning_rate": 5.882847360420104e-06,
      "loss": 0.1232,
      "step": 26605
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.6522699389091695,
      "learning_rate": 5.875457842129442e-06,
      "loss": 0.1097,
      "step": 26610
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.608592157775863,
      "learning_rate": 5.8680723497552385e-06,
      "loss": 0.1014,
      "step": 26615
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.7015928263690588,
      "learning_rate": 5.8606908848522075e-06,
      "loss": 0.13,
      "step": 26620
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.9573701161443582,
      "learning_rate": 5.853313448974223e-06,
      "loss": 0.116,
      "step": 26625
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.638785191618915,
      "learning_rate": 5.8459400436743194e-06,
      "loss": 0.1058,
      "step": 26630
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.6470159394474275,
      "learning_rate": 5.83857067050467e-06,
      "loss": 0.1163,
      "step": 26635
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.717140696406229,
      "learning_rate": 5.831205331016612e-06,
      "loss": 0.1044,
      "step": 26640
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.6791795053845087,
      "learning_rate": 5.82384402676063e-06,
      "loss": 0.1302,
      "step": 26645
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.7550541743212152,
      "learning_rate": 5.816486759286346e-06,
      "loss": 0.0997,
      "step": 26650
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.6621251421228375,
      "learning_rate": 5.809133530142541e-06,
      "loss": 0.1239,
      "step": 26655
    },
    {
      "epoch": 3.11,
      "grad_norm": 0.6896644489241749,
      "learning_rate": 5.8017843408771505e-06,
      "loss": 0.0954,
      "step": 26660
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.7413372527055853,
      "learning_rate": 5.794439193037254e-06,
      "loss": 0.0983,
      "step": 26665
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.8539530848760365,
      "learning_rate": 5.787098088169093e-06,
      "loss": 0.0882,
      "step": 26670
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.675353532582595,
      "learning_rate": 5.779761027818032e-06,
      "loss": 0.0891,
      "step": 26675
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.7077296112373209,
      "learning_rate": 5.772428013528603e-06,
      "loss": 0.0894,
      "step": 26680
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6178922942418108,
      "learning_rate": 5.765099046844491e-06,
      "loss": 0.086,
      "step": 26685
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6442620754744803,
      "learning_rate": 5.7577741293085155e-06,
      "loss": 0.0868,
      "step": 26690
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.9776893155592145,
      "learning_rate": 5.750453262462638e-06,
      "loss": 0.1255,
      "step": 26695
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.7794271144464824,
      "learning_rate": 5.743136447847988e-06,
      "loss": 0.1163,
      "step": 26700
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6175694692614657,
      "learning_rate": 5.735823687004829e-06,
      "loss": 0.0804,
      "step": 26705
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6939106951586549,
      "learning_rate": 5.728514981472574e-06,
      "loss": 0.1122,
      "step": 26710
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6671432513878683,
      "learning_rate": 5.721210332789789e-06,
      "loss": 0.1415,
      "step": 26715
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.7205003142859444,
      "learning_rate": 5.713909742494175e-06,
      "loss": 0.0876,
      "step": 26720
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6369795426666236,
      "learning_rate": 5.706613212122572e-06,
      "loss": 0.0915,
      "step": 26725
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.725033362553751,
      "learning_rate": 5.699320743210984e-06,
      "loss": 0.1124,
      "step": 26730
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.7360469854393888,
      "learning_rate": 5.692032337294553e-06,
      "loss": 0.0984,
      "step": 26735
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.8521217209399783,
      "learning_rate": 5.684747995907572e-06,
      "loss": 0.1223,
      "step": 26740
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.6698252790546645,
      "learning_rate": 5.677467720583454e-06,
      "loss": 0.0991,
      "step": 26745
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.7834146283020887,
      "learning_rate": 5.670191512854794e-06,
      "loss": 0.1075,
      "step": 26750
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.7869042545390015,
      "learning_rate": 5.66291937425329e-06,
      "loss": 0.1018,
      "step": 26755
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.6237457533117062,
      "learning_rate": 5.655651306309817e-06,
      "loss": 0.1058,
      "step": 26760
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.7257566097352954,
      "learning_rate": 5.64838731055437e-06,
      "loss": 0.1152,
      "step": 26765
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.686365040716021,
      "learning_rate": 5.641127388516099e-06,
      "loss": 0.0947,
      "step": 26770
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.8668466677381959,
      "learning_rate": 5.633871541723295e-06,
      "loss": 0.1146,
      "step": 26775
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.8206974661503591,
      "learning_rate": 5.626619771703401e-06,
      "loss": 0.1099,
      "step": 26780
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.6511487152495449,
      "learning_rate": 5.619372079982977e-06,
      "loss": 0.1099,
      "step": 26785
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.7182798907040946,
      "learning_rate": 5.612128468087733e-06,
      "loss": 0.1012,
      "step": 26790
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.6051376427094062,
      "learning_rate": 5.6048889375425325e-06,
      "loss": 0.0854,
      "step": 26795
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.7423645166512383,
      "learning_rate": 5.597653489871374e-06,
      "loss": 0.1057,
      "step": 26800
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.7291798969335211,
      "learning_rate": 5.590422126597392e-06,
      "loss": 0.1016,
      "step": 26805
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.7614474823627243,
      "learning_rate": 5.583194849242876e-06,
      "loss": 0.0896,
      "step": 26810
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.8200355087782505,
      "learning_rate": 5.575971659329232e-06,
      "loss": 0.0968,
      "step": 26815
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.7595931128737672,
      "learning_rate": 5.5687525583770135e-06,
      "loss": 0.1163,
      "step": 26820
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.6852212704750229,
      "learning_rate": 5.56153754790592e-06,
      "loss": 0.1037,
      "step": 26825
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.9401996968414855,
      "learning_rate": 5.5543266294347975e-06,
      "loss": 0.099,
      "step": 26830
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.7019136730149339,
      "learning_rate": 5.547119804481602e-06,
      "loss": 0.1147,
      "step": 26835
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.8750223399812699,
      "learning_rate": 5.539917074563458e-06,
      "loss": 0.1165,
      "step": 26840
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.7394121315107945,
      "learning_rate": 5.532718441196619e-06,
      "loss": 0.0969,
      "step": 26845
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.6738828238811119,
      "learning_rate": 5.525523905896457e-06,
      "loss": 0.1151,
      "step": 26850
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.9018949667894144,
      "learning_rate": 5.518333470177514e-06,
      "loss": 0.1119,
      "step": 26855
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.8344878140872373,
      "learning_rate": 5.51114713555344e-06,
      "loss": 0.113,
      "step": 26860
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.6732499314882177,
      "learning_rate": 5.503964903537037e-06,
      "loss": 0.0905,
      "step": 26865
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.7580865065964432,
      "learning_rate": 5.496786775640239e-06,
      "loss": 0.1009,
      "step": 26870
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.5485585342289443,
      "learning_rate": 5.489612753374129e-06,
      "loss": 0.0816,
      "step": 26875
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.6132192215312308,
      "learning_rate": 5.482442838248902e-06,
      "loss": 0.1045,
      "step": 26880
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.6994951119398936,
      "learning_rate": 5.4752770317738986e-06,
      "loss": 0.1101,
      "step": 26885
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.7652090036615278,
      "learning_rate": 5.4681153354575965e-06,
      "loss": 0.1112,
      "step": 26890
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.7267449347499928,
      "learning_rate": 5.460957750807613e-06,
      "loss": 0.1203,
      "step": 26895
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.6964480910368923,
      "learning_rate": 5.453804279330699e-06,
      "loss": 0.0961,
      "step": 26900
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.6601239058338471,
      "learning_rate": 5.446654922532723e-06,
      "loss": 0.1176,
      "step": 26905
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.5339339899569211,
      "learning_rate": 5.439509681918703e-06,
      "loss": 0.0828,
      "step": 26910
    },
    {
      "epoch": 3.14,
      "grad_norm": 0.7847670398120649,
      "learning_rate": 5.432368558992798e-06,
      "loss": 0.1142,
      "step": 26915
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.6715354267693744,
      "learning_rate": 5.425231555258275e-06,
      "loss": 0.096,
      "step": 26920
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.6980572843695022,
      "learning_rate": 5.418098672217559e-06,
      "loss": 0.0914,
      "step": 26925
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.6421570779050814,
      "learning_rate": 5.410969911372185e-06,
      "loss": 0.0976,
      "step": 26930
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.6982240700427049,
      "learning_rate": 5.403845274222838e-06,
      "loss": 0.1152,
      "step": 26935
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.8689530704203923,
      "learning_rate": 5.396724762269331e-06,
      "loss": 0.125,
      "step": 26940
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.7526612680374339,
      "learning_rate": 5.389608377010608e-06,
      "loss": 0.0912,
      "step": 26945
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.7315746698970573,
      "learning_rate": 5.382496119944741e-06,
      "loss": 0.0819,
      "step": 26950
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.7457490956056376,
      "learning_rate": 5.375387992568926e-06,
      "loss": 0.1192,
      "step": 26955
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.6396251754410884,
      "learning_rate": 5.368283996379503e-06,
      "loss": 0.1259,
      "step": 26960
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.8173946270816257,
      "learning_rate": 5.361184132871938e-06,
      "loss": 0.1138,
      "step": 26965
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.7156755356941285,
      "learning_rate": 5.354088403540839e-06,
      "loss": 0.1077,
      "step": 26970
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.586096916571345,
      "learning_rate": 5.346996809879909e-06,
      "loss": 0.0875,
      "step": 26975
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.6724035558098741,
      "learning_rate": 5.339909353382025e-06,
      "loss": 0.0946,
      "step": 26980
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.748002460811208,
      "learning_rate": 5.332826035539151e-06,
      "loss": 0.1245,
      "step": 26985
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.7630746176250409,
      "learning_rate": 5.325746857842409e-06,
      "loss": 0.1296,
      "step": 26990
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.9542456042326904,
      "learning_rate": 5.318671821782045e-06,
      "loss": 0.105,
      "step": 26995
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.6647030524377335,
      "learning_rate": 5.3116009288474184e-06,
      "loss": 0.104,
      "step": 27000
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.6342066903485182,
      "learning_rate": 5.3045341805270295e-06,
      "loss": 0.1013,
      "step": 27005
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.623039292805288,
      "learning_rate": 5.297471578308514e-06,
      "loss": 0.0856,
      "step": 27010
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.5249978434384504,
      "learning_rate": 5.2904131236786035e-06,
      "loss": 0.0962,
      "step": 27015
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.6838930500034113,
      "learning_rate": 5.283358818123194e-06,
      "loss": 0.1043,
      "step": 27020
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.6466364545351454,
      "learning_rate": 5.276308663127275e-06,
      "loss": 0.097,
      "step": 27025
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.8560642212933232,
      "learning_rate": 5.2692626601749896e-06,
      "loss": 0.0941,
      "step": 27030
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.7567922570859831,
      "learning_rate": 5.262220810749588e-06,
      "loss": 0.1198,
      "step": 27035
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.5690760130912845,
      "learning_rate": 5.255183116333468e-06,
      "loss": 0.1139,
      "step": 27040
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.7396968849244785,
      "learning_rate": 5.2481495784081266e-06,
      "loss": 0.0928,
      "step": 27045
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.8090046760330366,
      "learning_rate": 5.241120198454191e-06,
      "loss": 0.1196,
      "step": 27050
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.7455128271891508,
      "learning_rate": 5.234094977951426e-06,
      "loss": 0.111,
      "step": 27055
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.790042469115102,
      "learning_rate": 5.2270739183787135e-06,
      "loss": 0.1078,
      "step": 27060
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.6974573320143602,
      "learning_rate": 5.22005702121407e-06,
      "loss": 0.1046,
      "step": 27065
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.786268622734536,
      "learning_rate": 5.213044287934607e-06,
      "loss": 0.115,
      "step": 27070
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.670783263341755,
      "learning_rate": 5.206035720016597e-06,
      "loss": 0.0853,
      "step": 27075
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.773883624157871,
      "learning_rate": 5.199031318935401e-06,
      "loss": 0.1154,
      "step": 27080
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.6617728403963338,
      "learning_rate": 5.192031086165527e-06,
      "loss": 0.1146,
      "step": 27085
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.7619043365471327,
      "learning_rate": 5.185035023180601e-06,
      "loss": 0.1106,
      "step": 27090
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.5899245509212457,
      "learning_rate": 5.178043131453356e-06,
      "loss": 0.1113,
      "step": 27095
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.6256910143154276,
      "learning_rate": 5.171055412455666e-06,
      "loss": 0.1054,
      "step": 27100
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.7309779170390247,
      "learning_rate": 5.1640718676585266e-06,
      "loss": 0.1239,
      "step": 27105
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.7315159673523742,
      "learning_rate": 5.157092498532037e-06,
      "loss": 0.0994,
      "step": 27110
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.6441877490629239,
      "learning_rate": 5.150117306545421e-06,
      "loss": 0.097,
      "step": 27115
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.8021058180272631,
      "learning_rate": 5.143146293167042e-06,
      "loss": 0.1179,
      "step": 27120
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.8419191250545054,
      "learning_rate": 5.136179459864365e-06,
      "loss": 0.1086,
      "step": 27125
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.6973020320305778,
      "learning_rate": 5.129216808103982e-06,
      "loss": 0.1042,
      "step": 27130
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.8417271559701516,
      "learning_rate": 5.122258339351615e-06,
      "loss": 0.1116,
      "step": 27135
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.8639719780933128,
      "learning_rate": 5.115304055072076e-06,
      "loss": 0.1234,
      "step": 27140
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.7724986819807443,
      "learning_rate": 5.108353956729334e-06,
      "loss": 0.1058,
      "step": 27145
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.7591267969376034,
      "learning_rate": 5.101408045786437e-06,
      "loss": 0.1192,
      "step": 27150
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.905319124257068,
      "learning_rate": 5.094466323705585e-06,
      "loss": 0.113,
      "step": 27155
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.730739666838174,
      "learning_rate": 5.087528791948087e-06,
      "loss": 0.1413,
      "step": 27160
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.7460351126458739,
      "learning_rate": 5.0805954519743545e-06,
      "loss": 0.1128,
      "step": 27165
    },
    {
      "epoch": 3.17,
      "grad_norm": 0.8485064216318304,
      "learning_rate": 5.073666305243932e-06,
      "loss": 0.1419,
      "step": 27170
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.7356958177883444,
      "learning_rate": 5.066741353215485e-06,
      "loss": 0.1172,
      "step": 27175
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.7050655369783162,
      "learning_rate": 5.059820597346784e-06,
      "loss": 0.1218,
      "step": 27180
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.6576658350405158,
      "learning_rate": 5.0529040390947104e-06,
      "loss": 0.1034,
      "step": 27185
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.7458185831059889,
      "learning_rate": 5.045991679915282e-06,
      "loss": 0.0967,
      "step": 27190
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.7324996519904005,
      "learning_rate": 5.039083521263621e-06,
      "loss": 0.0875,
      "step": 27195
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.648054663546603,
      "learning_rate": 5.03217956459397e-06,
      "loss": 0.1005,
      "step": 27200
    },
    {
      "epoch": 3.18,
      "grad_norm": 1.4949461433351363,
      "learning_rate": 5.025279811359685e-06,
      "loss": 0.1248,
      "step": 27205
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.8264578900501282,
      "learning_rate": 5.018384263013234e-06,
      "loss": 0.0889,
      "step": 27210
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.7877416501333684,
      "learning_rate": 5.011492921006197e-06,
      "loss": 0.129,
      "step": 27215
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.6481943119871718,
      "learning_rate": 5.004605786789276e-06,
      "loss": 0.0917,
      "step": 27220
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.8508648860382823,
      "learning_rate": 4.9977228618122905e-06,
      "loss": 0.1144,
      "step": 27225
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.7705185755863887,
      "learning_rate": 4.990844147524168e-06,
      "loss": 0.1145,
      "step": 27230
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.7134886485696199,
      "learning_rate": 4.983969645372941e-06,
      "loss": 0.1056,
      "step": 27235
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.7419276364071465,
      "learning_rate": 4.977099356805776e-06,
      "loss": 0.12,
      "step": 27240
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.6391129925150747,
      "learning_rate": 4.9702332832689296e-06,
      "loss": 0.1058,
      "step": 27245
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.6182474754521315,
      "learning_rate": 4.963371426207792e-06,
      "loss": 0.1131,
      "step": 27250
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.747832014556228,
      "learning_rate": 4.956513787066844e-06,
      "loss": 0.1122,
      "step": 27255
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.8018221106196859,
      "learning_rate": 4.949660367289699e-06,
      "loss": 0.0928,
      "step": 27260
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.6773045912995305,
      "learning_rate": 4.9428111683190724e-06,
      "loss": 0.0964,
      "step": 27265
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.7642994700303932,
      "learning_rate": 4.9359661915968e-06,
      "loss": 0.1308,
      "step": 27270
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.8148137456643133,
      "learning_rate": 4.9291254385638115e-06,
      "loss": 0.1162,
      "step": 27275
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.6189231669299485,
      "learning_rate": 4.9222889106601555e-06,
      "loss": 0.1088,
      "step": 27280
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.6594079789690681,
      "learning_rate": 4.915456609324992e-06,
      "loss": 0.1121,
      "step": 27285
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.649369583688146,
      "learning_rate": 4.908628535996598e-06,
      "loss": 0.0915,
      "step": 27290
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.947305512989576,
      "learning_rate": 4.901804692112355e-06,
      "loss": 0.1085,
      "step": 27295
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.6132927610783767,
      "learning_rate": 4.894985079108755e-06,
      "loss": 0.0797,
      "step": 27300
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.7964776300632762,
      "learning_rate": 4.888169698421394e-06,
      "loss": 0.1067,
      "step": 27305
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.7696520997622317,
      "learning_rate": 4.881358551484977e-06,
      "loss": 0.1306,
      "step": 27310
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.7592114319955823,
      "learning_rate": 4.874551639733324e-06,
      "loss": 0.1057,
      "step": 27315
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.7532520958374271,
      "learning_rate": 4.86774896459937e-06,
      "loss": 0.1221,
      "step": 27320
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.593577090320849,
      "learning_rate": 4.860950527515137e-06,
      "loss": 0.0982,
      "step": 27325
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.6969294665036798,
      "learning_rate": 4.854156329911771e-06,
      "loss": 0.1239,
      "step": 27330
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.7178381604216698,
      "learning_rate": 4.847366373219533e-06,
      "loss": 0.1006,
      "step": 27335
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.7814375234050084,
      "learning_rate": 4.840580658867763e-06,
      "loss": 0.0829,
      "step": 27340
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.667215134841802,
      "learning_rate": 4.833799188284938e-06,
      "loss": 0.0946,
      "step": 27345
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.6603972568708967,
      "learning_rate": 4.827021962898618e-06,
      "loss": 0.0892,
      "step": 27350
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.6877643207029196,
      "learning_rate": 4.820248984135484e-06,
      "loss": 0.1139,
      "step": 27355
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.7350312212434802,
      "learning_rate": 4.813480253421321e-06,
      "loss": 0.1018,
      "step": 27360
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.6445812729815801,
      "learning_rate": 4.806715772181017e-06,
      "loss": 0.1136,
      "step": 27365
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.657302089898427,
      "learning_rate": 4.799955541838571e-06,
      "loss": 0.0912,
      "step": 27370
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.7606474669808397,
      "learning_rate": 4.793199563817077e-06,
      "loss": 0.1072,
      "step": 27375
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.6174379246275021,
      "learning_rate": 4.786447839538735e-06,
      "loss": 0.0736,
      "step": 27380
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.8589451756160479,
      "learning_rate": 4.779700370424859e-06,
      "loss": 0.1285,
      "step": 27385
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.7252507841928352,
      "learning_rate": 4.772957157895863e-06,
      "loss": 0.092,
      "step": 27390
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.669912688672878,
      "learning_rate": 4.766218203371259e-06,
      "loss": 0.0736,
      "step": 27395
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.6631870987168231,
      "learning_rate": 4.759483508269669e-06,
      "loss": 0.081,
      "step": 27400
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.586173927746635,
      "learning_rate": 4.7527530740088255e-06,
      "loss": 0.1007,
      "step": 27405
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.8162239229999849,
      "learning_rate": 4.746026902005543e-06,
      "loss": 0.0804,
      "step": 27410
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.8529158843599705,
      "learning_rate": 4.7393049936757615e-06,
      "loss": 0.1179,
      "step": 27415
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.6451636090274976,
      "learning_rate": 4.7325873504345e-06,
      "loss": 0.0918,
      "step": 27420
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.6442376937288511,
      "learning_rate": 4.725873973695901e-06,
      "loss": 0.114,
      "step": 27425
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.7265220580977397,
      "learning_rate": 4.719164864873202e-06,
      "loss": 0.1152,
      "step": 27430
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.8066954147334778,
      "learning_rate": 4.712460025378745e-06,
      "loss": 0.1083,
      "step": 27435
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.8562634903931027,
      "learning_rate": 4.705759456623962e-06,
      "loss": 0.1008,
      "step": 27440
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.6539493485674843,
      "learning_rate": 4.69906316001939e-06,
      "loss": 0.1035,
      "step": 27445
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.6934335428071814,
      "learning_rate": 4.692371136974671e-06,
      "loss": 0.0889,
      "step": 27450
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.8194032096332868,
      "learning_rate": 4.685683388898549e-06,
      "loss": 0.105,
      "step": 27455
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.8242818311171684,
      "learning_rate": 4.67899991719887e-06,
      "loss": 0.0952,
      "step": 27460
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.68509853483034,
      "learning_rate": 4.672320723282567e-06,
      "loss": 0.1006,
      "step": 27465
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.646136344114196,
      "learning_rate": 4.665645808555685e-06,
      "loss": 0.0946,
      "step": 27470
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.9226283805520535,
      "learning_rate": 4.658975174423355e-06,
      "loss": 0.1278,
      "step": 27475
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.6413821967411327,
      "learning_rate": 4.652308822289822e-06,
      "loss": 0.0929,
      "step": 27480
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.9290877306279111,
      "learning_rate": 4.645646753558427e-06,
      "loss": 0.0905,
      "step": 27485
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.6284747443256513,
      "learning_rate": 4.638988969631597e-06,
      "loss": 0.0745,
      "step": 27490
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.582407361883294,
      "learning_rate": 4.632335471910867e-06,
      "loss": 0.0865,
      "step": 27495
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.7124413487525606,
      "learning_rate": 4.625686261796877e-06,
      "loss": 0.103,
      "step": 27500
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.8082164147629005,
      "learning_rate": 4.619041340689342e-06,
      "loss": 0.0911,
      "step": 27505
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.9446964085254289,
      "learning_rate": 4.6124007099871e-06,
      "loss": 0.132,
      "step": 27510
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.8165742292714081,
      "learning_rate": 4.605764371088059e-06,
      "loss": 0.0943,
      "step": 27515
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.6491200237377126,
      "learning_rate": 4.599132325389246e-06,
      "loss": 0.1182,
      "step": 27520
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.702653031198587,
      "learning_rate": 4.5925045742867795e-06,
      "loss": 0.0969,
      "step": 27525
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.7558146522110792,
      "learning_rate": 4.58588111917587e-06,
      "loss": 0.1003,
      "step": 27530
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.6964839434435839,
      "learning_rate": 4.5792619614508205e-06,
      "loss": 0.1387,
      "step": 27535
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.6307877078793908,
      "learning_rate": 4.572647102505029e-06,
      "loss": 0.0984,
      "step": 27540
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.6002784628232989,
      "learning_rate": 4.566036543730998e-06,
      "loss": 0.1157,
      "step": 27545
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.5956414211799476,
      "learning_rate": 4.559430286520316e-06,
      "loss": 0.1011,
      "step": 27550
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.8924788513454035,
      "learning_rate": 4.552828332263681e-06,
      "loss": 0.1026,
      "step": 27555
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.7150669248264977,
      "learning_rate": 4.546230682350858e-06,
      "loss": 0.0952,
      "step": 27560
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.6112310103405998,
      "learning_rate": 4.5396373381707325e-06,
      "loss": 0.1015,
      "step": 27565
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.6999709408215883,
      "learning_rate": 4.533048301111261e-06,
      "loss": 0.1045,
      "step": 27570
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.6177048309235855,
      "learning_rate": 4.526463572559514e-06,
      "loss": 0.0955,
      "step": 27575
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.7049165061116903,
      "learning_rate": 4.5198831539016495e-06,
      "loss": 0.1115,
      "step": 27580
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.8279660873808702,
      "learning_rate": 4.513307046522902e-06,
      "loss": 0.1382,
      "step": 27585
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.729804069750821,
      "learning_rate": 4.50673525180762e-06,
      "loss": 0.0993,
      "step": 27590
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.702232755288587,
      "learning_rate": 4.50016777113923e-06,
      "loss": 0.108,
      "step": 27595
    },
    {
      "epoch": 3.22,
      "grad_norm": 0.6043014054154779,
      "learning_rate": 4.493604605900265e-06,
      "loss": 0.1242,
      "step": 27600
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.9120160607162382,
      "learning_rate": 4.487045757472333e-06,
      "loss": 0.1235,
      "step": 27605
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.7529335024242906,
      "learning_rate": 4.4804912272361335e-06,
      "loss": 0.0977,
      "step": 27610
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.7767946440583706,
      "learning_rate": 4.473941016571473e-06,
      "loss": 0.1156,
      "step": 27615
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.720918558122027,
      "learning_rate": 4.467395126857235e-06,
      "loss": 0.1189,
      "step": 27620
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.7390866302083426,
      "learning_rate": 4.460853559471409e-06,
      "loss": 0.1005,
      "step": 27625
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.692631963637318,
      "learning_rate": 4.4543163157910465e-06,
      "loss": 0.1001,
      "step": 27630
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.8751217070481909,
      "learning_rate": 4.44778339719232e-06,
      "loss": 0.1019,
      "step": 27635
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.676708125136999,
      "learning_rate": 4.441254805050468e-06,
      "loss": 0.0994,
      "step": 27640
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.8743354861161313,
      "learning_rate": 4.434730540739826e-06,
      "loss": 0.1397,
      "step": 27645
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.8309006449711872,
      "learning_rate": 4.428210605633834e-06,
      "loss": 0.0949,
      "step": 27650
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.8199344137038984,
      "learning_rate": 4.421695001104992e-06,
      "loss": 0.1037,
      "step": 27655
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.9577009355195544,
      "learning_rate": 4.4151837285249046e-06,
      "loss": 0.1262,
      "step": 27660
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.6913118427731747,
      "learning_rate": 4.408676789264277e-06,
      "loss": 0.0916,
      "step": 27665
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.5849149667509002,
      "learning_rate": 4.402174184692876e-06,
      "loss": 0.1063,
      "step": 27670
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.6410479779252873,
      "learning_rate": 4.395675916179562e-06,
      "loss": 0.1253,
      "step": 27675
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.7934768652877257,
      "learning_rate": 4.389181985092297e-06,
      "loss": 0.1021,
      "step": 27680
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.7515742225417876,
      "learning_rate": 4.382692392798121e-06,
      "loss": 0.1055,
      "step": 27685
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.8122932922837836,
      "learning_rate": 4.3762071406631605e-06,
      "loss": 0.1194,
      "step": 27690
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.6175770177498836,
      "learning_rate": 4.369726230052637e-06,
      "loss": 0.0922,
      "step": 27695
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.72023934967413,
      "learning_rate": 4.3632496623308425e-06,
      "loss": 0.1074,
      "step": 27700
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.5931993310959411,
      "learning_rate": 4.3567774388611564e-06,
      "loss": 0.1234,
      "step": 27705
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.8489762207115195,
      "learning_rate": 4.350309561006052e-06,
      "loss": 0.128,
      "step": 27710
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.6640435514765165,
      "learning_rate": 4.343846030127094e-06,
      "loss": 0.1126,
      "step": 27715
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.764850100630166,
      "learning_rate": 4.337386847584921e-06,
      "loss": 0.1036,
      "step": 27720
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.7684789331928018,
      "learning_rate": 4.33093201473925e-06,
      "loss": 0.1259,
      "step": 27725
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.644490431888392,
      "learning_rate": 4.324481532948904e-06,
      "loss": 0.0929,
      "step": 27730
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.6882481887072169,
      "learning_rate": 4.318035403571763e-06,
      "loss": 0.1002,
      "step": 27735
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.8834753509239305,
      "learning_rate": 4.3115936279648165e-06,
      "loss": 0.1199,
      "step": 27740
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.6479994512780162,
      "learning_rate": 4.305156207484115e-06,
      "loss": 0.0928,
      "step": 27745
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.7513403539592943,
      "learning_rate": 4.298723143484809e-06,
      "loss": 0.1028,
      "step": 27750
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.7625194523488714,
      "learning_rate": 4.292294437321126e-06,
      "loss": 0.1298,
      "step": 27755
    },
    {
      "epoch": 3.24,
      "grad_norm": 1.0784639149206594,
      "learning_rate": 4.28587009034638e-06,
      "loss": 0.1114,
      "step": 27760
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.748527229005044,
      "learning_rate": 4.27945010391296e-06,
      "loss": 0.0996,
      "step": 27765
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.6424390744174807,
      "learning_rate": 4.273034479372334e-06,
      "loss": 0.1007,
      "step": 27770
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.6343905500050343,
      "learning_rate": 4.266623218075061e-06,
      "loss": 0.1134,
      "step": 27775
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.71423066113107,
      "learning_rate": 4.260216321370783e-06,
      "loss": 0.1068,
      "step": 27780
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.7915180587399318,
      "learning_rate": 4.253813790608216e-06,
      "loss": 0.1215,
      "step": 27785
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.7141301407900196,
      "learning_rate": 4.247415627135168e-06,
      "loss": 0.1154,
      "step": 27790
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.7255746806542855,
      "learning_rate": 4.2410218322985115e-06,
      "loss": 0.0934,
      "step": 27795
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.6306130921518569,
      "learning_rate": 4.2346324074442e-06,
      "loss": 0.096,
      "step": 27800
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.6052389191725739,
      "learning_rate": 4.228247353917283e-06,
      "loss": 0.0877,
      "step": 27805
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.7184656926850397,
      "learning_rate": 4.221866673061889e-06,
      "loss": 0.1185,
      "step": 27810
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.7437902535805804,
      "learning_rate": 4.215490366221201e-06,
      "loss": 0.0917,
      "step": 27815
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.5551879590931311,
      "learning_rate": 4.209118434737508e-06,
      "loss": 0.1443,
      "step": 27820
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.6216844687697246,
      "learning_rate": 4.202750879952167e-06,
      "loss": 0.1009,
      "step": 27825
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.6760739931594562,
      "learning_rate": 4.196387703205621e-06,
      "loss": 0.1115,
      "step": 27830
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.6945406611582664,
      "learning_rate": 4.190028905837382e-06,
      "loss": 0.097,
      "step": 27835
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.639885643623021,
      "learning_rate": 4.183674489186032e-06,
      "loss": 0.0884,
      "step": 27840
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.742319783732264,
      "learning_rate": 4.177324454589252e-06,
      "loss": 0.1095,
      "step": 27845
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.7420810909770709,
      "learning_rate": 4.1709788033837905e-06,
      "loss": 0.1096,
      "step": 27850
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.9133393405230974,
      "learning_rate": 4.164637536905472e-06,
      "loss": 0.1309,
      "step": 27855
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.5642246154062726,
      "learning_rate": 4.1583006564892084e-06,
      "loss": 0.0685,
      "step": 27860
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.7951090429763442,
      "learning_rate": 4.1519681634689725e-06,
      "loss": 0.1087,
      "step": 27865
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.6162920868713485,
      "learning_rate": 4.14564005917781e-06,
      "loss": 0.1056,
      "step": 27870
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.7408872322545546,
      "learning_rate": 4.139316344947866e-06,
      "loss": 0.1178,
      "step": 27875
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.8148778293857757,
      "learning_rate": 4.132997022110352e-06,
      "loss": 0.1163,
      "step": 27880
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.6475694152389248,
      "learning_rate": 4.1266820919955375e-06,
      "loss": 0.1058,
      "step": 27885
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.8219073220725632,
      "learning_rate": 4.120371555932786e-06,
      "loss": 0.084,
      "step": 27890
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.661840268827891,
      "learning_rate": 4.114065415250545e-06,
      "loss": 0.1113,
      "step": 27895
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.8958507627503823,
      "learning_rate": 4.107763671276305e-06,
      "loss": 0.1501,
      "step": 27900
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.6915926013371538,
      "learning_rate": 4.101466325336661e-06,
      "loss": 0.0841,
      "step": 27905
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.6496701901388301,
      "learning_rate": 4.0951733787572635e-06,
      "loss": 0.0987,
      "step": 27910
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.8739538308905007,
      "learning_rate": 4.088884832862844e-06,
      "loss": 0.1324,
      "step": 27915
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.8178434259171943,
      "learning_rate": 4.082600688977209e-06,
      "loss": 0.1168,
      "step": 27920
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.8809897556690398,
      "learning_rate": 4.0763209484232425e-06,
      "loss": 0.1277,
      "step": 27925
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.742955650754623,
      "learning_rate": 4.070045612522891e-06,
      "loss": 0.091,
      "step": 27930
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.7630920473057111,
      "learning_rate": 4.063774682597174e-06,
      "loss": 0.0967,
      "step": 27935
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.689831566834404,
      "learning_rate": 4.057508159966189e-06,
      "loss": 0.1049,
      "step": 27940
    },
    {
      "epoch": 3.26,
      "grad_norm": 0.8296147421207558,
      "learning_rate": 4.051246045949106e-06,
      "loss": 0.1253,
      "step": 27945
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.67284792786023,
      "learning_rate": 4.044988341864175e-06,
      "loss": 0.1007,
      "step": 27950
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.8211630670759348,
      "learning_rate": 4.038735049028694e-06,
      "loss": 0.1109,
      "step": 27955
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.7889608098422833,
      "learning_rate": 4.032486168759059e-06,
      "loss": 0.1215,
      "step": 27960
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.6847364483082103,
      "learning_rate": 4.026241702370712e-06,
      "loss": 0.0921,
      "step": 27965
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.7284210583509032,
      "learning_rate": 4.020001651178182e-06,
      "loss": 0.112,
      "step": 27970
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.8095645977967392,
      "learning_rate": 4.013766016495077e-06,
      "loss": 0.1221,
      "step": 27975
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.7618551184614817,
      "learning_rate": 4.00753479963405e-06,
      "loss": 0.1274,
      "step": 27980
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.7932976114361806,
      "learning_rate": 4.001308001906839e-06,
      "loss": 0.1248,
      "step": 27985
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.7605060396957194,
      "learning_rate": 3.995085624624262e-06,
      "loss": 0.1027,
      "step": 27990
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.7354698511348152,
      "learning_rate": 3.98886766909618e-06,
      "loss": 0.0931,
      "step": 27995
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.70975155276137,
      "learning_rate": 3.982654136631547e-06,
      "loss": 0.1035,
      "step": 28000
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.7110277812308022,
      "learning_rate": 3.976445028538367e-06,
      "loss": 0.0913,
      "step": 28005
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.758231518178429,
      "learning_rate": 3.970240346123732e-06,
      "loss": 0.1016,
      "step": 28010
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.6528700462826965,
      "learning_rate": 3.964040090693785e-06,
      "loss": 0.1034,
      "step": 28015
    },
    {
      "epoch": 3.27,
      "grad_norm": 1.109410882650879,
      "learning_rate": 3.9578442635537586e-06,
      "loss": 0.1029,
      "step": 28020
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.6821703378736803,
      "learning_rate": 3.951652866007926e-06,
      "loss": 0.0851,
      "step": 28025
    },
    {
      "epoch": 3.27,
      "grad_norm": 0.6878920530103287,
      "learning_rate": 3.945465899359643e-06,
      "loss": 0.0851,
      "step": 28030
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.6855964259262746,
      "learning_rate": 3.939283364911331e-06,
      "loss": 0.0847,
      "step": 28035
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.7523315156156533,
      "learning_rate": 3.933105263964479e-06,
      "loss": 0.0862,
      "step": 28040
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.8325013214330024,
      "learning_rate": 3.92693159781965e-06,
      "loss": 0.0942,
      "step": 28045
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.7470833577632429,
      "learning_rate": 3.920762367776451e-06,
      "loss": 0.0894,
      "step": 28050
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.6105296101043943,
      "learning_rate": 3.914597575133577e-06,
      "loss": 0.0938,
      "step": 28055
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.9751688959200584,
      "learning_rate": 3.9084372211887875e-06,
      "loss": 0.1404,
      "step": 28060
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.7088108020323782,
      "learning_rate": 3.9022813072388895e-06,
      "loss": 0.1071,
      "step": 28065
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.7931428650114524,
      "learning_rate": 3.896129834579778e-06,
      "loss": 0.1217,
      "step": 28070
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.6318974791436598,
      "learning_rate": 3.889982804506395e-06,
      "loss": 0.1038,
      "step": 28075
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.7251789213608979,
      "learning_rate": 3.883840218312757e-06,
      "loss": 0.0805,
      "step": 28080
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.6574597475711454,
      "learning_rate": 3.877702077291939e-06,
      "loss": 0.0894,
      "step": 28085
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.8217694309033453,
      "learning_rate": 3.871568382736099e-06,
      "loss": 0.1343,
      "step": 28090
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.6296160237840762,
      "learning_rate": 3.86543913593643e-06,
      "loss": 0.1017,
      "step": 28095
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.9733008740232575,
      "learning_rate": 3.859314338183201e-06,
      "loss": 0.1015,
      "step": 28100
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.8336248904635234,
      "learning_rate": 3.853193990765749e-06,
      "loss": 0.1226,
      "step": 28105
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.8732939090618191,
      "learning_rate": 3.8470780949724755e-06,
      "loss": 0.1345,
      "step": 28110
    },
    {
      "epoch": 3.28,
      "grad_norm": 0.7173714317798201,
      "learning_rate": 3.840966652090841e-06,
      "loss": 0.1023,
      "step": 28115
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.7318872194025108,
      "learning_rate": 3.834859663407361e-06,
      "loss": 0.1151,
      "step": 28120
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.7035034621776279,
      "learning_rate": 3.828757130207633e-06,
      "loss": 0.0881,
      "step": 28125
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.7824267856465141,
      "learning_rate": 3.8226590537762865e-06,
      "loss": 0.104,
      "step": 28130
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.6988643708450567,
      "learning_rate": 3.816565435397043e-06,
      "loss": 0.1138,
      "step": 28135
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.7216208500758892,
      "learning_rate": 3.810476276352676e-06,
      "loss": 0.0869,
      "step": 28140
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.853636524374595,
      "learning_rate": 3.804391577925004e-06,
      "loss": 0.1302,
      "step": 28145
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.8574143648342926,
      "learning_rate": 3.798311341394928e-06,
      "loss": 0.0974,
      "step": 28150
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.6646070219220201,
      "learning_rate": 3.7922355680424043e-06,
      "loss": 0.0931,
      "step": 28155
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.7554673561955634,
      "learning_rate": 3.786164259146438e-06,
      "loss": 0.1055,
      "step": 28160
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.5975903415718572,
      "learning_rate": 3.7800974159851176e-06,
      "loss": 0.1076,
      "step": 28165
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.7166386805167545,
      "learning_rate": 3.7740350398355577e-06,
      "loss": 0.1191,
      "step": 28170
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.7235675483820421,
      "learning_rate": 3.7679771319739654e-06,
      "loss": 0.0917,
      "step": 28175
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.6122845853775039,
      "learning_rate": 3.76192369367559e-06,
      "loss": 0.1167,
      "step": 28180
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.8014150079940554,
      "learning_rate": 3.7558747262147478e-06,
      "loss": 0.107,
      "step": 28185
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.715272868162659,
      "learning_rate": 3.749830230864809e-06,
      "loss": 0.0876,
      "step": 28190
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.726670229347444,
      "learning_rate": 3.743790208898193e-06,
      "loss": 0.1069,
      "step": 28195
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.9572182692267391,
      "learning_rate": 3.7377546615863956e-06,
      "loss": 0.1199,
      "step": 28200
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.759306046843035,
      "learning_rate": 3.731723590199962e-06,
      "loss": 0.0988,
      "step": 28205
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.7367485391109551,
      "learning_rate": 3.7256969960085015e-06,
      "loss": 0.0974,
      "step": 28210
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.6373062728382746,
      "learning_rate": 3.719674880280663e-06,
      "loss": 0.0823,
      "step": 28215
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.6748088665471876,
      "learning_rate": 3.71365724428418e-06,
      "loss": 0.137,
      "step": 28220
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.7394675782194076,
      "learning_rate": 3.707644089285814e-06,
      "loss": 0.116,
      "step": 28225
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.7095203871798705,
      "learning_rate": 3.7016354165514023e-06,
      "loss": 0.0966,
      "step": 28230
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.6416884365890312,
      "learning_rate": 3.6956312273458386e-06,
      "loss": 0.0925,
      "step": 28235
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.9913771035590639,
      "learning_rate": 3.6896315229330568e-06,
      "loss": 0.0859,
      "step": 28240
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.7768635680095392,
      "learning_rate": 3.6836363045760656e-06,
      "loss": 0.0956,
      "step": 28245
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.6909818548395155,
      "learning_rate": 3.6776455735369255e-06,
      "loss": 0.0925,
      "step": 28250
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.7961110282872308,
      "learning_rate": 3.671659331076743e-06,
      "loss": 0.1227,
      "step": 28255
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.6560022020241033,
      "learning_rate": 3.665677578455676e-06,
      "loss": 0.0912,
      "step": 28260
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.6939255407942331,
      "learning_rate": 3.659700316932954e-06,
      "loss": 0.1064,
      "step": 28265
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.6773516481523861,
      "learning_rate": 3.6537275477668553e-06,
      "loss": 0.0826,
      "step": 28270
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.6924984935034643,
      "learning_rate": 3.647759272214707e-06,
      "loss": 0.0986,
      "step": 28275
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.633980236253523,
      "learning_rate": 3.6417954915329013e-06,
      "loss": 0.1136,
      "step": 28280
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.7523943599154672,
      "learning_rate": 3.6358362069768613e-06,
      "loss": 0.1082,
      "step": 28285
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.625307175686647,
      "learning_rate": 3.6298814198010956e-06,
      "loss": 0.1086,
      "step": 28290
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.6645045567046022,
      "learning_rate": 3.6239311312591367e-06,
      "loss": 0.0922,
      "step": 28295
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.6528998271115086,
      "learning_rate": 3.617985342603586e-06,
      "loss": 0.1064,
      "step": 28300
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.6265223743295981,
      "learning_rate": 3.6120440550861e-06,
      "loss": 0.0786,
      "step": 28305
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.6686967521700447,
      "learning_rate": 3.6061072699573735e-06,
      "loss": 0.1055,
      "step": 28310
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.0386554672662853,
      "learning_rate": 3.6001749884671672e-06,
      "loss": 0.103,
      "step": 28315
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.6439533182825401,
      "learning_rate": 3.5942472118642905e-06,
      "loss": 0.1174,
      "step": 28320
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.7497856409960862,
      "learning_rate": 3.5883239413966015e-06,
      "loss": 0.1023,
      "step": 28325
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.6791858942497785,
      "learning_rate": 3.582405178310999e-06,
      "loss": 0.0966,
      "step": 28330
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.5653042669623298,
      "learning_rate": 3.5764909238534566e-06,
      "loss": 0.0856,
      "step": 28335
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.7874656910578491,
      "learning_rate": 3.5705811792689835e-06,
      "loss": 0.1108,
      "step": 28340
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.7339945506210517,
      "learning_rate": 3.564675945801646e-06,
      "loss": 0.1097,
      "step": 28345
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.7367749406803198,
      "learning_rate": 3.5587752246945577e-06,
      "loss": 0.106,
      "step": 28350
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.6448960373103251,
      "learning_rate": 3.5528790171898806e-06,
      "loss": 0.1022,
      "step": 28355
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.7017291207586156,
      "learning_rate": 3.546987324528822e-06,
      "loss": 0.105,
      "step": 28360
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.6891959880441897,
      "learning_rate": 3.541100147951651e-06,
      "loss": 0.0844,
      "step": 28365
    },
    {
      "epoch": 3.31,
      "grad_norm": 0.8467842946050073,
      "learning_rate": 3.5352174886976786e-06,
      "loss": 0.12,
      "step": 28370
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.7064643668473771,
      "learning_rate": 3.52933934800527e-06,
      "loss": 0.0957,
      "step": 28375
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.8760332734549043,
      "learning_rate": 3.52346572711183e-06,
      "loss": 0.1171,
      "step": 28380
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.7109978048818296,
      "learning_rate": 3.517596627253822e-06,
      "loss": 0.1295,
      "step": 28385
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.7434335016570166,
      "learning_rate": 3.5117320496667465e-06,
      "loss": 0.1049,
      "step": 28390
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.7395214464760852,
      "learning_rate": 3.5058719955851667e-06,
      "loss": 0.1171,
      "step": 28395
    },
    {
      "epoch": 3.32,
      "grad_norm": 46.628418668526784,
      "learning_rate": 3.500016466242678e-06,
      "loss": 0.1812,
      "step": 28400
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.829512945499508,
      "learning_rate": 3.4941654628719317e-06,
      "loss": 0.1094,
      "step": 28405
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.8133826051556866,
      "learning_rate": 3.488318986704628e-06,
      "loss": 0.1262,
      "step": 28410
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.7473918140585176,
      "learning_rate": 3.4824770389715155e-06,
      "loss": 0.1072,
      "step": 28415
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.9698470878744179,
      "learning_rate": 3.4766396209023807e-06,
      "loss": 0.1339,
      "step": 28420
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.9618886651414976,
      "learning_rate": 3.4708067337260564e-06,
      "loss": 0.1022,
      "step": 28425
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.703730759439064,
      "learning_rate": 3.4649783786704304e-06,
      "loss": 0.0988,
      "step": 28430
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.9484169541283957,
      "learning_rate": 3.459154556962435e-06,
      "loss": 0.1103,
      "step": 28435
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.6923825211374073,
      "learning_rate": 3.4533352698280424e-06,
      "loss": 0.0841,
      "step": 28440
    },
    {
      "epoch": 3.32,
      "grad_norm": 1.151036733018734,
      "learning_rate": 3.447520518492281e-06,
      "loss": 0.0932,
      "step": 28445
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.7057042108508013,
      "learning_rate": 3.4417103041792103e-06,
      "loss": 0.0882,
      "step": 28450
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.7327663201075238,
      "learning_rate": 3.4359046281119357e-06,
      "loss": 0.0912,
      "step": 28455
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.7802611355170154,
      "learning_rate": 3.4301034915126206e-06,
      "loss": 0.1109,
      "step": 28460
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6617619079776523,
      "learning_rate": 3.4243068956024692e-06,
      "loss": 0.117,
      "step": 28465
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6453268036726225,
      "learning_rate": 3.418514841601714e-06,
      "loss": 0.0936,
      "step": 28470
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6348505617563418,
      "learning_rate": 3.412727330729648e-06,
      "loss": 0.1211,
      "step": 28475
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.7756873705446905,
      "learning_rate": 3.406944364204609e-06,
      "loss": 0.0945,
      "step": 28480
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6345505585427157,
      "learning_rate": 3.401165943243964e-06,
      "loss": 0.0979,
      "step": 28485
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.8463425298075826,
      "learning_rate": 3.395392069064138e-06,
      "loss": 0.1534,
      "step": 28490
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.7317160188867926,
      "learning_rate": 3.38962274288058e-06,
      "loss": 0.1046,
      "step": 28495
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.7495061458301996,
      "learning_rate": 3.3838579659078052e-06,
      "loss": 0.1253,
      "step": 28500
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.7484124777400966,
      "learning_rate": 3.378097739359354e-06,
      "loss": 0.1281,
      "step": 28505
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.7896845924974772,
      "learning_rate": 3.372342064447817e-06,
      "loss": 0.103,
      "step": 28510
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6806777040521867,
      "learning_rate": 3.3665909423848287e-06,
      "loss": 0.1045,
      "step": 28515
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.9745860705600712,
      "learning_rate": 3.360844374381056e-06,
      "loss": 0.1173,
      "step": 28520
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.722399156603803,
      "learning_rate": 3.3551023616462034e-06,
      "loss": 0.1054,
      "step": 28525
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6534491395575144,
      "learning_rate": 3.3493649053890326e-06,
      "loss": 0.0869,
      "step": 28530
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6696850162519702,
      "learning_rate": 3.343632006817346e-06,
      "loss": 0.0852,
      "step": 28535
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6424707775122042,
      "learning_rate": 3.3379036671379623e-06,
      "loss": 0.0988,
      "step": 28540
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.6005995174874295,
      "learning_rate": 3.3321798875567662e-06,
      "loss": 0.0962,
      "step": 28545
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.799612994702888,
      "learning_rate": 3.3264606692786774e-06,
      "loss": 0.1165,
      "step": 28550
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.7275850439751429,
      "learning_rate": 3.320746013507642e-06,
      "loss": 0.1024,
      "step": 28555
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.6859576681219415,
      "learning_rate": 3.3150359214466657e-06,
      "loss": 0.1111,
      "step": 28560
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.7557470833944618,
      "learning_rate": 3.3093303942977687e-06,
      "loss": 0.0927,
      "step": 28565
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.7975180596742816,
      "learning_rate": 3.3036294332620317e-06,
      "loss": 0.1026,
      "step": 28570
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.7370316545547761,
      "learning_rate": 3.2979330395395662e-06,
      "loss": 0.0918,
      "step": 28575
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.6524360551785489,
      "learning_rate": 3.2922412143295308e-06,
      "loss": 0.1163,
      "step": 28580
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.605049105321238,
      "learning_rate": 3.2865539588301046e-06,
      "loss": 0.1485,
      "step": 28585
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.8167607212334527,
      "learning_rate": 3.2808712742385143e-06,
      "loss": 0.0952,
      "step": 28590
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.6721525492901864,
      "learning_rate": 3.2751931617510223e-06,
      "loss": 0.0904,
      "step": 28595
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.8248765600577694,
      "learning_rate": 3.269519622562939e-06,
      "loss": 0.1235,
      "step": 28600
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.5608091027475581,
      "learning_rate": 3.263850657868606e-06,
      "loss": 0.0811,
      "step": 28605
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.7019272348132117,
      "learning_rate": 3.2581862688613852e-06,
      "loss": 0.0965,
      "step": 28610
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.8745104263853927,
      "learning_rate": 3.2525264567337067e-06,
      "loss": 0.1065,
      "step": 28615
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.6045150586551867,
      "learning_rate": 3.246871222677006e-06,
      "loss": 0.1209,
      "step": 28620
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.5933104381777957,
      "learning_rate": 3.2412205678817754e-06,
      "loss": 0.0892,
      "step": 28625
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.9128926987384887,
      "learning_rate": 3.2355744935375408e-06,
      "loss": 0.1221,
      "step": 28630
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.6301476160604672,
      "learning_rate": 3.229933000832852e-06,
      "loss": 0.1034,
      "step": 28635
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.8105424576133822,
      "learning_rate": 3.2242960909553034e-06,
      "loss": 0.1232,
      "step": 28640
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.9823096751582189,
      "learning_rate": 3.218663765091534e-06,
      "loss": 0.1121,
      "step": 28645
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.6567971919147917,
      "learning_rate": 3.213036024427196e-06,
      "loss": 0.104,
      "step": 28650
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.6363796489399048,
      "learning_rate": 3.207412870146995e-06,
      "loss": 0.0938,
      "step": 28655
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.6292393997725017,
      "learning_rate": 3.2017943034346547e-06,
      "loss": 0.1189,
      "step": 28660
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.8254562079331257,
      "learning_rate": 3.1961803254729473e-06,
      "loss": 0.1196,
      "step": 28665
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.6406829868749524,
      "learning_rate": 3.1905709374436766e-06,
      "loss": 0.0983,
      "step": 28670
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.677321685074616,
      "learning_rate": 3.1849661405276786e-06,
      "loss": 0.1029,
      "step": 28675
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.6561117157146823,
      "learning_rate": 3.179365935904821e-06,
      "loss": 0.11,
      "step": 28680
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.7956179799967459,
      "learning_rate": 3.173770324753997e-06,
      "loss": 0.0945,
      "step": 28685
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.5988166043308951,
      "learning_rate": 3.1681793082531467e-06,
      "loss": 0.0838,
      "step": 28690
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.5788063393337889,
      "learning_rate": 3.162592887579241e-06,
      "loss": 0.0913,
      "step": 28695
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.5552989395705786,
      "learning_rate": 3.1570110639082828e-06,
      "loss": 0.108,
      "step": 28700
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.7528522604751364,
      "learning_rate": 3.1514338384152907e-06,
      "loss": 0.1099,
      "step": 28705
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.775242964088523,
      "learning_rate": 3.1458612122743452e-06,
      "loss": 0.1105,
      "step": 28710
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.7445331715729363,
      "learning_rate": 3.14029318665853e-06,
      "loss": 0.105,
      "step": 28715
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.8185360665545963,
      "learning_rate": 3.134729762739977e-06,
      "loss": 0.132,
      "step": 28720
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.5629552780660295,
      "learning_rate": 3.129170941689849e-06,
      "loss": 0.0957,
      "step": 28725
    },
    {
      "epoch": 3.36,
      "grad_norm": 1.278820647104455,
      "learning_rate": 3.12361672467833e-06,
      "loss": 0.1104,
      "step": 28730
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.7438259220057238,
      "learning_rate": 3.1180671128746424e-06,
      "loss": 0.1046,
      "step": 28735
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.7834522284533213,
      "learning_rate": 3.1125221074470412e-06,
      "loss": 0.0937,
      "step": 28740
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.7097981764499439,
      "learning_rate": 3.106981709562809e-06,
      "loss": 0.1082,
      "step": 28745
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.6641091932527232,
      "learning_rate": 3.1014459203882547e-06,
      "loss": 0.0942,
      "step": 28750
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.7569603173961842,
      "learning_rate": 3.095914741088715e-06,
      "loss": 0.1162,
      "step": 28755
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.6410382781028555,
      "learning_rate": 3.0903881728285676e-06,
      "loss": 0.0954,
      "step": 28760
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.6319997471040962,
      "learning_rate": 3.0848662167712093e-06,
      "loss": 0.1101,
      "step": 28765
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.6311186394777405,
      "learning_rate": 3.0793488740790772e-06,
      "loss": 0.0963,
      "step": 28770
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.6514567768433571,
      "learning_rate": 3.0738361459136193e-06,
      "loss": 0.075,
      "step": 28775
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.864989351619348,
      "learning_rate": 3.0683280334353303e-06,
      "loss": 0.0934,
      "step": 28780
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.7283768040490727,
      "learning_rate": 3.0628245378037213e-06,
      "loss": 0.1051,
      "step": 28785
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.7430287648124071,
      "learning_rate": 3.0573256601773347e-06,
      "loss": 0.115,
      "step": 28790
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.8345508745104409,
      "learning_rate": 3.0518314017137507e-06,
      "loss": 0.1105,
      "step": 28795
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.9637332157980986,
      "learning_rate": 3.046341763569557e-06,
      "loss": 0.1283,
      "step": 28800
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.7005262897299117,
      "learning_rate": 3.0408567469003833e-06,
      "loss": 0.0845,
      "step": 28805
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.7161662199125126,
      "learning_rate": 3.0353763528608912e-06,
      "loss": 0.1013,
      "step": 28810
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.7400550628093231,
      "learning_rate": 3.029900582604753e-06,
      "loss": 0.1053,
      "step": 28815
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.8866578364353414,
      "learning_rate": 3.024429437284673e-06,
      "loss": 0.1204,
      "step": 28820
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.7800555378250028,
      "learning_rate": 3.01896291805239e-06,
      "loss": 0.0977,
      "step": 28825
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.6619530551831401,
      "learning_rate": 3.0135010260586597e-06,
      "loss": 0.1142,
      "step": 28830
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.5808083178837569,
      "learning_rate": 3.008043762453272e-06,
      "loss": 0.0902,
      "step": 28835
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.7287917742169926,
      "learning_rate": 3.0025911283850407e-06,
      "loss": 0.0971,
      "step": 28840
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.8441276554314261,
      "learning_rate": 2.997143125001797e-06,
      "loss": 0.1166,
      "step": 28845
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.8818960550511538,
      "learning_rate": 2.9916997534504e-06,
      "loss": 0.0992,
      "step": 28850
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.8366504717913369,
      "learning_rate": 2.98626101487674e-06,
      "loss": 0.1344,
      "step": 28855
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.8771492738864335,
      "learning_rate": 2.980826910425727e-06,
      "loss": 0.1195,
      "step": 28860
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.7389778485725227,
      "learning_rate": 2.9753974412413064e-06,
      "loss": 0.0978,
      "step": 28865
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.7682188432319635,
      "learning_rate": 2.969972608466423e-06,
      "loss": 0.1097,
      "step": 28870
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.6878324031177027,
      "learning_rate": 2.964552413243074e-06,
      "loss": 0.0935,
      "step": 28875
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.7648388262874948,
      "learning_rate": 2.9591368567122552e-06,
      "loss": 0.1329,
      "step": 28880
    },
    {
      "epoch": 3.37,
      "grad_norm": 0.7243896712392304,
      "learning_rate": 2.9537259400140098e-06,
      "loss": 0.0986,
      "step": 28885
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.8949618611385405,
      "learning_rate": 2.948319664287383e-06,
      "loss": 0.115,
      "step": 28890
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.6975126638492564,
      "learning_rate": 2.9429180306704577e-06,
      "loss": 0.1144,
      "step": 28895
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.563270986491268,
      "learning_rate": 2.937521040300331e-06,
      "loss": 0.0917,
      "step": 28900
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.684725148392325,
      "learning_rate": 2.932128694313133e-06,
      "loss": 0.1057,
      "step": 28905
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.8012474732674006,
      "learning_rate": 2.926740993844004e-06,
      "loss": 0.1391,
      "step": 28910
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.7043712683169141,
      "learning_rate": 2.9213579400271084e-06,
      "loss": 0.0895,
      "step": 28915
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.8109038973939088,
      "learning_rate": 2.915979533995636e-06,
      "loss": 0.1369,
      "step": 28920
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.6234423968702772,
      "learning_rate": 2.9106057768818e-06,
      "loss": 0.0923,
      "step": 28925
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.6418320256885169,
      "learning_rate": 2.9052366698168325e-06,
      "loss": 0.0971,
      "step": 28930
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.8288007080766526,
      "learning_rate": 2.8998722139309933e-06,
      "loss": 0.1039,
      "step": 28935
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.7728739172187259,
      "learning_rate": 2.89451241035355e-06,
      "loss": 0.0915,
      "step": 28940
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.6284288967122575,
      "learning_rate": 2.889157260212791e-06,
      "loss": 0.1303,
      "step": 28945
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.6832618380112833,
      "learning_rate": 2.8838067646360406e-06,
      "loss": 0.0998,
      "step": 28950
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.7799895755027861,
      "learning_rate": 2.8784609247496345e-06,
      "loss": 0.1046,
      "step": 28955
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.5902164516015337,
      "learning_rate": 2.87311974167892e-06,
      "loss": 0.0735,
      "step": 28960
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.799705596574271,
      "learning_rate": 2.867783216548278e-06,
      "loss": 0.1094,
      "step": 28965
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.841515992106704,
      "learning_rate": 2.8624513504811016e-06,
      "loss": 0.1146,
      "step": 28970
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.5646634517917594,
      "learning_rate": 2.85712414459981e-06,
      "loss": 0.1143,
      "step": 28975
    },
    {
      "epoch": 3.39,
      "grad_norm": 1.1905755291134354,
      "learning_rate": 2.85180160002583e-06,
      "loss": 0.1095,
      "step": 28980
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.7485076791360326,
      "learning_rate": 2.846483717879611e-06,
      "loss": 0.1291,
      "step": 28985
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.7274379661646979,
      "learning_rate": 2.8411704992806235e-06,
      "loss": 0.1035,
      "step": 28990
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.6748241732764865,
      "learning_rate": 2.8358619453473595e-06,
      "loss": 0.11,
      "step": 28995
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.7836737650211666,
      "learning_rate": 2.8305580571973217e-06,
      "loss": 0.1087,
      "step": 29000
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.6059980465913074,
      "learning_rate": 2.8252588359470423e-06,
      "loss": 0.1126,
      "step": 29005
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.8912021457147173,
      "learning_rate": 2.8199642827120565e-06,
      "loss": 0.11,
      "step": 29010
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.7152104094778989,
      "learning_rate": 2.8146743986069173e-06,
      "loss": 0.1047,
      "step": 29015
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.7644796147224163,
      "learning_rate": 2.8093891847452093e-06,
      "loss": 0.1094,
      "step": 29020
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.7180270545351612,
      "learning_rate": 2.804108642239525e-06,
      "loss": 0.0986,
      "step": 29025
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.6626397345100121,
      "learning_rate": 2.7988327722014677e-06,
      "loss": 0.1025,
      "step": 29030
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.6646506326763864,
      "learning_rate": 2.7935615757416657e-06,
      "loss": 0.0965,
      "step": 29035
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.7728865547213659,
      "learning_rate": 2.7882950539697695e-06,
      "loss": 0.0853,
      "step": 29040
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.6209600407248309,
      "learning_rate": 2.783033207994426e-06,
      "loss": 0.0893,
      "step": 29045
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.7401401218026368,
      "learning_rate": 2.777776038923316e-06,
      "loss": 0.0974,
      "step": 29050
    },
    {
      "epoch": 3.39,
      "grad_norm": 0.6757958408727438,
      "learning_rate": 2.7725235478631208e-06,
      "loss": 0.0932,
      "step": 29055
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.8297523385247889,
      "learning_rate": 2.767275735919553e-06,
      "loss": 0.1313,
      "step": 29060
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.7640702041099631,
      "learning_rate": 2.762032604197326e-06,
      "loss": 0.1029,
      "step": 29065
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.721447562342801,
      "learning_rate": 2.7567941538001796e-06,
      "loss": 0.1044,
      "step": 29070
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.6158025124774261,
      "learning_rate": 2.7515603858308646e-06,
      "loss": 0.0855,
      "step": 29075
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.8049845982201027,
      "learning_rate": 2.746331301391131e-06,
      "loss": 0.114,
      "step": 29080
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.7012783328623803,
      "learning_rate": 2.7411069015817637e-06,
      "loss": 0.1054,
      "step": 29085
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.7509113185769443,
      "learning_rate": 2.7358871875025533e-06,
      "loss": 0.136,
      "step": 29090
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.629958538376477,
      "learning_rate": 2.7306721602523122e-06,
      "loss": 0.0869,
      "step": 29095
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.7068033603025061,
      "learning_rate": 2.7254618209288428e-06,
      "loss": 0.081,
      "step": 29100
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.5573870894706169,
      "learning_rate": 2.7202561706289902e-06,
      "loss": 0.0771,
      "step": 29105
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.6748362995080456,
      "learning_rate": 2.7150552104485887e-06,
      "loss": 0.1073,
      "step": 29110
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.7418188309902358,
      "learning_rate": 2.709858941482496e-06,
      "loss": 0.1463,
      "step": 29115
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.6778664681816945,
      "learning_rate": 2.7046673648245924e-06,
      "loss": 0.1142,
      "step": 29120
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.6360195370630303,
      "learning_rate": 2.69948048156774e-06,
      "loss": 0.0961,
      "step": 29125
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.8104118949415474,
      "learning_rate": 2.6942982928038456e-06,
      "loss": 0.0832,
      "step": 29130
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.6628470983511452,
      "learning_rate": 2.6891207996238147e-06,
      "loss": 0.1023,
      "step": 29135
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.6188057428595316,
      "learning_rate": 2.6839480031175563e-06,
      "loss": 0.0923,
      "step": 29140
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.6751220481125374,
      "learning_rate": 2.678779904374007e-06,
      "loss": 0.0875,
      "step": 29145
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.7905414691782526,
      "learning_rate": 2.6736165044810968e-06,
      "loss": 0.1414,
      "step": 29150
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.5906334431066138,
      "learning_rate": 2.6684578045257756e-06,
      "loss": 0.1158,
      "step": 29155
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9755394176692816,
      "learning_rate": 2.663303805594011e-06,
      "loss": 0.1081,
      "step": 29160
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.7795644246251584,
      "learning_rate": 2.6581545087707745e-06,
      "loss": 0.0934,
      "step": 29165
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.7290049032443088,
      "learning_rate": 2.6530099151400424e-06,
      "loss": 0.1041,
      "step": 29170
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.9267338127754627,
      "learning_rate": 2.6478700257848015e-06,
      "loss": 0.0951,
      "step": 29175
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.762222417964181,
      "learning_rate": 2.642734841787059e-06,
      "loss": 0.1197,
      "step": 29180
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.8189430335745673,
      "learning_rate": 2.6376043642278194e-06,
      "loss": 0.1311,
      "step": 29185
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.7403126685475722,
      "learning_rate": 2.632478594187113e-06,
      "loss": 0.0748,
      "step": 29190
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.6530098855201154,
      "learning_rate": 2.6273575327439543e-06,
      "loss": 0.0957,
      "step": 29195
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.6657590050695665,
      "learning_rate": 2.622241180976384e-06,
      "loss": 0.1083,
      "step": 29200
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.828161588253386,
      "learning_rate": 2.6171295399614603e-06,
      "loss": 0.0944,
      "step": 29205
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.6644647074876299,
      "learning_rate": 2.612022610775219e-06,
      "loss": 0.1031,
      "step": 29210
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.7970219720616535,
      "learning_rate": 2.6069203944927344e-06,
      "loss": 0.0947,
      "step": 29215
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.7416239119729081,
      "learning_rate": 2.6018228921880716e-06,
      "loss": 0.1097,
      "step": 29220
    },
    {
      "epoch": 3.41,
      "grad_norm": 0.6856097941178884,
      "learning_rate": 2.5967301049343057e-06,
      "loss": 0.101,
      "step": 29225
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.6421768561224025,
      "learning_rate": 2.591642033803526e-06,
      "loss": 0.1171,
      "step": 29230
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.7121098570384933,
      "learning_rate": 2.5865586798668295e-06,
      "loss": 0.094,
      "step": 29235
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.6344778805389947,
      "learning_rate": 2.581480044194309e-06,
      "loss": 0.0722,
      "step": 29240
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.8119722711457465,
      "learning_rate": 2.576406127855066e-06,
      "loss": 0.1026,
      "step": 29245
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.7264250483786848,
      "learning_rate": 2.5713369319172203e-06,
      "loss": 0.0967,
      "step": 29250
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.7592911933932857,
      "learning_rate": 2.566272457447888e-06,
      "loss": 0.0968,
      "step": 29255
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.7267884548447344,
      "learning_rate": 2.5612127055131994e-06,
      "loss": 0.1136,
      "step": 29260
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.6628811294532958,
      "learning_rate": 2.5561576771782753e-06,
      "loss": 0.0923,
      "step": 29265
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.9113579569054251,
      "learning_rate": 2.551107373507264e-06,
      "loss": 0.098,
      "step": 29270
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.740175511306563,
      "learning_rate": 2.5460617955632927e-06,
      "loss": 0.1172,
      "step": 29275
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.6891898154441953,
      "learning_rate": 2.5410209444085202e-06,
      "loss": 0.1019,
      "step": 29280
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.7693895953309321,
      "learning_rate": 2.535984821104098e-06,
      "loss": 0.1377,
      "step": 29285
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.5597343743842762,
      "learning_rate": 2.5309534267101755e-06,
      "loss": 0.1272,
      "step": 29290
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.8711712768550239,
      "learning_rate": 2.5259267622859182e-06,
      "loss": 0.0971,
      "step": 29295
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.6343153191358625,
      "learning_rate": 2.5209048288894947e-06,
      "loss": 0.1164,
      "step": 29300
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.6573692866518238,
      "learning_rate": 2.515887627578073e-06,
      "loss": 0.1283,
      "step": 29305
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.7636692752731608,
      "learning_rate": 2.5108751594078215e-06,
      "loss": 0.118,
      "step": 29310
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.6975976270822801,
      "learning_rate": 2.505867425433919e-06,
      "loss": 0.1004,
      "step": 29315
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.7305641733284647,
      "learning_rate": 2.5008644267105475e-06,
      "loss": 0.1408,
      "step": 29320
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.749871081140077,
      "learning_rate": 2.4958661642908916e-06,
      "loss": 0.1075,
      "step": 29325
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.6998715082856394,
      "learning_rate": 2.4908726392271405e-06,
      "loss": 0.1105,
      "step": 29330
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.7163083322383815,
      "learning_rate": 2.485883852570481e-06,
      "loss": 0.1093,
      "step": 29335
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.6692625496174391,
      "learning_rate": 2.4808998053711007e-06,
      "loss": 0.1036,
      "step": 29340
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.6574895050472488,
      "learning_rate": 2.475920498678194e-06,
      "loss": 0.0909,
      "step": 29345
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.7502253731469911,
      "learning_rate": 2.4709459335399644e-06,
      "loss": 0.1232,
      "step": 29350
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.6703918768763688,
      "learning_rate": 2.4659761110036067e-06,
      "loss": 0.084,
      "step": 29355
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.697285688473299,
      "learning_rate": 2.461011032115315e-06,
      "loss": 0.0834,
      "step": 29360
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.792383437010644,
      "learning_rate": 2.4560506979203037e-06,
      "loss": 0.0916,
      "step": 29365
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.7460563190654858,
      "learning_rate": 2.4510951094627587e-06,
      "loss": 0.1296,
      "step": 29370
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.8121431572883597,
      "learning_rate": 2.4461442677858966e-06,
      "loss": 0.1557,
      "step": 29375
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.8172526067249357,
      "learning_rate": 2.441198173931911e-06,
      "loss": 0.1044,
      "step": 29380
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.7242549091305366,
      "learning_rate": 2.4362568289420103e-06,
      "loss": 0.1015,
      "step": 29385
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.962713573667299,
      "learning_rate": 2.431320233856402e-06,
      "loss": 0.1008,
      "step": 29390
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.627634598651306,
      "learning_rate": 2.426388389714293e-06,
      "loss": 0.0898,
      "step": 29395
    },
    {
      "epoch": 3.43,
      "grad_norm": 0.7161684131859233,
      "learning_rate": 2.421461297553887e-06,
      "loss": 0.1115,
      "step": 29400
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.6172124523067426,
      "learning_rate": 2.4165389584123786e-06,
      "loss": 0.1103,
      "step": 29405
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.7832300611814855,
      "learning_rate": 2.4116213733259813e-06,
      "loss": 0.1029,
      "step": 29410
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.6723344207683406,
      "learning_rate": 2.406708543329897e-06,
      "loss": 0.1063,
      "step": 29415
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.669013782229076,
      "learning_rate": 2.401800469458326e-06,
      "loss": 0.1289,
      "step": 29420
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.9427092079307101,
      "learning_rate": 2.396897152744476e-06,
      "loss": 0.1314,
      "step": 29425
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.886625223129027,
      "learning_rate": 2.3919985942205347e-06,
      "loss": 0.1271,
      "step": 29430
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.7342224755835731,
      "learning_rate": 2.3871047949177134e-06,
      "loss": 0.1081,
      "step": 29435
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.6425947907024737,
      "learning_rate": 2.3822157558661936e-06,
      "loss": 0.1078,
      "step": 29440
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.742880312146628,
      "learning_rate": 2.377331478095182e-06,
      "loss": 0.1013,
      "step": 29445
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.7553991079315303,
      "learning_rate": 2.37245196263286e-06,
      "loss": 0.0997,
      "step": 29450
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.7379032736592557,
      "learning_rate": 2.3675772105064246e-06,
      "loss": 0.1003,
      "step": 29455
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.7791272004835175,
      "learning_rate": 2.3627072227420555e-06,
      "loss": 0.1044,
      "step": 29460
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.7825039822147692,
      "learning_rate": 2.357842000364946e-06,
      "loss": 0.1085,
      "step": 29465
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.7846339138558028,
      "learning_rate": 2.3529815443992716e-06,
      "loss": 0.1428,
      "step": 29470
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.9272745116826714,
      "learning_rate": 2.3481258558681993e-06,
      "loss": 0.1009,
      "step": 29475
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.6641186732339446,
      "learning_rate": 2.3432749357939154e-06,
      "loss": 0.1272,
      "step": 29480
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.6713945118911927,
      "learning_rate": 2.3384287851975828e-06,
      "loss": 0.1016,
      "step": 29485
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.6932855509169203,
      "learning_rate": 2.3335874050993662e-06,
      "loss": 0.0812,
      "step": 29490
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.6408136630860655,
      "learning_rate": 2.328750796518439e-06,
      "loss": 0.1055,
      "step": 29495
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.6771248487102963,
      "learning_rate": 2.3239189604729456e-06,
      "loss": 0.1063,
      "step": 29500
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.7345077934824498,
      "learning_rate": 2.3190918979800384e-06,
      "loss": 0.1136,
      "step": 29505
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.6127066578573924,
      "learning_rate": 2.3142696100558666e-06,
      "loss": 0.1164,
      "step": 29510
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.6199968957723145,
      "learning_rate": 2.3094520977155724e-06,
      "loss": 0.0906,
      "step": 29515
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.6928113091544492,
      "learning_rate": 2.3046393619732988e-06,
      "loss": 0.1117,
      "step": 29520
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.7205811573603301,
      "learning_rate": 2.2998314038421647e-06,
      "loss": 0.0875,
      "step": 29525
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.8272049674631475,
      "learning_rate": 2.295028224334311e-06,
      "loss": 0.1228,
      "step": 29530
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.6431489298814284,
      "learning_rate": 2.2902298244608404e-06,
      "loss": 0.1092,
      "step": 29535
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.8655063556552018,
      "learning_rate": 2.2854362052318805e-06,
      "loss": 0.1011,
      "step": 29540
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.5948747205967501,
      "learning_rate": 2.28064736765653e-06,
      "loss": 0.0963,
      "step": 29545
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.7627693110926421,
      "learning_rate": 2.2758633127428902e-06,
      "loss": 0.1307,
      "step": 29550
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.8005795290374716,
      "learning_rate": 2.271084041498056e-06,
      "loss": 0.1125,
      "step": 29555
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.7562957756354073,
      "learning_rate": 2.2663095549281198e-06,
      "loss": 0.0894,
      "step": 29560
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.7367859080830766,
      "learning_rate": 2.2615398540381582e-06,
      "loss": 0.1403,
      "step": 29565
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.7087112407472577,
      "learning_rate": 2.256774939832235e-06,
      "loss": 0.0939,
      "step": 29570
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.8423252212658027,
      "learning_rate": 2.2520148133134188e-06,
      "loss": 0.1172,
      "step": 29575
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.753557258574696,
      "learning_rate": 2.2472594754837677e-06,
      "loss": 0.1306,
      "step": 29580
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.7164448453210157,
      "learning_rate": 2.2425089273443308e-06,
      "loss": 0.1154,
      "step": 29585
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.681736563524342,
      "learning_rate": 2.2377631698951517e-06,
      "loss": 0.1043,
      "step": 29590
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.8388401309111846,
      "learning_rate": 2.23302220413526e-06,
      "loss": 0.0977,
      "step": 29595
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.7027937256903803,
      "learning_rate": 2.228286031062671e-06,
      "loss": 0.1101,
      "step": 29600
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.7960660650406883,
      "learning_rate": 2.223554651674403e-06,
      "loss": 0.0855,
      "step": 29605
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.7798220483638502,
      "learning_rate": 2.218828066966466e-06,
      "loss": 0.1124,
      "step": 29610
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.7520684507516269,
      "learning_rate": 2.21410627793385e-06,
      "loss": 0.1127,
      "step": 29615
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.7982058092565384,
      "learning_rate": 2.2093892855705427e-06,
      "loss": 0.0905,
      "step": 29620
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.7402156315502857,
      "learning_rate": 2.2046770908695225e-06,
      "loss": 0.0936,
      "step": 29625
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.7553460326717047,
      "learning_rate": 2.199969694822751e-06,
      "loss": 0.1177,
      "step": 29630
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.7223576707840773,
      "learning_rate": 2.1952670984211916e-06,
      "loss": 0.0923,
      "step": 29635
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.9972742622300764,
      "learning_rate": 2.1905693026547824e-06,
      "loss": 0.1251,
      "step": 29640
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.6135068448064825,
      "learning_rate": 2.1858763085124606e-06,
      "loss": 0.0995,
      "step": 29645
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.8142846116876807,
      "learning_rate": 2.1811881169821525e-06,
      "loss": 0.1065,
      "step": 29650
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.6410774447950901,
      "learning_rate": 2.176504729050771e-06,
      "loss": 0.1286,
      "step": 29655
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.6528806528996806,
      "learning_rate": 2.1718261457042198e-06,
      "loss": 0.1027,
      "step": 29660
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.9261994985098858,
      "learning_rate": 2.1671523679273913e-06,
      "loss": 0.0942,
      "step": 29665
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.873180430554436,
      "learning_rate": 2.162483396704157e-06,
      "loss": 0.0828,
      "step": 29670
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.7506590214474583,
      "learning_rate": 2.1578192330173884e-06,
      "loss": 0.1154,
      "step": 29675
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.5714440020364331,
      "learning_rate": 2.153159877848945e-06,
      "loss": 0.0952,
      "step": 29680
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.6545279794856892,
      "learning_rate": 2.148505332179662e-06,
      "loss": 0.1016,
      "step": 29685
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.6759975362827372,
      "learning_rate": 2.143855596989375e-06,
      "loss": 0.1018,
      "step": 29690
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.8175998225899073,
      "learning_rate": 2.139210673256906e-06,
      "loss": 0.0752,
      "step": 29695
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.781744949817547,
      "learning_rate": 2.1345705619600505e-06,
      "loss": 0.1122,
      "step": 29700
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.6541858555385428,
      "learning_rate": 2.129935264075608e-06,
      "loss": 0.1195,
      "step": 29705
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.8150222126809816,
      "learning_rate": 2.1253047805793512e-06,
      "loss": 0.1123,
      "step": 29710
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.5947498118902488,
      "learning_rate": 2.1206791124460475e-06,
      "loss": 0.1115,
      "step": 29715
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.7376340811720784,
      "learning_rate": 2.1160582606494504e-06,
      "loss": 0.1145,
      "step": 29720
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.6914320295135671,
      "learning_rate": 2.1114422261623024e-06,
      "loss": 0.0882,
      "step": 29725
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.8912485210262087,
      "learning_rate": 2.1068310099563197e-06,
      "loss": 0.1109,
      "step": 29730
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.7760332751416525,
      "learning_rate": 2.10222461300221e-06,
      "loss": 0.1077,
      "step": 29735
    },
    {
      "epoch": 3.47,
      "grad_norm": 0.7159138146194893,
      "learning_rate": 2.0976230362696717e-06,
      "loss": 0.1135,
      "step": 29740
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.8504758711423386,
      "learning_rate": 2.0930262807273838e-06,
      "loss": 0.0995,
      "step": 29745
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.6697499728590433,
      "learning_rate": 2.0884343473430187e-06,
      "loss": 0.1078,
      "step": 29750
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.8488219778616212,
      "learning_rate": 2.083847237083214e-06,
      "loss": 0.1043,
      "step": 29755
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.6269326423906878,
      "learning_rate": 2.0792649509136176e-06,
      "loss": 0.1143,
      "step": 29760
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.5635992543444682,
      "learning_rate": 2.074687489798835e-06,
      "loss": 0.0879,
      "step": 29765
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.6632833663663561,
      "learning_rate": 2.070114854702479e-06,
      "loss": 0.1017,
      "step": 29770
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.8563487899051208,
      "learning_rate": 2.065547046587135e-06,
      "loss": 0.112,
      "step": 29775
    },
    {
      "epoch": 3.48,
      "grad_norm": 1.9769101649494136,
      "learning_rate": 2.060984066414373e-06,
      "loss": 0.1344,
      "step": 29780
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.6512013725430853,
      "learning_rate": 2.0564259151447495e-06,
      "loss": 0.1033,
      "step": 29785
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.7692582266715774,
      "learning_rate": 2.051872593737808e-06,
      "loss": 0.0995,
      "step": 29790
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.80599170768257,
      "learning_rate": 2.0473241031520595e-06,
      "loss": 0.1028,
      "step": 29795
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.6177769690351818,
      "learning_rate": 2.042780444345019e-06,
      "loss": 0.089,
      "step": 29800
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.6704388565558964,
      "learning_rate": 2.038241618273165e-06,
      "loss": 0.1069,
      "step": 29805
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.7145042613746281,
      "learning_rate": 2.033707625891973e-06,
      "loss": 0.1002,
      "step": 29810
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.7255153904857279,
      "learning_rate": 2.029178468155896e-06,
      "loss": 0.1113,
      "step": 29815
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.6002068071953629,
      "learning_rate": 2.024654146018373e-06,
      "loss": 0.1069,
      "step": 29820
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.6618288339805396,
      "learning_rate": 2.020134660431819e-06,
      "loss": 0.0954,
      "step": 29825
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.7243744673563712,
      "learning_rate": 2.015620012347627e-06,
      "loss": 0.1057,
      "step": 29830
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.6837794056184231,
      "learning_rate": 2.0111102027161795e-06,
      "loss": 0.128,
      "step": 29835
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.7147791612819262,
      "learning_rate": 2.006605232486841e-06,
      "loss": 0.112,
      "step": 29840
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.6344571573075815,
      "learning_rate": 2.0021051026079625e-06,
      "loss": 0.1153,
      "step": 29845
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.8699061730027201,
      "learning_rate": 1.9976098140268547e-06,
      "loss": 0.1049,
      "step": 29850
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.8577047826517152,
      "learning_rate": 1.993119367689833e-06,
      "loss": 0.1252,
      "step": 29855
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.8373719155975242,
      "learning_rate": 1.9886337645421748e-06,
      "loss": 0.0908,
      "step": 29860
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.81009924448923,
      "learning_rate": 1.9841530055281527e-06,
      "loss": 0.1044,
      "step": 29865
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.7653155678713628,
      "learning_rate": 1.979677091591017e-06,
      "loss": 0.1066,
      "step": 29870
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.8845095186235438,
      "learning_rate": 1.9752060236729847e-06,
      "loss": 0.1106,
      "step": 29875
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.6219240422414486,
      "learning_rate": 1.9707398027152657e-06,
      "loss": 0.0809,
      "step": 29880
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.6679866263327529,
      "learning_rate": 1.96627842965805e-06,
      "loss": 0.1104,
      "step": 29885
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.7040350693426078,
      "learning_rate": 1.961821905440503e-06,
      "loss": 0.113,
      "step": 29890
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.5605961104647257,
      "learning_rate": 1.9573702310007685e-06,
      "loss": 0.0878,
      "step": 29895
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.741879989123289,
      "learning_rate": 1.9529234072759666e-06,
      "loss": 0.1033,
      "step": 29900
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.7520960961749575,
      "learning_rate": 1.948481435202204e-06,
      "loss": 0.1207,
      "step": 29905
    },
    {
      "epoch": 3.49,
      "grad_norm": 0.6861962547627196,
      "learning_rate": 1.944044315714563e-06,
      "loss": 0.0865,
      "step": 29910
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.8830763845265285,
      "learning_rate": 1.9396120497471055e-06,
      "loss": 0.1127,
      "step": 29915
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.7228421517754152,
      "learning_rate": 1.9351846382328647e-06,
      "loss": 0.1066,
      "step": 29920
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.7289905994994464,
      "learning_rate": 1.9307620821038624e-06,
      "loss": 0.1057,
      "step": 29925
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.6549743339898696,
      "learning_rate": 1.926344382291084e-06,
      "loss": 0.1148,
      "step": 29930
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.6763036977054995,
      "learning_rate": 1.9219315397245112e-06,
      "loss": 0.0913,
      "step": 29935
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.729354571440642,
      "learning_rate": 1.917523555333092e-06,
      "loss": 0.0936,
      "step": 29940
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.7220866756089067,
      "learning_rate": 1.9131204300447485e-06,
      "loss": 0.0981,
      "step": 29945
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.641621699347034,
      "learning_rate": 1.908722164786389e-06,
      "loss": 0.0905,
      "step": 29950
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.829459591491851,
      "learning_rate": 1.904328760483895e-06,
      "loss": 0.1013,
      "step": 29955
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.8684300945979784,
      "learning_rate": 1.899940218062124e-06,
      "loss": 0.1315,
      "step": 29960
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.7534157393863317,
      "learning_rate": 1.8955565384449009e-06,
      "loss": 0.1049,
      "step": 29965
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.9170822053165137,
      "learning_rate": 1.891177722555046e-06,
      "loss": 0.1291,
      "step": 29970
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.7201914176973393,
      "learning_rate": 1.8868037713143438e-06,
      "loss": 0.0883,
      "step": 29975
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.5993128329341708,
      "learning_rate": 1.8824346856435554e-06,
      "loss": 0.0814,
      "step": 29980
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.6684869037525144,
      "learning_rate": 1.8780704664624254e-06,
      "loss": 0.1095,
      "step": 29985
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.6591917817028597,
      "learning_rate": 1.8737111146896636e-06,
      "loss": 0.1021,
      "step": 29990
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.8337658355263574,
      "learning_rate": 1.8693566312429523e-06,
      "loss": 0.1213,
      "step": 29995
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.7030770025677868,
      "learning_rate": 1.8650070170389638e-06,
      "loss": 0.0988,
      "step": 30000
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.6724310530097579,
      "learning_rate": 1.8606622729933327e-06,
      "loss": 0.0939,
      "step": 30005
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.601669461990693,
      "learning_rate": 1.8563224000206797e-06,
      "loss": 0.121,
      "step": 30010
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.6694623566725991,
      "learning_rate": 1.851987399034588e-06,
      "loss": 0.1158,
      "step": 30015
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.7579723088424573,
      "learning_rate": 1.8476572709476252e-06,
      "loss": 0.1136,
      "step": 30020
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.7066572789129003,
      "learning_rate": 1.843332016671323e-06,
      "loss": 0.1164,
      "step": 30025
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.6848351202689783,
      "learning_rate": 1.8390116371161975e-06,
      "loss": 0.1333,
      "step": 30030
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.6090232475076448,
      "learning_rate": 1.8346961331917268e-06,
      "loss": 0.0904,
      "step": 30035
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.8349273496802704,
      "learning_rate": 1.8303855058063785e-06,
      "loss": 0.1,
      "step": 30040
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.5877918752062457,
      "learning_rate": 1.8260797558675795e-06,
      "loss": 0.0795,
      "step": 30045
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.6970295666938696,
      "learning_rate": 1.8217788842817385e-06,
      "loss": 0.1142,
      "step": 30050
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.6864493678625406,
      "learning_rate": 1.8174828919542365e-06,
      "loss": 0.0817,
      "step": 30055
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.7567386165997011,
      "learning_rate": 1.8131917797894137e-06,
      "loss": 0.0951,
      "step": 30060
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.7009500549872775,
      "learning_rate": 1.808905548690601e-06,
      "loss": 0.1221,
      "step": 30065
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.6867681321454739,
      "learning_rate": 1.8046241995600955e-06,
      "loss": 0.0993,
      "step": 30070
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.8387130800187012,
      "learning_rate": 1.800347733299168e-06,
      "loss": 0.1073,
      "step": 30075
    },
    {
      "epoch": 3.51,
      "grad_norm": 0.7515601317918266,
      "learning_rate": 1.7960761508080598e-06,
      "loss": 0.1131,
      "step": 30080
    },
    {
      "epoch": 3.52,
      "grad_norm": 1.007349960605761,
      "learning_rate": 1.791809452985979e-06,
      "loss": 0.1212,
      "step": 30085
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.5772775531085205,
      "learning_rate": 1.7875476407311104e-06,
      "loss": 0.1129,
      "step": 30090
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.8278259993212991,
      "learning_rate": 1.7832907149406107e-06,
      "loss": 0.1233,
      "step": 30095
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.69576482445141,
      "learning_rate": 1.7790386765106136e-06,
      "loss": 0.122,
      "step": 30100
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.764394656390215,
      "learning_rate": 1.7747915263362087e-06,
      "loss": 0.1031,
      "step": 30105
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.7131820499632343,
      "learning_rate": 1.7705492653114692e-06,
      "loss": 0.127,
      "step": 30110
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.7916639532196048,
      "learning_rate": 1.7663118943294366e-06,
      "loss": 0.0958,
      "step": 30115
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.6555928940259859,
      "learning_rate": 1.7620794142821252e-06,
      "loss": 0.0971,
      "step": 30120
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.7408316574870435,
      "learning_rate": 1.7578518260605136e-06,
      "loss": 0.1075,
      "step": 30125
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.7308757022576423,
      "learning_rate": 1.753629130554546e-06,
      "loss": 0.111,
      "step": 30130
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.6917739341510492,
      "learning_rate": 1.74941132865315e-06,
      "loss": 0.09,
      "step": 30135
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.6396529175396755,
      "learning_rate": 1.745198421244218e-06,
      "loss": 0.0849,
      "step": 30140
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.7181317820142172,
      "learning_rate": 1.740990409214613e-06,
      "loss": 0.1148,
      "step": 30145
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.7552779098443219,
      "learning_rate": 1.736787293450165e-06,
      "loss": 0.1099,
      "step": 30150
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.6042706706691103,
      "learning_rate": 1.732589074835672e-06,
      "loss": 0.106,
      "step": 30155
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.7693145189900836,
      "learning_rate": 1.7283957542548967e-06,
      "loss": 0.1032,
      "step": 30160
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.5757751422567693,
      "learning_rate": 1.7242073325905855e-06,
      "loss": 0.0904,
      "step": 30165
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.7345532378064857,
      "learning_rate": 1.720023810724447e-06,
      "loss": 0.0947,
      "step": 30170
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.672432548764249,
      "learning_rate": 1.7158451895371492e-06,
      "loss": 0.1209,
      "step": 30175
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.78512066957067,
      "learning_rate": 1.711671469908338e-06,
      "loss": 0.1142,
      "step": 30180
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.753435965669234,
      "learning_rate": 1.7075026527166276e-06,
      "loss": 0.1242,
      "step": 30185
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.6875550180403542,
      "learning_rate": 1.7033387388395939e-06,
      "loss": 0.1147,
      "step": 30190
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.6858820948959818,
      "learning_rate": 1.699179729153791e-06,
      "loss": 0.1085,
      "step": 30195
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.8546619948360874,
      "learning_rate": 1.695025624534724e-06,
      "loss": 0.1319,
      "step": 30200
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.7589788955019657,
      "learning_rate": 1.6908764258568798e-06,
      "loss": 0.0927,
      "step": 30205
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.7109319974877252,
      "learning_rate": 1.6867321339937097e-06,
      "loss": 0.0883,
      "step": 30210
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.7844257336172835,
      "learning_rate": 1.6825927498176352e-06,
      "loss": 0.1026,
      "step": 30215
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.7270447619891282,
      "learning_rate": 1.6784582742000344e-06,
      "loss": 0.1074,
      "step": 30220
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.6296880993887636,
      "learning_rate": 1.6743287080112551e-06,
      "loss": 0.0765,
      "step": 30225
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.7570993723475119,
      "learning_rate": 1.6702040521206164e-06,
      "loss": 0.111,
      "step": 30230
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.6729880671579386,
      "learning_rate": 1.666084307396401e-06,
      "loss": 0.1044,
      "step": 30235
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.7117127035916897,
      "learning_rate": 1.6619694747058655e-06,
      "loss": 0.1186,
      "step": 30240
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.8451615844004793,
      "learning_rate": 1.6578595549152143e-06,
      "loss": 0.1144,
      "step": 30245
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.6799578625344473,
      "learning_rate": 1.6537545488896383e-06,
      "loss": 0.1051,
      "step": 30250
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.7528467138154912,
      "learning_rate": 1.6496544574932742e-06,
      "loss": 0.1656,
      "step": 30255
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.632563571183305,
      "learning_rate": 1.6455592815892429e-06,
      "loss": 0.0849,
      "step": 30260
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.7278141213811052,
      "learning_rate": 1.6414690220396212e-06,
      "loss": 0.0943,
      "step": 30265
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.6980445332990993,
      "learning_rate": 1.6373836797054454e-06,
      "loss": 0.0979,
      "step": 30270
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.7086109283789179,
      "learning_rate": 1.6333032554467247e-06,
      "loss": 0.1256,
      "step": 30275
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.6240411305109839,
      "learning_rate": 1.629227750122439e-06,
      "loss": 0.0865,
      "step": 30280
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.7338511459381485,
      "learning_rate": 1.6251571645905155e-06,
      "loss": 0.104,
      "step": 30285
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.5954368010784094,
      "learning_rate": 1.6210914997078608e-06,
      "loss": 0.0997,
      "step": 30290
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.7219874291456158,
      "learning_rate": 1.6170307563303345e-06,
      "loss": 0.1178,
      "step": 30295
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.663865683842347,
      "learning_rate": 1.6129749353127666e-06,
      "loss": 0.1089,
      "step": 30300
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.6370785721988429,
      "learning_rate": 1.6089240375089548e-06,
      "loss": 0.1013,
      "step": 30305
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.6977146089209333,
      "learning_rate": 1.604878063771656e-06,
      "loss": 0.0833,
      "step": 30310
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.728816747907338,
      "learning_rate": 1.600837014952586e-06,
      "loss": 0.135,
      "step": 30315
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.6861784533880604,
      "learning_rate": 1.5968008919024252e-06,
      "loss": 0.1003,
      "step": 30320
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.7388362126691856,
      "learning_rate": 1.592769695470822e-06,
      "loss": 0.1037,
      "step": 30325
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.6320539051673353,
      "learning_rate": 1.5887434265063867e-06,
      "loss": 0.1011,
      "step": 30330
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.7347120604634856,
      "learning_rate": 1.5847220858566963e-06,
      "loss": 0.106,
      "step": 30335
    },
    {
      "epoch": 3.54,
      "grad_norm": 0.7565692857938016,
      "learning_rate": 1.5807056743682768e-06,
      "loss": 0.0995,
      "step": 30340
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7274958811806206,
      "learning_rate": 1.5766941928866268e-06,
      "loss": 0.1176,
      "step": 30345
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7476399689038891,
      "learning_rate": 1.5726876422562097e-06,
      "loss": 0.099,
      "step": 30350
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7032509082437083,
      "learning_rate": 1.5686860233204426e-06,
      "loss": 0.1046,
      "step": 30355
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7185373977541134,
      "learning_rate": 1.5646893369217098e-06,
      "loss": 0.1116,
      "step": 30360
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.8439230053912808,
      "learning_rate": 1.5606975839013493e-06,
      "loss": 0.1347,
      "step": 30365
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.8656226453840884,
      "learning_rate": 1.5567107650996755e-06,
      "loss": 0.1486,
      "step": 30370
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7463729970352376,
      "learning_rate": 1.55272888135595e-06,
      "loss": 0.0896,
      "step": 30375
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.8349109123694637,
      "learning_rate": 1.5487519335084077e-06,
      "loss": 0.1267,
      "step": 30380
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.6105206625714796,
      "learning_rate": 1.5447799223942322e-06,
      "loss": 0.1269,
      "step": 30385
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7396446408486453,
      "learning_rate": 1.5408128488495738e-06,
      "loss": 0.1163,
      "step": 30390
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.6189592696357049,
      "learning_rate": 1.5368507137095395e-06,
      "loss": 0.103,
      "step": 30395
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.8032044614354412,
      "learning_rate": 1.5328935178082093e-06,
      "loss": 0.0941,
      "step": 30400
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.695375524556541,
      "learning_rate": 1.5289412619786087e-06,
      "loss": 0.1015,
      "step": 30405
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7058744521661271,
      "learning_rate": 1.5249939470527303e-06,
      "loss": 0.1198,
      "step": 30410
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.7093769818338022,
      "learning_rate": 1.521051573861526e-06,
      "loss": 0.1108,
      "step": 30415
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.6873656152446026,
      "learning_rate": 1.5171141432349017e-06,
      "loss": 0.1198,
      "step": 30420
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.8086205495645065,
      "learning_rate": 1.5131816560017332e-06,
      "loss": 0.1017,
      "step": 30425
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.622626190627697,
      "learning_rate": 1.5092541129898496e-06,
      "loss": 0.093,
      "step": 30430
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.8382484078300015,
      "learning_rate": 1.5053315150260372e-06,
      "loss": 0.1166,
      "step": 30435
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.7707883733747339,
      "learning_rate": 1.5014138629360464e-06,
      "loss": 0.1226,
      "step": 30440
    },
    {
      "epoch": 3.56,
      "grad_norm": 1.1690520415423127,
      "learning_rate": 1.4975011575445868e-06,
      "loss": 0.1357,
      "step": 30445
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.7405758833093162,
      "learning_rate": 1.4935933996753193e-06,
      "loss": 0.1058,
      "step": 30450
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.7929422386184488,
      "learning_rate": 1.4896905901508634e-06,
      "loss": 0.0947,
      "step": 30455
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.5315884543493301,
      "learning_rate": 1.4857927297928093e-06,
      "loss": 0.0993,
      "step": 30460
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.6672226288493984,
      "learning_rate": 1.481899819421695e-06,
      "loss": 0.1018,
      "step": 30465
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.7490504824315871,
      "learning_rate": 1.4780118598570176e-06,
      "loss": 0.0798,
      "step": 30470
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.8520425237413395,
      "learning_rate": 1.4741288519172392e-06,
      "loss": 0.1076,
      "step": 30475
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.9415960741560183,
      "learning_rate": 1.4702507964197698e-06,
      "loss": 0.1146,
      "step": 30480
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.854722336985771,
      "learning_rate": 1.4663776941809758e-06,
      "loss": 0.1306,
      "step": 30485
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.6425322403902617,
      "learning_rate": 1.4625095460161886e-06,
      "loss": 0.085,
      "step": 30490
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.6880677266207403,
      "learning_rate": 1.4586463527396953e-06,
      "loss": 0.0874,
      "step": 30495
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.760682069787075,
      "learning_rate": 1.454788115164743e-06,
      "loss": 0.1349,
      "step": 30500
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.7350827242572235,
      "learning_rate": 1.4509348341035234e-06,
      "loss": 0.1473,
      "step": 30505
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.5100135971975714,
      "learning_rate": 1.4470865103671988e-06,
      "loss": 0.0787,
      "step": 30510
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.6221786234969559,
      "learning_rate": 1.4432431447658768e-06,
      "loss": 0.1011,
      "step": 30515
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.7495511918024548,
      "learning_rate": 1.4394047381086296e-06,
      "loss": 0.1126,
      "step": 30520
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.7170330828502082,
      "learning_rate": 1.4355712912034747e-06,
      "loss": 0.0987,
      "step": 30525
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.8918494796035314,
      "learning_rate": 1.4317428048574028e-06,
      "loss": 0.1231,
      "step": 30530
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.6702582202625147,
      "learning_rate": 1.4279192798763442e-06,
      "loss": 0.1157,
      "step": 30535
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.7101520738050748,
      "learning_rate": 1.4241007170651966e-06,
      "loss": 0.1013,
      "step": 30540
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.749895158624273,
      "learning_rate": 1.4202871172277976e-06,
      "loss": 0.1,
      "step": 30545
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.7012710684574495,
      "learning_rate": 1.4164784811669634e-06,
      "loss": 0.0837,
      "step": 30550
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.7544510955121172,
      "learning_rate": 1.4126748096844389e-06,
      "loss": 0.1127,
      "step": 30555
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.799453078811553,
      "learning_rate": 1.4088761035809418e-06,
      "loss": 0.1139,
      "step": 30560
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.671300088548873,
      "learning_rate": 1.4050823636561405e-06,
      "loss": 0.0998,
      "step": 30565
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.6814785351247556,
      "learning_rate": 1.4012935907086604e-06,
      "loss": 0.1028,
      "step": 30570
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.6033503820126753,
      "learning_rate": 1.3975097855360719e-06,
      "loss": 0.1141,
      "step": 30575
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.8607250181611813,
      "learning_rate": 1.3937309489349098e-06,
      "loss": 0.1136,
      "step": 30580
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.7260714249231374,
      "learning_rate": 1.3899570817006518e-06,
      "loss": 0.0724,
      "step": 30585
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.0134082384505552,
      "learning_rate": 1.3861881846277457e-06,
      "loss": 0.1057,
      "step": 30590
    },
    {
      "epoch": 3.57,
      "grad_norm": 0.6903353111328807,
      "learning_rate": 1.3824242585095765e-06,
      "loss": 0.0904,
      "step": 30595
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.6479093169972459,
      "learning_rate": 1.3786653041384934e-06,
      "loss": 0.1253,
      "step": 30600
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.706618052323915,
      "learning_rate": 1.374911322305794e-06,
      "loss": 0.0928,
      "step": 30605
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.7730806081105609,
      "learning_rate": 1.3711623138017382e-06,
      "loss": 0.1093,
      "step": 30610
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.8110782556336296,
      "learning_rate": 1.3674182794155221e-06,
      "loss": 0.1241,
      "step": 30615
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.7101880388993977,
      "learning_rate": 1.3636792199353072e-06,
      "loss": 0.134,
      "step": 30620
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.7609645809319204,
      "learning_rate": 1.3599451361482023e-06,
      "loss": 0.0994,
      "step": 30625
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.7251122545405657,
      "learning_rate": 1.3562160288402765e-06,
      "loss": 0.104,
      "step": 30630
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.701560935746696,
      "learning_rate": 1.352491898796543e-06,
      "loss": 0.1069,
      "step": 30635
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.6962299398585544,
      "learning_rate": 1.348772746800972e-06,
      "loss": 0.0824,
      "step": 30640
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.6115203717305909,
      "learning_rate": 1.3450585736364845e-06,
      "loss": 0.1069,
      "step": 30645
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.7932417908654995,
      "learning_rate": 1.341349380084947e-06,
      "loss": 0.0944,
      "step": 30650
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.7533833079263604,
      "learning_rate": 1.3376451669271843e-06,
      "loss": 0.1146,
      "step": 30655
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.8226205256475028,
      "learning_rate": 1.3339459349429817e-06,
      "loss": 0.0796,
      "step": 30660
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.9158840146963627,
      "learning_rate": 1.3302516849110575e-06,
      "loss": 0.1128,
      "step": 30665
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.9103873327014819,
      "learning_rate": 1.3265624176090896e-06,
      "loss": 0.1097,
      "step": 30670
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.9007267389169074,
      "learning_rate": 1.3228781338137154e-06,
      "loss": 0.1209,
      "step": 30675
    },
    {
      "epoch": 3.58,
      "grad_norm": 0.6968928320298946,
      "learning_rate": 1.3191988343005057e-06,
      "loss": 0.1139,
      "step": 30680
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7790285046756766,
      "learning_rate": 1.3155245198439998e-06,
      "loss": 0.1107,
      "step": 30685
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.8161993660905291,
      "learning_rate": 1.3118551912176702e-06,
      "loss": 0.1258,
      "step": 30690
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7584225985260754,
      "learning_rate": 1.3081908491939548e-06,
      "loss": 0.0953,
      "step": 30695
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.6214847642514225,
      "learning_rate": 1.3045314945442365e-06,
      "loss": 0.1364,
      "step": 30700
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7803490102127154,
      "learning_rate": 1.3008771280388493e-06,
      "loss": 0.0973,
      "step": 30705
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.8423252221511995,
      "learning_rate": 1.2972277504470747e-06,
      "loss": 0.1052,
      "step": 30710
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.8632699019500358,
      "learning_rate": 1.2935833625371402e-06,
      "loss": 0.1539,
      "step": 30715
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.6369688793359027,
      "learning_rate": 1.289943965076229e-06,
      "loss": 0.1059,
      "step": 30720
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7977223320859471,
      "learning_rate": 1.286309558830473e-06,
      "loss": 0.1083,
      "step": 30725
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7072128610398377,
      "learning_rate": 1.2826801445649572e-06,
      "loss": 0.0926,
      "step": 30730
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.8339533938867757,
      "learning_rate": 1.2790557230437067e-06,
      "loss": 0.1155,
      "step": 30735
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7905407541306227,
      "learning_rate": 1.2754362950297056e-06,
      "loss": 0.1047,
      "step": 30740
    },
    {
      "epoch": 3.59,
      "grad_norm": 1.0314729179383137,
      "learning_rate": 1.271821861284872e-06,
      "loss": 0.1384,
      "step": 30745
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7513239790318796,
      "learning_rate": 1.2682124225700864e-06,
      "loss": 0.1019,
      "step": 30750
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.8367626738826848,
      "learning_rate": 1.2646079796451799e-06,
      "loss": 0.1216,
      "step": 30755
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7439746037629938,
      "learning_rate": 1.2610085332689148e-06,
      "loss": 0.081,
      "step": 30760
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.6648709530113841,
      "learning_rate": 1.2574140841990183e-06,
      "loss": 0.0994,
      "step": 30765
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.6738018534951443,
      "learning_rate": 1.2538246331921633e-06,
      "loss": 0.0976,
      "step": 30770
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.7370853713720266,
      "learning_rate": 1.250240181003956e-06,
      "loss": 0.1048,
      "step": 30775
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.7794566942866542,
      "learning_rate": 1.2466607283889735e-06,
      "loss": 0.1225,
      "step": 30780
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.7887026521849617,
      "learning_rate": 1.2430862761007162e-06,
      "loss": 0.1136,
      "step": 30785
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.8797796784147511,
      "learning_rate": 1.239516824891651e-06,
      "loss": 0.1018,
      "step": 30790
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.7385919099057635,
      "learning_rate": 1.235952375513183e-06,
      "loss": 0.1094,
      "step": 30795
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.7626768551121854,
      "learning_rate": 1.23239292871567e-06,
      "loss": 0.0979,
      "step": 30800
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.7017793073840625,
      "learning_rate": 1.228838485248407e-06,
      "loss": 0.0754,
      "step": 30805
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.6892827610660984,
      "learning_rate": 1.2252890458596456e-06,
      "loss": 0.1122,
      "step": 30810
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.7292032464870041,
      "learning_rate": 1.2217446112965742e-06,
      "loss": 0.1065,
      "step": 30815
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.8284638103765318,
      "learning_rate": 1.2182051823053403e-06,
      "loss": 0.1117,
      "step": 30820
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.852184232673036,
      "learning_rate": 1.2146707596310287e-06,
      "loss": 0.1252,
      "step": 30825
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.8685632630499958,
      "learning_rate": 1.2111413440176689e-06,
      "loss": 0.1355,
      "step": 30830
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.740040692250554,
      "learning_rate": 1.2076169362082417e-06,
      "loss": 0.1038,
      "step": 30835
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.6154452377136472,
      "learning_rate": 1.204097536944679e-06,
      "loss": 0.0991,
      "step": 30840
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.8404543237894199,
      "learning_rate": 1.2005831469678375e-06,
      "loss": 0.1006,
      "step": 30845
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.6151444357096342,
      "learning_rate": 1.197073767017548e-06,
      "loss": 0.12,
      "step": 30850
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.7415547058952063,
      "learning_rate": 1.1935693978325586e-06,
      "loss": 0.0957,
      "step": 30855
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.7183084712073893,
      "learning_rate": 1.1900700401505815e-06,
      "loss": 0.1165,
      "step": 30860
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.6758753993057471,
      "learning_rate": 1.1865756947082668e-06,
      "loss": 0.0821,
      "step": 30865
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.788278272907019,
      "learning_rate": 1.1830863622412174e-06,
      "loss": 0.1298,
      "step": 30870
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.6406636438402642,
      "learning_rate": 1.1796020434839684e-06,
      "loss": 0.1086,
      "step": 30875
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.7689486724984816,
      "learning_rate": 1.176122739170002e-06,
      "loss": 0.0959,
      "step": 30880
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.8118827275917105,
      "learning_rate": 1.1726484500317519e-06,
      "loss": 0.1128,
      "step": 30885
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.6521569657206522,
      "learning_rate": 1.1691791768005944e-06,
      "loss": 0.0924,
      "step": 30890
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.7462813556491096,
      "learning_rate": 1.165714920206848e-06,
      "loss": 0.1038,
      "step": 30895
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.7121505231635527,
      "learning_rate": 1.1622556809797709e-06,
      "loss": 0.1019,
      "step": 30900
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.6794821530992015,
      "learning_rate": 1.158801459847575e-06,
      "loss": 0.1053,
      "step": 30905
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.5513109753605802,
      "learning_rate": 1.155352257537401e-06,
      "loss": 0.0969,
      "step": 30910
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.752400586844115,
      "learning_rate": 1.1519080747753485e-06,
      "loss": 0.0939,
      "step": 30915
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.7123151205222714,
      "learning_rate": 1.1484689122864572e-06,
      "loss": 0.0925,
      "step": 30920
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.7062978269609498,
      "learning_rate": 1.145034770794698e-06,
      "loss": 0.1241,
      "step": 30925
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.7806735515865889,
      "learning_rate": 1.1416056510230006e-06,
      "loss": 0.103,
      "step": 30930
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.8264858524728509,
      "learning_rate": 1.1381815536932294e-06,
      "loss": 0.1148,
      "step": 30935
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.6920881588928678,
      "learning_rate": 1.1347624795261913e-06,
      "loss": 0.1,
      "step": 30940
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.7409309185540295,
      "learning_rate": 1.1313484292416355e-06,
      "loss": 0.1208,
      "step": 30945
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.6981410730097878,
      "learning_rate": 1.127939403558259e-06,
      "loss": 0.095,
      "step": 30950
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.706329616500167,
      "learning_rate": 1.124535403193694e-06,
      "loss": 0.1085,
      "step": 30955
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.5929923159217567,
      "learning_rate": 1.121136428864522e-06,
      "loss": 0.0813,
      "step": 30960
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.8325646783131384,
      "learning_rate": 1.117742481286263e-06,
      "loss": 0.0972,
      "step": 30965
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.6234694109559572,
      "learning_rate": 1.1143535611733756e-06,
      "loss": 0.1117,
      "step": 30970
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.7597598893184861,
      "learning_rate": 1.1109696692392618e-06,
      "loss": 0.1132,
      "step": 30975
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.776733663886879,
      "learning_rate": 1.107590806196268e-06,
      "loss": 0.0935,
      "step": 30980
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.7362651475659323,
      "learning_rate": 1.1042169727556784e-06,
      "loss": 0.1042,
      "step": 30985
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.7399682046729068,
      "learning_rate": 1.1008481696277273e-06,
      "loss": 0.1365,
      "step": 30990
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.6785790904653346,
      "learning_rate": 1.0974843975215755e-06,
      "loss": 0.0971,
      "step": 30995
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.6813358393585447,
      "learning_rate": 1.0941256571453395e-06,
      "loss": 0.1009,
      "step": 31000
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.7985176997914822,
      "learning_rate": 1.0907719492060598e-06,
      "loss": 0.1002,
      "step": 31005
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.7796320693181589,
      "learning_rate": 1.0874232744097324e-06,
      "loss": 0.1017,
      "step": 31010
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.7565294050199212,
      "learning_rate": 1.0840796334612908e-06,
      "loss": 0.0981,
      "step": 31015
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.9640351906112309,
      "learning_rate": 1.0807410270645995e-06,
      "loss": 0.1374,
      "step": 31020
    },
    {
      "epoch": 3.62,
      "grad_norm": 0.7208860018110056,
      "learning_rate": 1.0774074559224745e-06,
      "loss": 0.1014,
      "step": 31025
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.7140557847340524,
      "learning_rate": 1.0740789207366681e-06,
      "loss": 0.0931,
      "step": 31030
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.8788224816679496,
      "learning_rate": 1.0707554222078753e-06,
      "loss": 0.0992,
      "step": 31035
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.7157146461960749,
      "learning_rate": 1.0674369610357226e-06,
      "loss": 0.0951,
      "step": 31040
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.7104996632753755,
      "learning_rate": 1.0641235379187765e-06,
      "loss": 0.0758,
      "step": 31045
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.7603019890925156,
      "learning_rate": 1.0608151535545536e-06,
      "loss": 0.129,
      "step": 31050
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.8831543879602979,
      "learning_rate": 1.0575118086394998e-06,
      "loss": 0.1026,
      "step": 31055
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.6477463640915135,
      "learning_rate": 1.0542135038690115e-06,
      "loss": 0.0931,
      "step": 31060
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.6321805796937292,
      "learning_rate": 1.0509202399374057e-06,
      "loss": 0.1031,
      "step": 31065
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.7702906780146669,
      "learning_rate": 1.0476320175379584e-06,
      "loss": 0.1059,
      "step": 31070
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.5245198782706144,
      "learning_rate": 1.0443488373628657e-06,
      "loss": 0.0924,
      "step": 31075
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.7050140212425805,
      "learning_rate": 1.0410707001032778e-06,
      "loss": 0.0896,
      "step": 31080
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.9136205760240936,
      "learning_rate": 1.0377976064492784e-06,
      "loss": 0.1036,
      "step": 31085
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.753894979494184,
      "learning_rate": 1.0345295570898833e-06,
      "loss": 0.1029,
      "step": 31090
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.880739181097662,
      "learning_rate": 1.0312665527130505e-06,
      "loss": 0.1206,
      "step": 31095
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.6196072565620092,
      "learning_rate": 1.0280085940056834e-06,
      "loss": 0.1209,
      "step": 31100
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.7404682146724354,
      "learning_rate": 1.0247556816536108e-06,
      "loss": 0.121,
      "step": 31105
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.674463419867539,
      "learning_rate": 1.0215078163416019e-06,
      "loss": 0.1032,
      "step": 31110
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.7085558959853813,
      "learning_rate": 1.0182649987533737e-06,
      "loss": 0.1305,
      "step": 31115
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.7633816213228541,
      "learning_rate": 1.0150272295715662e-06,
      "loss": 0.1187,
      "step": 31120
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.8208019223665899,
      "learning_rate": 1.0117945094777703e-06,
      "loss": 0.0906,
      "step": 31125
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.5707036933825403,
      "learning_rate": 1.0085668391525056e-06,
      "loss": 0.0876,
      "step": 31130
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.8106112387888617,
      "learning_rate": 1.005344219275231e-06,
      "loss": 0.1268,
      "step": 31135
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.6529543234238515,
      "learning_rate": 1.0021266505243375e-06,
      "loss": 0.1105,
      "step": 31140
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.7479011668431952,
      "learning_rate": 9.989141335771585e-07,
      "loss": 0.0945,
      "step": 31145
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.7200662064724209,
      "learning_rate": 9.957066691099665e-07,
      "loss": 0.1242,
      "step": 31150
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.7598052155566435,
      "learning_rate": 9.925042577979632e-07,
      "loss": 0.1062,
      "step": 31155
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.7696118458374727,
      "learning_rate": 9.893069003152898e-07,
      "loss": 0.1092,
      "step": 31160
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.651026528529035,
      "learning_rate": 9.86114597335025e-07,
      "loss": 0.1112,
      "step": 31165
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.6929494746884063,
      "learning_rate": 9.829273495291808e-07,
      "loss": 0.108,
      "step": 31170
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.6084206607832694,
      "learning_rate": 9.797451575687067e-07,
      "loss": 0.0921,
      "step": 31175
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.8976307820162673,
      "learning_rate": 9.765680221234836e-07,
      "loss": 0.1069,
      "step": 31180
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.709830016301572,
      "learning_rate": 9.733959438623374e-07,
      "loss": 0.1144,
      "step": 31185
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.7192244443408814,
      "learning_rate": 9.702289234530204e-07,
      "loss": 0.1064,
      "step": 31190
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.7429228989070715,
      "learning_rate": 9.670669615622297e-07,
      "loss": 0.0999,
      "step": 31195
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.930482859875252,
      "learning_rate": 9.639100588555855e-07,
      "loss": 0.1034,
      "step": 31200
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.7764541793833177,
      "learning_rate": 9.60758215997648e-07,
      "loss": 0.1011,
      "step": 31205
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.740458995237549,
      "learning_rate": 9.576114336519115e-07,
      "loss": 0.0846,
      "step": 31210
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.7112168281038594,
      "learning_rate": 9.544697124808127e-07,
      "loss": 0.0967,
      "step": 31215
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.7646635169320589,
      "learning_rate": 9.51333053145717e-07,
      "loss": 0.1114,
      "step": 31220
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.8011725676063679,
      "learning_rate": 9.482014563069214e-07,
      "loss": 0.0872,
      "step": 31225
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.741812880068649,
      "learning_rate": 9.450749226236594e-07,
      "loss": 0.1101,
      "step": 31230
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.6852096825791646,
      "learning_rate": 9.419534527540991e-07,
      "loss": 0.0986,
      "step": 31235
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.7256183959011522,
      "learning_rate": 9.388370473553398e-07,
      "loss": 0.0835,
      "step": 31240
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.837669727743677,
      "learning_rate": 9.357257070834264e-07,
      "loss": 0.1152,
      "step": 31245
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.4832255391006528,
      "learning_rate": 9.326194325933179e-07,
      "loss": 0.1037,
      "step": 31250
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.5585812776062311,
      "learning_rate": 9.29518224538925e-07,
      "loss": 0.0975,
      "step": 31255
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.5704524566223363,
      "learning_rate": 9.264220835730808e-07,
      "loss": 0.1043,
      "step": 31260
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.6979544046940148,
      "learning_rate": 9.233310103475612e-07,
      "loss": 0.1074,
      "step": 31265
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.6515405559149345,
      "learning_rate": 9.202450055130623e-07,
      "loss": 0.099,
      "step": 31270
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.7728655197938917,
      "learning_rate": 9.17164069719223e-07,
      "loss": 0.0949,
      "step": 31275
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.6407515661204616,
      "learning_rate": 9.140882036146103e-07,
      "loss": 0.0668,
      "step": 31280
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.7749618913566683,
      "learning_rate": 9.110174078467315e-07,
      "loss": 0.1059,
      "step": 31285
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.7454509444142788,
      "learning_rate": 9.079516830620166e-07,
      "loss": 0.1173,
      "step": 31290
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.7000106844419353,
      "learning_rate": 9.048910299058383e-07,
      "loss": 0.1052,
      "step": 31295
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.7239052084244525,
      "learning_rate": 9.018354490224923e-07,
      "loss": 0.1181,
      "step": 31300
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.7089461885977787,
      "learning_rate": 8.987849410552085e-07,
      "loss": 0.1136,
      "step": 31305
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.8107318611294542,
      "learning_rate": 8.95739506646151e-07,
      "loss": 0.1004,
      "step": 31310
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.6556299897288846,
      "learning_rate": 8.926991464364209e-07,
      "loss": 0.087,
      "step": 31315
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.6825216426856706,
      "learning_rate": 8.896638610660396e-07,
      "loss": 0.0992,
      "step": 31320
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.6728781173135941,
      "learning_rate": 8.866336511739709e-07,
      "loss": 0.1033,
      "step": 31325
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.7835423788270992,
      "learning_rate": 8.836085173981046e-07,
      "loss": 0.1101,
      "step": 31330
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.5644290505735234,
      "learning_rate": 8.805884603752618e-07,
      "loss": 0.1095,
      "step": 31335
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.6575360225949405,
      "learning_rate": 8.775734807412006e-07,
      "loss": 0.0892,
      "step": 31340
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.6739638964744041,
      "learning_rate": 8.745635791305967e-07,
      "loss": 0.0839,
      "step": 31345
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.5604147421155383,
      "learning_rate": 8.715587561770739e-07,
      "loss": 0.0873,
      "step": 31350
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.5821479040472021,
      "learning_rate": 8.685590125131732e-07,
      "loss": 0.093,
      "step": 31355
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.7177866912022105,
      "learning_rate": 8.655643487703813e-07,
      "loss": 0.117,
      "step": 31360
    },
    {
      "epoch": 3.66,
      "grad_norm": 0.515857735156103,
      "learning_rate": 8.625747655790994e-07,
      "loss": 0.104,
      "step": 31365
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.7684893506185952,
      "learning_rate": 8.595902635686631e-07,
      "loss": 0.0877,
      "step": 31370
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.6116212074920049,
      "learning_rate": 8.566108433673448e-07,
      "loss": 0.0924,
      "step": 31375
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.6616138923328854,
      "learning_rate": 8.536365056023454e-07,
      "loss": 0.113,
      "step": 31380
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.7857435117126611,
      "learning_rate": 8.506672508997948e-07,
      "loss": 0.1145,
      "step": 31385
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.8801971501001228,
      "learning_rate": 8.477030798847485e-07,
      "loss": 0.0844,
      "step": 31390
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.6597655440928749,
      "learning_rate": 8.44743993181199e-07,
      "loss": 0.1184,
      "step": 31395
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.7267353538771502,
      "learning_rate": 8.417899914120591e-07,
      "loss": 0.0907,
      "step": 31400
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.8222255970681178,
      "learning_rate": 8.388410751991843e-07,
      "loss": 0.0977,
      "step": 31405
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.6174506661697344,
      "learning_rate": 8.358972451633501e-07,
      "loss": 0.1041,
      "step": 31410
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.6644366043972187,
      "learning_rate": 8.329585019242609e-07,
      "loss": 0.0994,
      "step": 31415
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.8194103486338397,
      "learning_rate": 8.300248461005521e-07,
      "loss": 0.1171,
      "step": 31420
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.6706162633012751,
      "learning_rate": 8.270962783097968e-07,
      "loss": 0.0892,
      "step": 31425
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.748968476989497,
      "learning_rate": 8.241727991684823e-07,
      "loss": 0.1021,
      "step": 31430
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.6647751647930565,
      "learning_rate": 8.212544092920354e-07,
      "loss": 0.1087,
      "step": 31435
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.7662992417188828,
      "learning_rate": 8.18341109294804e-07,
      "loss": 0.1357,
      "step": 31440
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.721200236585321,
      "learning_rate": 8.154328997900668e-07,
      "loss": 0.0932,
      "step": 31445
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.7380514291941953,
      "learning_rate": 8.125297813900396e-07,
      "loss": 0.0895,
      "step": 31450
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.7081270897732465,
      "learning_rate": 8.096317547058557e-07,
      "loss": 0.1009,
      "step": 31455
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.8165014134913714,
      "learning_rate": 8.067388203475801e-07,
      "loss": 0.0856,
      "step": 31460
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.8569513419803297,
      "learning_rate": 8.038509789242005e-07,
      "loss": 0.1236,
      "step": 31465
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.8692520356919332,
      "learning_rate": 8.009682310436445e-07,
      "loss": 0.1087,
      "step": 31470
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.6750348123895584,
      "learning_rate": 7.980905773127572e-07,
      "loss": 0.0991,
      "step": 31475
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.8982958265958888,
      "learning_rate": 7.952180183373209e-07,
      "loss": 0.0977,
      "step": 31480
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.6588766502930193,
      "learning_rate": 7.923505547220322e-07,
      "loss": 0.0861,
      "step": 31485
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.5565799674616073,
      "learning_rate": 7.894881870705251e-07,
      "loss": 0.092,
      "step": 31490
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.7884827987896651,
      "learning_rate": 7.866309159853591e-07,
      "loss": 0.1012,
      "step": 31495
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.6243882843330959,
      "learning_rate": 7.83778742068017e-07,
      "loss": 0.0924,
      "step": 31500
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.7287092352528951,
      "learning_rate": 7.809316659189131e-07,
      "loss": 0.1041,
      "step": 31505
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.741149182688481,
      "learning_rate": 7.780896881373872e-07,
      "loss": 0.102,
      "step": 31510
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.6576854112107415,
      "learning_rate": 7.752528093217026e-07,
      "loss": 0.0977,
      "step": 31515
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.8299192735893729,
      "learning_rate": 7.724210300690538e-07,
      "loss": 0.1001,
      "step": 31520
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.7424029451038306,
      "learning_rate": 7.695943509755609e-07,
      "loss": 0.1225,
      "step": 31525
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.7973278921384976,
      "learning_rate": 7.667727726362705e-07,
      "loss": 0.1031,
      "step": 31530
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.7818614369903614,
      "learning_rate": 7.639562956451491e-07,
      "loss": 0.1352,
      "step": 31535
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.6567330468296246,
      "learning_rate": 7.611449205950999e-07,
      "loss": 0.0998,
      "step": 31540
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.6997190924182812,
      "learning_rate": 7.583386480779442e-07,
      "loss": 0.0955,
      "step": 31545
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7549684899034701,
      "learning_rate": 7.555374786844344e-07,
      "loss": 0.1239,
      "step": 31550
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7230958954969856,
      "learning_rate": 7.527414130042403e-07,
      "loss": 0.12,
      "step": 31555
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.6872266124706731,
      "learning_rate": 7.49950451625972e-07,
      "loss": 0.1011,
      "step": 31560
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.8932201897522624,
      "learning_rate": 7.471645951371481e-07,
      "loss": 0.0991,
      "step": 31565
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7296096098265074,
      "learning_rate": 7.443838441242218e-07,
      "loss": 0.1152,
      "step": 31570
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7188277105664748,
      "learning_rate": 7.416081991725748e-07,
      "loss": 0.0987,
      "step": 31575
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7514901226980588,
      "learning_rate": 7.388376608665037e-07,
      "loss": 0.0982,
      "step": 31580
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.6221105917830667,
      "learning_rate": 7.36072229789242e-07,
      "loss": 0.0934,
      "step": 31585
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7605546615709045,
      "learning_rate": 7.333119065229377e-07,
      "loss": 0.0985,
      "step": 31590
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7088352797716291,
      "learning_rate": 7.305566916486678e-07,
      "loss": 0.1231,
      "step": 31595
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.6991949017852305,
      "learning_rate": 7.27806585746435e-07,
      "loss": 0.0995,
      "step": 31600
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.9342096300297633,
      "learning_rate": 7.250615893951651e-07,
      "loss": 0.1018,
      "step": 31605
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7993146714026558,
      "learning_rate": 7.22321703172707e-07,
      "loss": 0.1312,
      "step": 31610
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.9210544337818863,
      "learning_rate": 7.195869276558381e-07,
      "loss": 0.1313,
      "step": 31615
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7288024236976424,
      "learning_rate": 7.168572634202592e-07,
      "loss": 0.0798,
      "step": 31620
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.7668314603172326,
      "learning_rate": 7.141327110405882e-07,
      "loss": 0.0948,
      "step": 31625
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.6733367562177196,
      "learning_rate": 7.11413271090372e-07,
      "loss": 0.0868,
      "step": 31630
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.794689203530448,
      "learning_rate": 7.08698944142086e-07,
      "loss": 0.0939,
      "step": 31635
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.002349815481805,
      "learning_rate": 7.059897307671176e-07,
      "loss": 0.0908,
      "step": 31640
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.7176324922817938,
      "learning_rate": 7.032856315357938e-07,
      "loss": 0.0952,
      "step": 31645
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.967835312412662,
      "learning_rate": 7.005866470173484e-07,
      "loss": 0.1021,
      "step": 31650
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.6344779328448761,
      "learning_rate": 6.978927777799488e-07,
      "loss": 0.1005,
      "step": 31655
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.7548874723433963,
      "learning_rate": 6.952040243906804e-07,
      "loss": 0.0823,
      "step": 31660
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.8022687325241348,
      "learning_rate": 6.925203874155601e-07,
      "loss": 0.1017,
      "step": 31665
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.6464930726035433,
      "learning_rate": 6.898418674195106e-07,
      "loss": 0.0902,
      "step": 31670
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.6920613426884838,
      "learning_rate": 6.871684649663979e-07,
      "loss": 0.0837,
      "step": 31675
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.6195165995257804,
      "learning_rate": 6.845001806189994e-07,
      "loss": 0.1149,
      "step": 31680
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.7221992648829098,
      "learning_rate": 6.818370149390157e-07,
      "loss": 0.1098,
      "step": 31685
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.6597585306554329,
      "learning_rate": 6.791789684870681e-07,
      "loss": 0.0942,
      "step": 31690
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.6231016938905666,
      "learning_rate": 6.765260418227115e-07,
      "loss": 0.0784,
      "step": 31695
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.9862723356322458,
      "learning_rate": 6.738782355044049e-07,
      "loss": 0.1291,
      "step": 31700
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.7602967534371786,
      "learning_rate": 6.71235550089544e-07,
      "loss": 0.0976,
      "step": 31705
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.7055854913640962,
      "learning_rate": 6.685979861344422e-07,
      "loss": 0.1187,
      "step": 31710
    },
    {
      "epoch": 3.71,
      "grad_norm": 1.402533773406938,
      "learning_rate": 6.659655441943357e-07,
      "loss": 0.1497,
      "step": 31715
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7086175003531568,
      "learning_rate": 6.633382248233788e-07,
      "loss": 0.0856,
      "step": 31720
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7558513054407868,
      "learning_rate": 6.607160285746511e-07,
      "loss": 0.1102,
      "step": 31725
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.8424433960132696,
      "learning_rate": 6.580989560001472e-07,
      "loss": 0.1075,
      "step": 31730
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.71752907496591,
      "learning_rate": 6.554870076507985e-07,
      "loss": 0.0685,
      "step": 31735
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7754307549127908,
      "learning_rate": 6.528801840764348e-07,
      "loss": 0.1,
      "step": 31740
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.6130146678101475,
      "learning_rate": 6.50278485825831e-07,
      "loss": 0.0978,
      "step": 31745
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7311569641789891,
      "learning_rate": 6.476819134466627e-07,
      "loss": 0.1075,
      "step": 31750
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.8774468706453505,
      "learning_rate": 6.450904674855457e-07,
      "loss": 0.1079,
      "step": 31755
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.6682794645991317,
      "learning_rate": 6.425041484880017e-07,
      "loss": 0.1039,
      "step": 31760
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.599137427065413,
      "learning_rate": 6.399229569984733e-07,
      "loss": 0.0907,
      "step": 31765
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7520552828935392,
      "learning_rate": 6.37346893560331e-07,
      "loss": 0.0927,
      "step": 31770
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.897157424920915,
      "learning_rate": 6.347759587158664e-07,
      "loss": 0.1275,
      "step": 31775
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7211398917763695,
      "learning_rate": 6.322101530062851e-07,
      "loss": 0.0667,
      "step": 31780
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7601362803643443,
      "learning_rate": 6.296494769717192e-07,
      "loss": 0.1082,
      "step": 31785
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.7497018239373299,
      "learning_rate": 6.27093931151218e-07,
      "loss": 0.0946,
      "step": 31790
    },
    {
      "epoch": 3.71,
      "grad_norm": 0.788625277471891,
      "learning_rate": 6.245435160827456e-07,
      "loss": 0.1182,
      "step": 31795
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.7166816562564375,
      "learning_rate": 6.21998232303192e-07,
      "loss": 0.1055,
      "step": 31800
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.7289494593735245,
      "learning_rate": 6.194580803483729e-07,
      "loss": 0.1234,
      "step": 31805
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.6962758016708448,
      "learning_rate": 6.16923060753008e-07,
      "loss": 0.0874,
      "step": 31810
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.7105918346150539,
      "learning_rate": 6.143931740507508e-07,
      "loss": 0.1231,
      "step": 31815
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.6369738757187089,
      "learning_rate": 6.118684207741671e-07,
      "loss": 0.0857,
      "step": 31820
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.7475843133510652,
      "learning_rate": 6.093488014547454e-07,
      "loss": 0.095,
      "step": 31825
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.8053850669258545,
      "learning_rate": 6.068343166228924e-07,
      "loss": 0.1068,
      "step": 31830
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.675416035558085,
      "learning_rate": 6.043249668079287e-07,
      "loss": 0.1035,
      "step": 31835
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.6617614436126136,
      "learning_rate": 6.018207525381043e-07,
      "loss": 0.1003,
      "step": 31840
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.8191196748517496,
      "learning_rate": 5.993216743405805e-07,
      "loss": 0.0966,
      "step": 31845
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.778793782037446,
      "learning_rate": 5.968277327414396e-07,
      "loss": 0.1214,
      "step": 31850
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.9787277871929254,
      "learning_rate": 5.943389282656836e-07,
      "loss": 0.1245,
      "step": 31855
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.5940157034284285,
      "learning_rate": 5.918552614372269e-07,
      "loss": 0.0912,
      "step": 31860
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.7904920113539996,
      "learning_rate": 5.893767327789123e-07,
      "loss": 0.1017,
      "step": 31865
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.770788055170638,
      "learning_rate": 5.869033428124949e-07,
      "loss": 0.1005,
      "step": 31870
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.7748995685414579,
      "learning_rate": 5.844350920586494e-07,
      "loss": 0.1096,
      "step": 31875
    },
    {
      "epoch": 3.72,
      "grad_norm": 0.7501230595308362,
      "learning_rate": 5.81971981036969e-07,
      "loss": 0.1078,
      "step": 31880
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.729095357781294,
      "learning_rate": 5.795140102659635e-07,
      "loss": 0.0934,
      "step": 31885
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.6928355964505368,
      "learning_rate": 5.770611802630582e-07,
      "loss": 0.1274,
      "step": 31890
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.6009133560066421,
      "learning_rate": 5.746134915446038e-07,
      "loss": 0.0884,
      "step": 31895
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.766552774339302,
      "learning_rate": 5.72170944625866e-07,
      "loss": 0.1207,
      "step": 31900
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.7177780762936741,
      "learning_rate": 5.697335400210224e-07,
      "loss": 0.1001,
      "step": 31905
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.7226168733491731,
      "learning_rate": 5.673012782431736e-07,
      "loss": 0.1013,
      "step": 31910
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.6662601906310749,
      "learning_rate": 5.648741598043378e-07,
      "loss": 0.0877,
      "step": 31915
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.8466086073922562,
      "learning_rate": 5.624521852154452e-07,
      "loss": 0.121,
      "step": 31920
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.7274178543233253,
      "learning_rate": 5.600353549863518e-07,
      "loss": 0.101,
      "step": 31925
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.8168205245693064,
      "learning_rate": 5.576236696258174e-07,
      "loss": 0.1153,
      "step": 31930
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.7560861537779024,
      "learning_rate": 5.55217129641536e-07,
      "loss": 0.1237,
      "step": 31935
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.6426173752321725,
      "learning_rate": 5.528157355401048e-07,
      "loss": 0.1,
      "step": 31940
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.7950875421713026,
      "learning_rate": 5.504194878270447e-07,
      "loss": 0.1065,
      "step": 31945
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.7621534245877817,
      "learning_rate": 5.480283870067909e-07,
      "loss": 0.0896,
      "step": 31950
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.8779140943706185,
      "learning_rate": 5.456424335826938e-07,
      "loss": 0.1027,
      "step": 31955
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.7790190175684206,
      "learning_rate": 5.43261628057018e-07,
      "loss": 0.098,
      "step": 31960
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.5752335770154818,
      "learning_rate": 5.408859709309543e-07,
      "loss": 0.1155,
      "step": 31965
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.7171955695246557,
      "learning_rate": 5.38515462704603e-07,
      "loss": 0.0917,
      "step": 31970
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.6756969120921441,
      "learning_rate": 5.361501038769757e-07,
      "loss": 0.0935,
      "step": 31975
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.696641107065719,
      "learning_rate": 5.337898949460074e-07,
      "loss": 0.0882,
      "step": 31980
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.7664429656739975,
      "learning_rate": 5.314348364085536e-07,
      "loss": 0.0939,
      "step": 31985
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.7360982228554642,
      "learning_rate": 5.290849287603677e-07,
      "loss": 0.1476,
      "step": 31990
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.7287834259358721,
      "learning_rate": 5.267401724961374e-07,
      "loss": 0.1105,
      "step": 31995
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.7474724704349867,
      "learning_rate": 5.244005681094566e-07,
      "loss": 0.102,
      "step": 32000
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.5805564856057654,
      "learning_rate": 5.220661160928341e-07,
      "loss": 0.0928,
      "step": 32005
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.6680307774440886,
      "learning_rate": 5.197368169376965e-07,
      "loss": 0.0959,
      "step": 32010
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.6567285003555864,
      "learning_rate": 5.174126711343929e-07,
      "loss": 0.0723,
      "step": 32015
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.75056860111149,
      "learning_rate": 5.150936791721738e-07,
      "loss": 0.0803,
      "step": 32020
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.6198175822490691,
      "learning_rate": 5.127798415392126e-07,
      "loss": 0.0927,
      "step": 32025
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.68919247441596,
      "learning_rate": 5.104711587225946e-07,
      "loss": 0.0897,
      "step": 32030
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.6058948482392801,
      "learning_rate": 5.081676312083227e-07,
      "loss": 0.1101,
      "step": 32035
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.7077953911942385,
      "learning_rate": 5.058692594813174e-07,
      "loss": 0.096,
      "step": 32040
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.6550449890185702,
      "learning_rate": 5.035760440254055e-07,
      "loss": 0.0891,
      "step": 32045
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.6897821749729027,
      "learning_rate": 5.01287985323337e-07,
      "loss": 0.1231,
      "step": 32050
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.6312566097100962,
      "learning_rate": 4.990050838567656e-07,
      "loss": 0.0998,
      "step": 32055
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.7908750295842191,
      "learning_rate": 4.967273401062706e-07,
      "loss": 0.0943,
      "step": 32060
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.7165356925541896,
      "learning_rate": 4.944547545513434e-07,
      "loss": 0.1101,
      "step": 32065
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.6855069235440959,
      "learning_rate": 4.921873276703793e-07,
      "loss": 0.1121,
      "step": 32070
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.6598458109315708,
      "learning_rate": 4.899250599407018e-07,
      "loss": 0.0992,
      "step": 32075
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.8019533892863002,
      "learning_rate": 4.876679518385441e-07,
      "loss": 0.1243,
      "step": 32080
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.5490616878274747,
      "learning_rate": 4.854160038390454e-07,
      "loss": 0.0902,
      "step": 32085
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.7606295165897898,
      "learning_rate": 4.831692164162627e-07,
      "loss": 0.1196,
      "step": 32090
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.6654441845897996,
      "learning_rate": 4.809275900431731e-07,
      "loss": 0.1134,
      "step": 32095
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.7055330103260731,
      "learning_rate": 4.786911251916632e-07,
      "loss": 0.1089,
      "step": 32100
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.6216197575243764,
      "learning_rate": 4.764598223325284e-07,
      "loss": 0.0952,
      "step": 32105
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.7659172735607368,
      "learning_rate": 4.742336819354848e-07,
      "loss": 0.1004,
      "step": 32110
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.6941872468872129,
      "learning_rate": 4.720127044691547e-07,
      "loss": 0.1041,
      "step": 32115
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.6535215367952291,
      "learning_rate": 4.697968904010808e-07,
      "loss": 0.0922,
      "step": 32120
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.6553835309409535,
      "learning_rate": 4.6758624019771214e-07,
      "loss": 0.0904,
      "step": 32125
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.6325256204357212,
      "learning_rate": 4.6538075432441265e-07,
      "loss": 0.1047,
      "step": 32130
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.8063852630655081,
      "learning_rate": 4.631804332454665e-07,
      "loss": 0.0986,
      "step": 32135
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.7026017866628933,
      "learning_rate": 4.609852774240586e-07,
      "loss": 0.1181,
      "step": 32140
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.9341510778916376,
      "learning_rate": 4.587952873222945e-07,
      "loss": 0.1094,
      "step": 32145
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.8396831801570894,
      "learning_rate": 4.5661046340118863e-07,
      "loss": 0.1388,
      "step": 32150
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.7897132713379599,
      "learning_rate": 4.544308061206731e-07,
      "loss": 0.1176,
      "step": 32155
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.8918041132470432,
      "learning_rate": 4.5225631593958096e-07,
      "loss": 0.127,
      "step": 32160
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.8481475239619473,
      "learning_rate": 4.5008699331567094e-07,
      "loss": 0.1175,
      "step": 32165
    },
    {
      "epoch": 3.76,
      "grad_norm": 1.037197264147563,
      "learning_rate": 4.479228387056084e-07,
      "loss": 0.1244,
      "step": 32170
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.6201156288530035,
      "learning_rate": 4.4576385256497043e-07,
      "loss": 0.0869,
      "step": 32175
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.5389787897735214,
      "learning_rate": 4.4361003534824363e-07,
      "loss": 0.1034,
      "step": 32180
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.6958170324794998,
      "learning_rate": 4.414613875088347e-07,
      "loss": 0.0853,
      "step": 32185
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.7305058019977781,
      "learning_rate": 4.3931790949904574e-07,
      "loss": 0.0859,
      "step": 32190
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.8097256084435239,
      "learning_rate": 4.371796017701102e-07,
      "loss": 0.1063,
      "step": 32195
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.8237693426801642,
      "learning_rate": 4.350464647721625e-07,
      "loss": 0.1093,
      "step": 32200
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.8031535617941814,
      "learning_rate": 4.3291849895425175e-07,
      "loss": 0.1064,
      "step": 32205
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.5219197508824569,
      "learning_rate": 4.307957047643335e-07,
      "loss": 0.093,
      "step": 32210
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.878516332890269,
      "learning_rate": 4.286780826492809e-07,
      "loss": 0.1161,
      "step": 32215
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.6921775188321084,
      "learning_rate": 4.265656330548734e-07,
      "loss": 0.0914,
      "step": 32220
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.6746299030826209,
      "learning_rate": 4.244583564258053e-07,
      "loss": 0.0735,
      "step": 32225
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.696905826905185,
      "learning_rate": 4.223562532056802e-07,
      "loss": 0.1361,
      "step": 32230
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.689137537666753,
      "learning_rate": 4.202593238370106e-07,
      "loss": 0.1075,
      "step": 32235
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.7473092294054602,
      "learning_rate": 4.1816756876122407e-07,
      "loss": 0.1089,
      "step": 32240
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.729034784265019,
      "learning_rate": 4.1608098841866005e-07,
      "loss": 0.0915,
      "step": 32245
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.7484673214344195,
      "learning_rate": 4.1399958324856436e-07,
      "loss": 0.1093,
      "step": 32250
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.6085416620023498,
      "learning_rate": 4.1192335368908654e-07,
      "loss": 0.1186,
      "step": 32255
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.7167815382190764,
      "learning_rate": 4.098523001773047e-07,
      "loss": 0.0945,
      "step": 32260
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.5231131116256913,
      "learning_rate": 4.077864231491896e-07,
      "loss": 0.0925,
      "step": 32265
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.677433455664288,
      "learning_rate": 4.057257230396377e-07,
      "loss": 0.1055,
      "step": 32270
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.6689239647252212,
      "learning_rate": 4.036702002824438e-07,
      "loss": 0.113,
      "step": 32275
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.7964204649183307,
      "learning_rate": 4.016198553103173e-07,
      "loss": 0.1325,
      "step": 32280
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.66506983266089,
      "learning_rate": 3.9957468855487676e-07,
      "loss": 0.0958,
      "step": 32285
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.7439628542562258,
      "learning_rate": 3.975347004466501e-07,
      "loss": 0.1228,
      "step": 32290
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.628035276525356,
      "learning_rate": 3.954998914150826e-07,
      "loss": 0.1026,
      "step": 32295
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.8237452201719537,
      "learning_rate": 3.9347026188851224e-07,
      "loss": 0.0874,
      "step": 32300
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.8599742567001131,
      "learning_rate": 3.9144581229420554e-07,
      "loss": 0.1077,
      "step": 32305
    },
    {
      "epoch": 3.77,
      "grad_norm": 0.6495778697538434,
      "learning_rate": 3.8942654305832994e-07,
      "loss": 0.0862,
      "step": 32310
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.8468017153386671,
      "learning_rate": 3.8741245460595933e-07,
      "loss": 0.1273,
      "step": 32315
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7284267593418271,
      "learning_rate": 3.854035473610823e-07,
      "loss": 0.106,
      "step": 32320
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7357282763768429,
      "learning_rate": 3.833998217465912e-07,
      "loss": 0.1109,
      "step": 32325
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.6074583455708537,
      "learning_rate": 3.814012781842985e-07,
      "loss": 0.11,
      "step": 32330
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.6939829432836436,
      "learning_rate": 3.794079170949122e-07,
      "loss": 0.1032,
      "step": 32335
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7779167688886567,
      "learning_rate": 3.774197388980605e-07,
      "loss": 0.1013,
      "step": 32340
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7098688463691589,
      "learning_rate": 3.754367440122725e-07,
      "loss": 0.1277,
      "step": 32345
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7668796203291967,
      "learning_rate": 3.7345893285498643e-07,
      "loss": 0.0975,
      "step": 32350
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7766788071318416,
      "learning_rate": 3.714863058425583e-07,
      "loss": 0.1094,
      "step": 32355
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.5571161753711943,
      "learning_rate": 3.6951886339024464e-07,
      "loss": 0.0916,
      "step": 32360
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.6805847714632857,
      "learning_rate": 3.675566059122143e-07,
      "loss": 0.1712,
      "step": 32365
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7480091049755091,
      "learning_rate": 3.655995338215396e-07,
      "loss": 0.084,
      "step": 32370
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7666541441240888,
      "learning_rate": 3.636476475302075e-07,
      "loss": 0.0922,
      "step": 32375
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.8174045648653325,
      "learning_rate": 3.61700947449109e-07,
      "loss": 0.1427,
      "step": 32380
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.8975254307081054,
      "learning_rate": 3.597594339880439e-07,
      "loss": 0.0932,
      "step": 32385
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.7851768562536167,
      "learning_rate": 3.5782310755572424e-07,
      "loss": 0.0993,
      "step": 32390
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.6238493419521051,
      "learning_rate": 3.5589196855976557e-07,
      "loss": 0.1218,
      "step": 32395
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6440162629752348,
      "learning_rate": 3.5396601740669e-07,
      "loss": 0.1035,
      "step": 32400
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6845897160311911,
      "learning_rate": 3.520452545019343e-07,
      "loss": 0.1258,
      "step": 32405
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.7304107013891382,
      "learning_rate": 3.5012968024983886e-07,
      "loss": 0.1149,
      "step": 32410
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.7161864661783787,
      "learning_rate": 3.482192950536506e-07,
      "loss": 0.0788,
      "step": 32415
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.641249480587915,
      "learning_rate": 3.463140993155256e-07,
      "loss": 0.1133,
      "step": 32420
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6667119593137242,
      "learning_rate": 3.4441409343652645e-07,
      "loss": 0.0917,
      "step": 32425
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6802432431372306,
      "learning_rate": 3.4251927781662753e-07,
      "loss": 0.1215,
      "step": 32430
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6121738068271202,
      "learning_rate": 3.406296528547071e-07,
      "loss": 0.1124,
      "step": 32435
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.7707344658469334,
      "learning_rate": 3.3874521894854683e-07,
      "loss": 0.1031,
      "step": 32440
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.7543138390164132,
      "learning_rate": 3.368659764948462e-07,
      "loss": 0.0887,
      "step": 32445
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6678721222443602,
      "learning_rate": 3.3499192588919704e-07,
      "loss": 0.0853,
      "step": 32450
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6952611342488365,
      "learning_rate": 3.331230675261143e-07,
      "loss": 0.1365,
      "step": 32455
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6713618547812098,
      "learning_rate": 3.3125940179901117e-07,
      "loss": 0.0874,
      "step": 32460
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6875983823508983,
      "learning_rate": 3.2940092910020427e-07,
      "loss": 0.1038,
      "step": 32465
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.6112247853868717,
      "learning_rate": 3.275476498209251e-07,
      "loss": 0.0925,
      "step": 32470
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.7269089250932869,
      "learning_rate": 3.256995643513089e-07,
      "loss": 0.1063,
      "step": 32475
    },
    {
      "epoch": 3.79,
      "grad_norm": 0.7565875614682014,
      "learning_rate": 3.2385667308039157e-07,
      "loss": 0.1121,
      "step": 32480
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.6802430124661524,
      "learning_rate": 3.2201897639612953e-07,
      "loss": 0.1107,
      "step": 32485
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.638711612266897,
      "learning_rate": 3.2018647468537166e-07,
      "loss": 0.1049,
      "step": 32490
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.709719443730476,
      "learning_rate": 3.183591683338788e-07,
      "loss": 0.0968,
      "step": 32495
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.7325362835943887,
      "learning_rate": 3.165370577263183e-07,
      "loss": 0.1145,
      "step": 32500
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.6702909787447334,
      "learning_rate": 3.147201432462665e-07,
      "loss": 0.1119,
      "step": 32505
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.7473171509810815,
      "learning_rate": 3.12908425276201e-07,
      "loss": 0.1135,
      "step": 32510
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.6613029331545832,
      "learning_rate": 3.1110190419750553e-07,
      "loss": 0.0883,
      "step": 32515
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.6391911362908526,
      "learning_rate": 3.093005803904758e-07,
      "loss": 0.11,
      "step": 32520
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.790014312046305,
      "learning_rate": 3.0750445423430306e-07,
      "loss": 0.1226,
      "step": 32525
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.7103440016776329,
      "learning_rate": 3.0571352610709867e-07,
      "loss": 0.1102,
      "step": 32530
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.8647429918507457,
      "learning_rate": 3.0392779638586665e-07,
      "loss": 0.1196,
      "step": 32535
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.6848698310642378,
      "learning_rate": 3.0214726544652297e-07,
      "loss": 0.0824,
      "step": 32540
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.9288260149324713,
      "learning_rate": 3.0037193366388717e-07,
      "loss": 0.1169,
      "step": 32545
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.625657519376986,
      "learning_rate": 2.986018014116854e-07,
      "loss": 0.0858,
      "step": 32550
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.7673256417289757,
      "learning_rate": 2.9683686906255e-07,
      "loss": 0.1239,
      "step": 32555
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.9064260430582168,
      "learning_rate": 2.950771369880173e-07,
      "loss": 0.1157,
      "step": 32560
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.7211034657064757,
      "learning_rate": 2.933226055585298e-07,
      "loss": 0.1021,
      "step": 32565
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.902949273593796,
      "learning_rate": 2.9157327514343357e-07,
      "loss": 0.0935,
      "step": 32570
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.557179738459025,
      "learning_rate": 2.8982914611098145e-07,
      "loss": 0.1153,
      "step": 32575
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.7169991489878155,
      "learning_rate": 2.8809021882833245e-07,
      "loss": 0.0974,
      "step": 32580
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.8178308296262314,
      "learning_rate": 2.863564936615465e-07,
      "loss": 0.0984,
      "step": 32585
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.754432967088495,
      "learning_rate": 2.8462797097559277e-07,
      "loss": 0.1055,
      "step": 32590
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.7991279121977617,
      "learning_rate": 2.829046511343414e-07,
      "loss": 0.1178,
      "step": 32595
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.6101320785201979,
      "learning_rate": 2.8118653450057156e-07,
      "loss": 0.0894,
      "step": 32600
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.6675669973010184,
      "learning_rate": 2.794736214359661e-07,
      "loss": 0.083,
      "step": 32605
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.6948785367932933,
      "learning_rate": 2.7776591230110616e-07,
      "loss": 0.1013,
      "step": 32610
    },
    {
      "epoch": 3.81,
      "grad_norm": 1.1985140712048135,
      "learning_rate": 2.7606340745548463e-07,
      "loss": 0.1315,
      "step": 32615
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.664158543894892,
      "learning_rate": 2.743661072574955e-07,
      "loss": 0.0871,
      "step": 32620
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.5929488520729995,
      "learning_rate": 2.7267401206444445e-07,
      "loss": 0.1048,
      "step": 32625
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.7556457158394289,
      "learning_rate": 2.709871222325272e-07,
      "loss": 0.1284,
      "step": 32630
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.5821284737285009,
      "learning_rate": 2.6930543811685417e-07,
      "loss": 0.0976,
      "step": 32635
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.7919901912087645,
      "learning_rate": 2.676289600714421e-07,
      "loss": 0.0996,
      "step": 32640
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.7236007923326127,
      "learning_rate": 2.6595768844920043e-07,
      "loss": 0.122,
      "step": 32645
    },
    {
      "epoch": 3.81,
      "grad_norm": 0.8350896875559715,
      "learning_rate": 2.6429162360195604e-07,
      "loss": 0.1068,
      "step": 32650
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.7259744368333533,
      "learning_rate": 2.6263076588042557e-07,
      "loss": 0.1211,
      "step": 32655
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.7469267122403975,
      "learning_rate": 2.609751156342405e-07,
      "loss": 0.1173,
      "step": 32660
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.6845252081357407,
      "learning_rate": 2.593246732119359e-07,
      "loss": 0.1084,
      "step": 32665
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.5837892758380301,
      "learning_rate": 2.5767943896094214e-07,
      "loss": 0.102,
      "step": 32670
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.6398196226972974,
      "learning_rate": 2.5603941322760447e-07,
      "loss": 0.1148,
      "step": 32675
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.6639789541658405,
      "learning_rate": 2.544045963571551e-07,
      "loss": 0.0907,
      "step": 32680
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.6943677216364776,
      "learning_rate": 2.5277498869374915e-07,
      "loss": 0.1023,
      "step": 32685
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.6715262922582725,
      "learning_rate": 2.5115059058043175e-07,
      "loss": 0.0813,
      "step": 32690
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.7556016083695952,
      "learning_rate": 2.4953140235915996e-07,
      "loss": 0.0926,
      "step": 32695
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.685707078716426,
      "learning_rate": 2.47917424370786e-07,
      "loss": 0.0894,
      "step": 32700
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.6267829794940367,
      "learning_rate": 2.4630865695507156e-07,
      "loss": 0.0989,
      "step": 32705
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.7271071170557324,
      "learning_rate": 2.447051004506762e-07,
      "loss": 0.1027,
      "step": 32710
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.73891620426844,
      "learning_rate": 2.43106755195166e-07,
      "loss": 0.0942,
      "step": 32715
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.7630890231537586,
      "learning_rate": 2.4151362152501077e-07,
      "loss": 0.1151,
      "step": 32720
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.6415328811103522,
      "learning_rate": 2.399256997755811e-07,
      "loss": 0.1045,
      "step": 32725
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.6457011037497107,
      "learning_rate": 2.3834299028115125e-07,
      "loss": 0.1107,
      "step": 32730
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.7563047492549106,
      "learning_rate": 2.3676549337490195e-07,
      "loss": 0.11,
      "step": 32735
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.6943494751661062,
      "learning_rate": 2.3519320938890644e-07,
      "loss": 0.0843,
      "step": 32740
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.7397622650658459,
      "learning_rate": 2.3362613865414994e-07,
      "loss": 0.1445,
      "step": 32745
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.7719907620459707,
      "learning_rate": 2.3206428150051584e-07,
      "loss": 0.1243,
      "step": 32750
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.6483475507050217,
      "learning_rate": 2.305076382567911e-07,
      "loss": 0.0964,
      "step": 32755
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.9620139163266311,
      "learning_rate": 2.2895620925066918e-07,
      "loss": 0.0997,
      "step": 32760
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.703196161486108,
      "learning_rate": 2.2740999480874158e-07,
      "loss": 0.0931,
      "step": 32765
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.6959990049701891,
      "learning_rate": 2.2586899525650074e-07,
      "loss": 0.1041,
      "step": 32770
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.7188801905720508,
      "learning_rate": 2.243332109183399e-07,
      "loss": 0.1149,
      "step": 32775
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.6900215615479066,
      "learning_rate": 2.2280264211756162e-07,
      "loss": 0.1108,
      "step": 32780
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.8582597865770255,
      "learning_rate": 2.2127728917636925e-07,
      "loss": 0.1074,
      "step": 32785
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.7723227820547632,
      "learning_rate": 2.197571524158615e-07,
      "loss": 0.0841,
      "step": 32790
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.7325119649035282,
      "learning_rate": 2.1824223215604345e-07,
      "loss": 0.0934,
      "step": 32795
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.6585334310683003,
      "learning_rate": 2.1673252871582394e-07,
      "loss": 0.0915,
      "step": 32800
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.6829520818192908,
      "learning_rate": 2.152280424130071e-07,
      "loss": 0.1232,
      "step": 32805
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.8933620592736912,
      "learning_rate": 2.1372877356430898e-07,
      "loss": 0.139,
      "step": 32810
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.670699260948708,
      "learning_rate": 2.1223472248533828e-07,
      "loss": 0.1057,
      "step": 32815
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.7181288777438867,
      "learning_rate": 2.1074588949060447e-07,
      "loss": 0.1071,
      "step": 32820
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.8206283163313677,
      "learning_rate": 2.0926227489352913e-07,
      "loss": 0.1011,
      "step": 32825
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.7898996390125568,
      "learning_rate": 2.0778387900642627e-07,
      "loss": 0.1171,
      "step": 32830
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.7397473588449304,
      "learning_rate": 2.0631070214051086e-07,
      "loss": 0.0908,
      "step": 32835
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.6019278201804931,
      "learning_rate": 2.0484274460590702e-07,
      "loss": 0.0967,
      "step": 32840
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.7294917106704344,
      "learning_rate": 2.0338000671163148e-07,
      "loss": 0.0931,
      "step": 32845
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.735366965535307,
      "learning_rate": 2.0192248876560738e-07,
      "loss": 0.1015,
      "step": 32850
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.695523670843052,
      "learning_rate": 2.0047019107465592e-07,
      "loss": 0.0972,
      "step": 32855
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.717702936640991,
      "learning_rate": 1.990231139445048e-07,
      "loss": 0.1001,
      "step": 32860
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.6931280547082822,
      "learning_rate": 1.9758125767977698e-07,
      "loss": 0.12,
      "step": 32865
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.920092984059423,
      "learning_rate": 1.9614462258399636e-07,
      "loss": 0.0989,
      "step": 32870
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.7544529122596434,
      "learning_rate": 1.9471320895959044e-07,
      "loss": 0.0813,
      "step": 32875
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.6353310379459537,
      "learning_rate": 1.9328701710788766e-07,
      "loss": 0.1283,
      "step": 32880
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.6008702900711697,
      "learning_rate": 1.918660473291145e-07,
      "loss": 0.097,
      "step": 32885
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.8212571569984326,
      "learning_rate": 1.9045029992240394e-07,
      "loss": 0.109,
      "step": 32890
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.6573232738310341,
      "learning_rate": 1.890397751857814e-07,
      "loss": 0.1119,
      "step": 32895
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.7396375184239369,
      "learning_rate": 1.8763447341618166e-07,
      "loss": 0.0988,
      "step": 32900
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.6706843706478215,
      "learning_rate": 1.8623439490943196e-07,
      "loss": 0.0977,
      "step": 32905
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.6518495048785922,
      "learning_rate": 1.8483953996026315e-07,
      "loss": 0.0894,
      "step": 32910
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.853311570942987,
      "learning_rate": 1.8344990886230707e-07,
      "loss": 0.1016,
      "step": 32915
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.6822120818833445,
      "learning_rate": 1.8206550190809913e-07,
      "loss": 0.0849,
      "step": 32920
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.808763867838551,
      "learning_rate": 1.8068631938906455e-07,
      "loss": 0.1065,
      "step": 32925
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.6777801634803904,
      "learning_rate": 1.793123615955461e-07,
      "loss": 0.121,
      "step": 32930
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.5216664008856158,
      "learning_rate": 1.7794362881676795e-07,
      "loss": 0.1038,
      "step": 32935
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.733465640691882,
      "learning_rate": 1.7658012134086355e-07,
      "loss": 0.1186,
      "step": 32940
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.6690813153046264,
      "learning_rate": 1.7522183945486715e-07,
      "loss": 0.0959,
      "step": 32945
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.6527260330200685,
      "learning_rate": 1.7386878344471113e-07,
      "loss": 0.0969,
      "step": 32950
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.7548421301919747,
      "learning_rate": 1.725209535952288e-07,
      "loss": 0.1092,
      "step": 32955
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.7434471422746521,
      "learning_rate": 1.7117835019014873e-07,
      "loss": 0.1014,
      "step": 32960
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.7781774749965156,
      "learning_rate": 1.698409735121087e-07,
      "loss": 0.1114,
      "step": 32965
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.8529976801774993,
      "learning_rate": 1.6850882384263633e-07,
      "loss": 0.13,
      "step": 32970
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.6729666114275548,
      "learning_rate": 1.6718190146216561e-07,
      "loss": 0.1044,
      "step": 32975
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.820076386360636,
      "learning_rate": 1.6586020665002588e-07,
      "loss": 0.1169,
      "step": 32980
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.5692615674621903,
      "learning_rate": 1.6454373968444736e-07,
      "loss": 0.1005,
      "step": 32985
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.7497492165580306,
      "learning_rate": 1.6323250084256115e-07,
      "loss": 0.112,
      "step": 32990
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.6672835548870172,
      "learning_rate": 1.619264904003992e-07,
      "loss": 0.1029,
      "step": 32995
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.0023162133682435,
      "learning_rate": 1.6062570863288608e-07,
      "loss": 0.0922,
      "step": 33000
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.7626385054677094,
      "learning_rate": 1.593301558138527e-07,
      "loss": 0.124,
      "step": 33005
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.5707166728478943,
      "learning_rate": 1.5803983221602257e-07,
      "loss": 0.0954,
      "step": 33010
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.6640522393751017,
      "learning_rate": 1.5675473811102848e-07,
      "loss": 0.117,
      "step": 33015
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.7930275495634006,
      "learning_rate": 1.5547487376939007e-07,
      "loss": 0.1109,
      "step": 33020
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.6476419683914998,
      "learning_rate": 1.542002394605363e-07,
      "loss": 0.1042,
      "step": 33025
    },
    {
      "epoch": 3.86,
      "grad_norm": 1.2257779445973738,
      "learning_rate": 1.529308354527914e-07,
      "loss": 0.1506,
      "step": 33030
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.8766695120797283,
      "learning_rate": 1.51666662013375e-07,
      "loss": 0.1324,
      "step": 33035
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.7746484476213903,
      "learning_rate": 1.5040771940841024e-07,
      "loss": 0.1222,
      "step": 33040
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.7162793865115763,
      "learning_rate": 1.4915400790292133e-07,
      "loss": 0.1192,
      "step": 33045
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.7436189803687565,
      "learning_rate": 1.4790552776081935e-07,
      "loss": 0.1252,
      "step": 33050
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.512078692715849,
      "learning_rate": 1.4666227924493014e-07,
      "loss": 0.0928,
      "step": 33055
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.6412273309909928,
      "learning_rate": 1.454242626169694e-07,
      "loss": 0.0957,
      "step": 33060
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.7453721356989709,
      "learning_rate": 1.4419147813754808e-07,
      "loss": 0.0928,
      "step": 33065
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.6044719042014551,
      "learning_rate": 1.4296392606618359e-07,
      "loss": 0.0829,
      "step": 33070
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.6579268397350826,
      "learning_rate": 1.4174160666128865e-07,
      "loss": 0.0967,
      "step": 33075
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.6337095541512262,
      "learning_rate": 1.4052452018017415e-07,
      "loss": 0.0898,
      "step": 33080
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.6854527863497736,
      "learning_rate": 1.3931266687904897e-07,
      "loss": 0.1107,
      "step": 33085
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.8285005601543393,
      "learning_rate": 1.381060470130202e-07,
      "loss": 0.0973,
      "step": 33090
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.7835418171330808,
      "learning_rate": 1.3690466083609577e-07,
      "loss": 0.1163,
      "step": 33095
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.7789627426880127,
      "learning_rate": 1.3570850860118166e-07,
      "loss": 0.1078,
      "step": 33100
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.8177357433154049,
      "learning_rate": 1.3451759056007373e-07,
      "loss": 0.1,
      "step": 33105
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.6553833711563116,
      "learning_rate": 1.3333190696347697e-07,
      "loss": 0.1216,
      "step": 33110
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.8986849612898306,
      "learning_rate": 1.3215145806099172e-07,
      "loss": 0.0992,
      "step": 33115
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.5771019353697012,
      "learning_rate": 1.3097624410111364e-07,
      "loss": 0.1003,
      "step": 33120
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.6467013077509892,
      "learning_rate": 1.2980626533123652e-07,
      "loss": 0.0971,
      "step": 33125
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.6947407302025335,
      "learning_rate": 1.28641521997655e-07,
      "loss": 0.1084,
      "step": 33130
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.6018842177853012,
      "learning_rate": 1.2748201434555628e-07,
      "loss": 0.0741,
      "step": 33135
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.7536956818890865,
      "learning_rate": 1.2632774261903403e-07,
      "loss": 0.1032,
      "step": 33140
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.7462745606244382,
      "learning_rate": 1.2517870706106883e-07,
      "loss": 0.089,
      "step": 33145
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.6993265220807361,
      "learning_rate": 1.2403490791354776e-07,
      "loss": 0.1259,
      "step": 33150
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.7986671695766824,
      "learning_rate": 1.2289634541725325e-07,
      "loss": 0.1147,
      "step": 33155
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.8410296199753822,
      "learning_rate": 1.21763019811863e-07,
      "loss": 0.1169,
      "step": 33160
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.8784415726983966,
      "learning_rate": 1.2063493133595837e-07,
      "loss": 0.1129,
      "step": 33165
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.7451154941953951,
      "learning_rate": 1.19512080227005e-07,
      "loss": 0.1062,
      "step": 33170
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.7734901166213242,
      "learning_rate": 1.1839446672138043e-07,
      "loss": 0.1009,
      "step": 33175
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.7809789727997014,
      "learning_rate": 1.1728209105435484e-07,
      "loss": 0.1001,
      "step": 33180
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.6281136783223772,
      "learning_rate": 1.1617495346009367e-07,
      "loss": 0.1,
      "step": 33185
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.7723672394865146,
      "learning_rate": 1.1507305417166048e-07,
      "loss": 0.0941,
      "step": 33190
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.7205157090910749,
      "learning_rate": 1.1397639342101696e-07,
      "loss": 0.114,
      "step": 33195
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.6665552215993625,
      "learning_rate": 1.1288497143902288e-07,
      "loss": 0.0916,
      "step": 33200
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.8077242572567152,
      "learning_rate": 1.1179878845543056e-07,
      "loss": 0.1105,
      "step": 33205
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.725589908230325,
      "learning_rate": 1.10717844698896e-07,
      "loss": 0.1069,
      "step": 33210
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.6961444627918473,
      "learning_rate": 1.0964214039696774e-07,
      "loss": 0.1029,
      "step": 33215
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.7568406385243643,
      "learning_rate": 1.0857167577609517e-07,
      "loss": 0.0901,
      "step": 33220
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.9051158833564148,
      "learning_rate": 1.0750645106162028e-07,
      "loss": 0.1322,
      "step": 33225
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.8357552113691215,
      "learning_rate": 1.0644646647778312e-07,
      "loss": 0.1192,
      "step": 33230
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.8965697828998737,
      "learning_rate": 1.0539172224772465e-07,
      "loss": 0.1306,
      "step": 33235
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.8219282601721526,
      "learning_rate": 1.043422185934756e-07,
      "loss": 0.0919,
      "step": 33240
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.8386350469888751,
      "learning_rate": 1.0329795573596756e-07,
      "loss": 0.1242,
      "step": 33245
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.6588818579812464,
      "learning_rate": 1.0225893389503304e-07,
      "loss": 0.0925,
      "step": 33250
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.6571316830726425,
      "learning_rate": 1.0122515328939709e-07,
      "loss": 0.0848,
      "step": 33255
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.7358825850984076,
      "learning_rate": 1.0019661413667725e-07,
      "loss": 0.0879,
      "step": 33260
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.6644260693294591,
      "learning_rate": 9.917331665339202e-08,
      "loss": 0.0966,
      "step": 33265
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.6325696920490801,
      "learning_rate": 9.815526105495799e-08,
      "loss": 0.0727,
      "step": 33270
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.7947258279183854,
      "learning_rate": 9.714244755568702e-08,
      "loss": 0.1053,
      "step": 33275
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.9447115517050975,
      "learning_rate": 9.613487636878915e-08,
      "loss": 0.1182,
      "step": 33280
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.6856283036419851,
      "learning_rate": 9.513254770636137e-08,
      "loss": 0.086,
      "step": 33285
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.5332081866270109,
      "learning_rate": 9.413546177941269e-08,
      "loss": 0.0799,
      "step": 33290
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.5446089743953006,
      "learning_rate": 9.314361879783351e-08,
      "loss": 0.0836,
      "step": 33295
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.7627020285780585,
      "learning_rate": 9.215701897042072e-08,
      "loss": 0.1232,
      "step": 33300
    },
    {
      "epoch": 3.89,
      "grad_norm": 2.257405302459278,
      "learning_rate": 9.117566250486098e-08,
      "loss": 0.0944,
      "step": 33305
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.7979373104809461,
      "learning_rate": 9.019954960774457e-08,
      "loss": 0.1111,
      "step": 33310
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.6369170698961653,
      "learning_rate": 8.92286804845488e-08,
      "loss": 0.1126,
      "step": 33315
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.6965647196067069,
      "learning_rate": 8.82630553396574e-08,
      "loss": 0.1002,
      "step": 33320
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.7328705376593999,
      "learning_rate": 8.730267437633832e-08,
      "loss": 0.1044,
      "step": 33325
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.7695604713234119,
      "learning_rate": 8.634753779676596e-08,
      "loss": 0.1126,
      "step": 33330
    },
    {
      "epoch": 3.89,
      "grad_norm": 0.7168317718667033,
      "learning_rate": 8.539764580200726e-08,
      "loss": 0.1076,
      "step": 33335
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.6860980668540384,
      "learning_rate": 8.445299859201894e-08,
      "loss": 0.0887,
      "step": 33340
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.7178454389519474,
      "learning_rate": 8.351359636566414e-08,
      "loss": 0.1001,
      "step": 33345
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.6627330711064285,
      "learning_rate": 8.257943932069856e-08,
      "loss": 0.1023,
      "step": 33350
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.6314070593844423,
      "learning_rate": 8.165052765376491e-08,
      "loss": 0.0941,
      "step": 33355
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.2514996973935248,
      "learning_rate": 8.07268615604151e-08,
      "loss": 0.1229,
      "step": 33360
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.8448608013988977,
      "learning_rate": 7.980844123508524e-08,
      "loss": 0.1244,
      "step": 33365
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.5392703455846917,
      "learning_rate": 7.88952668711207e-08,
      "loss": 0.1034,
      "step": 33370
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.5959426998468517,
      "learning_rate": 7.798733866074548e-08,
      "loss": 0.0821,
      "step": 33375
    },
    {
      "epoch": 3.9,
      "grad_norm": 1.0519091864583134,
      "learning_rate": 7.708465679509003e-08,
      "loss": 0.1207,
      "step": 33380
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.6156119003200068,
      "learning_rate": 7.61872214641829e-08,
      "loss": 0.0956,
      "step": 33385
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.8278157176081044,
      "learning_rate": 7.529503285693962e-08,
      "loss": 0.0971,
      "step": 33390
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.8911182785867745,
      "learning_rate": 7.440809116117664e-08,
      "loss": 0.1542,
      "step": 33395
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.8142086442637851,
      "learning_rate": 7.352639656360294e-08,
      "loss": 0.1412,
      "step": 33400
    },
    {
      "epoch": 3.9,
      "grad_norm": 3.9884926239661724,
      "learning_rate": 7.26499492498256e-08,
      "loss": 0.1485,
      "step": 33405
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.6543922954498517,
      "learning_rate": 7.177874940434703e-08,
      "loss": 0.1086,
      "step": 33410
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.9101625988160381,
      "learning_rate": 7.091279721055943e-08,
      "loss": 0.1162,
      "step": 33415
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.7435678301335883,
      "learning_rate": 7.005209285076141e-08,
      "loss": 0.1384,
      "step": 33420
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.6626040543619331,
      "learning_rate": 6.919663650613862e-08,
      "loss": 0.0875,
      "step": 33425
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.7360348438866782,
      "learning_rate": 6.834642835676919e-08,
      "loss": 0.0923,
      "step": 33430
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.722797405398677,
      "learning_rate": 6.750146858163498e-08,
      "loss": 0.0983,
      "step": 33435
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.7893377507024466,
      "learning_rate": 6.666175735860758e-08,
      "loss": 0.1026,
      "step": 33440
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.7249850498666216,
      "learning_rate": 6.582729486445393e-08,
      "loss": 0.1107,
      "step": 33445
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.8121679770889151,
      "learning_rate": 6.49980812748363e-08,
      "loss": 0.1138,
      "step": 33450
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.6601798958973863,
      "learning_rate": 6.417411676431784e-08,
      "loss": 0.0777,
      "step": 33455
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.607472502744155,
      "learning_rate": 6.335540150634589e-08,
      "loss": 0.0795,
      "step": 33460
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.6922592549521205,
      "learning_rate": 6.254193567327427e-08,
      "loss": 0.1412,
      "step": 33465
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.6649560521868016,
      "learning_rate": 6.173371943634099e-08,
      "loss": 0.1155,
      "step": 33470
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.6744963317035093,
      "learning_rate": 6.093075296568774e-08,
      "loss": 0.0819,
      "step": 33475
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.861428374146387,
      "learning_rate": 6.013303643034319e-08,
      "loss": 0.0977,
      "step": 33480
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.8426083832818959,
      "learning_rate": 5.934056999823967e-08,
      "loss": 0.0984,
      "step": 33485
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.3346263487238814,
      "learning_rate": 5.8553353836196535e-08,
      "loss": 0.0991,
      "step": 33490
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.6448161012973965,
      "learning_rate": 5.777138810993399e-08,
      "loss": 0.1077,
      "step": 33495
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.6667483839701805,
      "learning_rate": 5.699467298405925e-08,
      "loss": 0.0831,
      "step": 33500
    },
    {
      "epoch": 3.91,
      "grad_norm": 0.8138717103727736,
      "learning_rate": 5.622320862208319e-08,
      "loss": 0.0894,
      "step": 33505
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.6845526280905971,
      "learning_rate": 5.5456995186406455e-08,
      "loss": 0.1004,
      "step": 33510
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.7395078166880183,
      "learning_rate": 5.469603283832225e-08,
      "loss": 0.0987,
      "step": 33515
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.6214036595764711,
      "learning_rate": 5.3940321738024656e-08,
      "loss": 0.0847,
      "step": 33520
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.6401316669431774,
      "learning_rate": 5.3189862044597526e-08,
      "loss": 0.0926,
      "step": 33525
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.6217032885242302,
      "learning_rate": 5.2444653916020045e-08,
      "loss": 0.095,
      "step": 33530
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.7248023953012112,
      "learning_rate": 5.170469750916396e-08,
      "loss": 0.1178,
      "step": 33535
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.8369934764451854,
      "learning_rate": 5.096999297980465e-08,
      "loss": 0.0985,
      "step": 33540
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.6431345849655774,
      "learning_rate": 5.0240540482596187e-08,
      "loss": 0.1211,
      "step": 33545
    },
    {
      "epoch": 3.92,
      "grad_norm": 1.026422262899042,
      "learning_rate": 4.9516340171104626e-08,
      "loss": 0.1182,
      "step": 33550
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.625611866226964,
      "learning_rate": 4.879739219777746e-08,
      "loss": 0.0848,
      "step": 33555
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.6293556494239031,
      "learning_rate": 4.808369671396029e-08,
      "loss": 0.081,
      "step": 33560
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.7686242653350159,
      "learning_rate": 4.737525386989405e-08,
      "loss": 0.1037,
      "step": 33565
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.7330377126756894,
      "learning_rate": 4.667206381471223e-08,
      "loss": 0.0987,
      "step": 33570
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.6797606005008716,
      "learning_rate": 4.597412669644641e-08,
      "loss": 0.1156,
      "step": 33575
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.7835310593315485,
      "learning_rate": 4.5281442662017945e-08,
      "loss": 0.1243,
      "step": 33580
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.8678003874081279,
      "learning_rate": 4.459401185724355e-08,
      "loss": 0.1289,
      "step": 33585
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.6305765323751037,
      "learning_rate": 4.3911834426838e-08,
      "loss": 0.0855,
      "step": 33590
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.663003353273071,
      "learning_rate": 4.323491051440309e-08,
      "loss": 0.1084,
      "step": 33595
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7415541684729337,
      "learning_rate": 4.256324026243874e-08,
      "loss": 0.1085,
      "step": 33600
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7920274582707633,
      "learning_rate": 4.1896823812340166e-08,
      "loss": 0.1192,
      "step": 33605
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7914292127569258,
      "learning_rate": 4.1235661304395154e-08,
      "loss": 0.0941,
      "step": 33610
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.6800069258927826,
      "learning_rate": 4.057975287778404e-08,
      "loss": 0.0962,
      "step": 33615
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.662411293180192,
      "learning_rate": 3.9929098670582476e-08,
      "loss": 0.1008,
      "step": 33620
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.5597369403391514,
      "learning_rate": 3.9283698819758684e-08,
      "loss": 0.1129,
      "step": 33625
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7887612776867691,
      "learning_rate": 3.8643553461178985e-08,
      "loss": 0.1054,
      "step": 33630
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.6394196479168663,
      "learning_rate": 3.8008662729599464e-08,
      "loss": 0.0704,
      "step": 33635
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.6397274259824062,
      "learning_rate": 3.737902675867155e-08,
      "loss": 0.0939,
      "step": 33640
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7653282387371841,
      "learning_rate": 3.6754645680942e-08,
      "loss": 0.0771,
      "step": 33645
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.8695060810047941,
      "learning_rate": 3.613551962784456e-08,
      "loss": 0.1264,
      "step": 33650
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7124541455814115,
      "learning_rate": 3.552164872971664e-08,
      "loss": 0.1311,
      "step": 33655
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7483825871614354,
      "learning_rate": 3.491303311577987e-08,
      "loss": 0.1237,
      "step": 33660
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7314402274775395,
      "learning_rate": 3.430967291415954e-08,
      "loss": 0.1047,
      "step": 33665
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.784191002070497,
      "learning_rate": 3.3711568251865166e-08,
      "loss": 0.1036,
      "step": 33670
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.7002832022753681,
      "learning_rate": 3.311871925480714e-08,
      "loss": 0.089,
      "step": 33675
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7157808569863947,
      "learning_rate": 3.253112604778286e-08,
      "loss": 0.0944,
      "step": 33680
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7690302242905498,
      "learning_rate": 3.194878875449059e-08,
      "loss": 0.1001,
      "step": 33685
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7414198306593133,
      "learning_rate": 3.1371707497512836e-08,
      "loss": 0.1064,
      "step": 33690
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7107395518265439,
      "learning_rate": 3.079988239833853e-08,
      "loss": 0.1095,
      "step": 33695
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7577010613976696,
      "learning_rate": 3.0233313577338055e-08,
      "loss": 0.0867,
      "step": 33700
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.9477436716739325,
      "learning_rate": 2.967200115378266e-08,
      "loss": 0.1254,
      "step": 33705
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7070504141193993,
      "learning_rate": 2.911594524583061e-08,
      "loss": 0.1116,
      "step": 33710
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7098825236377333,
      "learning_rate": 2.856514597054383e-08,
      "loss": 0.1208,
      "step": 33715
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.737516373092627,
      "learning_rate": 2.801960344386567e-08,
      "loss": 0.1021,
      "step": 33720
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7687391928011947,
      "learning_rate": 2.747931778064039e-08,
      "loss": 0.0962,
      "step": 33725
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.6901272284902701,
      "learning_rate": 2.6944289094602005e-08,
      "loss": 0.119,
      "step": 33730
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.6559284554910177,
      "learning_rate": 2.641451749838264e-08,
      "loss": 0.1188,
      "step": 33735
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.6720122791796105,
      "learning_rate": 2.58900031035042e-08,
      "loss": 0.1079,
      "step": 33740
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.6682742709753189,
      "learning_rate": 2.5370746020381142e-08,
      "loss": 0.0857,
      "step": 33745
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7208439681675699,
      "learning_rate": 2.4856746358326023e-08,
      "loss": 0.1074,
      "step": 33750
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.815113495890879,
      "learning_rate": 2.4348004225535627e-08,
      "loss": 0.1263,
      "step": 33755
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.6423649019025683,
      "learning_rate": 2.3844519729107616e-08,
      "loss": 0.0894,
      "step": 33760
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.7536765704891965,
      "learning_rate": 2.3346292975032212e-08,
      "loss": 0.1248,
      "step": 33765
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.7319238420791185,
      "learning_rate": 2.2853324068192184e-08,
      "loss": 0.1095,
      "step": 33770
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.5632626053157793,
      "learning_rate": 2.2365613112360074e-08,
      "loss": 0.0788,
      "step": 33775
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.6740306505306219,
      "learning_rate": 2.1883160210203756e-08,
      "loss": 0.1018,
      "step": 33780
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.7522175455704624,
      "learning_rate": 2.1405965463289212e-08,
      "loss": 0.1205,
      "step": 33785
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.784204697416689,
      "learning_rate": 2.0934028972063867e-08,
      "loss": 0.1062,
      "step": 33790
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.6753449597633256,
      "learning_rate": 2.0467350835881583e-08,
      "loss": 0.1027,
      "step": 33795
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.6617045140518762,
      "learning_rate": 2.000593115298044e-08,
      "loss": 0.0929,
      "step": 33800
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.6234472419640817,
      "learning_rate": 1.9549770020491076e-08,
      "loss": 0.0814,
      "step": 33805
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.8697063755447425,
      "learning_rate": 1.9098867534447785e-08,
      "loss": 0.1052,
      "step": 33810
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.7352054725543583,
      "learning_rate": 1.865322378976353e-08,
      "loss": 0.1034,
      "step": 33815
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.6781023912643896,
      "learning_rate": 1.821283888025216e-08,
      "loss": 0.0973,
      "step": 33820
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.9892892763916536,
      "learning_rate": 1.7777712898622845e-08,
      "loss": 0.1115,
      "step": 33825
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.6928806081834832,
      "learning_rate": 1.7347845936468987e-08,
      "loss": 0.0986,
      "step": 33830
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.7609546731271315,
      "learning_rate": 1.692323808428764e-08,
      "loss": 0.118,
      "step": 33835
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.7410730759366381,
      "learning_rate": 1.6503889431460085e-08,
      "loss": 0.1541,
      "step": 33840
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.7218414487697394,
      "learning_rate": 1.608980006626015e-08,
      "loss": 0.1066,
      "step": 33845
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.8822990443595471,
      "learning_rate": 1.568097007586533e-08,
      "loss": 0.1025,
      "step": 33850
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.8202464436402083,
      "learning_rate": 1.527739954633456e-08,
      "loss": 0.0982,
      "step": 33855
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.7192033692752401,
      "learning_rate": 1.4879088562624877e-08,
      "loss": 0.1053,
      "step": 33860
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.705012222162766,
      "learning_rate": 1.4486037208583102e-08,
      "loss": 0.1185,
      "step": 33865
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.7513719561032955,
      "learning_rate": 1.4098245566951385e-08,
      "loss": 0.1005,
      "step": 33870
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.8403034307299907,
      "learning_rate": 1.371571371936442e-08,
      "loss": 0.1173,
      "step": 33875
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.8237935847068727,
      "learning_rate": 1.3338441746346687e-08,
      "loss": 0.1038,
      "step": 33880
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.6510598233606743,
      "learning_rate": 1.2966429727323537e-08,
      "loss": 0.0766,
      "step": 33885
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.7207365751696249,
      "learning_rate": 1.2599677740604554e-08,
      "loss": 0.1144,
      "step": 33890
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.6489953454447048,
      "learning_rate": 1.2238185863391871e-08,
      "loss": 0.127,
      "step": 33895
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.6161927058919678,
      "learning_rate": 1.1881954171785725e-08,
      "loss": 0.0904,
      "step": 33900
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.6284960074667723,
      "learning_rate": 1.1530982740778906e-08,
      "loss": 0.093,
      "step": 33905
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.7343434090453264,
      "learning_rate": 1.1185271644251206e-08,
      "loss": 0.1004,
      "step": 33910
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.6523001321962878,
      "learning_rate": 1.0844820954980516e-08,
      "loss": 0.0959,
      "step": 33915
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.683588832166793,
      "learning_rate": 1.0509630744634513e-08,
      "loss": 0.099,
      "step": 33920
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.7001910580164512,
      "learning_rate": 1.0179701083770643e-08,
      "loss": 0.0876,
      "step": 33925
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.65851734533734,
      "learning_rate": 9.85503204185001e-09,
      "loss": 0.075,
      "step": 33930
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.7039173929392916,
      "learning_rate": 9.535623687212391e-09,
      "loss": 0.0772,
      "step": 33935
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.7712174304185561,
      "learning_rate": 9.221476087101222e-09,
      "loss": 0.1143,
      "step": 33940
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.7134411916766169,
      "learning_rate": 8.91258930764416e-09,
      "loss": 0.089,
      "step": 33945
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.6945707367005447,
      "learning_rate": 8.608963413866966e-09,
      "loss": 0.1038,
      "step": 33950
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.6810045447227157,
      "learning_rate": 8.310598469687958e-09,
      "loss": 0.0833,
      "step": 33955
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.1836911040275282,
      "learning_rate": 8.017494537909676e-09,
      "loss": 0.1162,
      "step": 33960
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.7143943769664722,
      "learning_rate": 7.72965168024109e-09,
      "loss": 0.0786,
      "step": 33965
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.7195874144744634,
      "learning_rate": 7.447069957272624e-09,
      "loss": 0.0971,
      "step": 33970
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.6224476831249508,
      "learning_rate": 7.1697494284900245e-09,
      "loss": 0.0873,
      "step": 33975
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.6470236046022421,
      "learning_rate": 6.897690152271596e-09,
      "loss": 0.0964,
      "step": 33980
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.604727612901551,
      "learning_rate": 6.630892185890969e-09,
      "loss": 0.0898,
      "step": 33985
    },
    {
      "epoch": 3.97,
      "grad_norm": 1.0460214148557467,
      "learning_rate": 6.369355585511549e-09,
      "loss": 0.1192,
      "step": 33990
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.7126613373807706,
      "learning_rate": 6.113080406189298e-09,
      "loss": 0.111,
      "step": 33995
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.7555909151058204,
      "learning_rate": 5.862066701872726e-09,
      "loss": 0.1019,
      "step": 34000
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.7451304178457472,
      "learning_rate": 5.616314525402899e-09,
      "loss": 0.1142,
      "step": 34005
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.6690123178253058,
      "learning_rate": 5.375823928510659e-09,
      "loss": 0.0772,
      "step": 34010
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.6222457397145524,
      "learning_rate": 5.140594961827727e-09,
      "loss": 0.1098,
      "step": 34015
    },
    {
      "epoch": 3.97,
      "grad_norm": 0.7833964491785993,
      "learning_rate": 4.910627674867274e-09,
      "loss": 0.1043,
      "step": 34020
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.6505869625534229,
      "learning_rate": 4.685922116040575e-09,
      "loss": 0.0757,
      "step": 34025
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.8965448975155454,
      "learning_rate": 4.466478332651458e-09,
      "loss": 0.1063,
      "step": 34030
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.7700486995481837,
      "learning_rate": 4.252296370893527e-09,
      "loss": 0.1088,
      "step": 34035
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.7656239378842655,
      "learning_rate": 4.043376275855715e-09,
      "loss": 0.104,
      "step": 34040
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.545221838353672,
      "learning_rate": 3.839718091519506e-09,
      "loss": 0.079,
      "step": 34045
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.5706935403414807,
      "learning_rate": 3.641321860756164e-09,
      "loss": 0.0998,
      "step": 34050
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.6710423199561348,
      "learning_rate": 3.4481876253267263e-09,
      "loss": 0.102,
      "step": 34055
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.673991634828044,
      "learning_rate": 3.2603154258931122e-09,
      "loss": 0.1019,
      "step": 34060
    },
    {
      "epoch": 3.98,
      "grad_norm": 1.172901444146483,
      "learning_rate": 3.0777053019986903e-09,
      "loss": 0.1553,
      "step": 34065
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.723033328228414,
      "learning_rate": 2.900357292090483e-09,
      "loss": 0.1211,
      "step": 34070
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.8122987070651587,
      "learning_rate": 2.7282714334997405e-09,
      "loss": 0.0984,
      "step": 34075
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.7597744751992892,
      "learning_rate": 2.5614477624530397e-09,
      "loss": 0.0995,
      "step": 34080
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.6960762295450328,
      "learning_rate": 2.3998863140639594e-09,
      "loss": 0.1072,
      "step": 34085
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.7525631690355659,
      "learning_rate": 2.2435871223497328e-09,
      "loss": 0.1076,
      "step": 34090
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.7563564478825053,
      "learning_rate": 2.092550220209044e-09,
      "loss": 0.1279,
      "step": 34095
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.9037530226918882,
      "learning_rate": 1.9467756394386804e-09,
      "loss": 0.1157,
      "step": 34100
    },
    {
      "epoch": 3.98,
      "grad_norm": 0.7520334279010105,
      "learning_rate": 1.8062634107224308e-09,
      "loss": 0.1144,
      "step": 34105
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.7719614821902274,
      "learning_rate": 1.671013563644963e-09,
      "loss": 0.097,
      "step": 34110
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.6740552218025032,
      "learning_rate": 1.5410261266723958e-09,
      "loss": 0.1128,
      "step": 34115
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.7173180353504346,
      "learning_rate": 1.4163011271689508e-09,
      "loss": 0.0998,
      "step": 34120
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.7838965630235181,
      "learning_rate": 1.2968385913941784e-09,
      "loss": 0.1062,
      "step": 34125
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.9710404510554173,
      "learning_rate": 1.1826385444946297e-09,
      "loss": 0.1151,
      "step": 34130
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.7472071809720408,
      "learning_rate": 1.0737010105094091e-09,
      "loss": 0.0923,
      "step": 34135
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.6630474952740282,
      "learning_rate": 9.700260123729488e-10,
      "loss": 0.1119,
      "step": 34140
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.755434665642974,
      "learning_rate": 8.716135719094576e-10,
      "loss": 0.1038,
      "step": 34145
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.6704277119315952,
      "learning_rate": 7.784637098329217e-10,
      "loss": 0.0867,
      "step": 34150
    },
    {
      "epoch": 3.99,
      "grad_norm": 1.039383158977122,
      "learning_rate": 6.905764457554309e-10,
      "loss": 0.096,
      "step": 34155
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.8253035942519326,
      "learning_rate": 6.079517981788519e-10,
      "loss": 0.1033,
      "step": 34160
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.6901537550325304,
      "learning_rate": 5.305897844920527e-10,
      "loss": 0.0944,
      "step": 34165
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.6065534322052565,
      "learning_rate": 4.584904209847807e-10,
      "loss": 0.0915,
      "step": 34170
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.725501993545981,
      "learning_rate": 3.9165372283378467e-10,
      "loss": 0.1113,
      "step": 34175
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.6751466793515113,
      "learning_rate": 3.300797041083659e-10,
      "loss": 0.0942,
      "step": 34180
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.8102656387388302,
      "learning_rate": 2.737683777731537e-10,
      "loss": 0.1247,
      "step": 34185
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.7669960411364124,
      "learning_rate": 2.227197556770033e-10,
      "loss": 0.0901,
      "step": 34190
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.6535970616670373,
      "learning_rate": 1.7693384856964923e-10,
      "loss": 0.0972,
      "step": 34195
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7958982069060069,
      "learning_rate": 1.3641066608782728e-10,
      "loss": 0.1079,
      "step": 34200
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.6170226473389493,
      "learning_rate": 1.0115021676360137e-10,
      "loss": 0.0906,
      "step": 34205
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7297000805729437,
      "learning_rate": 7.115250801881246e-11,
      "loss": 0.0991,
      "step": 34210
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.6705272964389867,
      "learning_rate": 4.641754617062954e-11,
      "loss": 0.0744,
      "step": 34215
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7694244317658472,
      "learning_rate": 2.6945336423223054e-11,
      "loss": 0.133,
      "step": 34220
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7535135737427812,
      "learning_rate": 1.2735882873315952e-11,
      "loss": 0.142,
      "step": 34225
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.731573104772695,
      "learning_rate": 3.7891885185104004e-12,
      "loss": 0.1157,
      "step": 34230
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.7739159439442312,
      "learning_rate": 1.0525523785887004e-13,
      "loss": 0.112,
      "step": 34235
    },
    {
      "epoch": 4.0,
      "step": 34236,
      "total_flos": 3463540144865280.0,
      "train_loss": 0.4881485020316365,
      "train_runtime": 142779.1917,
      "train_samples_per_second": 0.959,
      "train_steps_per_second": 0.24
    }
  ],
  "logging_steps": 5,
  "max_steps": 34236,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 8900,
  "total_flos": 3463540144865280.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}