{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 70729,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00014138472196694425,
      "grad_norm": 14.262463569641113,
      "learning_rate": 9.998868922224264e-06,
      "loss": 0.0293,
      "step": 10
    },
    {
      "epoch": 0.0002827694439338885,
      "grad_norm": 0.261907696723938,
      "learning_rate": 9.997596459726563e-06,
      "loss": 0.022,
      "step": 20
    },
    {
      "epoch": 0.0004241541659008328,
      "grad_norm": 0.008495091460645199,
      "learning_rate": 9.996182612506893e-06,
      "loss": 0.0029,
      "step": 30
    },
    {
      "epoch": 0.000565538887867777,
      "grad_norm": 0.6746492385864258,
      "learning_rate": 9.994768765287224e-06,
      "loss": 0.0137,
      "step": 40
    },
    {
      "epoch": 0.0007069236098347213,
      "grad_norm": 0.005497786682099104,
      "learning_rate": 9.993354918067555e-06,
      "loss": 0.0012,
      "step": 50
    },
    {
      "epoch": 0.0008483083318016656,
      "grad_norm": 24.002790451049805,
      "learning_rate": 9.991941070847884e-06,
      "loss": 0.0209,
      "step": 60
    },
    {
      "epoch": 0.0009896930537686098,
      "grad_norm": 0.0002481811388861388,
      "learning_rate": 9.990527223628215e-06,
      "loss": 0.0121,
      "step": 70
    },
    {
      "epoch": 0.001131077775735554,
      "grad_norm": NaN,
      "learning_rate": 9.98939614585248e-06,
      "loss": 0.1261,
      "step": 80
    },
    {
      "epoch": 0.0012724624977024982,
      "grad_norm": 6.186431710375473e-05,
      "learning_rate": 9.98798229863281e-06,
      "loss": 0.0001,
      "step": 90
    },
    {
      "epoch": 0.0014138472196694426,
      "grad_norm": 0.03662971407175064,
      "learning_rate": 9.98656845141314e-06,
      "loss": 0.0015,
      "step": 100
    },
    {
      "epoch": 0.0015552319416363867,
      "grad_norm": 0.003412543796002865,
      "learning_rate": 9.985154604193471e-06,
      "loss": 0.0028,
      "step": 110
    },
    {
      "epoch": 0.001696616663603331,
      "grad_norm": 3.710695743560791,
      "learning_rate": 9.983740756973802e-06,
      "loss": 0.0097,
      "step": 120
    },
    {
      "epoch": 0.0018380013855702753,
      "grad_norm": 6.216172187123448e-05,
      "learning_rate": 9.982326909754133e-06,
      "loss": 0.0028,
      "step": 130
    },
    {
      "epoch": 0.0019793861075372197,
      "grad_norm": 0.0015779079403728247,
      "learning_rate": 9.980913062534462e-06,
      "loss": 0.0043,
      "step": 140
    },
    {
      "epoch": 0.002120770829504164,
      "grad_norm": 0.0011654989793896675,
      "learning_rate": 9.979499215314795e-06,
      "loss": 0.0012,
      "step": 150
    },
    {
      "epoch": 0.002262155551471108,
      "grad_norm": 21.838481903076172,
      "learning_rate": 9.978226752817091e-06,
      "loss": 0.9202,
      "step": 160
    },
    {
      "epoch": 0.002403540273438052,
      "grad_norm": 0.001239921199157834,
      "learning_rate": 9.976812905597422e-06,
      "loss": 0.0001,
      "step": 170
    },
    {
      "epoch": 0.0025449249954049963,
      "grad_norm": 45.4626350402832,
      "learning_rate": 9.975399058377753e-06,
      "loss": 0.0955,
      "step": 180
    },
    {
      "epoch": 0.002686309717371941,
      "grad_norm": 0.010774602182209492,
      "learning_rate": 9.973985211158082e-06,
      "loss": 0.0319,
      "step": 190
    },
    {
      "epoch": 0.002827694439338885,
      "grad_norm": 0.02312907762825489,
      "learning_rate": 9.972571363938413e-06,
      "loss": 0.0006,
      "step": 200
    },
    {
      "epoch": 0.0029690791613058293,
      "grad_norm": 0.0063919625245034695,
      "learning_rate": 9.971157516718744e-06,
      "loss": 0.0174,
      "step": 210
    },
    {
      "epoch": 0.0031104638832727734,
      "grad_norm": 0.0008793229353614151,
      "learning_rate": 9.969743669499075e-06,
      "loss": 0.1126,
      "step": 220
    },
    {
      "epoch": 0.0032518486052397176,
      "grad_norm": 43.13256072998047,
      "learning_rate": 9.968329822279406e-06,
      "loss": 0.0084,
      "step": 230
    },
    {
      "epoch": 0.003393233327206662,
      "grad_norm": 0.000771598017308861,
      "learning_rate": 9.966915975059735e-06,
      "loss": 0.0001,
      "step": 240
    },
    {
      "epoch": 0.0035346180491736064,
      "grad_norm": 0.005171002354472876,
      "learning_rate": 9.965502127840067e-06,
      "loss": 0.0016,
      "step": 250
    },
    {
      "epoch": 0.0036760027711405505,
      "grad_norm": 0.0008054971694946289,
      "learning_rate": 9.964088280620396e-06,
      "loss": 0.0002,
      "step": 260
    },
    {
      "epoch": 0.0038173874931074947,
      "grad_norm": 0.009220878593623638,
      "learning_rate": 9.962674433400727e-06,
      "loss": 0.0,
      "step": 270
    },
    {
      "epoch": 0.003958772215074439,
      "grad_norm": 4.363506178606258e-08,
      "learning_rate": 9.961260586181058e-06,
      "loss": 0.0004,
      "step": 280
    },
    {
      "epoch": 0.0041001569370413835,
      "grad_norm": 0.0033621739130467176,
      "learning_rate": 9.959846738961389e-06,
      "loss": 0.0002,
      "step": 290
    },
    {
      "epoch": 0.004241541659008328,
      "grad_norm": 0.00027329399017617106,
      "learning_rate": 9.95843289174172e-06,
      "loss": 0.0002,
      "step": 300
    },
    {
      "epoch": 0.004382926380975272,
      "grad_norm": 5.605544117770478e-08,
      "learning_rate": 9.957019044522049e-06,
      "loss": 0.0001,
      "step": 310
    },
    {
      "epoch": 0.004524311102942216,
      "grad_norm": 0.0004279848944861442,
      "learning_rate": 9.95560519730238e-06,
      "loss": 0.0,
      "step": 320
    },
    {
      "epoch": 0.00466569582490916,
      "grad_norm": 0.0,
      "learning_rate": 9.95419135008271e-06,
      "loss": 0.0,
      "step": 330
    },
    {
      "epoch": 0.004807080546876104,
      "grad_norm": 2.337314786160505e-09,
      "learning_rate": 9.952777502863042e-06,
      "loss": 0.0264,
      "step": 340
    },
    {
      "epoch": 0.0049484652688430485,
      "grad_norm": 0.0,
      "learning_rate": 9.951363655643372e-06,
      "loss": 0.0125,
      "step": 350
    },
    {
      "epoch": 0.005089849990809993,
      "grad_norm": 2.531542087069738e-08,
      "learning_rate": 9.949949808423702e-06,
      "loss": 0.001,
      "step": 360
    },
    {
      "epoch": 0.005231234712776938,
      "grad_norm": 0.020365700125694275,
      "learning_rate": 9.948535961204032e-06,
      "loss": 0.0002,
      "step": 370
    },
    {
      "epoch": 0.005372619434743882,
      "grad_norm": 0.0011259489692747593,
      "learning_rate": 9.947122113984363e-06,
      "loss": 0.0007,
      "step": 380
    },
    {
      "epoch": 0.005514004156710826,
      "grad_norm": 4.708147048950195,
      "learning_rate": 9.945708266764694e-06,
      "loss": 0.005,
      "step": 390
    },
    {
      "epoch": 0.00565538887867777,
      "grad_norm": 0.45584020018577576,
      "learning_rate": 9.944294419545025e-06,
      "loss": 0.0001,
      "step": 400
    },
    {
      "epoch": 0.005796773600644714,
      "grad_norm": 0.004194090608507395,
      "learning_rate": 9.942880572325356e-06,
      "loss": 0.0179,
      "step": 410
    },
    {
      "epoch": 0.0059381583226116585,
      "grad_norm": 6.750839020241983e-06,
      "learning_rate": 9.941466725105687e-06,
      "loss": 0.0044,
      "step": 420
    },
    {
      "epoch": 0.006079543044578603,
      "grad_norm": 0.00268085696734488,
      "learning_rate": 9.940052877886016e-06,
      "loss": 0.0004,
      "step": 430
    },
    {
      "epoch": 0.006220927766545547,
      "grad_norm": 0.005731990095227957,
      "learning_rate": 9.938639030666347e-06,
      "loss": 0.0017,
      "step": 440
    },
    {
      "epoch": 0.006362312488512491,
      "grad_norm": 1.0801550160977058e-06,
      "learning_rate": 9.937225183446678e-06,
      "loss": 0.0001,
      "step": 450
    },
    {
      "epoch": 0.006503697210479435,
      "grad_norm": 1.3552018174323166e-07,
      "learning_rate": 9.935811336227008e-06,
      "loss": 0.0001,
      "step": 460
    },
    {
      "epoch": 0.00664508193244638,
      "grad_norm": 0.012302791699767113,
      "learning_rate": 9.93439748900734e-06,
      "loss": 0.0073,
      "step": 470
    },
    {
      "epoch": 0.006786466654413324,
      "grad_norm": 9.02082347869873,
      "learning_rate": 9.932983641787669e-06,
      "loss": 0.0017,
      "step": 480
    },
    {
      "epoch": 0.006927851376380269,
      "grad_norm": 5.626787697110558e-06,
      "learning_rate": 9.931569794568e-06,
      "loss": 0.0,
      "step": 490
    },
    {
      "epoch": 0.007069236098347213,
      "grad_norm": 0.009398400783538818,
      "learning_rate": 9.93015594734833e-06,
      "loss": 0.0,
      "step": 500
    },
    {
      "epoch": 0.007210620820314157,
      "grad_norm": 0.03133725747466087,
      "learning_rate": 9.928742100128661e-06,
      "loss": 0.0002,
      "step": 510
    },
    {
      "epoch": 0.007352005542281101,
      "grad_norm": 0.002769770799204707,
      "learning_rate": 9.92732825290899e-06,
      "loss": 0.0001,
      "step": 520
    },
    {
      "epoch": 0.007493390264248045,
      "grad_norm": 0.030245915055274963,
      "learning_rate": 9.925914405689323e-06,
      "loss": 0.0044,
      "step": 530
    },
    {
      "epoch": 0.0076347749862149894,
      "grad_norm": 2.06316417461494e-05,
      "learning_rate": 9.924500558469654e-06,
      "loss": 0.0008,
      "step": 540
    },
    {
      "epoch": 0.007776159708181934,
      "grad_norm": 0.004097342025488615,
      "learning_rate": 9.923086711249983e-06,
      "loss": 0.0844,
      "step": 550
    },
    {
      "epoch": 0.007917544430148879,
      "grad_norm": 0.0,
      "learning_rate": 9.921672864030314e-06,
      "loss": 0.041,
      "step": 560
    },
    {
      "epoch": 0.008058929152115823,
      "grad_norm": 0.05861283466219902,
      "learning_rate": 9.920259016810645e-06,
      "loss": 0.0011,
      "step": 570
    },
    {
      "epoch": 0.008200313874082767,
      "grad_norm": 1.375386545987567e-06,
      "learning_rate": 9.918845169590975e-06,
      "loss": 0.003,
      "step": 580
    },
    {
      "epoch": 0.008341698596049711,
      "grad_norm": 0.06908536702394485,
      "learning_rate": 9.917431322371306e-06,
      "loss": 0.0254,
      "step": 590
    },
    {
      "epoch": 0.008483083318016655,
      "grad_norm": 9.233574019162916e-06,
      "learning_rate": 9.916017475151635e-06,
      "loss": 0.0,
      "step": 600
    },
    {
      "epoch": 0.0086244680399836,
      "grad_norm": 126.84648895263672,
      "learning_rate": 9.914603627931966e-06,
      "loss": 0.5102,
      "step": 610
    },
    {
      "epoch": 0.008765852761950544,
      "grad_norm": 1.834599424910266e-05,
      "learning_rate": 9.913189780712297e-06,
      "loss": 0.0009,
      "step": 620
    },
    {
      "epoch": 0.008907237483917488,
      "grad_norm": 0.09086769819259644,
      "learning_rate": 9.911775933492628e-06,
      "loss": 0.0,
      "step": 630
    },
    {
      "epoch": 0.009048622205884432,
      "grad_norm": 0.005661682225763798,
      "learning_rate": 9.910362086272957e-06,
      "loss": 0.002,
      "step": 640
    },
    {
      "epoch": 0.009190006927851376,
      "grad_norm": 1.5837044884392526e-06,
      "learning_rate": 9.908948239053288e-06,
      "loss": 0.0,
      "step": 650
    },
    {
      "epoch": 0.00933139164981832,
      "grad_norm": 8.83768643689109e-06,
      "learning_rate": 9.90753439183362e-06,
      "loss": 0.0012,
      "step": 660
    },
    {
      "epoch": 0.009472776371785264,
      "grad_norm": 0.0,
      "learning_rate": 9.90612054461395e-06,
      "loss": 0.0,
      "step": 670
    },
    {
      "epoch": 0.009614161093752209,
      "grad_norm": 1.2138000329287024e-06,
      "learning_rate": 9.90470669739428e-06,
      "loss": 0.0006,
      "step": 680
    },
    {
      "epoch": 0.009755545815719153,
      "grad_norm": 0.0002607766364235431,
      "learning_rate": 9.903292850174611e-06,
      "loss": 0.0502,
      "step": 690
    },
    {
      "epoch": 0.009896930537686097,
      "grad_norm": 0.0,
      "learning_rate": 9.901879002954942e-06,
      "loss": 0.0006,
      "step": 700
    },
    {
      "epoch": 0.010038315259653041,
      "grad_norm": 1.0272954398260481e-07,
      "learning_rate": 9.900465155735271e-06,
      "loss": 0.0094,
      "step": 710
    },
    {
      "epoch": 0.010179699981619985,
      "grad_norm": 4.787257257277133e-09,
      "learning_rate": 9.899051308515602e-06,
      "loss": 0.0001,
      "step": 720
    },
    {
      "epoch": 0.010321084703586931,
      "grad_norm": 9.609036445617676,
      "learning_rate": 9.897637461295933e-06,
      "loss": 0.1762,
      "step": 730
    },
    {
      "epoch": 0.010462469425553875,
      "grad_norm": 4.663838204521653e-09,
      "learning_rate": 9.896223614076264e-06,
      "loss": 0.0,
      "step": 740
    },
    {
      "epoch": 0.01060385414752082,
      "grad_norm": 1.970793483963007e-08,
      "learning_rate": 9.894809766856595e-06,
      "loss": 0.0,
      "step": 750
    },
    {
      "epoch": 0.010745238869487764,
      "grad_norm": 3.562535047531128,
      "learning_rate": 9.893395919636924e-06,
      "loss": 0.0012,
      "step": 760
    },
    {
      "epoch": 0.010886623591454708,
      "grad_norm": 2.3147062222506065e-07,
      "learning_rate": 9.891982072417255e-06,
      "loss": 0.0,
      "step": 770
    },
    {
      "epoch": 0.011028008313421652,
      "grad_norm": 0.00020299811149016023,
      "learning_rate": 9.890568225197586e-06,
      "loss": 0.0004,
      "step": 780
    },
    {
      "epoch": 0.011169393035388596,
      "grad_norm": 0.0033032905776053667,
      "learning_rate": 9.889154377977917e-06,
      "loss": 0.2141,
      "step": 790
    },
    {
      "epoch": 0.01131077775735554,
      "grad_norm": 0.0005747198592871428,
      "learning_rate": 9.887740530758247e-06,
      "loss": 0.0002,
      "step": 800
    },
    {
      "epoch": 0.011452162479322485,
      "grad_norm": 5.074514319858281e-08,
      "learning_rate": 9.886326683538577e-06,
      "loss": 0.1234,
      "step": 810
    },
    {
      "epoch": 0.011593547201289429,
      "grad_norm": 2.83273675449891e-06,
      "learning_rate": 9.88491283631891e-06,
      "loss": 0.0099,
      "step": 820
    },
    {
      "epoch": 0.011734931923256373,
      "grad_norm": 5.2760519793082494e-06,
      "learning_rate": 9.883498989099238e-06,
      "loss": 0.0006,
      "step": 830
    },
    {
      "epoch": 0.011876316645223317,
      "grad_norm": 389.735595703125,
      "learning_rate": 9.88208514187957e-06,
      "loss": 0.3943,
      "step": 840
    },
    {
      "epoch": 0.012017701367190261,
      "grad_norm": 1.1932961285765487e-07,
      "learning_rate": 9.8806712946599e-06,
      "loss": 0.0008,
      "step": 850
    },
    {
      "epoch": 0.012159086089157205,
      "grad_norm": 0.007641827221959829,
      "learning_rate": 9.879257447440231e-06,
      "loss": 0.2881,
      "step": 860
    },
    {
      "epoch": 0.01230047081112415,
      "grad_norm": 5.116144730088479e-10,
      "learning_rate": 9.877843600220562e-06,
      "loss": 0.0512,
      "step": 870
    },
    {
      "epoch": 0.012441855533091094,
      "grad_norm": 0.0,
      "learning_rate": 9.876429753000891e-06,
      "loss": 0.0013,
      "step": 880
    },
    {
      "epoch": 0.012583240255058038,
      "grad_norm": 0.002561827888712287,
      "learning_rate": 9.875015905781222e-06,
      "loss": 0.0,
      "step": 890
    },
    {
      "epoch": 0.012724624977024982,
      "grad_norm": 0.0,
      "learning_rate": 9.873602058561553e-06,
      "loss": 0.0,
      "step": 900
    },
    {
      "epoch": 0.012866009698991926,
      "grad_norm": 1.2506252527236938,
      "learning_rate": 9.872188211341883e-06,
      "loss": 0.0004,
      "step": 910
    },
    {
      "epoch": 0.01300739442095887,
      "grad_norm": 0.0007419384201057255,
      "learning_rate": 9.870774364122214e-06,
      "loss": 0.0001,
      "step": 920
    },
    {
      "epoch": 0.013148779142925815,
      "grad_norm": 3.7602691804750066e-07,
      "learning_rate": 9.869360516902543e-06,
      "loss": 0.0,
      "step": 930
    },
    {
      "epoch": 0.01329016386489276,
      "grad_norm": 7.051665761537151e-06,
      "learning_rate": 9.867946669682874e-06,
      "loss": 0.0,
      "step": 940
    },
    {
      "epoch": 0.013431548586859705,
      "grad_norm": 12.585866928100586,
      "learning_rate": 9.866532822463205e-06,
      "loss": 0.0094,
      "step": 950
    },
    {
      "epoch": 0.013572933308826649,
      "grad_norm": 0.0018373159691691399,
      "learning_rate": 9.865118975243536e-06,
      "loss": 0.0002,
      "step": 960
    },
    {
      "epoch": 0.013714318030793593,
      "grad_norm": 0.5143635272979736,
      "learning_rate": 9.863705128023865e-06,
      "loss": 0.0008,
      "step": 970
    },
    {
      "epoch": 0.013855702752760537,
      "grad_norm": 404.3897705078125,
      "learning_rate": 9.862291280804198e-06,
      "loss": 0.1753,
      "step": 980
    },
    {
      "epoch": 0.013997087474727481,
      "grad_norm": 0.0004114680632483214,
      "learning_rate": 9.860877433584529e-06,
      "loss": 0.0033,
      "step": 990
    },
    {
      "epoch": 0.014138472196694426,
      "grad_norm": 0.0,
      "learning_rate": 9.859463586364858e-06,
      "loss": 0.0004,
      "step": 1000
    },
    {
      "epoch": 0.01427985691866137,
      "grad_norm": 5.697476694876968e-07,
      "learning_rate": 9.858049739145189e-06,
      "loss": 0.0529,
      "step": 1010
    },
    {
      "epoch": 0.014421241640628314,
      "grad_norm": 0.10179593414068222,
      "learning_rate": 9.85663589192552e-06,
      "loss": 0.0,
      "step": 1020
    },
    {
      "epoch": 0.014562626362595258,
      "grad_norm": 0.003961221780627966,
      "learning_rate": 9.85522204470585e-06,
      "loss": 0.0057,
      "step": 1030
    },
    {
      "epoch": 0.014704011084562202,
      "grad_norm": 1.481937670178013e-05,
      "learning_rate": 9.853808197486181e-06,
      "loss": 0.0005,
      "step": 1040
    },
    {
      "epoch": 0.014845395806529146,
      "grad_norm": 0.0,
      "learning_rate": 9.85239435026651e-06,
      "loss": 0.0322,
      "step": 1050
    },
    {
      "epoch": 0.01498678052849609,
      "grad_norm": 0.0,
      "learning_rate": 9.850980503046841e-06,
      "loss": 0.0,
      "step": 1060
    },
    {
      "epoch": 0.015128165250463035,
      "grad_norm": 0.0175678338855505,
      "learning_rate": 9.849566655827172e-06,
      "loss": 0.0072,
      "step": 1070
    },
    {
      "epoch": 0.015269549972429979,
      "grad_norm": 8.276358016701124e-07,
      "learning_rate": 9.848152808607503e-06,
      "loss": 0.0,
      "step": 1080
    },
    {
      "epoch": 0.015410934694396923,
      "grad_norm": 7.149280918383738e-07,
      "learning_rate": 9.846738961387832e-06,
      "loss": 0.0002,
      "step": 1090
    },
    {
      "epoch": 0.015552319416363867,
      "grad_norm": 4.5070461140994666e-08,
      "learning_rate": 9.845325114168165e-06,
      "loss": 0.009,
      "step": 1100
    },
    {
      "epoch": 0.01569370413833081,
      "grad_norm": 0.0015607335371896625,
      "learning_rate": 9.843911266948496e-06,
      "loss": 0.0,
      "step": 1110
    },
    {
      "epoch": 0.015835088860297757,
      "grad_norm": 0.020774630829691887,
      "learning_rate": 9.842497419728825e-06,
      "loss": 0.0,
      "step": 1120
    },
    {
      "epoch": 0.0159764735822647,
      "grad_norm": 0.0013689767802134156,
      "learning_rate": 9.841083572509156e-06,
      "loss": 0.0293,
      "step": 1130
    },
    {
      "epoch": 0.016117858304231646,
      "grad_norm": 0.19884544610977173,
      "learning_rate": 9.839669725289486e-06,
      "loss": 0.0003,
      "step": 1140
    },
    {
      "epoch": 0.016259243026198588,
      "grad_norm": 8.745597028791963e-07,
      "learning_rate": 9.838255878069817e-06,
      "loss": 0.0179,
      "step": 1150
    },
    {
      "epoch": 0.016400627748165534,
      "grad_norm": 0.00747031532227993,
      "learning_rate": 9.836842030850146e-06,
      "loss": 0.0003,
      "step": 1160
    },
    {
      "epoch": 0.016542012470132476,
      "grad_norm": 0.00021051010116934776,
      "learning_rate": 9.835428183630477e-06,
      "loss": 0.0001,
      "step": 1170
    },
    {
      "epoch": 0.016683397192099422,
      "grad_norm": 6.286531117893901e-08,
      "learning_rate": 9.834014336410808e-06,
      "loss": 0.0077,
      "step": 1180
    },
    {
      "epoch": 0.016824781914066365,
      "grad_norm": 0.0003884889301843941,
      "learning_rate": 9.832600489191139e-06,
      "loss": 0.0,
      "step": 1190
    },
    {
      "epoch": 0.01696616663603331,
      "grad_norm": 5.879924325213892e-10,
      "learning_rate": 9.83118664197147e-06,
      "loss": 0.0037,
      "step": 1200
    },
    {
      "epoch": 0.017107551358000253,
      "grad_norm": 9.322138616596476e-10,
      "learning_rate": 9.829772794751799e-06,
      "loss": 0.0005,
      "step": 1210
    },
    {
      "epoch": 0.0172489360799672,
      "grad_norm": 0.0001175247089122422,
      "learning_rate": 9.82835894753213e-06,
      "loss": 0.0011,
      "step": 1220
    },
    {
      "epoch": 0.01739032080193414,
      "grad_norm": 0.0,
      "learning_rate": 9.82694510031246e-06,
      "loss": 0.0002,
      "step": 1230
    },
    {
      "epoch": 0.017531705523901087,
      "grad_norm": 0.004724294878542423,
      "learning_rate": 9.825531253092792e-06,
      "loss": 0.0005,
      "step": 1240
    },
    {
      "epoch": 0.017673090245868033,
      "grad_norm": 5.34083943737329e-10,
      "learning_rate": 9.824117405873122e-06,
      "loss": 0.0005,
      "step": 1250
    },
    {
      "epoch": 0.017814474967834976,
      "grad_norm": 0.0,
      "learning_rate": 9.822703558653453e-06,
      "loss": 0.0001,
      "step": 1260
    },
    {
      "epoch": 0.01795585968980192,
      "grad_norm": 0.0,
      "learning_rate": 9.821289711433784e-06,
      "loss": 0.0313,
      "step": 1270
    },
    {
      "epoch": 0.018097244411768864,
      "grad_norm": 5.57953516722165e-10,
      "learning_rate": 9.819875864214113e-06,
      "loss": 0.0005,
      "step": 1280
    },
    {
      "epoch": 0.01823862913373581,
      "grad_norm": 5.946174042037455e-07,
      "learning_rate": 9.818462016994444e-06,
      "loss": 0.0005,
      "step": 1290
    },
    {
      "epoch": 0.018380013855702752,
      "grad_norm": 0.00025980654754675925,
      "learning_rate": 9.817048169774775e-06,
      "loss": 0.0003,
      "step": 1300
    },
    {
      "epoch": 0.018521398577669698,
      "grad_norm": 1.0725130117350545e-08,
      "learning_rate": 9.815634322555106e-06,
      "loss": 0.0,
      "step": 1310
    },
    {
      "epoch": 0.01866278329963664,
      "grad_norm": 0.00037257836083881557,
      "learning_rate": 9.814220475335437e-06,
      "loss": 0.0002,
      "step": 1320
    },
    {
      "epoch": 0.018804168021603587,
      "grad_norm": 1.98855842903356e-09,
      "learning_rate": 9.812806628115766e-06,
      "loss": 0.0009,
      "step": 1330
    },
    {
      "epoch": 0.01894555274357053,
      "grad_norm": 0.00026435984182171524,
      "learning_rate": 9.811392780896097e-06,
      "loss": 0.0078,
      "step": 1340
    },
    {
      "epoch": 0.019086937465537475,
      "grad_norm": 0.0,
      "learning_rate": 9.809978933676428e-06,
      "loss": 0.0,
      "step": 1350
    },
    {
      "epoch": 0.019228322187504417,
      "grad_norm": 0.0,
      "learning_rate": 9.808565086456758e-06,
      "loss": 0.0161,
      "step": 1360
    },
    {
      "epoch": 0.019369706909471363,
      "grad_norm": 0.003482252126559615,
      "learning_rate": 9.80715123923709e-06,
      "loss": 0.0157,
      "step": 1370
    },
    {
      "epoch": 0.019511091631438306,
      "grad_norm": 6.263466275413521e-06,
      "learning_rate": 9.805737392017418e-06,
      "loss": 0.0,
      "step": 1380
    },
    {
      "epoch": 0.01965247635340525,
      "grad_norm": 2.329007387161255,
      "learning_rate": 9.804323544797751e-06,
      "loss": 0.0081,
      "step": 1390
    },
    {
      "epoch": 0.019793861075372194,
      "grad_norm": 0.0,
      "learning_rate": 9.80290969757808e-06,
      "loss": 0.1483,
      "step": 1400
    },
    {
      "epoch": 0.01993524579733914,
      "grad_norm": 2.6276164054870605,
      "learning_rate": 9.801495850358411e-06,
      "loss": 0.1404,
      "step": 1410
    },
    {
      "epoch": 0.020076630519306082,
      "grad_norm": 0.0,
      "learning_rate": 9.800082003138742e-06,
      "loss": 0.0,
      "step": 1420
    },
    {
      "epoch": 0.020218015241273028,
      "grad_norm": 2.763486554613337e-05,
      "learning_rate": 9.798668155919073e-06,
      "loss": 0.0006,
      "step": 1430
    },
    {
      "epoch": 0.02035939996323997,
      "grad_norm": 0.0,
      "learning_rate": 9.797254308699404e-06,
      "loss": 0.0007,
      "step": 1440
    },
    {
      "epoch": 0.020500784685206917,
      "grad_norm": 3.4370737012068275e-07,
      "learning_rate": 9.795840461479733e-06,
      "loss": 0.0,
      "step": 1450
    },
    {
      "epoch": 0.020642169407173862,
      "grad_norm": 1.4225420841285086e-08,
      "learning_rate": 9.794426614260064e-06,
      "loss": 0.0014,
      "step": 1460
    },
    {
      "epoch": 0.020783554129140805,
      "grad_norm": 0.0,
      "learning_rate": 9.793012767040394e-06,
      "loss": 0.0003,
      "step": 1470
    },
    {
      "epoch": 0.02092493885110775,
      "grad_norm": 7.01696123428519e-08,
      "learning_rate": 9.791598919820725e-06,
      "loss": 0.0,
      "step": 1480
    },
    {
      "epoch": 0.021066323573074693,
      "grad_norm": 0.0,
      "learning_rate": 9.790185072601055e-06,
      "loss": 0.0006,
      "step": 1490
    },
    {
      "epoch": 0.02120770829504164,
      "grad_norm": 0.0,
      "learning_rate": 9.788771225381385e-06,
      "loss": 0.0,
      "step": 1500
    },
    {
      "epoch": 0.02134909301700858,
      "grad_norm": 5.084136489585944e-08,
      "learning_rate": 9.787357378161716e-06,
      "loss": 0.0001,
      "step": 1510
    },
    {
      "epoch": 0.021490477738975527,
      "grad_norm": 0.011734791100025177,
      "learning_rate": 9.785943530942047e-06,
      "loss": 0.0,
      "step": 1520
    },
    {
      "epoch": 0.02163186246094247,
      "grad_norm": 0.000401741242967546,
      "learning_rate": 9.784529683722378e-06,
      "loss": 0.0,
      "step": 1530
    },
    {
      "epoch": 0.021773247182909416,
      "grad_norm": 0.0,
      "learning_rate": 9.783115836502707e-06,
      "loss": 0.0,
      "step": 1540
    },
    {
      "epoch": 0.021914631904876358,
      "grad_norm": 0.6668468713760376,
      "learning_rate": 9.78170198928304e-06,
      "loss": 0.0142,
      "step": 1550
    },
    {
      "epoch": 0.022056016626843304,
      "grad_norm": 1.3297108969467786e-09,
      "learning_rate": 9.78028814206337e-06,
      "loss": 0.0,
      "step": 1560
    },
    {
      "epoch": 0.022197401348810247,
      "grad_norm": 0.0,
      "learning_rate": 9.7788742948437e-06,
      "loss": 0.0147,
      "step": 1570
    },
    {
      "epoch": 0.022338786070777192,
      "grad_norm": 1.904671087515908e-08,
      "learning_rate": 9.77746044762403e-06,
      "loss": 0.0003,
      "step": 1580
    },
    {
      "epoch": 0.022480170792744135,
      "grad_norm": 0.0,
      "learning_rate": 9.776046600404361e-06,
      "loss": 0.0,
      "step": 1590
    },
    {
      "epoch": 0.02262155551471108,
      "grad_norm": 0.0003447172639425844,
      "learning_rate": 9.774632753184692e-06,
      "loss": 0.0,
      "step": 1600
    },
    {
      "epoch": 0.022762940236678023,
      "grad_norm": 1.9294568573968718e-06,
      "learning_rate": 9.773218905965021e-06,
      "loss": 0.0019,
      "step": 1610
    },
    {
      "epoch": 0.02290432495864497,
      "grad_norm": 1.2258342504501343,
      "learning_rate": 9.771805058745352e-06,
      "loss": 0.0027,
      "step": 1620
    },
    {
      "epoch": 0.02304570968061191,
      "grad_norm": 1.6706542282918235e-07,
      "learning_rate": 9.770391211525683e-06,
      "loss": 0.0006,
      "step": 1630
    },
    {
      "epoch": 0.023187094402578858,
      "grad_norm": 191.74658203125,
      "learning_rate": 9.768977364306014e-06,
      "loss": 0.0326,
      "step": 1640
    },
    {
      "epoch": 0.0233284791245458,
      "grad_norm": 6.547410011291504,
      "learning_rate": 9.767563517086345e-06,
      "loss": 0.0013,
      "step": 1650
    },
    {
      "epoch": 0.023469863846512746,
      "grad_norm": 1.1773488850508329e-08,
      "learning_rate": 9.766149669866674e-06,
      "loss": 0.0,
      "step": 1660
    },
    {
      "epoch": 0.023611248568479692,
      "grad_norm": 0.0,
      "learning_rate": 9.764735822647007e-06,
      "loss": 0.0,
      "step": 1670
    },
    {
      "epoch": 0.023752633290446634,
      "grad_norm": 9.123692512512207,
      "learning_rate": 9.763321975427336e-06,
      "loss": 0.0782,
      "step": 1680
    },
    {
      "epoch": 0.02389401801241358,
      "grad_norm": 0.000251657678745687,
      "learning_rate": 9.761908128207667e-06,
      "loss": 0.0005,
      "step": 1690
    },
    {
      "epoch": 0.024035402734380523,
      "grad_norm": 0.2819625735282898,
      "learning_rate": 9.760494280987997e-06,
      "loss": 0.0001,
      "step": 1700
    },
    {
      "epoch": 0.02417678745634747,
      "grad_norm": 6.084712733667175e-10,
      "learning_rate": 9.759080433768328e-06,
      "loss": 0.0,
      "step": 1710
    },
    {
      "epoch": 0.02431817217831441,
      "grad_norm": 0.0,
      "learning_rate": 9.757666586548659e-06,
      "loss": 0.0,
      "step": 1720
    },
    {
      "epoch": 0.024459556900281357,
      "grad_norm": 0.02435947023332119,
      "learning_rate": 9.756252739328988e-06,
      "loss": 0.0001,
      "step": 1730
    },
    {
      "epoch": 0.0246009416222483,
      "grad_norm": 9.704025433165953e-05,
      "learning_rate": 9.754838892109319e-06,
      "loss": 0.0011,
      "step": 1740
    },
    {
      "epoch": 0.024742326344215245,
      "grad_norm": 496.8305969238281,
      "learning_rate": 9.75342504488965e-06,
      "loss": 0.2024,
      "step": 1750
    },
    {
      "epoch": 0.024883711066182188,
      "grad_norm": 0.0008469384629279375,
      "learning_rate": 9.752011197669981e-06,
      "loss": 0.0003,
      "step": 1760
    },
    {
      "epoch": 0.025025095788149133,
      "grad_norm": 0.0,
      "learning_rate": 9.750597350450312e-06,
      "loss": 0.0,
      "step": 1770
    },
    {
      "epoch": 0.025166480510116076,
      "grad_norm": 0.0,
      "learning_rate": 9.749183503230641e-06,
      "loss": 0.6485,
      "step": 1780
    },
    {
      "epoch": 0.025307865232083022,
      "grad_norm": 0.2815660238265991,
      "learning_rate": 9.747769656010972e-06,
      "loss": 0.0009,
      "step": 1790
    },
    {
      "epoch": 0.025449249954049964,
      "grad_norm": 0.050988614559173584,
      "learning_rate": 9.746355808791303e-06,
      "loss": 0.0029,
      "step": 1800
    },
    {
      "epoch": 0.02559063467601691,
      "grad_norm": 0.026839226484298706,
      "learning_rate": 9.744941961571633e-06,
      "loss": 0.0,
      "step": 1810
    },
    {
      "epoch": 0.025732019397983853,
      "grad_norm": 0.0014148855116218328,
      "learning_rate": 9.743528114351964e-06,
      "loss": 0.0001,
      "step": 1820
    },
    {
      "epoch": 0.0258734041199508,
      "grad_norm": 0.013273480348289013,
      "learning_rate": 9.742114267132295e-06,
      "loss": 0.0,
      "step": 1830
    },
    {
      "epoch": 0.02601478884191774,
      "grad_norm": 0.0,
      "learning_rate": 9.740700419912626e-06,
      "loss": 0.0,
      "step": 1840
    },
    {
      "epoch": 0.026156173563884687,
      "grad_norm": 0.0,
      "learning_rate": 9.739286572692955e-06,
      "loss": 0.0,
      "step": 1850
    },
    {
      "epoch": 0.02629755828585163,
      "grad_norm": 4.456249371287413e-06,
      "learning_rate": 9.737872725473286e-06,
      "loss": 0.0024,
      "step": 1860
    },
    {
      "epoch": 0.026438943007818575,
      "grad_norm": 0.022046441212296486,
      "learning_rate": 9.736458878253617e-06,
      "loss": 0.0017,
      "step": 1870
    },
    {
      "epoch": 0.02658032772978552,
      "grad_norm": 5.172800521258125e-10,
      "learning_rate": 9.735045031033948e-06,
      "loss": 0.1096,
      "step": 1880
    },
    {
      "epoch": 0.026721712451752463,
      "grad_norm": 0.9113357663154602,
      "learning_rate": 9.733631183814279e-06,
      "loss": 0.0002,
      "step": 1890
    },
    {
      "epoch": 0.02686309717371941,
      "grad_norm": 0.0,
      "learning_rate": 9.732217336594608e-06,
      "loss": 0.0,
      "step": 1900
    },
    {
      "epoch": 0.027004481895686352,
      "grad_norm": 0.0,
      "learning_rate": 9.730803489374939e-06,
      "loss": 0.0,
      "step": 1910
    },
    {
      "epoch": 0.027145866617653298,
      "grad_norm": 3.7537608577054016e-09,
      "learning_rate": 9.72938964215527e-06,
      "loss": 0.0,
      "step": 1920
    },
    {
      "epoch": 0.02728725133962024,
      "grad_norm": 0.0,
      "learning_rate": 9.7279757949356e-06,
      "loss": 0.0,
      "step": 1930
    },
    {
      "epoch": 0.027428636061587186,
      "grad_norm": 0.0,
      "learning_rate": 9.72656194771593e-06,
      "loss": 0.0027,
      "step": 1940
    },
    {
      "epoch": 0.02757002078355413,
      "grad_norm": 0.007947119884192944,
      "learning_rate": 9.72514810049626e-06,
      "loss": 0.0004,
      "step": 1950
    },
    {
      "epoch": 0.027711405505521074,
      "grad_norm": 0.0,
      "learning_rate": 9.723734253276593e-06,
      "loss": 0.0272,
      "step": 1960
    },
    {
      "epoch": 0.027852790227488017,
      "grad_norm": 0.0,
      "learning_rate": 9.722320406056922e-06,
      "loss": 1.0422,
      "step": 1970
    },
    {
      "epoch": 0.027994174949454963,
      "grad_norm": 3.040394631170784e-06,
      "learning_rate": 9.720906558837253e-06,
      "loss": 0.006,
      "step": 1980
    },
    {
      "epoch": 0.028135559671421905,
      "grad_norm": 0.00020326573576312512,
      "learning_rate": 9.719492711617584e-06,
      "loss": 0.0,
      "step": 1990
    },
    {
      "epoch": 0.02827694439338885,
      "grad_norm": 0.07844200730323792,
      "learning_rate": 9.718078864397915e-06,
      "loss": 0.1937,
      "step": 2000
    },
    {
      "epoch": 0.028418329115355793,
      "grad_norm": 2.4662628561600286e-07,
      "learning_rate": 9.716665017178246e-06,
      "loss": 0.0413,
      "step": 2010
    },
    {
      "epoch": 0.02855971383732274,
      "grad_norm": 0.013099907897412777,
      "learning_rate": 9.715251169958575e-06,
      "loss": 0.0023,
      "step": 2020
    },
    {
      "epoch": 0.028701098559289682,
      "grad_norm": 0.0,
      "learning_rate": 9.713837322738906e-06,
      "loss": 0.0,
      "step": 2030
    },
    {
      "epoch": 0.028842483281256628,
      "grad_norm": 8.683192572789267e-05,
      "learning_rate": 9.712423475519236e-06,
      "loss": 0.0008,
      "step": 2040
    },
    {
      "epoch": 0.02898386800322357,
      "grad_norm": 4.900627573078964e-06,
      "learning_rate": 9.711009628299567e-06,
      "loss": 0.0228,
      "step": 2050
    },
    {
      "epoch": 0.029125252725190516,
      "grad_norm": 2.4108876459649764e-06,
      "learning_rate": 9.709595781079896e-06,
      "loss": 0.0102,
      "step": 2060
    },
    {
      "epoch": 0.02926663744715746,
      "grad_norm": 0.0001849499822128564,
      "learning_rate": 9.708181933860227e-06,
      "loss": 0.0001,
      "step": 2070
    },
    {
      "epoch": 0.029408022169124404,
      "grad_norm": 2.161930751753971e-06,
      "learning_rate": 9.706768086640558e-06,
      "loss": 0.0,
      "step": 2080
    },
    {
      "epoch": 0.02954940689109135,
      "grad_norm": 0.0,
      "learning_rate": 9.705354239420889e-06,
      "loss": 0.0,
      "step": 2090
    },
    {
      "epoch": 0.029690791613058293,
      "grad_norm": 0.0006212295265868306,
      "learning_rate": 9.70394039220122e-06,
      "loss": 0.0,
      "step": 2100
    },
    {
      "epoch": 0.02983217633502524,
      "grad_norm": 0.0,
      "learning_rate": 9.702526544981549e-06,
      "loss": 0.0,
      "step": 2110
    },
    {
      "epoch": 0.02997356105699218,
      "grad_norm": 2.4256529808044434,
      "learning_rate": 9.701112697761882e-06,
      "loss": 0.0011,
      "step": 2120
    },
    {
      "epoch": 0.030114945778959127,
      "grad_norm": 2.3783270819421887e-07,
      "learning_rate": 9.69969885054221e-06,
      "loss": 0.0,
      "step": 2130
    },
    {
      "epoch": 0.03025633050092607,
      "grad_norm": 0.0,
      "learning_rate": 9.698285003322542e-06,
      "loss": 0.0,
      "step": 2140
    },
    {
      "epoch": 0.030397715222893015,
      "grad_norm": 2.2173534119929172e-08,
      "learning_rate": 9.696871156102872e-06,
      "loss": 0.0021,
      "step": 2150
    },
    {
      "epoch": 0.030539099944859958,
      "grad_norm": 9.328923411544565e-09,
      "learning_rate": 9.695457308883203e-06,
      "loss": 0.0012,
      "step": 2160
    },
    {
      "epoch": 0.030680484666826904,
      "grad_norm": 0.021909872069954872,
      "learning_rate": 9.694043461663534e-06,
      "loss": 0.0001,
      "step": 2170
    },
    {
      "epoch": 0.030821869388793846,
      "grad_norm": 0.0,
      "learning_rate": 9.692770999165832e-06,
      "loss": 1.332,
      "step": 2180
    },
    {
      "epoch": 0.030963254110760792,
      "grad_norm": 4.701133371476374e-10,
      "learning_rate": 9.691357151946161e-06,
      "loss": 1.1071,
      "step": 2190
    },
    {
      "epoch": 0.031104638832727734,
      "grad_norm": 0.08870384097099304,
      "learning_rate": 9.689943304726492e-06,
      "loss": 0.2522,
      "step": 2200
    },
    {
      "epoch": 0.03124602355469468,
      "grad_norm": 0.0,
      "learning_rate": 9.688529457506823e-06,
      "loss": 0.0001,
      "step": 2210
    },
    {
      "epoch": 0.03138740827666162,
      "grad_norm": 0.00017338985344395041,
      "learning_rate": 9.687115610287154e-06,
      "loss": 0.0,
      "step": 2220
    },
    {
      "epoch": 0.03152879299862857,
      "grad_norm": 0.0,
      "learning_rate": 9.685701763067483e-06,
      "loss": 0.0274,
      "step": 2230
    },
    {
      "epoch": 0.031670177720595515,
      "grad_norm": 3.30842375717566e-08,
      "learning_rate": 9.684287915847814e-06,
      "loss": 0.0,
      "step": 2240
    },
    {
      "epoch": 0.031811562442562454,
      "grad_norm": 0.00010456040763529018,
      "learning_rate": 9.682874068628145e-06,
      "loss": 0.0,
      "step": 2250
    },
    {
      "epoch": 0.0319529471645294,
      "grad_norm": 0.0,
      "learning_rate": 9.681460221408476e-06,
      "loss": 0.0001,
      "step": 2260
    },
    {
      "epoch": 0.032094331886496345,
      "grad_norm": 0.0,
      "learning_rate": 9.680046374188807e-06,
      "loss": 0.0354,
      "step": 2270
    },
    {
      "epoch": 0.03223571660846329,
      "grad_norm": 0.0,
      "learning_rate": 9.678632526969136e-06,
      "loss": 0.0,
      "step": 2280
    },
    {
      "epoch": 0.03237710133043024,
      "grad_norm": 0.00019240380788687617,
      "learning_rate": 9.677218679749467e-06,
      "loss": 0.6485,
      "step": 2290
    },
    {
      "epoch": 0.032518486052397176,
      "grad_norm": 0.0,
      "learning_rate": 9.675804832529797e-06,
      "loss": 0.0,
      "step": 2300
    },
    {
      "epoch": 0.03265987077436412,
      "grad_norm": 0.0,
      "learning_rate": 9.674390985310128e-06,
      "loss": 0.0,
      "step": 2310
    },
    {
      "epoch": 0.03280125549633107,
      "grad_norm": 2.814327444866649e-06,
      "learning_rate": 9.672977138090459e-06,
      "loss": 0.0001,
      "step": 2320
    },
    {
      "epoch": 0.032942640218298014,
      "grad_norm": 0.008589006029069424,
      "learning_rate": 9.671563290870788e-06,
      "loss": 0.6798,
      "step": 2330
    },
    {
      "epoch": 0.03308402494026495,
      "grad_norm": 0.0,
      "learning_rate": 9.670149443651121e-06,
      "loss": 0.0,
      "step": 2340
    },
    {
      "epoch": 0.0332254096622319,
      "grad_norm": 0.03232518583536148,
      "learning_rate": 9.66873559643145e-06,
      "loss": 0.0003,
      "step": 2350
    },
    {
      "epoch": 0.033366794384198845,
      "grad_norm": 282.4214782714844,
      "learning_rate": 9.667321749211781e-06,
      "loss": 0.0356,
      "step": 2360
    },
    {
      "epoch": 0.03350817910616579,
      "grad_norm": 1.9086559177594609e-07,
      "learning_rate": 9.665907901992112e-06,
      "loss": 0.0008,
      "step": 2370
    },
    {
      "epoch": 0.03364956382813273,
      "grad_norm": 0.0,
      "learning_rate": 9.664494054772443e-06,
      "loss": 0.0003,
      "step": 2380
    },
    {
      "epoch": 0.033790948550099675,
      "grad_norm": 0.0,
      "learning_rate": 9.663080207552773e-06,
      "loss": 0.0,
      "step": 2390
    },
    {
      "epoch": 0.03393233327206662,
      "grad_norm": 0.09482633322477341,
      "learning_rate": 9.661666360333103e-06,
      "loss": 0.0028,
      "step": 2400
    },
    {
      "epoch": 0.03407371799403357,
      "grad_norm": 0.00021386062144301832,
      "learning_rate": 9.660252513113433e-06,
      "loss": 0.0002,
      "step": 2410
    },
    {
      "epoch": 0.034215102716000506,
      "grad_norm": 0.004378439858555794,
      "learning_rate": 9.658838665893764e-06,
      "loss": 0.0029,
      "step": 2420
    },
    {
      "epoch": 0.03435648743796745,
      "grad_norm": 0.0,
      "learning_rate": 9.657424818674095e-06,
      "loss": 0.0,
      "step": 2430
    },
    {
      "epoch": 0.0344978721599344,
      "grad_norm": 0.0006046416237950325,
      "learning_rate": 9.656010971454426e-06,
      "loss": 0.0,
      "step": 2440
    },
    {
      "epoch": 0.034639256881901344,
      "grad_norm": 5.513738909890264e-10,
      "learning_rate": 9.654597124234755e-06,
      "loss": 0.0006,
      "step": 2450
    },
    {
      "epoch": 0.03478064160386828,
      "grad_norm": 0.0017032412579283118,
      "learning_rate": 9.653183277015086e-06,
      "loss": 0.0,
      "step": 2460
    },
    {
      "epoch": 0.03492202632583523,
      "grad_norm": 5.506002875854676e-10,
      "learning_rate": 9.651769429795417e-06,
      "loss": 0.129,
      "step": 2470
    },
    {
      "epoch": 0.035063411047802175,
      "grad_norm": 0.011998319067060947,
      "learning_rate": 9.650355582575748e-06,
      "loss": 0.0004,
      "step": 2480
    },
    {
      "epoch": 0.03520479576976912,
      "grad_norm": 1.1249517939404541e-07,
      "learning_rate": 9.648941735356077e-06,
      "loss": 0.0,
      "step": 2490
    },
    {
      "epoch": 0.035346180491736066,
      "grad_norm": 5.331224905980037e-10,
      "learning_rate": 9.64752788813641e-06,
      "loss": 0.0,
      "step": 2500
    },
    {
      "epoch": 0.035487565213703005,
      "grad_norm": 8.837211608886719,
      "learning_rate": 9.64611404091674e-06,
      "loss": 0.0022,
      "step": 2510
    },
    {
      "epoch": 0.03562894993566995,
      "grad_norm": 0.002820813562721014,
      "learning_rate": 9.64470019369707e-06,
      "loss": 0.0002,
      "step": 2520
    },
    {
      "epoch": 0.0357703346576369,
      "grad_norm": 9.579474102849872e-09,
      "learning_rate": 9.6432863464774e-06,
      "loss": 0.0455,
      "step": 2530
    },
    {
      "epoch": 0.03591171937960384,
      "grad_norm": 0.0,
      "learning_rate": 9.641872499257731e-06,
      "loss": 0.0029,
      "step": 2540
    },
    {
      "epoch": 0.03605310410157078,
      "grad_norm": 0.0,
      "learning_rate": 9.640458652038062e-06,
      "loss": 0.0005,
      "step": 2550
    },
    {
      "epoch": 0.03619448882353773,
      "grad_norm": 0.0,
      "learning_rate": 9.639044804818393e-06,
      "loss": 0.1095,
      "step": 2560
    },
    {
      "epoch": 0.036335873545504674,
      "grad_norm": 8.505355834960938,
      "learning_rate": 9.637630957598722e-06,
      "loss": 0.0475,
      "step": 2570
    },
    {
      "epoch": 0.03647725826747162,
      "grad_norm": 0.0,
      "learning_rate": 9.636217110379053e-06,
      "loss": 0.1213,
      "step": 2580
    },
    {
      "epoch": 0.03661864298943856,
      "grad_norm": 0.0,
      "learning_rate": 9.634803263159384e-06,
      "loss": 0.0017,
      "step": 2590
    },
    {
      "epoch": 0.036760027711405505,
      "grad_norm": 1.3102123830321943e-09,
      "learning_rate": 9.633389415939715e-06,
      "loss": 0.0098,
      "step": 2600
    },
    {
      "epoch": 0.03690141243337245,
      "grad_norm": 0.0,
      "learning_rate": 9.631975568720044e-06,
      "loss": 0.0,
      "step": 2610
    },
    {
      "epoch": 0.037042797155339396,
      "grad_norm": 0.009377282112836838,
      "learning_rate": 9.630561721500375e-06,
      "loss": 0.3913,
      "step": 2620
    },
    {
      "epoch": 0.037184181877306335,
      "grad_norm": 0.0,
      "learning_rate": 9.629289259002673e-06,
      "loss": 0.3047,
      "step": 2630
    },
    {
      "epoch": 0.03732556659927328,
      "grad_norm": 0.00014320886111818254,
      "learning_rate": 9.627875411783004e-06,
      "loss": 0.0923,
      "step": 2640
    },
    {
      "epoch": 0.03746695132124023,
      "grad_norm": 0.7454742193222046,
      "learning_rate": 9.626461564563334e-06,
      "loss": 0.0002,
      "step": 2650
    },
    {
      "epoch": 0.03760833604320717,
      "grad_norm": 0.0,
      "learning_rate": 9.625047717343665e-06,
      "loss": 0.0115,
      "step": 2660
    },
    {
      "epoch": 0.03774972076517411,
      "grad_norm": 0.0,
      "learning_rate": 9.623633870123995e-06,
      "loss": 0.0,
      "step": 2670
    },
    {
      "epoch": 0.03789110548714106,
      "grad_norm": 0.0,
      "learning_rate": 9.622220022904325e-06,
      "loss": 0.0003,
      "step": 2680
    },
    {
      "epoch": 0.038032490209108004,
      "grad_norm": 0.0017288008239120245,
      "learning_rate": 9.620806175684656e-06,
      "loss": 0.2109,
      "step": 2690
    },
    {
      "epoch": 0.03817387493107495,
      "grad_norm": 1.1233649566122494e-09,
      "learning_rate": 9.619392328464987e-06,
      "loss": 0.0002,
      "step": 2700
    },
    {
      "epoch": 0.038315259653041896,
      "grad_norm": 52.07058334350586,
      "learning_rate": 9.617978481245316e-06,
      "loss": 0.0414,
      "step": 2710
    },
    {
      "epoch": 0.038456644375008835,
      "grad_norm": 0.003680927911773324,
      "learning_rate": 9.616564634025649e-06,
      "loss": 0.0001,
      "step": 2720
    },
    {
      "epoch": 0.03859802909697578,
      "grad_norm": 0.0,
      "learning_rate": 9.61515078680598e-06,
      "loss": 0.0001,
      "step": 2730
    },
    {
      "epoch": 0.038739413818942726,
      "grad_norm": 2.5791834559640847e-05,
      "learning_rate": 9.613736939586309e-06,
      "loss": 0.0009,
      "step": 2740
    },
    {
      "epoch": 0.03888079854090967,
      "grad_norm": 1.1622791618037809e-09,
      "learning_rate": 9.61232309236664e-06,
      "loss": 0.0,
      "step": 2750
    },
    {
      "epoch": 0.03902218326287661,
      "grad_norm": 0.039670180529356,
      "learning_rate": 9.61090924514697e-06,
      "loss": 0.0008,
      "step": 2760
    },
    {
      "epoch": 0.03916356798484356,
      "grad_norm": 0.00018340743554290384,
      "learning_rate": 9.609495397927301e-06,
      "loss": 0.0,
      "step": 2770
    },
    {
      "epoch": 0.0393049527068105,
      "grad_norm": 0.28265610337257385,
      "learning_rate": 9.60808155070763e-06,
      "loss": 0.0145,
      "step": 2780
    },
    {
      "epoch": 0.03944633742877745,
      "grad_norm": 6.507893431262346e-06,
      "learning_rate": 9.606667703487961e-06,
      "loss": 0.2037,
      "step": 2790
    },
    {
      "epoch": 0.03958772215074439,
      "grad_norm": 215.5770721435547,
      "learning_rate": 9.605253856268292e-06,
      "loss": 0.0647,
      "step": 2800
    },
    {
      "epoch": 0.039729106872711334,
      "grad_norm": 0.0,
      "learning_rate": 9.603840009048623e-06,
      "loss": 0.0229,
      "step": 2810
    },
    {
      "epoch": 0.03987049159467828,
      "grad_norm": 0.009560000151395798,
      "learning_rate": 9.602426161828954e-06,
      "loss": 0.006,
      "step": 2820
    },
    {
      "epoch": 0.040011876316645226,
      "grad_norm": 2.817656240949873e-05,
      "learning_rate": 9.601012314609283e-06,
      "loss": 0.0004,
      "step": 2830
    },
    {
      "epoch": 0.040153261038612165,
      "grad_norm": 0.0,
      "learning_rate": 9.599598467389614e-06,
      "loss": 0.0001,
      "step": 2840
    },
    {
      "epoch": 0.04029464576057911,
      "grad_norm": 1.1046907832934494e-09,
      "learning_rate": 9.598184620169945e-06,
      "loss": 0.0002,
      "step": 2850
    },
    {
      "epoch": 0.040436030482546056,
      "grad_norm": 0.0,
      "learning_rate": 9.596770772950276e-06,
      "loss": 0.0,
      "step": 2860
    },
    {
      "epoch": 0.040577415204513,
      "grad_norm": 0.0,
      "learning_rate": 9.595356925730607e-06,
      "loss": 0.7674,
      "step": 2870
    },
    {
      "epoch": 0.04071879992647994,
      "grad_norm": 6.8876875047863e-07,
      "learning_rate": 9.593943078510937e-06,
      "loss": 0.0,
      "step": 2880
    },
    {
      "epoch": 0.04086018464844689,
      "grad_norm": 3.931523084640503,
      "learning_rate": 9.592529231291268e-06,
      "loss": 0.3281,
      "step": 2890
    },
    {
      "epoch": 0.04100156937041383,
      "grad_norm": 2.3089678506948985e-06,
      "learning_rate": 9.591256768793565e-06,
      "loss": 0.3242,
      "step": 2900
    },
    {
      "epoch": 0.04114295409238078,
      "grad_norm": 0.6936022043228149,
      "learning_rate": 9.589842921573896e-06,
      "loss": 0.1258,
      "step": 2910
    },
    {
      "epoch": 0.041284338814347725,
      "grad_norm": 0.0,
      "learning_rate": 9.588429074354226e-06,
      "loss": 0.0,
      "step": 2920
    },
    {
      "epoch": 0.041425723536314664,
      "grad_norm": 6.96157997026603e-07,
      "learning_rate": 9.587015227134556e-06,
      "loss": 0.0003,
      "step": 2930
    },
    {
      "epoch": 0.04156710825828161,
      "grad_norm": 0.08760424703359604,
      "learning_rate": 9.585601379914886e-06,
      "loss": 0.0021,
      "step": 2940
    },
    {
      "epoch": 0.041708492980248556,
      "grad_norm": 0.0,
      "learning_rate": 9.584187532695219e-06,
      "loss": 0.0004,
      "step": 2950
    },
    {
      "epoch": 0.0418498777022155,
      "grad_norm": 1.96960248111111e-09,
      "learning_rate": 9.582773685475548e-06,
      "loss": 0.0,
      "step": 2960
    },
    {
      "epoch": 0.04199126242418244,
      "grad_norm": 2.45501446723938,
      "learning_rate": 9.581359838255879e-06,
      "loss": 0.0005,
      "step": 2970
    },
    {
      "epoch": 0.042132647146149386,
      "grad_norm": 2.3398383230954778e-09,
      "learning_rate": 9.57994599103621e-06,
      "loss": 0.0122,
      "step": 2980
    },
    {
      "epoch": 0.04227403186811633,
      "grad_norm": 23.852636337280273,
      "learning_rate": 9.57853214381654e-06,
      "loss": 0.3369,
      "step": 2990
    },
    {
      "epoch": 0.04241541659008328,
      "grad_norm": 0.7257593274116516,
      "learning_rate": 9.57711829659687e-06,
      "loss": 0.0,
      "step": 3000
    },
    {
      "epoch": 0.04255680131205022,
      "grad_norm": 6.9064412117004395,
      "learning_rate": 9.5757044493772e-06,
      "loss": 0.0059,
      "step": 3010
    },
    {
      "epoch": 0.04269818603401716,
      "grad_norm": 0.0007753377431072295,
      "learning_rate": 9.574290602157532e-06,
      "loss": 0.0,
      "step": 3020
    },
    {
      "epoch": 0.04283957075598411,
      "grad_norm": 1.1414989842251089e-08,
      "learning_rate": 9.572876754937862e-06,
      "loss": 0.4975,
      "step": 3030
    },
    {
      "epoch": 0.042980955477951055,
      "grad_norm": 3.465475674602203e-05,
      "learning_rate": 9.571462907718193e-06,
      "loss": 0.0,
      "step": 3040
    },
    {
      "epoch": 0.043122340199917994,
      "grad_norm": 1.9957242012023926,
      "learning_rate": 9.570049060498522e-06,
      "loss": 0.0199,
      "step": 3050
    },
    {
      "epoch": 0.04326372492188494,
      "grad_norm": 423.8110656738281,
      "learning_rate": 9.568635213278853e-06,
      "loss": 0.0582,
      "step": 3060
    },
    {
      "epoch": 0.043405109643851886,
      "grad_norm": 1.4103166279255674e-07,
      "learning_rate": 9.567221366059184e-06,
      "loss": 0.0,
      "step": 3070
    },
    {
      "epoch": 0.04354649436581883,
      "grad_norm": 1.1404845956519694e-08,
      "learning_rate": 9.565807518839515e-06,
      "loss": 0.0049,
      "step": 3080
    },
    {
      "epoch": 0.04368787908778577,
      "grad_norm": 1.4100515421944237e-08,
      "learning_rate": 9.564393671619846e-06,
      "loss": 0.0,
      "step": 3090
    },
    {
      "epoch": 0.043829263809752717,
      "grad_norm": 0.0,
      "learning_rate": 9.562979824400177e-06,
      "loss": 0.0003,
      "step": 3100
    },
    {
      "epoch": 0.04397064853171966,
      "grad_norm": 0.0,
      "learning_rate": 9.561565977180508e-06,
      "loss": 0.0,
      "step": 3110
    },
    {
      "epoch": 0.04411203325368661,
      "grad_norm": 0.0,
      "learning_rate": 9.560152129960837e-06,
      "loss": 0.0033,
      "step": 3120
    },
    {
      "epoch": 0.044253417975653554,
      "grad_norm": 0.0007268844055943191,
      "learning_rate": 9.558738282741168e-06,
      "loss": 0.0,
      "step": 3130
    },
    {
      "epoch": 0.04439480269762049,
      "grad_norm": 0.008608253672719002,
      "learning_rate": 9.557324435521498e-06,
      "loss": 0.0039,
      "step": 3140
    },
    {
      "epoch": 0.04453618741958744,
      "grad_norm": 0.0002218587906099856,
      "learning_rate": 9.55591058830183e-06,
      "loss": 0.1629,
      "step": 3150
    },
    {
      "epoch": 0.044677572141554385,
      "grad_norm": 0.0,
      "learning_rate": 9.55449674108216e-06,
      "loss": 0.0,
      "step": 3160
    },
    {
      "epoch": 0.04481895686352133,
      "grad_norm": 0.0,
      "learning_rate": 9.55308289386249e-06,
      "loss": 0.0,
      "step": 3170
    },
    {
      "epoch": 0.04496034158548827,
      "grad_norm": 3.1843249814755836e-08,
      "learning_rate": 9.55166904664282e-06,
      "loss": 0.0,
      "step": 3180
    },
    {
      "epoch": 0.045101726307455216,
      "grad_norm": 0.0,
      "learning_rate": 9.550255199423151e-06,
      "loss": 0.0,
      "step": 3190
    },
    {
      "epoch": 0.04524311102942216,
      "grad_norm": 0.015468222089111805,
      "learning_rate": 9.548841352203482e-06,
      "loss": 0.0081,
      "step": 3200
    },
    {
      "epoch": 0.04538449575138911,
      "grad_norm": 0.0,
      "learning_rate": 9.547427504983813e-06,
      "loss": 0.0407,
      "step": 3210
    },
    {
      "epoch": 0.04552588047335605,
      "grad_norm": 0.0,
      "learning_rate": 9.546013657764142e-06,
      "loss": 0.0,
      "step": 3220
    },
    {
      "epoch": 0.04566726519532299,
      "grad_norm": 1.9383983307363906e-09,
      "learning_rate": 9.544599810544474e-06,
      "loss": 0.0,
      "step": 3230
    },
    {
      "epoch": 0.04580864991728994,
      "grad_norm": 210.33193969726562,
      "learning_rate": 9.543185963324804e-06,
      "loss": 0.0511,
      "step": 3240
    },
    {
      "epoch": 0.045950034639256884,
      "grad_norm": 2.6634568257577484e-06,
      "learning_rate": 9.541772116105134e-06,
      "loss": 0.0107,
      "step": 3250
    },
    {
      "epoch": 0.04609141936122382,
      "grad_norm": 0.0,
      "learning_rate": 9.540358268885465e-06,
      "loss": 0.0003,
      "step": 3260
    },
    {
      "epoch": 0.04623280408319077,
      "grad_norm": 0.36443179845809937,
      "learning_rate": 9.538944421665796e-06,
      "loss": 0.0002,
      "step": 3270
    },
    {
      "epoch": 0.046374188805157715,
      "grad_norm": 0.012288296595215797,
      "learning_rate": 9.537530574446127e-06,
      "loss": 0.0003,
      "step": 3280
    },
    {
      "epoch": 0.04651557352712466,
      "grad_norm": 0.00010773780377348885,
      "learning_rate": 9.536116727226456e-06,
      "loss": 0.003,
      "step": 3290
    },
    {
      "epoch": 0.0466569582490916,
      "grad_norm": 0.0,
      "learning_rate": 9.534702880006787e-06,
      "loss": 0.0,
      "step": 3300
    },
    {
      "epoch": 0.046798342971058546,
      "grad_norm": 0.0,
      "learning_rate": 9.533289032787118e-06,
      "loss": 0.7844,
      "step": 3310
    },
    {
      "epoch": 0.04693972769302549,
      "grad_norm": 0.0,
      "learning_rate": 9.531875185567449e-06,
      "loss": 0.1029,
      "step": 3320
    },
    {
      "epoch": 0.04708111241499244,
      "grad_norm": 1.3283617761317146e-08,
      "learning_rate": 9.530461338347778e-06,
      "loss": 0.0,
      "step": 3330
    },
    {
      "epoch": 0.047222497136959384,
      "grad_norm": 4.295999112002846e-09,
      "learning_rate": 9.529047491128109e-06,
      "loss": 0.0,
      "step": 3340
    },
    {
      "epoch": 0.04736388185892632,
      "grad_norm": 3.220621920263511e-07,
      "learning_rate": 9.52763364390844e-06,
      "loss": 0.0,
      "step": 3350
    },
    {
      "epoch": 0.04750526658089327,
      "grad_norm": 0.0,
      "learning_rate": 9.52621979668877e-06,
      "loss": 0.0104,
      "step": 3360
    },
    {
      "epoch": 0.047646651302860214,
      "grad_norm": 2.049269642867557e-09,
      "learning_rate": 9.524805949469101e-06,
      "loss": 0.0,
      "step": 3370
    },
    {
      "epoch": 0.04778803602482716,
      "grad_norm": 6.566257582107937e-08,
      "learning_rate": 9.52339210224943e-06,
      "loss": 0.0,
      "step": 3380
    },
    {
      "epoch": 0.0479294207467941,
      "grad_norm": 0.0,
      "learning_rate": 9.521978255029763e-06,
      "loss": 0.0,
      "step": 3390
    },
    {
      "epoch": 0.048070805468761045,
      "grad_norm": 0.0,
      "learning_rate": 9.520564407810092e-06,
      "loss": 0.0001,
      "step": 3400
    },
    {
      "epoch": 0.04821219019072799,
      "grad_norm": 0.0,
      "learning_rate": 9.519150560590423e-06,
      "loss": 0.0419,
      "step": 3410
    },
    {
      "epoch": 0.04835357491269494,
      "grad_norm": 2.0999557364120847e-06,
      "learning_rate": 9.517736713370754e-06,
      "loss": 0.0002,
      "step": 3420
    },
    {
      "epoch": 0.048494959634661876,
      "grad_norm": 0.0,
      "learning_rate": 9.516322866151085e-06,
      "loss": 0.0008,
      "step": 3430
    },
    {
      "epoch": 0.04863634435662882,
      "grad_norm": 5.052978053754487e-07,
      "learning_rate": 9.514909018931416e-06,
      "loss": 0.0045,
      "step": 3440
    },
    {
      "epoch": 0.04877772907859577,
      "grad_norm": 0.0,
      "learning_rate": 9.513495171711745e-06,
      "loss": 0.0003,
      "step": 3450
    },
    {
      "epoch": 0.048919113800562714,
      "grad_norm": 1.8915664590224424e-09,
      "learning_rate": 9.512081324492076e-06,
      "loss": 0.0,
      "step": 3460
    },
    {
      "epoch": 0.04906049852252965,
      "grad_norm": 7.829565583961084e-05,
      "learning_rate": 9.510667477272407e-06,
      "loss": 0.0,
      "step": 3470
    },
    {
      "epoch": 0.0492018832444966,
      "grad_norm": 0.0,
      "learning_rate": 9.509253630052737e-06,
      "loss": 0.0118,
      "step": 3480
    },
    {
      "epoch": 0.049343267966463544,
      "grad_norm": 0.0,
      "learning_rate": 9.507839782833068e-06,
      "loss": 0.0354,
      "step": 3490
    },
    {
      "epoch": 0.04948465268843049,
      "grad_norm": 0.0,
      "learning_rate": 9.506425935613397e-06,
      "loss": 0.0001,
      "step": 3500
    },
    {
      "epoch": 0.04962603741039743,
      "grad_norm": 0.0010522229131311178,
      "learning_rate": 9.505012088393728e-06,
      "loss": 0.3057,
      "step": 3510
    },
    {
      "epoch": 0.049767422132364375,
      "grad_norm": 0.0,
      "learning_rate": 9.50359824117406e-06,
      "loss": 0.009,
      "step": 3520
    },
    {
      "epoch": 0.04990880685433132,
      "grad_norm": 3.4080399018421303e-06,
      "learning_rate": 9.50218439395439e-06,
      "loss": 0.0,
      "step": 3530
    },
    {
      "epoch": 0.05005019157629827,
      "grad_norm": 2.2265640442498125e-09,
      "learning_rate": 9.500770546734721e-06,
      "loss": 0.0003,
      "step": 3540
    },
    {
      "epoch": 0.05019157629826521,
      "grad_norm": 5.472434043884277,
      "learning_rate": 9.499356699515052e-06,
      "loss": 0.0009,
      "step": 3550
    },
    {
      "epoch": 0.05033296102023215,
      "grad_norm": 1.054369363373553e-06,
      "learning_rate": 9.497942852295383e-06,
      "loss": 0.0,
      "step": 3560
    },
    {
      "epoch": 0.0504743457421991,
      "grad_norm": 2.3696426865171816e-07,
      "learning_rate": 9.496529005075712e-06,
      "loss": 0.0,
      "step": 3570
    },
    {
      "epoch": 0.050615730464166044,
      "grad_norm": 0.0,
      "learning_rate": 9.495115157856043e-06,
      "loss": 0.0792,
      "step": 3580
    },
    {
      "epoch": 0.05075711518613299,
      "grad_norm": 1.878618149930844e-09,
      "learning_rate": 9.493701310636373e-06,
      "loss": 0.4924,
      "step": 3590
    },
    {
      "epoch": 0.05089849990809993,
      "grad_norm": 0.24778468906879425,
      "learning_rate": 9.49242884813867e-06,
      "loss": 0.6251,
      "step": 3600
    },
    {
      "epoch": 0.051039884630066874,
      "grad_norm": 0.0,
      "learning_rate": 9.491015000919002e-06,
      "loss": 0.0002,
      "step": 3610
    },
    {
      "epoch": 0.05118126935203382,
      "grad_norm": 0.7385488748550415,
      "learning_rate": 9.489601153699332e-06,
      "loss": 0.0413,
      "step": 3620
    },
    {
      "epoch": 0.051322654074000766,
      "grad_norm": 0.0,
      "learning_rate": 9.488187306479662e-06,
      "loss": 0.0,
      "step": 3630
    },
    {
      "epoch": 0.051464038795967705,
      "grad_norm": 2.489666712790495e-07,
      "learning_rate": 9.486773459259993e-06,
      "loss": 0.0002,
      "step": 3640
    },
    {
      "epoch": 0.05160542351793465,
      "grad_norm": 8.07943045799675e-09,
      "learning_rate": 9.485359612040324e-06,
      "loss": 0.0,
      "step": 3650
    },
    {
      "epoch": 0.0517468082399016,
      "grad_norm": 41.394447326660156,
      "learning_rate": 9.483945764820655e-06,
      "loss": 0.0093,
      "step": 3660
    },
    {
      "epoch": 0.05188819296186854,
      "grad_norm": 0.0,
      "learning_rate": 9.482531917600984e-06,
      "loss": 0.0011,
      "step": 3670
    },
    {
      "epoch": 0.05202957768383548,
      "grad_norm": 1.0497948288445969e-08,
      "learning_rate": 9.481118070381315e-06,
      "loss": 0.0116,
      "step": 3680
    },
    {
      "epoch": 0.05217096240580243,
      "grad_norm": 2.766147417787579e-06,
      "learning_rate": 9.479704223161646e-06,
      "loss": 0.0,
      "step": 3690
    },
    {
      "epoch": 0.052312347127769374,
      "grad_norm": 1.0261932611465454,
      "learning_rate": 9.478290375941977e-06,
      "loss": 0.0003,
      "step": 3700
    },
    {
      "epoch": 0.05245373184973632,
      "grad_norm": 0.0,
      "learning_rate": 9.476876528722308e-06,
      "loss": 0.0001,
      "step": 3710
    },
    {
      "epoch": 0.05259511657170326,
      "grad_norm": 0.0,
      "learning_rate": 9.475462681502637e-06,
      "loss": 0.0002,
      "step": 3720
    },
    {
      "epoch": 0.052736501293670204,
      "grad_norm": 0.0,
      "learning_rate": 9.474048834282968e-06,
      "loss": 0.0,
      "step": 3730
    },
    {
      "epoch": 0.05287788601563715,
      "grad_norm": 3.891140742950938e-09,
      "learning_rate": 9.472634987063298e-06,
      "loss": 0.0,
      "step": 3740
    },
    {
      "epoch": 0.053019270737604096,
      "grad_norm": 1.3350552320480347,
      "learning_rate": 9.47122113984363e-06,
      "loss": 0.0002,
      "step": 3750
    },
    {
      "epoch": 0.05316065545957104,
      "grad_norm": 0.0,
      "learning_rate": 9.46980729262396e-06,
      "loss": 0.0,
      "step": 3760
    },
    {
      "epoch": 0.05330204018153798,
      "grad_norm": 0.0,
      "learning_rate": 9.468393445404291e-06,
      "loss": 0.0,
      "step": 3770
    },
    {
      "epoch": 0.05344342490350493,
      "grad_norm": 5.447713036232926e-08,
      "learning_rate": 9.466979598184622e-06,
      "loss": 0.0001,
      "step": 3780
    },
    {
      "epoch": 0.05358480962547187,
      "grad_norm": 0.0003477009595371783,
      "learning_rate": 9.465565750964951e-06,
      "loss": 0.0001,
      "step": 3790
    },
    {
      "epoch": 0.05372619434743882,
      "grad_norm": 2.5487068455731787e-08,
      "learning_rate": 9.464151903745282e-06,
      "loss": 0.1663,
      "step": 3800
    },
    {
      "epoch": 0.05386757906940576,
      "grad_norm": 0.0,
      "learning_rate": 9.462738056525613e-06,
      "loss": 0.0,
      "step": 3810
    },
    {
      "epoch": 0.054008963791372704,
      "grad_norm": 0.0,
      "learning_rate": 9.461324209305944e-06,
      "loss": 0.0,
      "step": 3820
    },
    {
      "epoch": 0.05415034851333965,
      "grad_norm": 0.00029216555412858725,
      "learning_rate": 9.459910362086274e-06,
      "loss": 0.0,
      "step": 3830
    },
    {
      "epoch": 0.054291733235306595,
      "grad_norm": 0.011084984056651592,
      "learning_rate": 9.458496514866604e-06,
      "loss": 0.0,
      "step": 3840
    },
    {
      "epoch": 0.054433117957273534,
      "grad_norm": 5.095547521705157e-07,
      "learning_rate": 9.457082667646935e-06,
      "loss": 0.058,
      "step": 3850
    },
    {
      "epoch": 0.05457450267924048,
      "grad_norm": 573.4053344726562,
      "learning_rate": 9.455668820427265e-06,
      "loss": 0.1407,
      "step": 3860
    },
    {
      "epoch": 0.054715887401207426,
      "grad_norm": 0.0,
      "learning_rate": 9.454254973207596e-06,
      "loss": 0.0,
      "step": 3870
    },
    {
      "epoch": 0.05485727212317437,
      "grad_norm": 4.3064241062040765e-09,
      "learning_rate": 9.452841125987925e-06,
      "loss": 0.0,
      "step": 3880
    },
    {
      "epoch": 0.05499865684514131,
      "grad_norm": 0.001053140382282436,
      "learning_rate": 9.451427278768256e-06,
      "loss": 0.0,
      "step": 3890
    },
    {
      "epoch": 0.05514004156710826,
      "grad_norm": 0.084711953997612,
      "learning_rate": 9.450013431548589e-06,
      "loss": 0.0017,
      "step": 3900
    },
    {
      "epoch": 0.0552814262890752,
      "grad_norm": 0.49367108941078186,
      "learning_rate": 9.448599584328918e-06,
      "loss": 0.0004,
      "step": 3910
    },
    {
      "epoch": 0.05542281101104215,
      "grad_norm": 0.0,
      "learning_rate": 9.447185737109249e-06,
      "loss": 0.0044,
      "step": 3920
    },
    {
      "epoch": 0.05556419573300909,
      "grad_norm": 8.152216679491175e-09,
      "learning_rate": 9.44577188988958e-06,
      "loss": 0.0,
      "step": 3930
    },
    {
      "epoch": 0.055705580454976034,
      "grad_norm": 0.0,
      "learning_rate": 9.44435804266991e-06,
      "loss": 0.0006,
      "step": 3940
    },
    {
      "epoch": 0.05584696517694298,
      "grad_norm": 0.0,
      "learning_rate": 9.44294419545024e-06,
      "loss": 0.0915,
      "step": 3950
    },
    {
      "epoch": 0.055988349898909925,
      "grad_norm": 5.259552926872857e-05,
      "learning_rate": 9.44153034823057e-06,
      "loss": 0.0,
      "step": 3960
    },
    {
      "epoch": 0.05612973462087687,
      "grad_norm": 1.2345886716502719e-05,
      "learning_rate": 9.440116501010901e-06,
      "loss": 0.0005,
      "step": 3970
    },
    {
      "epoch": 0.05627111934284381,
      "grad_norm": 0.0,
      "learning_rate": 9.438702653791232e-06,
      "loss": 0.0,
      "step": 3980
    },
    {
      "epoch": 0.056412504064810756,
      "grad_norm": 4.014895527149065e-09,
      "learning_rate": 9.437288806571563e-06,
      "loss": 0.0,
      "step": 3990
    },
    {
      "epoch": 0.0565538887867777,
      "grad_norm": 1067.65283203125,
      "learning_rate": 9.435874959351892e-06,
      "loss": 0.4969,
      "step": 4000
    },
    {
      "epoch": 0.05669527350874465,
      "grad_norm": 8.21577262878418,
      "learning_rate": 9.434461112132223e-06,
      "loss": 0.6439,
      "step": 4010
    },
    {
      "epoch": 0.05683665823071159,
      "grad_norm": 9.395941560796928e-06,
      "learning_rate": 9.433047264912556e-06,
      "loss": 0.0,
      "step": 4020
    },
    {
      "epoch": 0.05697804295267853,
      "grad_norm": 0.0,
      "learning_rate": 9.431633417692885e-06,
      "loss": 0.002,
      "step": 4030
    },
    {
      "epoch": 0.05711942767464548,
      "grad_norm": 0.0002742137876339257,
      "learning_rate": 9.430219570473216e-06,
      "loss": 0.0358,
      "step": 4040
    },
    {
      "epoch": 0.057260812396612425,
      "grad_norm": 2.2773804175812984e-07,
      "learning_rate": 9.428805723253545e-06,
      "loss": 0.0,
      "step": 4050
    },
    {
      "epoch": 0.057402197118579364,
      "grad_norm": 0.0,
      "learning_rate": 9.427391876033877e-06,
      "loss": 0.0,
      "step": 4060
    },
    {
      "epoch": 0.05754358184054631,
      "grad_norm": 0.0,
      "learning_rate": 9.425978028814207e-06,
      "loss": 0.0,
      "step": 4070
    },
    {
      "epoch": 0.057684966562513255,
      "grad_norm": 0.0,
      "learning_rate": 9.424564181594537e-06,
      "loss": 0.0001,
      "step": 4080
    },
    {
      "epoch": 0.0578263512844802,
      "grad_norm": 0.0,
      "learning_rate": 9.423150334374868e-06,
      "loss": 0.0003,
      "step": 4090
    },
    {
      "epoch": 0.05796773600644714,
      "grad_norm": 8.73043063620571e-07,
      "learning_rate": 9.421736487155199e-06,
      "loss": 0.0011,
      "step": 4100
    },
    {
      "epoch": 0.058109120728414086,
      "grad_norm": 0.0,
      "learning_rate": 9.42032263993553e-06,
      "loss": 0.0,
      "step": 4110
    },
    {
      "epoch": 0.05825050545038103,
      "grad_norm": 0.00033942231675609946,
      "learning_rate": 9.41890879271586e-06,
      "loss": 0.1044,
      "step": 4120
    },
    {
      "epoch": 0.05839189017234798,
      "grad_norm": 321.0074462890625,
      "learning_rate": 9.41749494549619e-06,
      "loss": 0.0955,
      "step": 4130
    },
    {
      "epoch": 0.05853327489431492,
      "grad_norm": 0.00022709951736032963,
      "learning_rate": 9.416081098276521e-06,
      "loss": 0.0,
      "step": 4140
    },
    {
      "epoch": 0.05867465961628186,
      "grad_norm": 0.0,
      "learning_rate": 9.414667251056852e-06,
      "loss": 0.0629,
      "step": 4150
    },
    {
      "epoch": 0.05881604433824881,
      "grad_norm": 0.0,
      "learning_rate": 9.413253403837183e-06,
      "loss": 0.0019,
      "step": 4160
    },
    {
      "epoch": 0.058957429060215755,
      "grad_norm": 0.0030140273738652468,
      "learning_rate": 9.411839556617512e-06,
      "loss": 0.0,
      "step": 4170
    },
    {
      "epoch": 0.0590988137821827,
      "grad_norm": 0.0,
      "learning_rate": 9.410425709397844e-06,
      "loss": 0.0,
      "step": 4180
    },
    {
      "epoch": 0.05924019850414964,
      "grad_norm": 0.0,
      "learning_rate": 9.409011862178173e-06,
      "loss": 0.0,
      "step": 4190
    },
    {
      "epoch": 0.059381583226116585,
      "grad_norm": 0.0,
      "learning_rate": 9.407598014958504e-06,
      "loss": 0.0,
      "step": 4200
    },
    {
      "epoch": 0.05952296794808353,
      "grad_norm": 0.0,
      "learning_rate": 9.406184167738835e-06,
      "loss": 0.0429,
      "step": 4210
    },
    {
      "epoch": 0.05966435267005048,
      "grad_norm": 0.0,
      "learning_rate": 9.404770320519166e-06,
      "loss": 0.0089,
      "step": 4220
    },
    {
      "epoch": 0.059805737392017416,
      "grad_norm": 0.0,
      "learning_rate": 9.403356473299497e-06,
      "loss": 0.0042,
      "step": 4230
    },
    {
      "epoch": 0.05994712211398436,
      "grad_norm": 0.0,
      "learning_rate": 9.401942626079826e-06,
      "loss": 0.0,
      "step": 4240
    },
    {
      "epoch": 0.06008850683595131,
      "grad_norm": 0.0,
      "learning_rate": 9.400528778860157e-06,
      "loss": 0.0,
      "step": 4250
    },
    {
      "epoch": 0.060229891557918254,
      "grad_norm": 0.0,
      "learning_rate": 9.399114931640488e-06,
      "loss": 0.0003,
      "step": 4260
    },
    {
      "epoch": 0.06037127627988519,
      "grad_norm": 4.61290472486553e-09,
      "learning_rate": 9.397701084420819e-06,
      "loss": 0.0,
      "step": 4270
    },
    {
      "epoch": 0.06051266100185214,
      "grad_norm": 4.991599666936963e-07,
      "learning_rate": 9.39628723720115e-06,
      "loss": 0.0246,
      "step": 4280
    },
    {
      "epoch": 0.060654045723819085,
      "grad_norm": 298.04254150390625,
      "learning_rate": 9.394873389981479e-06,
      "loss": 0.0582,
      "step": 4290
    },
    {
      "epoch": 0.06079543044578603,
      "grad_norm": 0.0,
      "learning_rate": 9.39345954276181e-06,
      "loss": 0.0,
      "step": 4300
    },
    {
      "epoch": 0.06093681516775297,
      "grad_norm": 0.0,
      "learning_rate": 9.39204569554214e-06,
      "loss": 0.0,
      "step": 4310
    },
    {
      "epoch": 0.061078199889719915,
      "grad_norm": 1.083411098079523e-05,
      "learning_rate": 9.390631848322471e-06,
      "loss": 0.0,
      "step": 4320
    },
    {
      "epoch": 0.06121958461168686,
      "grad_norm": 4.481075666262768e-05,
      "learning_rate": 9.3892180011028e-06,
      "loss": 0.0255,
      "step": 4330
    },
    {
      "epoch": 0.06136096933365381,
      "grad_norm": 2.330335746592027e-08,
      "learning_rate": 9.387804153883133e-06,
      "loss": 0.0,
      "step": 4340
    },
    {
      "epoch": 0.061502354055620746,
      "grad_norm": 0.0,
      "learning_rate": 9.386390306663464e-06,
      "loss": 0.0009,
      "step": 4350
    },
    {
      "epoch": 0.06164373877758769,
      "grad_norm": 0.0,
      "learning_rate": 9.384976459443793e-06,
      "loss": 0.0,
      "step": 4360
    },
    {
      "epoch": 0.06178512349955464,
      "grad_norm": 0.0013094049645587802,
      "learning_rate": 9.383562612224124e-06,
      "loss": 0.0001,
      "step": 4370
    },
    {
      "epoch": 0.061926508221521584,
      "grad_norm": 0.0,
      "learning_rate": 9.382148765004455e-06,
      "loss": 0.0,
      "step": 4380
    },
    {
      "epoch": 0.06206789294348853,
      "grad_norm": 0.07243893295526505,
      "learning_rate": 9.380734917784786e-06,
      "loss": 0.0,
      "step": 4390
    },
    {
      "epoch": 0.06220927766545547,
      "grad_norm": 0.01943160779774189,
      "learning_rate": 9.379321070565115e-06,
      "loss": 0.1044,
      "step": 4400
    },
    {
      "epoch": 0.062350662387422415,
      "grad_norm": 0.017362864688038826,
      "learning_rate": 9.377907223345446e-06,
      "loss": 0.0107,
      "step": 4410
    },
    {
      "epoch": 0.06249204710938936,
      "grad_norm": 0.00048605105257593095,
      "learning_rate": 9.376493376125776e-06,
      "loss": 0.0,
      "step": 4420
    },
    {
      "epoch": 0.0626334318313563,
      "grad_norm": 0.0,
      "learning_rate": 9.375079528906107e-06,
      "loss": 0.0594,
      "step": 4430
    },
    {
      "epoch": 0.06277481655332325,
      "grad_norm": 0.0,
      "learning_rate": 9.373665681686438e-06,
      "loss": 0.0138,
      "step": 4440
    },
    {
      "epoch": 0.0629162012752902,
      "grad_norm": 0.0038553213234990835,
      "learning_rate": 9.372251834466767e-06,
      "loss": 0.0151,
      "step": 4450
    },
    {
      "epoch": 0.06305758599725714,
      "grad_norm": 0.0,
      "learning_rate": 9.370837987247098e-06,
      "loss": 0.0002,
      "step": 4460
    },
    {
      "epoch": 0.06319897071922408,
      "grad_norm": 2.863740711234186e-08,
      "learning_rate": 9.36942414002743e-06,
      "loss": 0.0,
      "step": 4470
    },
    {
      "epoch": 0.06334035544119103,
      "grad_norm": 0.0,
      "learning_rate": 9.36801029280776e-06,
      "loss": 0.008,
      "step": 4480
    },
    {
      "epoch": 0.06348174016315797,
      "grad_norm": 0.0,
      "learning_rate": 9.36659644558809e-06,
      "loss": 0.002,
      "step": 4490
    },
    {
      "epoch": 0.06362312488512491,
      "grad_norm": 0.0,
      "learning_rate": 9.365182598368422e-06,
      "loss": 0.0,
      "step": 4500
    },
    {
      "epoch": 0.06376450960709186,
      "grad_norm": 0.0,
      "learning_rate": 9.363768751148752e-06,
      "loss": 0.0021,
      "step": 4510
    },
    {
      "epoch": 0.0639058943290588,
      "grad_norm": 0.013969212770462036,
      "learning_rate": 9.362354903929082e-06,
      "loss": 0.0005,
      "step": 4520
    },
    {
      "epoch": 0.06404727905102575,
      "grad_norm": 8.2640663094935e-06,
      "learning_rate": 9.360941056709412e-06,
      "loss": 0.0224,
      "step": 4530
    },
    {
      "epoch": 0.06418866377299269,
      "grad_norm": 1.150738989963429e-05,
      "learning_rate": 9.359527209489743e-06,
      "loss": 0.0,
      "step": 4540
    },
    {
      "epoch": 0.06433004849495963,
      "grad_norm": 0.0,
      "learning_rate": 9.358113362270074e-06,
      "loss": 0.0026,
      "step": 4550
    },
    {
      "epoch": 0.06447143321692658,
      "grad_norm": 0.1438608020544052,
      "learning_rate": 9.356699515050405e-06,
      "loss": 0.0,
      "step": 4560
    },
    {
      "epoch": 0.06461281793889352,
      "grad_norm": 0.0,
      "learning_rate": 9.355285667830734e-06,
      "loss": 0.0041,
      "step": 4570
    },
    {
      "epoch": 0.06475420266086047,
      "grad_norm": 0.0,
      "learning_rate": 9.353871820611065e-06,
      "loss": 0.0001,
      "step": 4580
    },
    {
      "epoch": 0.06489558738282741,
      "grad_norm": 0.0,
      "learning_rate": 9.352457973391396e-06,
      "loss": 0.0,
      "step": 4590
    },
    {
      "epoch": 0.06503697210479435,
      "grad_norm": 0.0,
      "learning_rate": 9.351044126171727e-06,
      "loss": 0.0,
      "step": 4600
    },
    {
      "epoch": 0.0651783568267613,
      "grad_norm": 0.0,
      "learning_rate": 9.349630278952058e-06,
      "loss": 0.0069,
      "step": 4610
    },
    {
      "epoch": 0.06531974154872824,
      "grad_norm": 1.2946651622769423e-05,
      "learning_rate": 9.348216431732387e-06,
      "loss": 0.0,
      "step": 4620
    },
    {
      "epoch": 0.06546112627069518,
      "grad_norm": 0.7804092168807983,
      "learning_rate": 9.34680258451272e-06,
      "loss": 0.1349,
      "step": 4630
    },
    {
      "epoch": 0.06560251099266214,
      "grad_norm": 0.00010204726277152076,
      "learning_rate": 9.345388737293048e-06,
      "loss": 0.0,
      "step": 4640
    },
    {
      "epoch": 0.06574389571462907,
      "grad_norm": 0.00012707467249128968,
      "learning_rate": 9.34397489007338e-06,
      "loss": 0.0,
      "step": 4650
    },
    {
      "epoch": 0.06588528043659603,
      "grad_norm": 0.0,
      "learning_rate": 9.34256104285371e-06,
      "loss": 0.0,
      "step": 4660
    },
    {
      "epoch": 0.06602666515856297,
      "grad_norm": 1.0338148968003225e-05,
      "learning_rate": 9.341147195634041e-06,
      "loss": 0.0,
      "step": 4670
    },
    {
      "epoch": 0.0661680498805299,
      "grad_norm": 0.0,
      "learning_rate": 9.339733348414372e-06,
      "loss": 0.0235,
      "step": 4680
    },
    {
      "epoch": 0.06630943460249686,
      "grad_norm": 0.0,
      "learning_rate": 9.338319501194701e-06,
      "loss": 0.0398,
      "step": 4690
    },
    {
      "epoch": 0.0664508193244638,
      "grad_norm": 0.0,
      "learning_rate": 9.336905653975032e-06,
      "loss": 0.0,
      "step": 4700
    },
    {
      "epoch": 0.06659220404643074,
      "grad_norm": 0.0,
      "learning_rate": 9.335491806755363e-06,
      "loss": 0.1233,
      "step": 4710
    },
    {
      "epoch": 0.06673358876839769,
      "grad_norm": 4.308437606681537e-09,
      "learning_rate": 9.334077959535694e-06,
      "loss": 0.0,
      "step": 4720
    },
    {
      "epoch": 0.06687497349036463,
      "grad_norm": 0.4781586825847626,
      "learning_rate": 9.332664112316024e-06,
      "loss": 0.0019,
      "step": 4730
    },
    {
      "epoch": 0.06701635821233158,
      "grad_norm": 0.0,
      "learning_rate": 9.331250265096354e-06,
      "loss": 0.0,
      "step": 4740
    },
    {
      "epoch": 0.06715774293429852,
      "grad_norm": 0.00017699146701488644,
      "learning_rate": 9.329836417876686e-06,
      "loss": 0.0,
      "step": 4750
    },
    {
      "epoch": 0.06729912765626546,
      "grad_norm": 3.988694263767911e-09,
      "learning_rate": 9.328422570657015e-06,
      "loss": 0.0,
      "step": 4760
    },
    {
      "epoch": 0.06744051237823241,
      "grad_norm": 1.0196766879744246e-06,
      "learning_rate": 9.327008723437346e-06,
      "loss": 0.0,
      "step": 4770
    },
    {
      "epoch": 0.06758189710019935,
      "grad_norm": 9.467953532293905e-09,
      "learning_rate": 9.325594876217677e-06,
      "loss": 0.0,
      "step": 4780
    },
    {
      "epoch": 0.0677232818221663,
      "grad_norm": 0.0,
      "learning_rate": 9.324181028998008e-06,
      "loss": 0.0,
      "step": 4790
    },
    {
      "epoch": 0.06786466654413324,
      "grad_norm": 3.832819395199749e-09,
      "learning_rate": 9.322767181778339e-06,
      "loss": 0.0569,
      "step": 4800
    },
    {
      "epoch": 0.06800605126610018,
      "grad_norm": 3.426166372832995e-08,
      "learning_rate": 9.321353334558668e-06,
      "loss": 0.0,
      "step": 4810
    },
    {
      "epoch": 0.06814743598806713,
      "grad_norm": 1031.758056640625,
      "learning_rate": 9.319939487338999e-06,
      "loss": 0.4028,
      "step": 4820
    },
    {
      "epoch": 0.06828882071003407,
      "grad_norm": 0.0,
      "learning_rate": 9.31852564011933e-06,
      "loss": 0.0001,
      "step": 4830
    },
    {
      "epoch": 0.06843020543200101,
      "grad_norm": 0.010599265806376934,
      "learning_rate": 9.31711179289966e-06,
      "loss": 0.0165,
      "step": 4840
    },
    {
      "epoch": 0.06857159015396797,
      "grad_norm": 0.0,
      "learning_rate": 9.31569794567999e-06,
      "loss": 0.0,
      "step": 4850
    },
    {
      "epoch": 0.0687129748759349,
      "grad_norm": 5.747398376464844,
      "learning_rate": 9.31428409846032e-06,
      "loss": 0.0009,
      "step": 4860
    },
    {
      "epoch": 0.06885435959790186,
      "grad_norm": 0.0,
      "learning_rate": 9.312870251240651e-06,
      "loss": 0.0002,
      "step": 4870
    },
    {
      "epoch": 0.0689957443198688,
      "grad_norm": 0.0,
      "learning_rate": 9.311456404020982e-06,
      "loss": 0.0,
      "step": 4880
    },
    {
      "epoch": 0.06913712904183573,
      "grad_norm": 172.14991760253906,
      "learning_rate": 9.310042556801313e-06,
      "loss": 0.0313,
      "step": 4890
    },
    {
      "epoch": 0.06927851376380269,
      "grad_norm": 4.345204196454233e-09,
      "learning_rate": 9.308628709581642e-06,
      "loss": 0.0323,
      "step": 4900
    },
    {
      "epoch": 0.06941989848576963,
      "grad_norm": 3.904961431544507e-06,
      "learning_rate": 9.307214862361975e-06,
      "loss": 0.0,
      "step": 4910
    },
    {
      "epoch": 0.06956128320773657,
      "grad_norm": 0.0,
      "learning_rate": 9.305801015142304e-06,
      "loss": 0.0026,
      "step": 4920
    },
    {
      "epoch": 0.06970266792970352,
      "grad_norm": 590.4439086914062,
      "learning_rate": 9.304387167922635e-06,
      "loss": 0.3771,
      "step": 4930
    },
    {
      "epoch": 0.06984405265167046,
      "grad_norm": 3.667596892853453e-09,
      "learning_rate": 9.302973320702966e-06,
      "loss": 0.0,
      "step": 4940
    },
    {
      "epoch": 0.06998543737363741,
      "grad_norm": 0.0,
      "learning_rate": 9.301559473483297e-06,
      "loss": 0.0226,
      "step": 4950
    },
    {
      "epoch": 0.07012682209560435,
      "grad_norm": 0.0,
      "learning_rate": 9.300145626263627e-06,
      "loss": 0.0001,
      "step": 4960
    },
    {
      "epoch": 0.07026820681757129,
      "grad_norm": 0.0,
      "learning_rate": 9.298731779043957e-06,
      "loss": 0.2361,
      "step": 4970
    },
    {
      "epoch": 0.07040959153953824,
      "grad_norm": 0.0,
      "learning_rate": 9.297317931824287e-06,
      "loss": 0.0,
      "step": 4980
    },
    {
      "epoch": 0.07055097626150518,
      "grad_norm": 0.0,
      "learning_rate": 9.295904084604618e-06,
      "loss": 0.0,
      "step": 4990
    },
    {
      "epoch": 0.07069236098347213,
      "grad_norm": 4.0842245141448075e-09,
      "learning_rate": 9.294490237384949e-06,
      "loss": 0.0048,
      "step": 5000
    },
    {
      "epoch": 0.07083374570543907,
      "grad_norm": 55.0610466003418,
      "learning_rate": 9.29307639016528e-06,
      "loss": 0.7905,
      "step": 5010
    },
    {
      "epoch": 0.07097513042740601,
      "grad_norm": 0.0,
      "learning_rate": 9.291662542945609e-06,
      "loss": 0.0016,
      "step": 5020
    },
    {
      "epoch": 0.07111651514937296,
      "grad_norm": 0.008345073089003563,
      "learning_rate": 9.29024869572594e-06,
      "loss": 0.0006,
      "step": 5030
    },
    {
      "epoch": 0.0712578998713399,
      "grad_norm": 0.0,
      "learning_rate": 9.288834848506271e-06,
      "loss": 0.0,
      "step": 5040
    },
    {
      "epoch": 0.07139928459330684,
      "grad_norm": 0.0,
      "learning_rate": 9.287421001286602e-06,
      "loss": 0.0016,
      "step": 5050
    },
    {
      "epoch": 0.0715406693152738,
      "grad_norm": 8.808092921697153e-09,
      "learning_rate": 9.286007154066933e-06,
      "loss": 0.0001,
      "step": 5060
    },
    {
      "epoch": 0.07168205403724073,
      "grad_norm": 0.03391960263252258,
      "learning_rate": 9.284593306847263e-06,
      "loss": 0.019,
      "step": 5070
    },
    {
      "epoch": 0.07182343875920769,
      "grad_norm": 0.0,
      "learning_rate": 9.283179459627594e-06,
      "loss": 0.0001,
      "step": 5080
    },
    {
      "epoch": 0.07196482348117463,
      "grad_norm": 4.067486543135601e-07,
      "learning_rate": 9.281765612407923e-06,
      "loss": 0.0,
      "step": 5090
    },
    {
      "epoch": 0.07210620820314156,
      "grad_norm": 3.318576318633859e-06,
      "learning_rate": 9.280351765188254e-06,
      "loss": 0.0,
      "step": 5100
    },
    {
      "epoch": 0.07224759292510852,
      "grad_norm": 14.880258560180664,
      "learning_rate": 9.278937917968585e-06,
      "loss": 0.0059,
      "step": 5110
    },
    {
      "epoch": 0.07238897764707546,
      "grad_norm": 0.002327380236238241,
      "learning_rate": 9.277524070748916e-06,
      "loss": 0.0004,
      "step": 5120
    },
    {
      "epoch": 0.0725303623690424,
      "grad_norm": 0.0,
      "learning_rate": 9.276110223529247e-06,
      "loss": 0.0001,
      "step": 5130
    },
    {
      "epoch": 0.07267174709100935,
      "grad_norm": 2.509698617814138e-07,
      "learning_rate": 9.274696376309576e-06,
      "loss": 0.0005,
      "step": 5140
    },
    {
      "epoch": 0.07281313181297629,
      "grad_norm": 0.03955850750207901,
      "learning_rate": 9.273282529089907e-06,
      "loss": 0.0001,
      "step": 5150
    },
    {
      "epoch": 0.07295451653494324,
      "grad_norm": 0.0,
      "learning_rate": 9.271868681870238e-06,
      "loss": 0.0072,
      "step": 5160
    },
    {
      "epoch": 0.07309590125691018,
      "grad_norm": 4.652776624425314e-06,
      "learning_rate": 9.270454834650569e-06,
      "loss": 0.011,
      "step": 5170
    },
    {
      "epoch": 0.07323728597887712,
      "grad_norm": 0.0,
      "learning_rate": 9.2690409874309e-06,
      "loss": 0.0,
      "step": 5180
    },
    {
      "epoch": 0.07337867070084407,
      "grad_norm": 0.0,
      "learning_rate": 9.267627140211229e-06,
      "loss": 0.0,
      "step": 5190
    },
    {
      "epoch": 0.07352005542281101,
      "grad_norm": 0.0,
      "learning_rate": 9.266213292991561e-06,
      "loss": 0.0,
      "step": 5200
    },
    {
      "epoch": 0.07366144014477796,
      "grad_norm": 0.0,
      "learning_rate": 9.26479944577189e-06,
      "loss": 0.0,
      "step": 5210
    },
    {
      "epoch": 0.0738028248667449,
      "grad_norm": 0.0,
      "learning_rate": 9.263385598552221e-06,
      "loss": 0.0005,
      "step": 5220
    },
    {
      "epoch": 0.07394420958871184,
      "grad_norm": 0.0,
      "learning_rate": 9.261971751332552e-06,
      "loss": 0.0,
      "step": 5230
    },
    {
      "epoch": 0.07408559431067879,
      "grad_norm": 0.0,
      "learning_rate": 9.260557904112883e-06,
      "loss": 0.0,
      "step": 5240
    },
    {
      "epoch": 0.07422697903264573,
      "grad_norm": 2.948551696135837e-07,
      "learning_rate": 9.259144056893214e-06,
      "loss": 0.3678,
      "step": 5250
    },
    {
      "epoch": 0.07436836375461267,
      "grad_norm": 0.0,
      "learning_rate": 9.257730209673543e-06,
      "loss": 0.0002,
      "step": 5260
    },
    {
      "epoch": 0.07450974847657962,
      "grad_norm": 3.574390211724676e-05,
      "learning_rate": 9.256316362453874e-06,
      "loss": 0.0066,
      "step": 5270
    },
    {
      "epoch": 0.07465113319854656,
      "grad_norm": 5.003475962439552e-06,
      "learning_rate": 9.254902515234205e-06,
      "loss": 0.0011,
      "step": 5280
    },
    {
      "epoch": 0.07479251792051352,
      "grad_norm": 0.0,
      "learning_rate": 9.253488668014535e-06,
      "loss": 0.0,
      "step": 5290
    },
    {
      "epoch": 0.07493390264248045,
      "grad_norm": 0.0006885546026751399,
      "learning_rate": 9.252074820794865e-06,
      "loss": 0.005,
      "step": 5300
    },
    {
      "epoch": 0.0750752873644474,
      "grad_norm": 7.0065175350464415e-06,
      "learning_rate": 9.250660973575196e-06,
      "loss": 0.1744,
      "step": 5310
    },
    {
      "epoch": 0.07521667208641435,
      "grad_norm": 7.414686198359277e-09,
      "learning_rate": 9.249247126355528e-06,
      "loss": 0.0,
      "step": 5320
    },
    {
      "epoch": 0.07535805680838129,
      "grad_norm": 2.3739634968933387e-07,
      "learning_rate": 9.247833279135857e-06,
      "loss": 0.0011,
      "step": 5330
    },
    {
      "epoch": 0.07549944153034822,
      "grad_norm": 0.0,
      "learning_rate": 9.246419431916188e-06,
      "loss": 0.0006,
      "step": 5340
    },
    {
      "epoch": 0.07564082625231518,
      "grad_norm": 1.5691990852355957,
      "learning_rate": 9.245005584696519e-06,
      "loss": 0.001,
      "step": 5350
    },
    {
      "epoch": 0.07578221097428212,
      "grad_norm": 0.0035488863941282034,
      "learning_rate": 9.24359173747685e-06,
      "loss": 0.0015,
      "step": 5360
    },
    {
      "epoch": 0.07592359569624907,
      "grad_norm": 0.0,
      "learning_rate": 9.242177890257179e-06,
      "loss": 0.0,
      "step": 5370
    },
    {
      "epoch": 0.07606498041821601,
      "grad_norm": 0.0,
      "learning_rate": 9.24076404303751e-06,
      "loss": 0.0,
      "step": 5380
    },
    {
      "epoch": 0.07620636514018295,
      "grad_norm": 0.0,
      "learning_rate": 9.23935019581784e-06,
      "loss": 0.0,
      "step": 5390
    },
    {
      "epoch": 0.0763477498621499,
      "grad_norm": 0.0,
      "learning_rate": 9.237936348598172e-06,
      "loss": 0.0001,
      "step": 5400
    },
    {
      "epoch": 0.07648913458411684,
      "grad_norm": 0.0,
      "learning_rate": 9.236522501378502e-06,
      "loss": 0.0,
      "step": 5410
    },
    {
      "epoch": 0.07663051930608379,
      "grad_norm": 0.019375190138816833,
      "learning_rate": 9.235108654158832e-06,
      "loss": 0.0,
      "step": 5420
    },
    {
      "epoch": 0.07677190402805073,
      "grad_norm": 0.0,
      "learning_rate": 9.233694806939162e-06,
      "loss": 0.0,
      "step": 5430
    },
    {
      "epoch": 0.07691328875001767,
      "grad_norm": 0.0,
      "learning_rate": 9.232280959719493e-06,
      "loss": 0.0031,
      "step": 5440
    },
    {
      "epoch": 0.07705467347198462,
      "grad_norm": 0.0,
      "learning_rate": 9.230867112499824e-06,
      "loss": 0.0007,
      "step": 5450
    },
    {
      "epoch": 0.07719605819395156,
      "grad_norm": 3.1814677186048357e-06,
      "learning_rate": 9.229453265280155e-06,
      "loss": 0.2436,
      "step": 5460
    },
    {
      "epoch": 0.0773374429159185,
      "grad_norm": 0.0,
      "learning_rate": 9.228039418060484e-06,
      "loss": 0.0001,
      "step": 5470
    },
    {
      "epoch": 0.07747882763788545,
      "grad_norm": 0.010283908806741238,
      "learning_rate": 9.226625570840817e-06,
      "loss": 0.0,
      "step": 5480
    },
    {
      "epoch": 0.07762021235985239,
      "grad_norm": 0.0,
      "learning_rate": 9.225211723621146e-06,
      "loss": 0.0001,
      "step": 5490
    },
    {
      "epoch": 0.07776159708181934,
      "grad_norm": 36.44510269165039,
      "learning_rate": 9.223797876401477e-06,
      "loss": 0.0095,
      "step": 5500
    },
    {
      "epoch": 0.07790298180378628,
      "grad_norm": 0.0,
      "learning_rate": 9.222384029181808e-06,
      "loss": 0.0,
      "step": 5510
    },
    {
      "epoch": 0.07804436652575322,
      "grad_norm": 4.728606224060059,
      "learning_rate": 9.220970181962138e-06,
      "loss": 0.0666,
      "step": 5520
    },
    {
      "epoch": 0.07818575124772018,
      "grad_norm": 0.0,
      "learning_rate": 9.21955633474247e-06,
      "loss": 0.1447,
      "step": 5530
    },
    {
      "epoch": 0.07832713596968711,
      "grad_norm": 0.11026112735271454,
      "learning_rate": 9.218142487522798e-06,
      "loss": 0.0,
      "step": 5540
    },
    {
      "epoch": 0.07846852069165405,
      "grad_norm": 0.0,
      "learning_rate": 9.21672864030313e-06,
      "loss": 0.1406,
      "step": 5550
    },
    {
      "epoch": 0.078609905413621,
      "grad_norm": 6.122015476226807,
      "learning_rate": 9.21531479308346e-06,
      "loss": 0.0011,
      "step": 5560
    },
    {
      "epoch": 0.07875129013558795,
      "grad_norm": 0.0,
      "learning_rate": 9.213900945863791e-06,
      "loss": 0.0001,
      "step": 5570
    },
    {
      "epoch": 0.0788926748575549,
      "grad_norm": 9.05463252820482e-07,
      "learning_rate": 9.212487098644122e-06,
      "loss": 0.0012,
      "step": 5580
    },
    {
      "epoch": 0.07903405957952184,
      "grad_norm": 0.00013452963321469724,
      "learning_rate": 9.211073251424451e-06,
      "loss": 0.0001,
      "step": 5590
    },
    {
      "epoch": 0.07917544430148878,
      "grad_norm": 0.0,
      "learning_rate": 9.209659404204782e-06,
      "loss": 0.0001,
      "step": 5600
    },
    {
      "epoch": 0.07931682902345573,
      "grad_norm": 0.0015298252692446113,
      "learning_rate": 9.208245556985113e-06,
      "loss": 0.0,
      "step": 5610
    },
    {
      "epoch": 0.07945821374542267,
      "grad_norm": 0.0,
      "learning_rate": 9.206831709765444e-06,
      "loss": 0.0017,
      "step": 5620
    },
    {
      "epoch": 0.07959959846738962,
      "grad_norm": 0.0001733494718791917,
      "learning_rate": 9.205417862545773e-06,
      "loss": 0.0001,
      "step": 5630
    },
    {
      "epoch": 0.07974098318935656,
      "grad_norm": 5.561549187405035e-05,
      "learning_rate": 9.204004015326105e-06,
      "loss": 0.0018,
      "step": 5640
    },
    {
      "epoch": 0.0798823679113235,
      "grad_norm": 24.140342712402344,
      "learning_rate": 9.202590168106436e-06,
      "loss": 0.5218,
      "step": 5650
    },
    {
      "epoch": 0.08002375263329045,
      "grad_norm": 0.0,
      "learning_rate": 9.201176320886765e-06,
      "loss": 0.0035,
      "step": 5660
    },
    {
      "epoch": 0.08016513735525739,
      "grad_norm": 0.0,
      "learning_rate": 9.199762473667096e-06,
      "loss": 0.0,
      "step": 5670
    },
    {
      "epoch": 0.08030652207722433,
      "grad_norm": 4.892332228223495e-08,
      "learning_rate": 9.198348626447427e-06,
      "loss": 0.0003,
      "step": 5680
    },
    {
      "epoch": 0.08044790679919128,
      "grad_norm": 9.629698070057202e-06,
      "learning_rate": 9.196934779227758e-06,
      "loss": 0.0,
      "step": 5690
    },
    {
      "epoch": 0.08058929152115822,
      "grad_norm": 18.207582473754883,
      "learning_rate": 9.195520932008089e-06,
      "loss": 0.0118,
      "step": 5700
    },
    {
      "epoch": 0.08073067624312517,
      "grad_norm": 3.4007939575531054e-06,
      "learning_rate": 9.194107084788418e-06,
      "loss": 0.0,
      "step": 5710
    },
    {
      "epoch": 0.08087206096509211,
      "grad_norm": 0.0,
      "learning_rate": 9.192693237568749e-06,
      "loss": 0.0,
      "step": 5720
    },
    {
      "epoch": 0.08101344568705905,
      "grad_norm": 0.013868841342628002,
      "learning_rate": 9.19127939034908e-06,
      "loss": 0.0011,
      "step": 5730
    },
    {
      "epoch": 0.081154830409026,
      "grad_norm": 319.5315246582031,
      "learning_rate": 9.18986554312941e-06,
      "loss": 0.1152,
      "step": 5740
    },
    {
      "epoch": 0.08129621513099294,
      "grad_norm": 5.3989069783710875e-06,
      "learning_rate": 9.18845169590974e-06,
      "loss": 0.0,
      "step": 5750
    },
    {
      "epoch": 0.08143759985295988,
      "grad_norm": 0.036415405571460724,
      "learning_rate": 9.18703784869007e-06,
      "loss": 0.0004,
      "step": 5760
    },
    {
      "epoch": 0.08157898457492684,
      "grad_norm": 0.0,
      "learning_rate": 9.185624001470403e-06,
      "loss": 0.0004,
      "step": 5770
    },
    {
      "epoch": 0.08172036929689377,
      "grad_norm": 0.0,
      "learning_rate": 9.184210154250732e-06,
      "loss": 0.0,
      "step": 5780
    },
    {
      "epoch": 0.08186175401886073,
      "grad_norm": 7.33847071998639e-09,
      "learning_rate": 9.182796307031063e-06,
      "loss": 0.0292,
      "step": 5790
    },
    {
      "epoch": 0.08200313874082767,
      "grad_norm": 0.0,
      "learning_rate": 9.181382459811394e-06,
      "loss": 0.0007,
      "step": 5800
    },
    {
      "epoch": 0.0821445234627946,
      "grad_norm": 4.394492407300277e-06,
      "learning_rate": 9.179968612591725e-06,
      "loss": 0.0,
      "step": 5810
    },
    {
      "epoch": 0.08228590818476156,
      "grad_norm": 0.0,
      "learning_rate": 9.178554765372054e-06,
      "loss": 0.3198,
      "step": 5820
    },
    {
      "epoch": 0.0824272929067285,
      "grad_norm": 4.816751442149325e-09,
      "learning_rate": 9.177140918152385e-06,
      "loss": 0.1634,
      "step": 5830
    },
    {
      "epoch": 0.08256867762869545,
      "grad_norm": 1.1750206795113627e-05,
      "learning_rate": 9.175727070932716e-06,
      "loss": 0.0053,
      "step": 5840
    },
    {
      "epoch": 0.08271006235066239,
      "grad_norm": 0.002070905640721321,
      "learning_rate": 9.174313223713047e-06,
      "loss": 0.0,
      "step": 5850
    },
    {
      "epoch": 0.08285144707262933,
      "grad_norm": 0.018424546346068382,
      "learning_rate": 9.172899376493377e-06,
      "loss": 0.0005,
      "step": 5860
    },
    {
      "epoch": 0.08299283179459628,
      "grad_norm": 0.0,
      "learning_rate": 9.171485529273707e-06,
      "loss": 0.0,
      "step": 5870
    },
    {
      "epoch": 0.08313421651656322,
      "grad_norm": 0.0035166051238775253,
      "learning_rate": 9.170071682054037e-06,
      "loss": 0.0,
      "step": 5880
    },
    {
      "epoch": 0.08327560123853016,
      "grad_norm": 8.502797572873533e-05,
      "learning_rate": 9.168657834834368e-06,
      "loss": 0.0318,
      "step": 5890
    },
    {
      "epoch": 0.08341698596049711,
      "grad_norm": 0.0,
      "learning_rate": 9.167243987614699e-06,
      "loss": 0.2748,
      "step": 5900
    },
    {
      "epoch": 0.08355837068246405,
      "grad_norm": 8.527028967364458e-07,
      "learning_rate": 9.16583014039503e-06,
      "loss": 0.0001,
      "step": 5910
    },
    {
      "epoch": 0.083699755404431,
      "grad_norm": 0.16508832573890686,
      "learning_rate": 9.16441629317536e-06,
      "loss": 0.0007,
      "step": 5920
    },
    {
      "epoch": 0.08384114012639794,
      "grad_norm": 0.0,
      "learning_rate": 9.163002445955692e-06,
      "loss": 0.0403,
      "step": 5930
    },
    {
      "epoch": 0.08398252484836488,
      "grad_norm": 7.320099830627441,
      "learning_rate": 9.16158859873602e-06,
      "loss": 0.0017,
      "step": 5940
    },
    {
      "epoch": 0.08412390957033183,
      "grad_norm": 0.09893850237131119,
      "learning_rate": 9.160174751516352e-06,
      "loss": 0.0,
      "step": 5950
    },
    {
      "epoch": 0.08426529429229877,
      "grad_norm": 6.104153271735413e-07,
      "learning_rate": 9.158760904296683e-06,
      "loss": 0.0001,
      "step": 5960
    },
    {
      "epoch": 0.08440667901426571,
      "grad_norm": 0.0,
      "learning_rate": 9.157347057077013e-06,
      "loss": 0.0092,
      "step": 5970
    },
    {
      "epoch": 0.08454806373623266,
      "grad_norm": 0.00018120958702638745,
      "learning_rate": 9.155933209857344e-06,
      "loss": 0.0,
      "step": 5980
    },
    {
      "epoch": 0.0846894484581996,
      "grad_norm": 2.163043973268941e-05,
      "learning_rate": 9.154519362637673e-06,
      "loss": 0.0679,
      "step": 5990
    },
    {
      "epoch": 0.08483083318016656,
      "grad_norm": 2.1402939420767098e-09,
      "learning_rate": 9.153105515418004e-06,
      "loss": 0.0,
      "step": 6000
    },
    {
      "epoch": 0.0849722179021335,
      "grad_norm": 0.0,
      "learning_rate": 9.151691668198335e-06,
      "loss": 0.0,
      "step": 6010
    },
    {
      "epoch": 0.08511360262410043,
      "grad_norm": 5.489574687089771e-05,
      "learning_rate": 9.150277820978666e-06,
      "loss": 0.0,
      "step": 6020
    },
    {
      "epoch": 0.08525498734606739,
      "grad_norm": 0.0,
      "learning_rate": 9.148863973758997e-06,
      "loss": 0.0092,
      "step": 6030
    },
    {
      "epoch": 0.08539637206803433,
      "grad_norm": 0.00018150756659451872,
      "learning_rate": 9.147450126539326e-06,
      "loss": 0.0,
      "step": 6040
    },
    {
      "epoch": 0.08553775679000128,
      "grad_norm": 5.9756022352530636e-09,
      "learning_rate": 9.146036279319659e-06,
      "loss": 0.0,
      "step": 6050
    },
    {
      "epoch": 0.08567914151196822,
      "grad_norm": 7.323568570427597e-05,
      "learning_rate": 9.144622432099988e-06,
      "loss": 0.0,
      "step": 6060
    },
    {
      "epoch": 0.08582052623393516,
      "grad_norm": 0.0,
      "learning_rate": 9.143208584880319e-06,
      "loss": 0.0018,
      "step": 6070
    },
    {
      "epoch": 0.08596191095590211,
      "grad_norm": 0.00011765229282900691,
      "learning_rate": 9.14179473766065e-06,
      "loss": 0.0008,
      "step": 6080
    },
    {
      "epoch": 0.08610329567786905,
      "grad_norm": 0.0,
      "learning_rate": 9.14038089044098e-06,
      "loss": 0.0036,
      "step": 6090
    },
    {
      "epoch": 0.08624468039983599,
      "grad_norm": 1.2071375010691554e-07,
      "learning_rate": 9.138967043221311e-06,
      "loss": 0.0,
      "step": 6100
    },
    {
      "epoch": 0.08638606512180294,
      "grad_norm": 0.0,
      "learning_rate": 9.13755319600164e-06,
      "loss": 0.0005,
      "step": 6110
    },
    {
      "epoch": 0.08652744984376988,
      "grad_norm": 0.0,
      "learning_rate": 9.136139348781971e-06,
      "loss": 0.0,
      "step": 6120
    },
    {
      "epoch": 0.08666883456573683,
      "grad_norm": 0.0,
      "learning_rate": 9.134725501562302e-06,
      "loss": 0.0,
      "step": 6130
    },
    {
      "epoch": 0.08681021928770377,
      "grad_norm": 5.8547477976844675e-09,
      "learning_rate": 9.133311654342633e-06,
      "loss": 0.0,
      "step": 6140
    },
    {
      "epoch": 0.08695160400967071,
      "grad_norm": 0.0,
      "learning_rate": 9.131897807122964e-06,
      "loss": 0.0024,
      "step": 6150
    },
    {
      "epoch": 0.08709298873163766,
      "grad_norm": 0.0,
      "learning_rate": 9.130483959903293e-06,
      "loss": 0.0013,
      "step": 6160
    },
    {
      "epoch": 0.0872343734536046,
      "grad_norm": 0.0,
      "learning_rate": 9.129070112683624e-06,
      "loss": 0.0002,
      "step": 6170
    },
    {
      "epoch": 0.08737575817557154,
      "grad_norm": 0.0,
      "learning_rate": 9.127656265463955e-06,
      "loss": 0.0,
      "step": 6180
    },
    {
      "epoch": 0.0875171428975385,
      "grad_norm": 640.6080322265625,
      "learning_rate": 9.126242418244285e-06,
      "loss": 0.2695,
      "step": 6190
    },
    {
      "epoch": 0.08765852761950543,
      "grad_norm": 0.09605782479047775,
      "learning_rate": 9.124828571024615e-06,
      "loss": 0.0612,
      "step": 6200
    },
    {
      "epoch": 0.08779991234147239,
      "grad_norm": 0.00018609756079968065,
      "learning_rate": 9.123414723804947e-06,
      "loss": 0.0072,
      "step": 6210
    },
    {
      "epoch": 0.08794129706343932,
      "grad_norm": 0.000495040905661881,
      "learning_rate": 9.122000876585278e-06,
      "loss": 0.0,
      "step": 6220
    },
    {
      "epoch": 0.08808268178540626,
      "grad_norm": 0.0,
      "learning_rate": 9.120587029365607e-06,
      "loss": 0.0021,
      "step": 6230
    },
    {
      "epoch": 0.08822406650737322,
      "grad_norm": 0.0,
      "learning_rate": 9.119173182145938e-06,
      "loss": 0.1588,
      "step": 6240
    },
    {
      "epoch": 0.08836545122934016,
      "grad_norm": 0.3799512982368469,
      "learning_rate": 9.117759334926269e-06,
      "loss": 0.0001,
      "step": 6250
    },
    {
      "epoch": 0.08850683595130711,
      "grad_norm": 0.0,
      "learning_rate": 9.1163454877066e-06,
      "loss": 0.0002,
      "step": 6260
    },
    {
      "epoch": 0.08864822067327405,
      "grad_norm": 0.0,
      "learning_rate": 9.114931640486929e-06,
      "loss": 0.4341,
      "step": 6270
    },
    {
      "epoch": 0.08878960539524099,
      "grad_norm": 0.0004980498924851418,
      "learning_rate": 9.11351779326726e-06,
      "loss": 0.0,
      "step": 6280
    },
    {
      "epoch": 0.08893099011720794,
      "grad_norm": 2.0102852715808694e-09,
      "learning_rate": 9.11210394604759e-06,
      "loss": 0.0,
      "step": 6290
    },
    {
      "epoch": 0.08907237483917488,
      "grad_norm": 0.0011616009287536144,
      "learning_rate": 9.110690098827921e-06,
      "loss": 0.02,
      "step": 6300
    },
    {
      "epoch": 0.08921375956114182,
      "grad_norm": 3.697834261018329e-09,
      "learning_rate": 9.109276251608252e-06,
      "loss": 0.0001,
      "step": 6310
    },
    {
      "epoch": 0.08935514428310877,
      "grad_norm": 0.0,
      "learning_rate": 9.107862404388582e-06,
      "loss": 0.0,
      "step": 6320
    },
    {
      "epoch": 0.08949652900507571,
      "grad_norm": 0.0,
      "learning_rate": 9.106448557168912e-06,
      "loss": 0.7193,
      "step": 6330
    },
    {
      "epoch": 0.08963791372704266,
      "grad_norm": 0.0,
      "learning_rate": 9.105034709949243e-06,
      "loss": 0.0008,
      "step": 6340
    },
    {
      "epoch": 0.0897792984490096,
      "grad_norm": 0.0018736617639660835,
      "learning_rate": 9.103620862729574e-06,
      "loss": 0.0001,
      "step": 6350
    },
    {
      "epoch": 0.08992068317097654,
      "grad_norm": 0.0,
      "learning_rate": 9.102207015509905e-06,
      "loss": 0.0,
      "step": 6360
    },
    {
      "epoch": 0.09006206789294349,
      "grad_norm": 0.0,
      "learning_rate": 9.100793168290236e-06,
      "loss": 0.0,
      "step": 6370
    },
    {
      "epoch": 0.09020345261491043,
      "grad_norm": 0.0,
      "learning_rate": 9.099379321070567e-06,
      "loss": 0.0007,
      "step": 6380
    },
    {
      "epoch": 0.09034483733687737,
      "grad_norm": 7.3409271240234375,
      "learning_rate": 9.097965473850896e-06,
      "loss": 0.0013,
      "step": 6390
    },
    {
      "epoch": 0.09048622205884432,
      "grad_norm": 0.0016764653846621513,
      "learning_rate": 9.096551626631227e-06,
      "loss": 0.1936,
      "step": 6400
    },
    {
      "epoch": 0.09062760678081126,
      "grad_norm": 0.0003498728619888425,
      "learning_rate": 9.095137779411558e-06,
      "loss": 0.0,
      "step": 6410
    },
    {
      "epoch": 0.09076899150277822,
      "grad_norm": 0.0,
      "learning_rate": 9.093723932191888e-06,
      "loss": 0.0001,
      "step": 6420
    },
    {
      "epoch": 0.09091037622474515,
      "grad_norm": 0.0004780996241606772,
      "learning_rate": 9.09231008497222e-06,
      "loss": 0.5094,
      "step": 6430
    },
    {
      "epoch": 0.0910517609467121,
      "grad_norm": 0.0,
      "learning_rate": 9.090896237752548e-06,
      "loss": 0.0,
      "step": 6440
    },
    {
      "epoch": 0.09119314566867905,
      "grad_norm": 0.0,
      "learning_rate": 9.08948239053288e-06,
      "loss": 0.0003,
      "step": 6450
    },
    {
      "epoch": 0.09133453039064598,
      "grad_norm": 0.0,
      "learning_rate": 9.08806854331321e-06,
      "loss": 0.1186,
      "step": 6460
    },
    {
      "epoch": 0.09147591511261294,
      "grad_norm": 0.0,
      "learning_rate": 9.086654696093541e-06,
      "loss": 0.0,
      "step": 6470
    },
    {
      "epoch": 0.09161729983457988,
      "grad_norm": 0.0,
      "learning_rate": 9.085240848873872e-06,
      "loss": 0.0002,
      "step": 6480
    },
    {
      "epoch": 0.09175868455654682,
      "grad_norm": 2.279153088480257e-09,
      "learning_rate": 9.083827001654201e-06,
      "loss": 0.0058,
      "step": 6490
    },
    {
      "epoch": 0.09190006927851377,
      "grad_norm": 0.0003048465878237039,
      "learning_rate": 9.082413154434534e-06,
      "loss": 0.0004,
      "step": 6500
    },
    {
      "epoch": 0.09204145400048071,
      "grad_norm": 0.0,
      "learning_rate": 9.080999307214863e-06,
      "loss": 0.0025,
      "step": 6510
    },
    {
      "epoch": 0.09218283872244765,
      "grad_norm": 0.0,
      "learning_rate": 9.079585459995194e-06,
      "loss": 0.0,
      "step": 6520
    },
    {
      "epoch": 0.0923242234444146,
      "grad_norm": 0.0,
      "learning_rate": 9.078171612775524e-06,
      "loss": 0.01,
      "step": 6530
    },
    {
      "epoch": 0.09246560816638154,
      "grad_norm": 0.0,
      "learning_rate": 9.076757765555855e-06,
      "loss": 0.0001,
      "step": 6540
    },
    {
      "epoch": 0.09260699288834849,
      "grad_norm": 0.0,
      "learning_rate": 9.075343918336186e-06,
      "loss": 0.0003,
      "step": 6550
    },
    {
      "epoch": 0.09274837761031543,
      "grad_norm": 9.049884101841599e-05,
      "learning_rate": 9.073930071116515e-06,
      "loss": 0.0006,
      "step": 6560
    },
    {
      "epoch": 0.09288976233228237,
      "grad_norm": 0.0,
      "learning_rate": 9.072516223896846e-06,
      "loss": 0.0397,
      "step": 6570
    },
    {
      "epoch": 0.09303114705424932,
      "grad_norm": 0.0,
      "learning_rate": 9.071102376677177e-06,
      "loss": 0.0001,
      "step": 6580
    },
    {
      "epoch": 0.09317253177621626,
      "grad_norm": 0.0,
      "learning_rate": 9.069688529457508e-06,
      "loss": 0.0,
      "step": 6590
    },
    {
      "epoch": 0.0933139164981832,
      "grad_norm": 0.0,
      "learning_rate": 9.068274682237837e-06,
      "loss": 0.0016,
      "step": 6600
    },
    {
      "epoch": 0.09345530122015015,
      "grad_norm": 0.003409323748201132,
      "learning_rate": 9.066860835018168e-06,
      "loss": 0.0,
      "step": 6610
    },
    {
      "epoch": 0.09359668594211709,
      "grad_norm": 0.0002991340297739953,
      "learning_rate": 9.0654469877985e-06,
      "loss": 0.0002,
      "step": 6620
    },
    {
      "epoch": 0.09373807066408404,
      "grad_norm": 0.0,
      "learning_rate": 9.06403314057883e-06,
      "loss": 0.0,
      "step": 6630
    },
    {
      "epoch": 0.09387945538605098,
      "grad_norm": 0.0,
      "learning_rate": 9.06261929335916e-06,
      "loss": 0.0,
      "step": 6640
    },
    {
      "epoch": 0.09402084010801792,
      "grad_norm": 0.0,
      "learning_rate": 9.061205446139491e-06,
      "loss": 0.0153,
      "step": 6650
    },
    {
      "epoch": 0.09416222482998488,
      "grad_norm": 0.0,
      "learning_rate": 9.059791598919822e-06,
      "loss": 0.0022,
      "step": 6660
    },
    {
      "epoch": 0.09430360955195181,
      "grad_norm": 0.0,
      "learning_rate": 9.058377751700153e-06,
      "loss": 0.0047,
      "step": 6670
    },
    {
      "epoch": 0.09444499427391877,
      "grad_norm": 3.5992965698242188,
      "learning_rate": 9.056963904480482e-06,
      "loss": 0.0012,
      "step": 6680
    },
    {
      "epoch": 0.0945863789958857,
      "grad_norm": 0.0,
      "learning_rate": 9.055550057260813e-06,
      "loss": 0.0,
      "step": 6690
    },
    {
      "epoch": 0.09472776371785264,
      "grad_norm": 0.0,
      "learning_rate": 9.054136210041144e-06,
      "loss": 0.0,
      "step": 6700
    },
    {
      "epoch": 0.0948691484398196,
      "grad_norm": 0.0,
      "learning_rate": 9.052722362821475e-06,
      "loss": 0.0141,
      "step": 6710
    },
    {
      "epoch": 0.09501053316178654,
      "grad_norm": 0.1770326793193817,
      "learning_rate": 9.051308515601804e-06,
      "loss": 0.0002,
      "step": 6720
    },
    {
      "epoch": 0.09515191788375348,
      "grad_norm": 0.0,
      "learning_rate": 9.049894668382135e-06,
      "loss": 0.0,
      "step": 6730
    },
    {
      "epoch": 0.09529330260572043,
      "grad_norm": 1.1506814416861744e-06,
      "learning_rate": 9.048480821162466e-06,
      "loss": 0.0021,
      "step": 6740
    },
    {
      "epoch": 0.09543468732768737,
      "grad_norm": 0.0,
      "learning_rate": 9.047066973942796e-06,
      "loss": 0.2842,
      "step": 6750
    },
    {
      "epoch": 0.09557607204965432,
      "grad_norm": 0.0011667362414300442,
      "learning_rate": 9.045653126723127e-06,
      "loss": 0.0,
      "step": 6760
    },
    {
      "epoch": 0.09571745677162126,
      "grad_norm": 0.0,
      "learning_rate": 9.044239279503457e-06,
      "loss": 0.297,
      "step": 6770
    },
    {
      "epoch": 0.0958588414935882,
      "grad_norm": 0.0,
      "learning_rate": 9.042825432283789e-06,
      "loss": 0.0,
      "step": 6780
    },
    {
      "epoch": 0.09600022621555515,
      "grad_norm": 0.0,
      "learning_rate": 9.041411585064118e-06,
      "loss": 0.0,
      "step": 6790
    },
    {
      "epoch": 0.09614161093752209,
      "grad_norm": 0.0,
      "learning_rate": 9.039997737844449e-06,
      "loss": 0.0006,
      "step": 6800
    },
    {
      "epoch": 0.09628299565948903,
      "grad_norm": 0.0,
      "learning_rate": 9.03858389062478e-06,
      "loss": 0.0,
      "step": 6810
    },
    {
      "epoch": 0.09642438038145598,
      "grad_norm": 2.6800991690834053e-05,
      "learning_rate": 9.03717004340511e-06,
      "loss": 0.0,
      "step": 6820
    },
    {
      "epoch": 0.09656576510342292,
      "grad_norm": 0.0,
      "learning_rate": 9.035756196185442e-06,
      "loss": 0.0,
      "step": 6830
    },
    {
      "epoch": 0.09670714982538987,
      "grad_norm": 0.0,
      "learning_rate": 9.03434234896577e-06,
      "loss": 0.0,
      "step": 6840
    },
    {
      "epoch": 0.09684853454735681,
      "grad_norm": 0.0,
      "learning_rate": 9.032928501746102e-06,
      "loss": 0.0,
      "step": 6850
    },
    {
      "epoch": 0.09698991926932375,
      "grad_norm": 0.009607921354472637,
      "learning_rate": 9.031514654526433e-06,
      "loss": 0.1257,
      "step": 6860
    },
    {
      "epoch": 0.0971313039912907,
      "grad_norm": 0.0,
      "learning_rate": 9.030100807306763e-06,
      "loss": 0.0,
      "step": 6870
    },
    {
      "epoch": 0.09727268871325764,
      "grad_norm": 0.0,
      "learning_rate": 9.028686960087094e-06,
      "loss": 0.0,
      "step": 6880
    },
    {
      "epoch": 0.0974140734352246,
      "grad_norm": 0.005908549297600985,
      "learning_rate": 9.027273112867423e-06,
      "loss": 0.0,
      "step": 6890
    },
    {
      "epoch": 0.09755545815719154,
      "grad_norm": 0.0009036064730025828,
      "learning_rate": 9.025859265647754e-06,
      "loss": 0.0,
      "step": 6900
    },
    {
      "epoch": 0.09769684287915847,
      "grad_norm": 0.00022175305639393628,
      "learning_rate": 9.024445418428085e-06,
      "loss": 0.0014,
      "step": 6910
    },
    {
      "epoch": 0.09783822760112543,
      "grad_norm": 0.0019995064940303564,
      "learning_rate": 9.023031571208416e-06,
      "loss": 0.0,
      "step": 6920
    },
    {
      "epoch": 0.09797961232309237,
      "grad_norm": 4.437997176864883e-06,
      "learning_rate": 9.021617723988747e-06,
      "loss": 0.0001,
      "step": 6930
    },
    {
      "epoch": 0.0981209970450593,
      "grad_norm": 0.0,
      "learning_rate": 9.020203876769078e-06,
      "loss": 0.0007,
      "step": 6940
    },
    {
      "epoch": 0.09826238176702626,
      "grad_norm": 1.891333756276481e-09,
      "learning_rate": 9.018790029549409e-06,
      "loss": 0.0077,
      "step": 6950
    },
    {
      "epoch": 0.0984037664889932,
      "grad_norm": 0.0,
      "learning_rate": 9.017376182329738e-06,
      "loss": 0.0,
      "step": 6960
    },
    {
      "epoch": 0.09854515121096015,
      "grad_norm": 7.700653270603652e-08,
      "learning_rate": 9.015962335110069e-06,
      "loss": 0.0,
      "step": 6970
    },
    {
      "epoch": 0.09868653593292709,
      "grad_norm": 0.0,
      "learning_rate": 9.0145484878904e-06,
      "loss": 0.0,
      "step": 6980
    },
    {
      "epoch": 0.09882792065489403,
      "grad_norm": 0.0,
      "learning_rate": 9.01313464067073e-06,
      "loss": 0.3547,
      "step": 6990
    },
    {
      "epoch": 0.09896930537686098,
      "grad_norm": 0.0,
      "learning_rate": 9.011720793451061e-06,
      "loss": 0.0889,
      "step": 7000
    },
    {
      "epoch": 0.09911069009882792,
      "grad_norm": 0.0,
      "learning_rate": 9.01030694623139e-06,
      "loss": 0.0033,
      "step": 7010
    },
    {
      "epoch": 0.09925207482079486,
      "grad_norm": 0.0004047443508170545,
      "learning_rate": 9.008893099011721e-06,
      "loss": 0.0,
      "step": 7020
    },
    {
      "epoch": 0.09939345954276181,
      "grad_norm": 0.0,
      "learning_rate": 9.007479251792052e-06,
      "loss": 0.0001,
      "step": 7030
    },
    {
      "epoch": 0.09953484426472875,
      "grad_norm": 1.933863069680797e-09,
      "learning_rate": 9.006065404572383e-06,
      "loss": 0.0,
      "step": 7040
    },
    {
      "epoch": 0.0996762289866957,
      "grad_norm": 7.813167758285999e-05,
      "learning_rate": 9.004651557352712e-06,
      "loss": 0.0004,
      "step": 7050
    },
    {
      "epoch": 0.09981761370866264,
      "grad_norm": 1.872244581591076e-09,
      "learning_rate": 9.003237710133043e-06,
      "loss": 0.0,
      "step": 7060
    },
    {
      "epoch": 0.09995899843062958,
      "grad_norm": 7.996840167834307e-07,
      "learning_rate": 9.001823862913375e-06,
      "loss": 0.0052,
      "step": 7070
    },
    {
      "epoch": 0.10010038315259653,
      "grad_norm": 0.0,
      "learning_rate": 9.000410015693705e-06,
      "loss": 0.0106,
      "step": 7080
    },
    {
      "epoch": 0.10024176787456347,
      "grad_norm": 5.659393787384033,
      "learning_rate": 8.998996168474035e-06,
      "loss": 0.0169,
      "step": 7090
    },
    {
      "epoch": 0.10038315259653043,
      "grad_norm": 0.0,
      "learning_rate": 8.997582321254366e-06,
      "loss": 0.0,
      "step": 7100
    },
    {
      "epoch": 0.10052453731849736,
      "grad_norm": 0.0,
      "learning_rate": 8.996168474034697e-06,
      "loss": 0.0,
      "step": 7110
    },
    {
      "epoch": 0.1006659220404643,
      "grad_norm": 1.1799809129797723e-07,
      "learning_rate": 8.994754626815028e-06,
      "loss": 0.4502,
      "step": 7120
    },
    {
      "epoch": 0.10080730676243126,
      "grad_norm": 1.992252363081093e-09,
      "learning_rate": 8.993340779595357e-06,
      "loss": 0.0106,
      "step": 7130
    },
    {
      "epoch": 0.1009486914843982,
      "grad_norm": 73.46823120117188,
      "learning_rate": 8.991926932375688e-06,
      "loss": 0.0061,
      "step": 7140
    },
    {
      "epoch": 0.10109007620636513,
      "grad_norm": 0.0,
      "learning_rate": 8.990513085156019e-06,
      "loss": 0.0001,
      "step": 7150
    },
    {
      "epoch": 0.10123146092833209,
      "grad_norm": 0.0,
      "learning_rate": 8.98909923793635e-06,
      "loss": 0.0,
      "step": 7160
    },
    {
      "epoch": 0.10137284565029903,
      "grad_norm": 0.0,
      "learning_rate": 8.987685390716679e-06,
      "loss": 0.0,
      "step": 7170
    },
    {
      "epoch": 0.10151423037226598,
      "grad_norm": 0.0,
      "learning_rate": 8.98627154349701e-06,
      "loss": 0.0034,
      "step": 7180
    },
    {
      "epoch": 0.10165561509423292,
      "grad_norm": 0.0,
      "learning_rate": 8.984857696277342e-06,
      "loss": 0.0926,
      "step": 7190
    },
    {
      "epoch": 0.10179699981619986,
      "grad_norm": 0.0,
      "learning_rate": 8.983443849057671e-06,
      "loss": 0.0,
      "step": 7200
    },
    {
      "epoch": 0.10193838453816681,
      "grad_norm": 0.09449486434459686,
      "learning_rate": 8.982030001838002e-06,
      "loss": 0.0,
      "step": 7210
    },
    {
      "epoch": 0.10207976926013375,
      "grad_norm": 0.0,
      "learning_rate": 8.980616154618333e-06,
      "loss": 0.0,
      "step": 7220
    },
    {
      "epoch": 0.10222115398210069,
      "grad_norm": 0.009425126947462559,
      "learning_rate": 8.979202307398664e-06,
      "loss": 0.0,
      "step": 7230
    },
    {
      "epoch": 0.10236253870406764,
      "grad_norm": 0.00436916621401906,
      "learning_rate": 8.977788460178993e-06,
      "loss": 0.0,
      "step": 7240
    },
    {
      "epoch": 0.10250392342603458,
      "grad_norm": 0.0,
      "learning_rate": 8.976374612959324e-06,
      "loss": 0.0006,
      "step": 7250
    },
    {
      "epoch": 0.10264530814800153,
      "grad_norm": 1.332576036453247,
      "learning_rate": 8.974960765739655e-06,
      "loss": 0.003,
      "step": 7260
    },
    {
      "epoch": 0.10278669286996847,
      "grad_norm": 0.0,
      "learning_rate": 8.973546918519986e-06,
      "loss": 0.0,
      "step": 7270
    },
    {
      "epoch": 0.10292807759193541,
      "grad_norm": 1116.1085205078125,
      "learning_rate": 8.972133071300317e-06,
      "loss": 0.7078,
      "step": 7280
    },
    {
      "epoch": 0.10306946231390236,
      "grad_norm": 0.0,
      "learning_rate": 8.970719224080646e-06,
      "loss": 0.0,
      "step": 7290
    },
    {
      "epoch": 0.1032108470358693,
      "grad_norm": 0.0,
      "learning_rate": 8.969305376860977e-06,
      "loss": 0.0,
      "step": 7300
    },
    {
      "epoch": 0.10335223175783625,
      "grad_norm": 4.075867199304639e-09,
      "learning_rate": 8.967891529641308e-06,
      "loss": 0.0036,
      "step": 7310
    },
    {
      "epoch": 0.1034936164798032,
      "grad_norm": 0.023792805150151253,
      "learning_rate": 8.966477682421638e-06,
      "loss": 0.0,
      "step": 7320
    },
    {
      "epoch": 0.10363500120177013,
      "grad_norm": 0.7075510621070862,
      "learning_rate": 8.96506383520197e-06,
      "loss": 0.0149,
      "step": 7330
    },
    {
      "epoch": 0.10377638592373709,
      "grad_norm": 0.0,
      "learning_rate": 8.963649987982298e-06,
      "loss": 0.0001,
      "step": 7340
    },
    {
      "epoch": 0.10391777064570402,
      "grad_norm": 0.0,
      "learning_rate": 8.962236140762631e-06,
      "loss": 0.0,
      "step": 7350
    },
    {
      "epoch": 0.10405915536767096,
      "grad_norm": 7.214489414764103e-07,
      "learning_rate": 8.96082229354296e-06,
      "loss": 0.0001,
      "step": 7360
    },
    {
      "epoch": 0.10420054008963792,
      "grad_norm": 0.0,
      "learning_rate": 8.959408446323291e-06,
      "loss": 0.0001,
      "step": 7370
    },
    {
      "epoch": 0.10434192481160486,
      "grad_norm": 0.0,
      "learning_rate": 8.957994599103622e-06,
      "loss": 0.0,
      "step": 7380
    },
    {
      "epoch": 0.10448330953357181,
      "grad_norm": 42.41314697265625,
      "learning_rate": 8.956580751883953e-06,
      "loss": 0.0962,
      "step": 7390
    },
    {
      "epoch": 0.10462469425553875,
      "grad_norm": 0.0,
      "learning_rate": 8.955166904664284e-06,
      "loss": 0.0003,
      "step": 7400
    },
    {
      "epoch": 0.10476607897750569,
      "grad_norm": 0.0,
      "learning_rate": 8.953753057444613e-06,
      "loss": 0.0,
      "step": 7410
    },
    {
      "epoch": 0.10490746369947264,
      "grad_norm": 0.0,
      "learning_rate": 8.952339210224944e-06,
      "loss": 0.0,
      "step": 7420
    },
    {
      "epoch": 0.10504884842143958,
      "grad_norm": 0.0,
      "learning_rate": 8.950925363005274e-06,
      "loss": 0.0,
      "step": 7430
    },
    {
      "epoch": 0.10519023314340652,
      "grad_norm": 5.973474344500573e-06,
      "learning_rate": 8.949511515785605e-06,
      "loss": 0.0745,
      "step": 7440
    },
    {
      "epoch": 0.10533161786537347,
      "grad_norm": 4.243062790010299e-09,
      "learning_rate": 8.948097668565936e-06,
      "loss": 1.2253,
      "step": 7450
    },
    {
      "epoch": 0.10547300258734041,
      "grad_norm": 0.006417067255824804,
      "learning_rate": 8.946683821346265e-06,
      "loss": 0.0134,
      "step": 7460
    },
    {
      "epoch": 0.10561438730930736,
      "grad_norm": 0.06298182904720306,
      "learning_rate": 8.945269974126596e-06,
      "loss": 0.0023,
      "step": 7470
    },
    {
      "epoch": 0.1057557720312743,
      "grad_norm": 23.770410537719727,
      "learning_rate": 8.943856126906927e-06,
      "loss": 0.0056,
      "step": 7480
    },
    {
      "epoch": 0.10589715675324124,
      "grad_norm": 0.0,
      "learning_rate": 8.942442279687258e-06,
      "loss": 0.0,
      "step": 7490
    },
    {
      "epoch": 0.10603854147520819,
      "grad_norm": 0.0,
      "learning_rate": 8.941028432467587e-06,
      "loss": 0.0013,
      "step": 7500
    },
    {
      "epoch": 0.10617992619717513,
      "grad_norm": 0.0,
      "learning_rate": 8.93961458524792e-06,
      "loss": 0.0018,
      "step": 7510
    },
    {
      "epoch": 0.10632131091914208,
      "grad_norm": 0.0,
      "learning_rate": 8.93820073802825e-06,
      "loss": 0.0,
      "step": 7520
    },
    {
      "epoch": 0.10646269564110902,
      "grad_norm": 0.00012723487452603877,
      "learning_rate": 8.93678689080858e-06,
      "loss": 0.0,
      "step": 7530
    },
    {
      "epoch": 0.10660408036307596,
      "grad_norm": 2.06458827811673e-09,
      "learning_rate": 8.93537304358891e-06,
      "loss": 0.0001,
      "step": 7540
    },
    {
      "epoch": 0.10674546508504291,
      "grad_norm": 1.7022749432271667e-08,
      "learning_rate": 8.933959196369241e-06,
      "loss": 0.0006,
      "step": 7550
    },
    {
      "epoch": 0.10688684980700985,
      "grad_norm": 0.0,
      "learning_rate": 8.932545349149572e-06,
      "loss": 0.0005,
      "step": 7560
    },
    {
      "epoch": 0.10702823452897679,
      "grad_norm": 0.0,
      "learning_rate": 8.931131501929901e-06,
      "loss": 0.0,
      "step": 7570
    },
    {
      "epoch": 0.10716961925094375,
      "grad_norm": 1.9426154018731268e-09,
      "learning_rate": 8.929717654710232e-06,
      "loss": 0.0105,
      "step": 7580
    },
    {
      "epoch": 0.10731100397291068,
      "grad_norm": 0.00011057715892093256,
      "learning_rate": 8.928303807490563e-06,
      "loss": 0.0001,
      "step": 7590
    },
    {
      "epoch": 0.10745238869487764,
      "grad_norm": 0.0,
      "learning_rate": 8.926889960270894e-06,
      "loss": 0.0,
      "step": 7600
    },
    {
      "epoch": 0.10759377341684458,
      "grad_norm": 0.0,
      "learning_rate": 8.925476113051225e-06,
      "loss": 0.0468,
      "step": 7610
    },
    {
      "epoch": 0.10773515813881152,
      "grad_norm": 4.303150177001953,
      "learning_rate": 8.924062265831554e-06,
      "loss": 0.0217,
      "step": 7620
    },
    {
      "epoch": 0.10787654286077847,
      "grad_norm": 0.0,
      "learning_rate": 8.922648418611885e-06,
      "loss": 0.0003,
      "step": 7630
    },
    {
      "epoch": 0.10801792758274541,
      "grad_norm": 0.0,
      "learning_rate": 8.921234571392217e-06,
      "loss": 0.0,
      "step": 7640
    },
    {
      "epoch": 0.10815931230471235,
      "grad_norm": 0.0,
      "learning_rate": 8.919820724172546e-06,
      "loss": 0.0,
      "step": 7650
    },
    {
      "epoch": 0.1083006970266793,
      "grad_norm": 0.0,
      "learning_rate": 8.918406876952877e-06,
      "loss": 0.0994,
      "step": 7660
    },
    {
      "epoch": 0.10844208174864624,
      "grad_norm": 0.0,
      "learning_rate": 8.916993029733208e-06,
      "loss": 0.0,
      "step": 7670
    },
    {
      "epoch": 0.10858346647061319,
      "grad_norm": 0.0,
      "learning_rate": 8.915579182513539e-06,
      "loss": 0.0006,
      "step": 7680
    },
    {
      "epoch": 0.10872485119258013,
      "grad_norm": 0.0,
      "learning_rate": 8.914165335293868e-06,
      "loss": 0.0033,
      "step": 7690
    },
    {
      "epoch": 0.10886623591454707,
      "grad_norm": 0.0,
      "learning_rate": 8.912751488074199e-06,
      "loss": 0.0017,
      "step": 7700
    },
    {
      "epoch": 0.10900762063651402,
      "grad_norm": 1.399100824528432e-06,
      "learning_rate": 8.91133764085453e-06,
      "loss": 0.0005,
      "step": 7710
    },
    {
      "epoch": 0.10914900535848096,
      "grad_norm": 0.0,
      "learning_rate": 8.90992379363486e-06,
      "loss": 0.0003,
      "step": 7720
    },
    {
      "epoch": 0.10929039008044791,
      "grad_norm": 0.0,
      "learning_rate": 8.908509946415192e-06,
      "loss": 0.0,
      "step": 7730
    },
    {
      "epoch": 0.10943177480241485,
      "grad_norm": 0.0,
      "learning_rate": 8.90709609919552e-06,
      "loss": 0.0,
      "step": 7740
    },
    {
      "epoch": 0.10957315952438179,
      "grad_norm": 0.0,
      "learning_rate": 8.905682251975852e-06,
      "loss": 0.0,
      "step": 7750
    },
    {
      "epoch": 0.10971454424634874,
      "grad_norm": 0.0,
      "learning_rate": 8.904268404756182e-06,
      "loss": 0.0,
      "step": 7760
    },
    {
      "epoch": 0.10985592896831568,
      "grad_norm": 0.0,
      "learning_rate": 8.902854557536513e-06,
      "loss": 0.0001,
      "step": 7770
    },
    {
      "epoch": 0.10999731369028262,
      "grad_norm": 0.0,
      "learning_rate": 8.901440710316844e-06,
      "loss": 0.3597,
      "step": 7780
    },
    {
      "epoch": 0.11013869841224957,
      "grad_norm": 0.0,
      "learning_rate": 8.900026863097175e-06,
      "loss": 0.0,
      "step": 7790
    },
    {
      "epoch": 0.11028008313421651,
      "grad_norm": 0.0,
      "learning_rate": 8.898613015877506e-06,
      "loss": 0.3546,
      "step": 7800
    },
    {
      "epoch": 0.11042146785618347,
      "grad_norm": 0.0,
      "learning_rate": 8.897199168657835e-06,
      "loss": 0.0,
      "step": 7810
    },
    {
      "epoch": 0.1105628525781504,
      "grad_norm": 0.000519371242262423,
      "learning_rate": 8.895785321438166e-06,
      "loss": 0.0,
      "step": 7820
    },
    {
      "epoch": 0.11070423730011734,
      "grad_norm": 0.0,
      "learning_rate": 8.894371474218497e-06,
      "loss": 0.0022,
      "step": 7830
    },
    {
      "epoch": 0.1108456220220843,
      "grad_norm": 0.0,
      "learning_rate": 8.892957626998828e-06,
      "loss": 0.0002,
      "step": 7840
    },
    {
      "epoch": 0.11098700674405124,
      "grad_norm": 0.0004642541171051562,
      "learning_rate": 8.891543779779159e-06,
      "loss": 0.0,
      "step": 7850
    },
    {
      "epoch": 0.11112839146601818,
      "grad_norm": 2.588130598724092e-08,
      "learning_rate": 8.890129932559488e-06,
      "loss": 0.0001,
      "step": 7860
    },
    {
      "epoch": 0.11126977618798513,
      "grad_norm": 4.820982212550007e-06,
      "learning_rate": 8.888716085339819e-06,
      "loss": 0.0041,
      "step": 7870
    },
    {
      "epoch": 0.11141116090995207,
      "grad_norm": 0.0,
      "learning_rate": 8.88730223812015e-06,
      "loss": 0.0008,
      "step": 7880
    },
    {
      "epoch": 0.11155254563191902,
      "grad_norm": 0.05200035497546196,
      "learning_rate": 8.88588839090048e-06,
      "loss": 0.0,
      "step": 7890
    },
    {
      "epoch": 0.11169393035388596,
      "grad_norm": 9.357337694382295e-05,
      "learning_rate": 8.884474543680811e-06,
      "loss": 0.0016,
      "step": 7900
    },
    {
      "epoch": 0.1118353150758529,
      "grad_norm": 0.00016884878277778625,
      "learning_rate": 8.88306069646114e-06,
      "loss": 0.0,
      "step": 7910
    },
    {
      "epoch": 0.11197669979781985,
      "grad_norm": 9.280868851213597e-10,
      "learning_rate": 8.881646849241473e-06,
      "loss": 0.0,
      "step": 7920
    },
    {
      "epoch": 0.11211808451978679,
      "grad_norm": 4.7229509353637695,
      "learning_rate": 8.880233002021802e-06,
      "loss": 0.0477,
      "step": 7930
    },
    {
      "epoch": 0.11225946924175374,
      "grad_norm": 0.00042999163269996643,
      "learning_rate": 8.878819154802133e-06,
      "loss": 0.0,
      "step": 7940
    },
    {
      "epoch": 0.11240085396372068,
      "grad_norm": 0.004880134016275406,
      "learning_rate": 8.877405307582464e-06,
      "loss": 0.0,
      "step": 7950
    },
    {
      "epoch": 0.11254223868568762,
      "grad_norm": 5.562976360321045,
      "learning_rate": 8.875991460362795e-06,
      "loss": 0.0204,
      "step": 7960
    },
    {
      "epoch": 0.11268362340765457,
      "grad_norm": 0.0,
      "learning_rate": 8.874577613143125e-06,
      "loss": 0.0,
      "step": 7970
    },
    {
      "epoch": 0.11282500812962151,
      "grad_norm": 0.0,
      "learning_rate": 8.873163765923455e-06,
      "loss": 0.0004,
      "step": 7980
    },
    {
      "epoch": 0.11296639285158845,
      "grad_norm": 0.00026229239301756024,
      "learning_rate": 8.871749918703785e-06,
      "loss": 0.0006,
      "step": 7990
    },
    {
      "epoch": 0.1131077775735554,
      "grad_norm": 0.0,
      "learning_rate": 8.870336071484116e-06,
      "loss": 0.0,
      "step": 8000
    },
    {
      "epoch": 0.11324916229552234,
      "grad_norm": 0.0,
      "learning_rate": 8.868922224264447e-06,
      "loss": 0.0,
      "step": 8010
    },
    {
      "epoch": 0.1133905470174893,
      "grad_norm": 0.0,
      "learning_rate": 8.867508377044776e-06,
      "loss": 0.0072,
      "step": 8020
    },
    {
      "epoch": 0.11353193173945623,
      "grad_norm": 0.0,
      "learning_rate": 8.866094529825107e-06,
      "loss": 0.0101,
      "step": 8030
    },
    {
      "epoch": 0.11367331646142317,
      "grad_norm": 26.843605041503906,
      "learning_rate": 8.864680682605438e-06,
      "loss": 0.0029,
      "step": 8040
    },
    {
      "epoch": 0.11381470118339013,
      "grad_norm": 0.07870955765247345,
      "learning_rate": 8.863266835385769e-06,
      "loss": 0.0014,
      "step": 8050
    },
    {
      "epoch": 0.11395608590535707,
      "grad_norm": 1.0043170277640456e-06,
      "learning_rate": 8.8618529881661e-06,
      "loss": 0.0,
      "step": 8060
    },
    {
      "epoch": 0.114097470627324,
      "grad_norm": 0.000460799434222281,
      "learning_rate": 8.860439140946429e-06,
      "loss": 0.008,
      "step": 8070
    },
    {
      "epoch": 0.11423885534929096,
      "grad_norm": 0.0002003031549975276,
      "learning_rate": 8.859025293726761e-06,
      "loss": 0.0,
      "step": 8080
    },
    {
      "epoch": 0.1143802400712579,
      "grad_norm": 3.728193769347854e-05,
      "learning_rate": 8.857611446507092e-06,
      "loss": 0.001,
      "step": 8090
    },
    {
      "epoch": 0.11452162479322485,
      "grad_norm": 6.952925559744472e-06,
      "learning_rate": 8.856197599287421e-06,
      "loss": 0.0147,
      "step": 8100
    },
    {
      "epoch": 0.11466300951519179,
      "grad_norm": 0.0,
      "learning_rate": 8.854783752067752e-06,
      "loss": 0.0,
      "step": 8110
    },
    {
      "epoch": 0.11480439423715873,
      "grad_norm": 2.771701979042973e-09,
      "learning_rate": 8.853369904848083e-06,
      "loss": 0.0,
      "step": 8120
    },
    {
      "epoch": 0.11494577895912568,
      "grad_norm": 0.0,
      "learning_rate": 8.851956057628414e-06,
      "loss": 0.0,
      "step": 8130
    },
    {
      "epoch": 0.11508716368109262,
      "grad_norm": 3.2319283491233364e-05,
      "learning_rate": 8.850542210408743e-06,
      "loss": 0.0,
      "step": 8140
    },
    {
      "epoch": 0.11522854840305957,
      "grad_norm": 9.285391344704408e-10,
      "learning_rate": 8.849128363189074e-06,
      "loss": 0.0001,
      "step": 8150
    },
    {
      "epoch": 0.11536993312502651,
      "grad_norm": 0.030695801600813866,
      "learning_rate": 8.847714515969405e-06,
      "loss": 0.0002,
      "step": 8160
    },
    {
      "epoch": 0.11551131784699345,
      "grad_norm": 0.0,
      "learning_rate": 8.846300668749736e-06,
      "loss": 0.0007,
      "step": 8170
    },
    {
      "epoch": 0.1156527025689604,
      "grad_norm": 0.0,
      "learning_rate": 8.844886821530067e-06,
      "loss": 0.0,
      "step": 8180
    },
    {
      "epoch": 0.11579408729092734,
      "grad_norm": 1.207237971811992e-07,
      "learning_rate": 8.843472974310396e-06,
      "loss": 0.0004,
      "step": 8190
    },
    {
      "epoch": 0.11593547201289428,
      "grad_norm": 0.0,
      "learning_rate": 8.842059127090727e-06,
      "loss": 0.0001,
      "step": 8200
    },
    {
      "epoch": 0.11607685673486123,
      "grad_norm": 0.0,
      "learning_rate": 8.840645279871057e-06,
      "loss": 0.0324,
      "step": 8210
    },
    {
      "epoch": 0.11621824145682817,
      "grad_norm": 1.5034032685434795e-07,
      "learning_rate": 8.839231432651388e-06,
      "loss": 0.0,
      "step": 8220
    },
    {
      "epoch": 0.11635962617879513,
      "grad_norm": 0.0,
      "learning_rate": 8.83781758543172e-06,
      "loss": 0.0,
      "step": 8230
    },
    {
      "epoch": 0.11650101090076206,
      "grad_norm": 0.0,
      "learning_rate": 8.83640373821205e-06,
      "loss": 0.0,
      "step": 8240
    },
    {
      "epoch": 0.116642395622729,
      "grad_norm": 0.0007509257411584258,
      "learning_rate": 8.834989890992381e-06,
      "loss": 0.0103,
      "step": 8250
    },
    {
      "epoch": 0.11678378034469596,
      "grad_norm": 0.0,
      "learning_rate": 8.83357604377271e-06,
      "loss": 0.8856,
      "step": 8260
    },
    {
      "epoch": 0.1169251650666629,
      "grad_norm": 0.0,
      "learning_rate": 8.832162196553041e-06,
      "loss": 0.0001,
      "step": 8270
    },
    {
      "epoch": 0.11706654978862983,
      "grad_norm": 0.0003011324442923069,
      "learning_rate": 8.830748349333372e-06,
      "loss": 0.0,
      "step": 8280
    },
    {
      "epoch": 0.11720793451059679,
      "grad_norm": 0.042716898024082184,
      "learning_rate": 8.829334502113703e-06,
      "loss": 0.0,
      "step": 8290
    },
    {
      "epoch": 0.11734931923256373,
      "grad_norm": 6.854095317976316e-07,
      "learning_rate": 8.827920654894033e-06,
      "loss": 0.0,
      "step": 8300
    },
    {
      "epoch": 0.11749070395453068,
      "grad_norm": 0.0,
      "learning_rate": 8.826506807674363e-06,
      "loss": 0.0,
      "step": 8310
    },
    {
      "epoch": 0.11763208867649762,
      "grad_norm": 0.0,
      "learning_rate": 8.825092960454694e-06,
      "loss": 0.124,
      "step": 8320
    },
    {
      "epoch": 0.11777347339846456,
      "grad_norm": 0.13539712131023407,
      "learning_rate": 8.823679113235024e-06,
      "loss": 0.0002,
      "step": 8330
    },
    {
      "epoch": 0.11791485812043151,
      "grad_norm": 0.00019054370932281017,
      "learning_rate": 8.822265266015355e-06,
      "loss": 0.0,
      "step": 8340
    },
    {
      "epoch": 0.11805624284239845,
      "grad_norm": 6.775285328330938e-06,
      "learning_rate": 8.820851418795686e-06,
      "loss": 0.0,
      "step": 8350
    },
    {
      "epoch": 0.1181976275643654,
      "grad_norm": 0.0,
      "learning_rate": 8.819437571576017e-06,
      "loss": 0.0,
      "step": 8360
    },
    {
      "epoch": 0.11833901228633234,
      "grad_norm": 704.3154907226562,
      "learning_rate": 8.818023724356348e-06,
      "loss": 0.7,
      "step": 8370
    },
    {
      "epoch": 0.11848039700829928,
      "grad_norm": 2.251065416203346e-05,
      "learning_rate": 8.816609877136677e-06,
      "loss": 0.0,
      "step": 8380
    },
    {
      "epoch": 0.11862178173026623,
      "grad_norm": 0.0,
      "learning_rate": 8.815196029917008e-06,
      "loss": 0.0005,
      "step": 8390
    },
    {
      "epoch": 0.11876316645223317,
      "grad_norm": 0.0,
      "learning_rate": 8.813782182697339e-06,
      "loss": 0.0633,
      "step": 8400
    },
    {
      "epoch": 0.11890455117420011,
      "grad_norm": 0.0,
      "learning_rate": 8.81236833547767e-06,
      "loss": 0.2225,
      "step": 8410
    },
    {
      "epoch": 0.11904593589616706,
      "grad_norm": 0.0,
      "learning_rate": 8.810954488258e-06,
      "loss": 0.0,
      "step": 8420
    },
    {
      "epoch": 0.119187320618134,
      "grad_norm": 8.636416896479204e-05,
      "learning_rate": 8.80954064103833e-06,
      "loss": 0.0,
      "step": 8430
    },
    {
      "epoch": 0.11932870534010095,
      "grad_norm": 0.043671052902936935,
      "learning_rate": 8.80812679381866e-06,
      "loss": 0.0,
      "step": 8440
    },
    {
      "epoch": 0.1194700900620679,
      "grad_norm": 0.0,
      "learning_rate": 8.806712946598991e-06,
      "loss": 0.0,
      "step": 8450
    },
    {
      "epoch": 0.11961147478403483,
      "grad_norm": 0.0,
      "learning_rate": 8.805299099379322e-06,
      "loss": 0.0,
      "step": 8460
    },
    {
      "epoch": 0.11975285950600179,
      "grad_norm": 0.00024210558331105858,
      "learning_rate": 8.803885252159651e-06,
      "loss": 0.0,
      "step": 8470
    },
    {
      "epoch": 0.11989424422796872,
      "grad_norm": 0.0,
      "learning_rate": 8.802471404939982e-06,
      "loss": 0.0009,
      "step": 8480
    },
    {
      "epoch": 0.12003562894993566,
      "grad_norm": 0.0,
      "learning_rate": 8.801057557720315e-06,
      "loss": 0.0798,
      "step": 8490
    },
    {
      "epoch": 0.12017701367190262,
      "grad_norm": 1.1828554306703154e-05,
      "learning_rate": 8.799643710500644e-06,
      "loss": 0.0,
      "step": 8500
    },
    {
      "epoch": 0.12031839839386956,
      "grad_norm": 0.0,
      "learning_rate": 8.798229863280975e-06,
      "loss": 0.0001,
      "step": 8510
    },
    {
      "epoch": 0.12045978311583651,
      "grad_norm": 0.0,
      "learning_rate": 8.796816016061306e-06,
      "loss": 0.0,
      "step": 8520
    },
    {
      "epoch": 0.12060116783780345,
      "grad_norm": 0.008847554214298725,
      "learning_rate": 8.795402168841636e-06,
      "loss": 0.0,
      "step": 8530
    },
    {
      "epoch": 0.12074255255977039,
      "grad_norm": 4.0179052120947745e-06,
      "learning_rate": 8.793988321621966e-06,
      "loss": 0.0079,
      "step": 8540
    },
    {
      "epoch": 0.12088393728173734,
      "grad_norm": 1.4114423265709775e-06,
      "learning_rate": 8.792574474402296e-06,
      "loss": 0.0393,
      "step": 8550
    },
    {
      "epoch": 0.12102532200370428,
      "grad_norm": 0.0,
      "learning_rate": 8.791160627182627e-06,
      "loss": 0.0,
      "step": 8560
    },
    {
      "epoch": 0.12116670672567123,
      "grad_norm": 29.258621215820312,
      "learning_rate": 8.789746779962958e-06,
      "loss": 0.006,
      "step": 8570
    },
    {
      "epoch": 0.12130809144763817,
      "grad_norm": 0.0,
      "learning_rate": 8.788332932743289e-06,
      "loss": 0.0,
      "step": 8580
    },
    {
      "epoch": 0.12144947616960511,
      "grad_norm": 0.0,
      "learning_rate": 8.786919085523618e-06,
      "loss": 0.0,
      "step": 8590
    },
    {
      "epoch": 0.12159086089157206,
      "grad_norm": 0.004834381863474846,
      "learning_rate": 8.785505238303949e-06,
      "loss": 0.0,
      "step": 8600
    },
    {
      "epoch": 0.121732245613539,
      "grad_norm": 0.013689261861145496,
      "learning_rate": 8.78409139108428e-06,
      "loss": 0.0565,
      "step": 8610
    },
    {
      "epoch": 0.12187363033550594,
      "grad_norm": 0.0,
      "learning_rate": 8.78267754386461e-06,
      "loss": 0.0001,
      "step": 8620
    },
    {
      "epoch": 0.12201501505747289,
      "grad_norm": 0.0,
      "learning_rate": 8.781263696644942e-06,
      "loss": 0.0037,
      "step": 8630
    },
    {
      "epoch": 0.12215639977943983,
      "grad_norm": 0.0,
      "learning_rate": 8.77984984942527e-06,
      "loss": 0.0001,
      "step": 8640
    },
    {
      "epoch": 0.12229778450140678,
      "grad_norm": 2.0073818518540065e-07,
      "learning_rate": 8.778436002205603e-06,
      "loss": 0.0,
      "step": 8650
    },
    {
      "epoch": 0.12243916922337372,
      "grad_norm": 0.0,
      "learning_rate": 8.777022154985932e-06,
      "loss": 0.0,
      "step": 8660
    },
    {
      "epoch": 0.12258055394534066,
      "grad_norm": 0.00042410960304550827,
      "learning_rate": 8.775608307766263e-06,
      "loss": 0.0141,
      "step": 8670
    },
    {
      "epoch": 0.12272193866730761,
      "grad_norm": 0.0,
      "learning_rate": 8.774194460546594e-06,
      "loss": 0.0125,
      "step": 8680
    },
    {
      "epoch": 0.12286332338927455,
      "grad_norm": 0.0,
      "learning_rate": 8.772780613326925e-06,
      "loss": 1.208,
      "step": 8690
    },
    {
      "epoch": 0.12300470811124149,
      "grad_norm": 2.0089898109436035,
      "learning_rate": 8.771366766107256e-06,
      "loss": 0.0003,
      "step": 8700
    },
    {
      "epoch": 0.12314609283320845,
      "grad_norm": 0.0,
      "learning_rate": 8.769952918887585e-06,
      "loss": 0.0,
      "step": 8710
    },
    {
      "epoch": 0.12328747755517538,
      "grad_norm": 0.0,
      "learning_rate": 8.768539071667916e-06,
      "loss": 0.0464,
      "step": 8720
    },
    {
      "epoch": 0.12342886227714234,
      "grad_norm": 0.0,
      "learning_rate": 8.767125224448247e-06,
      "loss": 0.0,
      "step": 8730
    },
    {
      "epoch": 0.12357024699910928,
      "grad_norm": 0.0,
      "learning_rate": 8.765711377228578e-06,
      "loss": 0.0975,
      "step": 8740
    },
    {
      "epoch": 0.12371163172107622,
      "grad_norm": 0.0,
      "learning_rate": 8.764297530008908e-06,
      "loss": 0.0002,
      "step": 8750
    },
    {
      "epoch": 0.12385301644304317,
      "grad_norm": 0.0010065487585961819,
      "learning_rate": 8.762883682789238e-06,
      "loss": 0.0004,
      "step": 8760
    },
    {
      "epoch": 0.1239944011650101,
      "grad_norm": 0.0,
      "learning_rate": 8.761469835569569e-06,
      "loss": 0.1279,
      "step": 8770
    },
    {
      "epoch": 0.12413578588697706,
      "grad_norm": 9.082593011022766e-10,
      "learning_rate": 8.7600559883499e-06,
      "loss": 0.6715,
      "step": 8780
    },
    {
      "epoch": 0.124277170608944,
      "grad_norm": 0.0009641576907597482,
      "learning_rate": 8.75864214113023e-06,
      "loss": 0.0,
      "step": 8790
    },
    {
      "epoch": 0.12441855533091094,
      "grad_norm": 1.0508696504984982e-05,
      "learning_rate": 8.757228293910561e-06,
      "loss": 0.4586,
      "step": 8800
    },
    {
      "epoch": 0.12455994005287789,
      "grad_norm": 1.68985625350615e-05,
      "learning_rate": 8.755814446690892e-06,
      "loss": 0.0036,
      "step": 8810
    },
    {
      "epoch": 0.12470132477484483,
      "grad_norm": 0.0,
      "learning_rate": 8.754400599471223e-06,
      "loss": 0.0,
      "step": 8820
    },
    {
      "epoch": 0.12484270949681177,
      "grad_norm": 0.0,
      "learning_rate": 8.752986752251552e-06,
      "loss": 0.0002,
      "step": 8830
    },
    {
      "epoch": 0.12498409421877872,
      "grad_norm": 0.0,
      "learning_rate": 8.751572905031883e-06,
      "loss": 0.0008,
      "step": 8840
    },
    {
      "epoch": 0.12512547894074566,
      "grad_norm": 0.0,
      "learning_rate": 8.750159057812214e-06,
      "loss": 0.0,
      "step": 8850
    },
    {
      "epoch": 0.1252668636627126,
      "grad_norm": 5.045902184974693e-07,
      "learning_rate": 8.748745210592545e-06,
      "loss": 0.067,
      "step": 8860
    },
    {
      "epoch": 0.12540824838467954,
      "grad_norm": 1.4465963431575801e-05,
      "learning_rate": 8.747331363372875e-06,
      "loss": 0.0,
      "step": 8870
    },
    {
      "epoch": 0.1255496331066465,
      "grad_norm": 0.0,
      "learning_rate": 8.745917516153205e-06,
      "loss": 0.0003,
      "step": 8880
    },
    {
      "epoch": 0.12569101782861344,
      "grad_norm": 0.004725868348032236,
      "learning_rate": 8.744503668933535e-06,
      "loss": 0.0,
      "step": 8890
    },
    {
      "epoch": 0.1258324025505804,
      "grad_norm": 0.0,
      "learning_rate": 8.743089821713866e-06,
      "loss": 0.0,
      "step": 8900
    },
    {
      "epoch": 0.12597378727254732,
      "grad_norm": 0.0,
      "learning_rate": 8.741675974494197e-06,
      "loss": 0.0,
      "step": 8910
    },
    {
      "epoch": 0.12611517199451427,
      "grad_norm": 0.0,
      "learning_rate": 8.740262127274526e-06,
      "loss": 0.8584,
      "step": 8920
    },
    {
      "epoch": 0.12625655671648123,
      "grad_norm": 0.0,
      "learning_rate": 8.738848280054857e-06,
      "loss": 0.0005,
      "step": 8930
    },
    {
      "epoch": 0.12639794143844815,
      "grad_norm": 1.4877489805221558,
      "learning_rate": 8.73743443283519e-06,
      "loss": 0.0004,
      "step": 8940
    },
    {
      "epoch": 0.1265393261604151,
      "grad_norm": 0.016410617157816887,
      "learning_rate": 8.736020585615519e-06,
      "loss": 0.0,
      "step": 8950
    },
    {
      "epoch": 0.12668071088238206,
      "grad_norm": 0.00011500452819745988,
      "learning_rate": 8.73460673839585e-06,
      "loss": 0.0,
      "step": 8960
    },
    {
      "epoch": 0.12682209560434898,
      "grad_norm": 0.0,
      "learning_rate": 8.73319289117618e-06,
      "loss": 0.0,
      "step": 8970
    },
    {
      "epoch": 0.12696348032631594,
      "grad_norm": 9.64106789069774e-07,
      "learning_rate": 8.731779043956511e-06,
      "loss": 0.0,
      "step": 8980
    },
    {
      "epoch": 0.1271048650482829,
      "grad_norm": 0.003959037363529205,
      "learning_rate": 8.73036519673684e-06,
      "loss": 0.0,
      "step": 8990
    },
    {
      "epoch": 0.12724624977024981,
      "grad_norm": 2.928591698037053e-07,
      "learning_rate": 8.728951349517171e-06,
      "loss": 0.0001,
      "step": 9000
    },
    {
      "epoch": 0.12738763449221677,
      "grad_norm": 0.0,
      "learning_rate": 8.727537502297502e-06,
      "loss": 0.0,
      "step": 9010
    },
    {
      "epoch": 0.12752901921418372,
      "grad_norm": 0.0,
      "learning_rate": 8.726123655077833e-06,
      "loss": 0.2406,
      "step": 9020
    },
    {
      "epoch": 0.12767040393615067,
      "grad_norm": 0.00022562479716725647,
      "learning_rate": 8.724709807858164e-06,
      "loss": 0.0,
      "step": 9030
    },
    {
      "epoch": 0.1278117886581176,
      "grad_norm": 0.0,
      "learning_rate": 8.723295960638493e-06,
      "loss": 0.0539,
      "step": 9040
    },
    {
      "epoch": 0.12795317338008455,
      "grad_norm": 0.0,
      "learning_rate": 8.721882113418824e-06,
      "loss": 0.0005,
      "step": 9050
    },
    {
      "epoch": 0.1280945581020515,
      "grad_norm": 0.0,
      "learning_rate": 8.720468266199157e-06,
      "loss": 0.0,
      "step": 9060
    },
    {
      "epoch": 0.12823594282401843,
      "grad_norm": 0.0,
      "learning_rate": 8.719054418979486e-06,
      "loss": 0.0039,
      "step": 9070
    },
    {
      "epoch": 0.12837732754598538,
      "grad_norm": 1.7227227999683237e-06,
      "learning_rate": 8.717640571759817e-06,
      "loss": 0.0002,
      "step": 9080
    },
    {
      "epoch": 0.12851871226795233,
      "grad_norm": 0.0,
      "learning_rate": 8.716226724540147e-06,
      "loss": 0.0,
      "step": 9090
    },
    {
      "epoch": 0.12866009698991926,
      "grad_norm": 9.347219664945783e-10,
      "learning_rate": 8.714812877320478e-06,
      "loss": 0.0,
      "step": 9100
    },
    {
      "epoch": 0.1288014817118862,
      "grad_norm": 1.0221752972938702e-06,
      "learning_rate": 8.713399030100807e-06,
      "loss": 0.0,
      "step": 9110
    },
    {
      "epoch": 0.12894286643385316,
      "grad_norm": 0.0,
      "learning_rate": 8.711985182881138e-06,
      "loss": 0.0,
      "step": 9120
    },
    {
      "epoch": 0.1290842511558201,
      "grad_norm": 9.792204824776718e-10,
      "learning_rate": 8.71057133566147e-06,
      "loss": 0.0003,
      "step": 9130
    },
    {
      "epoch": 0.12922563587778704,
      "grad_norm": 0.0,
      "learning_rate": 8.7091574884418e-06,
      "loss": 0.0,
      "step": 9140
    },
    {
      "epoch": 0.129367020599754,
      "grad_norm": 0.0,
      "learning_rate": 8.707743641222131e-06,
      "loss": 0.0,
      "step": 9150
    },
    {
      "epoch": 0.12950840532172095,
      "grad_norm": 0.0,
      "learning_rate": 8.70632979400246e-06,
      "loss": 0.0,
      "step": 9160
    },
    {
      "epoch": 0.12964979004368787,
      "grad_norm": 1.0016569795823216e-09,
      "learning_rate": 8.704915946782791e-06,
      "loss": 0.0,
      "step": 9170
    },
    {
      "epoch": 0.12979117476565483,
      "grad_norm": 0.15401218831539154,
      "learning_rate": 8.703502099563122e-06,
      "loss": 0.0077,
      "step": 9180
    },
    {
      "epoch": 0.12993255948762178,
      "grad_norm": 1.0095965174983235e-09,
      "learning_rate": 8.702088252343453e-06,
      "loss": 0.1601,
      "step": 9190
    },
    {
      "epoch": 0.1300739442095887,
      "grad_norm": 0.0012683490058407187,
      "learning_rate": 8.700674405123783e-06,
      "loss": 0.0008,
      "step": 9200
    },
    {
      "epoch": 0.13021532893155566,
      "grad_norm": 2.024632522079628e-06,
      "learning_rate": 8.699260557904113e-06,
      "loss": 0.0,
      "step": 9210
    },
    {
      "epoch": 0.1303567136535226,
      "grad_norm": 0.0,
      "learning_rate": 8.697846710684445e-06,
      "loss": 0.0013,
      "step": 9220
    },
    {
      "epoch": 0.13049809837548954,
      "grad_norm": 0.0,
      "learning_rate": 8.696432863464774e-06,
      "loss": 0.0062,
      "step": 9230
    },
    {
      "epoch": 0.1306394830974565,
      "grad_norm": 0.0,
      "learning_rate": 8.695019016245105e-06,
      "loss": 0.0001,
      "step": 9240
    },
    {
      "epoch": 0.13078086781942344,
      "grad_norm": 0.0,
      "learning_rate": 8.693605169025436e-06,
      "loss": 0.0,
      "step": 9250
    },
    {
      "epoch": 0.13092225254139037,
      "grad_norm": 1.0535210321904742e-06,
      "learning_rate": 8.692191321805767e-06,
      "loss": 0.0016,
      "step": 9260
    },
    {
      "epoch": 0.13106363726335732,
      "grad_norm": 3.8424759623012505e-06,
      "learning_rate": 8.690777474586098e-06,
      "loss": 0.0001,
      "step": 9270
    },
    {
      "epoch": 0.13120502198532427,
      "grad_norm": 6.635727078929676e-09,
      "learning_rate": 8.689363627366427e-06,
      "loss": 0.0,
      "step": 9280
    },
    {
      "epoch": 0.1313464067072912,
      "grad_norm": 0.0,
      "learning_rate": 8.687949780146758e-06,
      "loss": 0.0,
      "step": 9290
    },
    {
      "epoch": 0.13148779142925815,
      "grad_norm": 0.5093436241149902,
      "learning_rate": 8.686535932927089e-06,
      "loss": 0.0001,
      "step": 9300
    },
    {
      "epoch": 0.1316291761512251,
      "grad_norm": 0.0,
      "learning_rate": 8.68512208570742e-06,
      "loss": 0.0001,
      "step": 9310
    },
    {
      "epoch": 0.13177056087319206,
      "grad_norm": 126.05400085449219,
      "learning_rate": 8.68370823848775e-06,
      "loss": 0.044,
      "step": 9320
    },
    {
      "epoch": 0.13191194559515898,
      "grad_norm": 2.3096465184835324e-08,
      "learning_rate": 8.68229439126808e-06,
      "loss": 0.0,
      "step": 9330
    },
    {
      "epoch": 0.13205333031712593,
      "grad_norm": 0.0,
      "learning_rate": 8.68088054404841e-06,
      "loss": 0.1778,
      "step": 9340
    },
    {
      "epoch": 0.13219471503909289,
      "grad_norm": 0.0,
      "learning_rate": 8.679466696828741e-06,
      "loss": 0.0001,
      "step": 9350
    },
    {
      "epoch": 0.1323360997610598,
      "grad_norm": 0.0,
      "learning_rate": 8.678052849609072e-06,
      "loss": 0.0,
      "step": 9360
    },
    {
      "epoch": 0.13247748448302676,
      "grad_norm": 0.0,
      "learning_rate": 8.676639002389401e-06,
      "loss": 0.0,
      "step": 9370
    },
    {
      "epoch": 0.13261886920499372,
      "grad_norm": 0.635364830493927,
      "learning_rate": 8.675225155169734e-06,
      "loss": 0.0026,
      "step": 9380
    },
    {
      "epoch": 0.13276025392696064,
      "grad_norm": 0.0014158926205709577,
      "learning_rate": 8.673811307950065e-06,
      "loss": 0.0065,
      "step": 9390
    },
    {
      "epoch": 0.1329016386489276,
      "grad_norm": 0.0,
      "learning_rate": 8.672397460730394e-06,
      "loss": 0.0001,
      "step": 9400
    },
    {
      "epoch": 0.13304302337089455,
      "grad_norm": 0.0,
      "learning_rate": 8.670983613510725e-06,
      "loss": 0.0,
      "step": 9410
    },
    {
      "epoch": 0.13318440809286147,
      "grad_norm": 0.0,
      "learning_rate": 8.669569766291056e-06,
      "loss": 0.0,
      "step": 9420
    },
    {
      "epoch": 0.13332579281482843,
      "grad_norm": 0.0,
      "learning_rate": 8.668155919071386e-06,
      "loss": 0.0,
      "step": 9430
    },
    {
      "epoch": 0.13346717753679538,
      "grad_norm": 0.0,
      "learning_rate": 8.666742071851716e-06,
      "loss": 0.0,
      "step": 9440
    },
    {
      "epoch": 0.13360856225876233,
      "grad_norm": 0.0,
      "learning_rate": 8.665328224632046e-06,
      "loss": 0.0004,
      "step": 9450
    },
    {
      "epoch": 0.13374994698072926,
      "grad_norm": 0.0,
      "learning_rate": 8.663914377412377e-06,
      "loss": 0.0,
      "step": 9460
    },
    {
      "epoch": 0.1338913317026962,
      "grad_norm": 0.0,
      "learning_rate": 8.662500530192708e-06,
      "loss": 0.0,
      "step": 9470
    },
    {
      "epoch": 0.13403271642466316,
      "grad_norm": 0.0,
      "learning_rate": 8.661086682973039e-06,
      "loss": 0.0,
      "step": 9480
    },
    {
      "epoch": 0.1341741011466301,
      "grad_norm": 1.3523161896955571e-06,
      "learning_rate": 8.659672835753368e-06,
      "loss": 0.0,
      "step": 9490
    },
    {
      "epoch": 0.13431548586859704,
      "grad_norm": 0.0,
      "learning_rate": 8.658258988533699e-06,
      "loss": 0.0016,
      "step": 9500
    },
    {
      "epoch": 0.134456870590564,
      "grad_norm": 0.005052709020674229,
      "learning_rate": 8.656845141314032e-06,
      "loss": 0.0,
      "step": 9510
    },
    {
      "epoch": 0.13459825531253092,
      "grad_norm": 3.048221941526208e-08,
      "learning_rate": 8.65543129409436e-06,
      "loss": 0.0,
      "step": 9520
    },
    {
      "epoch": 0.13473964003449787,
      "grad_norm": 0.0,
      "learning_rate": 8.654017446874692e-06,
      "loss": 0.0014,
      "step": 9530
    },
    {
      "epoch": 0.13488102475646482,
      "grad_norm": 0.003682431997731328,
      "learning_rate": 8.652603599655022e-06,
      "loss": 1.5797,
      "step": 9540
    },
    {
      "epoch": 0.13502240947843175,
      "grad_norm": 0.0,
      "learning_rate": 8.651189752435353e-06,
      "loss": 0.0284,
      "step": 9550
    },
    {
      "epoch": 0.1351637942003987,
      "grad_norm": 0.0,
      "learning_rate": 8.649775905215682e-06,
      "loss": 0.0,
      "step": 9560
    },
    {
      "epoch": 0.13530517892236565,
      "grad_norm": 0.0,
      "learning_rate": 8.648362057996013e-06,
      "loss": 0.0,
      "step": 9570
    },
    {
      "epoch": 0.1354465636443326,
      "grad_norm": 2.3245351258083247e-05,
      "learning_rate": 8.646948210776344e-06,
      "loss": 0.0,
      "step": 9580
    },
    {
      "epoch": 0.13558794836629953,
      "grad_norm": 0.009275789372622967,
      "learning_rate": 8.645534363556675e-06,
      "loss": 0.0682,
      "step": 9590
    },
    {
      "epoch": 0.13572933308826649,
      "grad_norm": 4.45172482210765e-10,
      "learning_rate": 8.644120516337006e-06,
      "loss": 0.02,
      "step": 9600
    },
    {
      "epoch": 0.13587071781023344,
      "grad_norm": 0.0,
      "learning_rate": 8.642706669117335e-06,
      "loss": 0.0,
      "step": 9610
    },
    {
      "epoch": 0.13601210253220036,
      "grad_norm": 0.0,
      "learning_rate": 8.641292821897666e-06,
      "loss": 0.0,
      "step": 9620
    },
    {
      "epoch": 0.13615348725416732,
      "grad_norm": 0.0,
      "learning_rate": 8.639878974677997e-06,
      "loss": 0.1147,
      "step": 9630
    },
    {
      "epoch": 0.13629487197613427,
      "grad_norm": 0.0,
      "learning_rate": 8.638465127458328e-06,
      "loss": 0.0004,
      "step": 9640
    },
    {
      "epoch": 0.1364362566981012,
      "grad_norm": 0.007070645224303007,
      "learning_rate": 8.637051280238658e-06,
      "loss": 0.0001,
      "step": 9650
    },
    {
      "epoch": 0.13657764142006815,
      "grad_norm": 0.0,
      "learning_rate": 8.63563743301899e-06,
      "loss": 0.0121,
      "step": 9660
    },
    {
      "epoch": 0.1367190261420351,
      "grad_norm": 0.00011837935744551942,
      "learning_rate": 8.63422358579932e-06,
      "loss": 0.0,
      "step": 9670
    },
    {
      "epoch": 0.13686041086400202,
      "grad_norm": 4.407257847560686e-08,
      "learning_rate": 8.63280973857965e-06,
      "loss": 0.0,
      "step": 9680
    },
    {
      "epoch": 0.13700179558596898,
      "grad_norm": 9.395802771905437e-05,
      "learning_rate": 8.63139589135998e-06,
      "loss": 0.0,
      "step": 9690
    },
    {
      "epoch": 0.13714318030793593,
      "grad_norm": 8.368814974346606e-09,
      "learning_rate": 8.629982044140311e-06,
      "loss": 0.0,
      "step": 9700
    },
    {
      "epoch": 0.13728456502990286,
      "grad_norm": 0.0,
      "learning_rate": 8.628568196920642e-06,
      "loss": 0.0,
      "step": 9710
    },
    {
      "epoch": 0.1374259497518698,
      "grad_norm": 0.0015980815514922142,
      "learning_rate": 8.627154349700973e-06,
      "loss": 0.0,
      "step": 9720
    },
    {
      "epoch": 0.13756733447383676,
      "grad_norm": 0.0,
      "learning_rate": 8.625740502481302e-06,
      "loss": 0.1663,
      "step": 9730
    },
    {
      "epoch": 0.1377087191958037,
      "grad_norm": 0.5180811882019043,
      "learning_rate": 8.624326655261633e-06,
      "loss": 0.0001,
      "step": 9740
    },
    {
      "epoch": 0.13785010391777064,
      "grad_norm": 0.6589863896369934,
      "learning_rate": 8.622912808041964e-06,
      "loss": 0.0116,
      "step": 9750
    },
    {
      "epoch": 0.1379914886397376,
      "grad_norm": 0.0,
      "learning_rate": 8.621498960822294e-06,
      "loss": 0.0001,
      "step": 9760
    },
    {
      "epoch": 0.13813287336170454,
      "grad_norm": 0.0,
      "learning_rate": 8.620085113602625e-06,
      "loss": 0.0,
      "step": 9770
    },
    {
      "epoch": 0.13827425808367147,
      "grad_norm": 3.1462888728128746e-05,
      "learning_rate": 8.618671266382955e-06,
      "loss": 0.0,
      "step": 9780
    },
    {
      "epoch": 0.13841564280563842,
      "grad_norm": 0.0,
      "learning_rate": 8.617257419163287e-06,
      "loss": 0.0259,
      "step": 9790
    },
    {
      "epoch": 0.13855702752760538,
      "grad_norm": 4.799990960258071e-10,
      "learning_rate": 8.615843571943616e-06,
      "loss": 0.0,
      "step": 9800
    },
    {
      "epoch": 0.1386984122495723,
      "grad_norm": 0.08505945652723312,
      "learning_rate": 8.614429724723947e-06,
      "loss": 0.0,
      "step": 9810
    },
    {
      "epoch": 0.13883979697153925,
      "grad_norm": 0.0,
      "learning_rate": 8.613015877504278e-06,
      "loss": 0.006,
      "step": 9820
    },
    {
      "epoch": 0.1389811816935062,
      "grad_norm": 4.4827778765466064e-05,
      "learning_rate": 8.611602030284609e-06,
      "loss": 0.2666,
      "step": 9830
    },
    {
      "epoch": 0.13912256641547313,
      "grad_norm": 101.53787231445312,
      "learning_rate": 8.61018818306494e-06,
      "loss": 0.0174,
      "step": 9840
    },
    {
      "epoch": 0.13926395113744008,
      "grad_norm": 9.474328544456512e-05,
      "learning_rate": 8.608774335845269e-06,
      "loss": 0.0001,
      "step": 9850
    },
    {
      "epoch": 0.13940533585940704,
      "grad_norm": 0.0,
      "learning_rate": 8.6073604886256e-06,
      "loss": 0.08,
      "step": 9860
    },
    {
      "epoch": 0.139546720581374,
      "grad_norm": 1.8411174096399918e-06,
      "learning_rate": 8.60594664140593e-06,
      "loss": 0.0001,
      "step": 9870
    },
    {
      "epoch": 0.13968810530334091,
      "grad_norm": 0.0,
      "learning_rate": 8.604532794186261e-06,
      "loss": 0.1274,
      "step": 9880
    },
    {
      "epoch": 0.13982949002530787,
      "grad_norm": 1.633659667277243e-05,
      "learning_rate": 8.60311894696659e-06,
      "loss": 0.0,
      "step": 9890
    },
    {
      "epoch": 0.13997087474727482,
      "grad_norm": 2.357857198376223e-08,
      "learning_rate": 8.601705099746921e-06,
      "loss": 0.0031,
      "step": 9900
    },
    {
      "epoch": 0.14011225946924175,
      "grad_norm": 4.62275218104935e-10,
      "learning_rate": 8.600291252527252e-06,
      "loss": 0.0,
      "step": 9910
    },
    {
      "epoch": 0.1402536441912087,
      "grad_norm": 7.185120921349153e-05,
      "learning_rate": 8.598877405307583e-06,
      "loss": 0.4777,
      "step": 9920
    },
    {
      "epoch": 0.14039502891317565,
      "grad_norm": 7.330666296212485e-09,
      "learning_rate": 8.597463558087914e-06,
      "loss": 0.0002,
      "step": 9930
    },
    {
      "epoch": 0.14053641363514258,
      "grad_norm": 0.00025888855452649295,
      "learning_rate": 8.596049710868243e-06,
      "loss": 0.0,
      "step": 9940
    },
    {
      "epoch": 0.14067779835710953,
      "grad_norm": 9.696777851786464e-06,
      "learning_rate": 8.594635863648576e-06,
      "loss": 0.0066,
      "step": 9950
    },
    {
      "epoch": 0.14081918307907648,
      "grad_norm": 0.5510509014129639,
      "learning_rate": 8.593222016428905e-06,
      "loss": 0.0001,
      "step": 9960
    },
    {
      "epoch": 0.1409605678010434,
      "grad_norm": 0.00021523560280911624,
      "learning_rate": 8.591808169209236e-06,
      "loss": 0.0,
      "step": 9970
    },
    {
      "epoch": 0.14110195252301036,
      "grad_norm": 0.0,
      "learning_rate": 8.590394321989567e-06,
      "loss": 0.0,
      "step": 9980
    },
    {
      "epoch": 0.1412433372449773,
      "grad_norm": 0.0,
      "learning_rate": 8.588980474769897e-06,
      "loss": 0.0004,
      "step": 9990
    },
    {
      "epoch": 0.14138472196694427,
      "grad_norm": 9.377777443475566e-10,
      "learning_rate": 8.587566627550228e-06,
      "loss": 0.0,
      "step": 10000
    },
    {
      "epoch": 0.1415261066889112,
      "grad_norm": 0.398966521024704,
      "learning_rate": 8.586152780330557e-06,
      "loss": 0.0932,
      "step": 10010
    },
    {
      "epoch": 0.14166749141087814,
      "grad_norm": 0.0,
      "learning_rate": 8.584738933110888e-06,
      "loss": 0.0081,
      "step": 10020
    },
    {
      "epoch": 0.1418088761328451,
      "grad_norm": 1.0152169105381859e-09,
      "learning_rate": 8.583325085891219e-06,
      "loss": 0.0003,
      "step": 10030
    },
    {
      "epoch": 0.14195026085481202,
      "grad_norm": 0.0005434475606307387,
      "learning_rate": 8.58191123867155e-06,
      "loss": 0.0,
      "step": 10040
    },
    {
      "epoch": 0.14209164557677897,
      "grad_norm": 0.008783343248069286,
      "learning_rate": 8.580497391451881e-06,
      "loss": 0.0003,
      "step": 10050
    },
    {
      "epoch": 0.14223303029874593,
      "grad_norm": 1.7802575826644897,
      "learning_rate": 8.57908354423221e-06,
      "loss": 0.013,
      "step": 10060
    },
    {
      "epoch": 0.14237441502071285,
      "grad_norm": 0.00012047711061313748,
      "learning_rate": 8.577669697012541e-06,
      "loss": 0.0,
      "step": 10070
    },
    {
      "epoch": 0.1425157997426798,
      "grad_norm": 3.574558320451615e-07,
      "learning_rate": 8.576255849792872e-06,
      "loss": 0.003,
      "step": 10080
    },
    {
      "epoch": 0.14265718446464676,
      "grad_norm": 0.0003613079898059368,
      "learning_rate": 8.574842002573203e-06,
      "loss": 0.0643,
      "step": 10090
    },
    {
      "epoch": 0.14279856918661368,
      "grad_norm": 0.00096643011784181,
      "learning_rate": 8.573428155353533e-06,
      "loss": 0.0,
      "step": 10100
    },
    {
      "epoch": 0.14293995390858064,
      "grad_norm": 0.0,
      "learning_rate": 8.572014308133864e-06,
      "loss": 0.0002,
      "step": 10110
    },
    {
      "epoch": 0.1430813386305476,
      "grad_norm": 0.0,
      "learning_rate": 8.570600460914195e-06,
      "loss": 0.0008,
      "step": 10120
    },
    {
      "epoch": 0.1432227233525145,
      "grad_norm": 0.0,
      "learning_rate": 8.569186613694524e-06,
      "loss": 0.0083,
      "step": 10130
    },
    {
      "epoch": 0.14336410807448147,
      "grad_norm": 0.0,
      "learning_rate": 8.567772766474855e-06,
      "loss": 0.0001,
      "step": 10140
    },
    {
      "epoch": 0.14350549279644842,
      "grad_norm": 0.0,
      "learning_rate": 8.566358919255186e-06,
      "loss": 0.0,
      "step": 10150
    },
    {
      "epoch": 0.14364687751841537,
      "grad_norm": 0.006315258797258139,
      "learning_rate": 8.564945072035517e-06,
      "loss": 0.0,
      "step": 10160
    },
    {
      "epoch": 0.1437882622403823,
      "grad_norm": 0.0005306534585542977,
      "learning_rate": 8.563531224815848e-06,
      "loss": 0.0,
      "step": 10170
    },
    {
      "epoch": 0.14392964696234925,
      "grad_norm": 4.786324114824936e-10,
      "learning_rate": 8.562117377596177e-06,
      "loss": 0.0036,
      "step": 10180
    },
    {
      "epoch": 0.1440710316843162,
      "grad_norm": 0.0,
      "learning_rate": 8.560703530376508e-06,
      "loss": 0.0,
      "step": 10190
    },
    {
      "epoch": 0.14421241640628313,
      "grad_norm": 1.065253911747277e-07,
      "learning_rate": 8.559289683156839e-06,
      "loss": 0.0003,
      "step": 10200
    },
    {
      "epoch": 0.14435380112825008,
      "grad_norm": 0.0,
      "learning_rate": 8.55787583593717e-06,
      "loss": 0.0,
      "step": 10210
    },
    {
      "epoch": 0.14449518585021703,
      "grad_norm": 0.00032365467632189393,
      "learning_rate": 8.556461988717499e-06,
      "loss": 0.0007,
      "step": 10220
    },
    {
      "epoch": 0.14463657057218396,
      "grad_norm": 0.1530914306640625,
      "learning_rate": 8.555048141497831e-06,
      "loss": 0.0,
      "step": 10230
    },
    {
      "epoch": 0.1447779552941509,
      "grad_norm": 0.0,
      "learning_rate": 8.553634294278162e-06,
      "loss": 0.0155,
      "step": 10240
    },
    {
      "epoch": 0.14491934001611786,
      "grad_norm": 395.3364562988281,
      "learning_rate": 8.552220447058491e-06,
      "loss": 0.165,
      "step": 10250
    },
    {
      "epoch": 0.1450607247380848,
      "grad_norm": 0.0,
      "learning_rate": 8.550806599838822e-06,
      "loss": 0.0032,
      "step": 10260
    },
    {
      "epoch": 0.14520210946005174,
      "grad_norm": 2.1677903987438185e-06,
      "learning_rate": 8.549392752619153e-06,
      "loss": 0.0272,
      "step": 10270
    },
    {
      "epoch": 0.1453434941820187,
      "grad_norm": 4.3858315734723874e-07,
      "learning_rate": 8.547978905399484e-06,
      "loss": 0.0,
      "step": 10280
    },
    {
      "epoch": 0.14548487890398565,
      "grad_norm": 0.0432489849627018,
      "learning_rate": 8.546565058179815e-06,
      "loss": 0.0001,
      "step": 10290
    },
    {
      "epoch": 0.14562626362595257,
      "grad_norm": 0.0,
      "learning_rate": 8.545151210960144e-06,
      "loss": 0.0,
      "step": 10300
    },
    {
      "epoch": 0.14576764834791953,
      "grad_norm": 14.852896690368652,
      "learning_rate": 8.543737363740475e-06,
      "loss": 0.0044,
      "step": 10310
    },
    {
      "epoch": 0.14590903306988648,
      "grad_norm": 0.0,
      "learning_rate": 8.542323516520806e-06,
      "loss": 0.0,
      "step": 10320
    },
    {
      "epoch": 0.1460504177918534,
      "grad_norm": 4.707149670046817e-10,
      "learning_rate": 8.540909669301136e-06,
      "loss": 0.0,
      "step": 10330
    },
    {
      "epoch": 0.14619180251382036,
      "grad_norm": 0.0,
      "learning_rate": 8.539495822081466e-06,
      "loss": 0.1792,
      "step": 10340
    },
    {
      "epoch": 0.1463331872357873,
      "grad_norm": 7.502579683205113e-05,
      "learning_rate": 8.538081974861796e-06,
      "loss": 0.0,
      "step": 10350
    },
    {
      "epoch": 0.14647457195775423,
      "grad_norm": 1.6078351109172218e-05,
      "learning_rate": 8.536668127642129e-06,
      "loss": 0.0002,
      "step": 10360
    },
    {
      "epoch": 0.1466159566797212,
      "grad_norm": 0.0,
      "learning_rate": 8.535254280422458e-06,
      "loss": 0.9297,
      "step": 10370
    },
    {
      "epoch": 0.14675734140168814,
      "grad_norm": 0.638262152671814,
      "learning_rate": 8.533840433202789e-06,
      "loss": 0.0001,
      "step": 10380
    },
    {
      "epoch": 0.14689872612365507,
      "grad_norm": 9.192243632938357e-10,
      "learning_rate": 8.53242658598312e-06,
      "loss": 0.0004,
      "step": 10390
    },
    {
      "epoch": 0.14704011084562202,
      "grad_norm": 0.0,
      "learning_rate": 8.53101273876345e-06,
      "loss": 0.0,
      "step": 10400
    },
    {
      "epoch": 0.14718149556758897,
      "grad_norm": 0.0,
      "learning_rate": 8.52959889154378e-06,
      "loss": 0.0005,
      "step": 10410
    },
    {
      "epoch": 0.14732288028955592,
      "grad_norm": 0.31660693883895874,
      "learning_rate": 8.52818504432411e-06,
      "loss": 0.0,
      "step": 10420
    },
    {
      "epoch": 0.14746426501152285,
      "grad_norm": 0.0,
      "learning_rate": 8.526771197104442e-06,
      "loss": 0.0006,
      "step": 10430
    },
    {
      "epoch": 0.1476056497334898,
      "grad_norm": 0.0,
      "learning_rate": 8.525357349884772e-06,
      "loss": 0.0,
      "step": 10440
    },
    {
      "epoch": 0.14774703445545675,
      "grad_norm": 1.7514629746528954e-07,
      "learning_rate": 8.523943502665103e-06,
      "loss": 0.0049,
      "step": 10450
    },
    {
      "epoch": 0.14788841917742368,
      "grad_norm": 0.0,
      "learning_rate": 8.522529655445432e-06,
      "loss": 0.0003,
      "step": 10460
    },
    {
      "epoch": 0.14802980389939063,
      "grad_norm": 0.0,
      "learning_rate": 8.521115808225763e-06,
      "loss": 0.0,
      "step": 10470
    },
    {
      "epoch": 0.14817118862135759,
      "grad_norm": 0.0,
      "learning_rate": 8.519701961006094e-06,
      "loss": 0.0,
      "step": 10480
    },
    {
      "epoch": 0.1483125733433245,
      "grad_norm": 4.982965151612007e-10,
      "learning_rate": 8.518288113786425e-06,
      "loss": 0.0002,
      "step": 10490
    },
    {
      "epoch": 0.14845395806529146,
      "grad_norm": 0.0,
      "learning_rate": 8.516874266566756e-06,
      "loss": 0.0,
      "step": 10500
    },
    {
      "epoch": 0.14859534278725842,
      "grad_norm": 412.45050048828125,
      "learning_rate": 8.515460419347085e-06,
      "loss": 0.3833,
      "step": 10510
    },
    {
      "epoch": 0.14873672750922534,
      "grad_norm": 1.5868599803070538e-06,
      "learning_rate": 8.514046572127418e-06,
      "loss": 0.4449,
      "step": 10520
    },
    {
      "epoch": 0.1488781122311923,
      "grad_norm": 0.1666027158498764,
      "learning_rate": 8.512632724907747e-06,
      "loss": 0.0,
      "step": 10530
    },
    {
      "epoch": 0.14901949695315925,
      "grad_norm": 54.533721923828125,
      "learning_rate": 8.511218877688078e-06,
      "loss": 0.0137,
      "step": 10540
    },
    {
      "epoch": 0.14916088167512617,
      "grad_norm": 0.0,
      "learning_rate": 8.509805030468408e-06,
      "loss": 0.0032,
      "step": 10550
    },
    {
      "epoch": 0.14930226639709313,
      "grad_norm": 5.962840987194795e-06,
      "learning_rate": 8.50839118324874e-06,
      "loss": 0.0434,
      "step": 10560
    },
    {
      "epoch": 0.14944365111906008,
      "grad_norm": 0.0,
      "learning_rate": 8.50697733602907e-06,
      "loss": 0.0,
      "step": 10570
    },
    {
      "epoch": 0.14958503584102703,
      "grad_norm": 7.437919703079388e-05,
      "learning_rate": 8.5055634888094e-06,
      "loss": 0.1315,
      "step": 10580
    },
    {
      "epoch": 0.14972642056299396,
      "grad_norm": 3.988351693351433e-08,
      "learning_rate": 8.50414964158973e-06,
      "loss": 0.0301,
      "step": 10590
    },
    {
      "epoch": 0.1498678052849609,
      "grad_norm": 6.421005878110009e-07,
      "learning_rate": 8.502735794370061e-06,
      "loss": 0.0,
      "step": 10600
    },
    {
      "epoch": 0.15000919000692786,
      "grad_norm": 0.0,
      "learning_rate": 8.501321947150392e-06,
      "loss": 0.0,
      "step": 10610
    },
    {
      "epoch": 0.1501505747288948,
      "grad_norm": 0.0,
      "learning_rate": 8.499908099930723e-06,
      "loss": 0.0,
      "step": 10620
    },
    {
      "epoch": 0.15029195945086174,
      "grad_norm": 0.0,
      "learning_rate": 8.498494252711052e-06,
      "loss": 0.0106,
      "step": 10630
    },
    {
      "epoch": 0.1504333441728287,
      "grad_norm": 0.0,
      "learning_rate": 8.497080405491383e-06,
      "loss": 0.0007,
      "step": 10640
    },
    {
      "epoch": 0.15057472889479562,
      "grad_norm": 0.0,
      "learning_rate": 8.495666558271714e-06,
      "loss": 0.0137,
      "step": 10650
    },
    {
      "epoch": 0.15071611361676257,
      "grad_norm": 0.0,
      "learning_rate": 8.494252711052044e-06,
      "loss": 0.9727,
      "step": 10660
    },
    {
      "epoch": 0.15085749833872952,
      "grad_norm": 4.914744722306352e-10,
      "learning_rate": 8.492838863832374e-06,
      "loss": 0.0,
      "step": 10670
    },
    {
      "epoch": 0.15099888306069645,
      "grad_norm": 0.0,
      "learning_rate": 8.491425016612706e-06,
      "loss": 0.0,
      "step": 10680
    },
    {
      "epoch": 0.1511402677826634,
      "grad_norm": 9.856275795527836e-10,
      "learning_rate": 8.490011169393037e-06,
      "loss": 0.0,
      "step": 10690
    },
    {
      "epoch": 0.15128165250463035,
      "grad_norm": 1.3460464288073126e-05,
      "learning_rate": 8.488597322173366e-06,
      "loss": 0.0,
      "step": 10700
    },
    {
      "epoch": 0.1514230372265973,
      "grad_norm": 0.0,
      "learning_rate": 8.487183474953697e-06,
      "loss": 0.2209,
      "step": 10710
    },
    {
      "epoch": 0.15156442194856423,
      "grad_norm": 0.0,
      "learning_rate": 8.485769627734028e-06,
      "loss": 0.0,
      "step": 10720
    },
    {
      "epoch": 0.15170580667053118,
      "grad_norm": 0.00025110342539846897,
      "learning_rate": 8.484355780514359e-06,
      "loss": 0.0486,
      "step": 10730
    },
    {
      "epoch": 0.15184719139249814,
      "grad_norm": 0.0004598846717271954,
      "learning_rate": 8.48294193329469e-06,
      "loss": 0.0002,
      "step": 10740
    },
    {
      "epoch": 0.15198857611446506,
      "grad_norm": 0.00042419761302880943,
      "learning_rate": 8.481528086075019e-06,
      "loss": 0.159,
      "step": 10750
    },
    {
      "epoch": 0.15212996083643202,
      "grad_norm": 0.0033416764345020056,
      "learning_rate": 8.48011423885535e-06,
      "loss": 0.0,
      "step": 10760
    },
    {
      "epoch": 0.15227134555839897,
      "grad_norm": 0.0,
      "learning_rate": 8.47870039163568e-06,
      "loss": 0.0,
      "step": 10770
    },
    {
      "epoch": 0.1524127302803659,
      "grad_norm": 0.28257912397384644,
      "learning_rate": 8.477286544416011e-06,
      "loss": 0.0001,
      "step": 10780
    },
    {
      "epoch": 0.15255411500233285,
      "grad_norm": 0.0,
      "learning_rate": 8.47587269719634e-06,
      "loss": 0.0335,
      "step": 10790
    },
    {
      "epoch": 0.1526954997242998,
      "grad_norm": 4.759774796525562e-10,
      "learning_rate": 8.474458849976673e-06,
      "loss": 0.0,
      "step": 10800
    },
    {
      "epoch": 0.15283688444626672,
      "grad_norm": 4.825278665521182e-07,
      "learning_rate": 8.473045002757004e-06,
      "loss": 0.8547,
      "step": 10810
    },
    {
      "epoch": 0.15297826916823368,
      "grad_norm": 3.7978688283146766e-07,
      "learning_rate": 8.471631155537333e-06,
      "loss": 0.0001,
      "step": 10820
    },
    {
      "epoch": 0.15311965389020063,
      "grad_norm": 0.0006714282208122313,
      "learning_rate": 8.470217308317664e-06,
      "loss": 0.0,
      "step": 10830
    },
    {
      "epoch": 0.15326103861216758,
      "grad_norm": 2.586531699932948e-08,
      "learning_rate": 8.468803461097995e-06,
      "loss": 0.0808,
      "step": 10840
    },
    {
      "epoch": 0.1534024233341345,
      "grad_norm": 0.0,
      "learning_rate": 8.467389613878326e-06,
      "loss": 0.0,
      "step": 10850
    },
    {
      "epoch": 0.15354380805610146,
      "grad_norm": 9.765932507121988e-10,
      "learning_rate": 8.465975766658655e-06,
      "loss": 0.0,
      "step": 10860
    },
    {
      "epoch": 0.1536851927780684,
      "grad_norm": 0.005827141460031271,
      "learning_rate": 8.464561919438986e-06,
      "loss": 0.0,
      "step": 10870
    },
    {
      "epoch": 0.15382657750003534,
      "grad_norm": 0.0,
      "learning_rate": 8.463148072219317e-06,
      "loss": 0.0057,
      "step": 10880
    },
    {
      "epoch": 0.1539679622220023,
      "grad_norm": 0.0006114965653978288,
      "learning_rate": 8.461734224999647e-06,
      "loss": 0.0,
      "step": 10890
    },
    {
      "epoch": 0.15410934694396924,
      "grad_norm": 0.004510736558586359,
      "learning_rate": 8.460320377779978e-06,
      "loss": 0.0,
      "step": 10900
    },
    {
      "epoch": 0.15425073166593617,
      "grad_norm": 0.001353011466562748,
      "learning_rate": 8.458906530560307e-06,
      "loss": 0.0,
      "step": 10910
    },
    {
      "epoch": 0.15439211638790312,
      "grad_norm": 3.44885449976573e-07,
      "learning_rate": 8.457492683340638e-06,
      "loss": 0.0009,
      "step": 10920
    },
    {
      "epoch": 0.15453350110987008,
      "grad_norm": 0.0,
      "learning_rate": 8.456078836120969e-06,
      "loss": 0.0002,
      "step": 10930
    },
    {
      "epoch": 0.154674885831837,
      "grad_norm": 0.033334385603666306,
      "learning_rate": 8.4546649889013e-06,
      "loss": 0.0001,
      "step": 10940
    },
    {
      "epoch": 0.15481627055380395,
      "grad_norm": 0.5134636759757996,
      "learning_rate": 8.45325114168163e-06,
      "loss": 0.0002,
      "step": 10950
    },
    {
      "epoch": 0.1549576552757709,
      "grad_norm": 9.370471065750507e-10,
      "learning_rate": 8.451837294461962e-06,
      "loss": 0.0,
      "step": 10960
    },
    {
      "epoch": 0.15509903999773783,
      "grad_norm": 5.51849019814199e-08,
      "learning_rate": 8.450423447242293e-06,
      "loss": 0.0994,
      "step": 10970
    },
    {
      "epoch": 0.15524042471970478,
      "grad_norm": 0.0,
      "learning_rate": 8.449009600022622e-06,
      "loss": 0.0,
      "step": 10980
    },
    {
      "epoch": 0.15538180944167174,
      "grad_norm": 5.385241141908637e-10,
      "learning_rate": 8.447595752802953e-06,
      "loss": 0.0,
      "step": 10990
    },
    {
      "epoch": 0.1555231941636387,
      "grad_norm": 0.000382145750336349,
      "learning_rate": 8.446181905583283e-06,
      "loss": 0.0,
      "step": 11000
    },
    {
      "epoch": 0.15566457888560561,
      "grad_norm": 0.0,
      "learning_rate": 8.444768058363614e-06,
      "loss": 0.0,
      "step": 11010
    },
    {
      "epoch": 0.15580596360757257,
      "grad_norm": 0.0,
      "learning_rate": 8.443354211143945e-06,
      "loss": 0.0026,
      "step": 11020
    },
    {
      "epoch": 0.15594734832953952,
      "grad_norm": 0.0,
      "learning_rate": 8.441940363924274e-06,
      "loss": 0.0023,
      "step": 11030
    },
    {
      "epoch": 0.15608873305150645,
      "grad_norm": 0.0008343369117937982,
      "learning_rate": 8.440526516704605e-06,
      "loss": 0.0099,
      "step": 11040
    },
    {
      "epoch": 0.1562301177734734,
      "grad_norm": 0.0,
      "learning_rate": 8.439112669484936e-06,
      "loss": 0.0436,
      "step": 11050
    },
    {
      "epoch": 0.15637150249544035,
      "grad_norm": 0.0,
      "learning_rate": 8.437698822265267e-06,
      "loss": 0.0004,
      "step": 11060
    },
    {
      "epoch": 0.15651288721740728,
      "grad_norm": 9.619511853742324e-10,
      "learning_rate": 8.436284975045598e-06,
      "loss": 0.0342,
      "step": 11070
    },
    {
      "epoch": 0.15665427193937423,
      "grad_norm": 1.5925898551940918,
      "learning_rate": 8.434871127825927e-06,
      "loss": 0.0003,
      "step": 11080
    },
    {
      "epoch": 0.15679565666134118,
      "grad_norm": 0.0,
      "learning_rate": 8.43345728060626e-06,
      "loss": 0.1043,
      "step": 11090
    },
    {
      "epoch": 0.1569370413833081,
      "grad_norm": 0.0,
      "learning_rate": 8.432043433386589e-06,
      "loss": 0.0,
      "step": 11100
    },
    {
      "epoch": 0.15707842610527506,
      "grad_norm": 4.4192387349539786e-07,
      "learning_rate": 8.43062958616692e-06,
      "loss": 0.0,
      "step": 11110
    },
    {
      "epoch": 0.157219810827242,
      "grad_norm": 0.0,
      "learning_rate": 8.42921573894725e-06,
      "loss": 0.0001,
      "step": 11120
    },
    {
      "epoch": 0.15736119554920897,
      "grad_norm": 0.0,
      "learning_rate": 8.427801891727581e-06,
      "loss": 0.0,
      "step": 11130
    },
    {
      "epoch": 0.1575025802711759,
      "grad_norm": 1.4114119039732032e-05,
      "learning_rate": 8.426388044507912e-06,
      "loss": 0.0002,
      "step": 11140
    },
    {
      "epoch": 0.15764396499314284,
      "grad_norm": 0.0,
      "learning_rate": 8.424974197288241e-06,
      "loss": 0.0,
      "step": 11150
    },
    {
      "epoch": 0.1577853497151098,
      "grad_norm": 0.0,
      "learning_rate": 8.423560350068572e-06,
      "loss": 0.0,
      "step": 11160
    },
    {
      "epoch": 0.15792673443707672,
      "grad_norm": 0.00039057974936440587,
      "learning_rate": 8.422146502848903e-06,
      "loss": 0.0794,
      "step": 11170
    },
    {
      "epoch": 0.15806811915904367,
      "grad_norm": 0.0,
      "learning_rate": 8.420732655629234e-06,
      "loss": 0.0,
      "step": 11180
    },
    {
      "epoch": 0.15820950388101063,
      "grad_norm": 0.0,
      "learning_rate": 8.419318808409565e-06,
      "loss": 0.0003,
      "step": 11190
    },
    {
      "epoch": 0.15835088860297755,
      "grad_norm": 0.0,
      "learning_rate": 8.417904961189894e-06,
      "loss": 0.0,
      "step": 11200
    },
    {
      "epoch": 0.1584922733249445,
      "grad_norm": 0.0,
      "learning_rate": 8.416491113970225e-06,
      "loss": 0.0558,
      "step": 11210
    },
    {
      "epoch": 0.15863365804691146,
      "grad_norm": 0.0,
      "learning_rate": 8.415077266750555e-06,
      "loss": 0.0,
      "step": 11220
    },
    {
      "epoch": 0.15877504276887838,
      "grad_norm": 0.0,
      "learning_rate": 8.413663419530886e-06,
      "loss": 0.0,
      "step": 11230
    },
    {
      "epoch": 0.15891642749084534,
      "grad_norm": 3.6504339551157727e-09,
      "learning_rate": 8.412249572311216e-06,
      "loss": 0.0188,
      "step": 11240
    },
    {
      "epoch": 0.1590578122128123,
      "grad_norm": 0.0,
      "learning_rate": 8.410835725091548e-06,
      "loss": 0.0,
      "step": 11250
    },
    {
      "epoch": 0.15919919693477924,
      "grad_norm": 0.0007553758914582431,
      "learning_rate": 8.409421877871879e-06,
      "loss": 0.0,
      "step": 11260
    },
    {
      "epoch": 0.15934058165674617,
      "grad_norm": 0.0007607428124174476,
      "learning_rate": 8.408008030652208e-06,
      "loss": 0.0,
      "step": 11270
    },
    {
      "epoch": 0.15948196637871312,
      "grad_norm": 3.187696165696252e-06,
      "learning_rate": 8.406594183432539e-06,
      "loss": 0.0216,
      "step": 11280
    },
    {
      "epoch": 0.15962335110068007,
      "grad_norm": 3.44413365382934e-06,
      "learning_rate": 8.40518033621287e-06,
      "loss": 0.0,
      "step": 11290
    },
    {
      "epoch": 0.159764735822647,
      "grad_norm": 0.00020598404807969928,
      "learning_rate": 8.4037664889932e-06,
      "loss": 0.0005,
      "step": 11300
    },
    {
      "epoch": 0.15990612054461395,
      "grad_norm": 1.8960994482040405,
      "learning_rate": 8.40235264177353e-06,
      "loss": 0.0006,
      "step": 11310
    },
    {
      "epoch": 0.1600475052665809,
      "grad_norm": 3.8525588024640456e-05,
      "learning_rate": 8.40093879455386e-06,
      "loss": 0.0,
      "step": 11320
    },
    {
      "epoch": 0.16018888998854783,
      "grad_norm": 1.0901359928539023e-05,
      "learning_rate": 8.399524947334192e-06,
      "loss": 0.0,
      "step": 11330
    },
    {
      "epoch": 0.16033027471051478,
      "grad_norm": 0.00036864320281893015,
      "learning_rate": 8.398111100114522e-06,
      "loss": 0.0,
      "step": 11340
    },
    {
      "epoch": 0.16047165943248173,
      "grad_norm": 4.317847015045118e-06,
      "learning_rate": 8.396697252894853e-06,
      "loss": 0.0,
      "step": 11350
    },
    {
      "epoch": 0.16061304415444866,
      "grad_norm": 0.0,
      "learning_rate": 8.395283405675182e-06,
      "loss": 0.0075,
      "step": 11360
    },
    {
      "epoch": 0.1607544288764156,
      "grad_norm": 9.381487808823863e-10,
      "learning_rate": 8.39401094317748e-06,
      "loss": 0.6693,
      "step": 11370
    },
    {
      "epoch": 0.16089581359838256,
      "grad_norm": 9.580465309966257e-10,
      "learning_rate": 8.392597095957811e-06,
      "loss": 0.0,
      "step": 11380
    },
    {
      "epoch": 0.1610371983203495,
      "grad_norm": 0.0,
      "learning_rate": 8.391183248738142e-06,
      "loss": 0.0,
      "step": 11390
    },
    {
      "epoch": 0.16117858304231644,
      "grad_norm": 3.580122864832447e-09,
      "learning_rate": 8.389769401518473e-06,
      "loss": 0.0,
      "step": 11400
    },
    {
      "epoch": 0.1613199677642834,
      "grad_norm": 0.010098010301589966,
      "learning_rate": 8.388355554298802e-06,
      "loss": 0.0468,
      "step": 11410
    },
    {
      "epoch": 0.16146135248625035,
      "grad_norm": 7.172251571319066e-06,
      "learning_rate": 8.386941707079133e-06,
      "loss": 0.0,
      "step": 11420
    },
    {
      "epoch": 0.16160273720821727,
      "grad_norm": 3.7942667404422536e-05,
      "learning_rate": 8.385527859859464e-06,
      "loss": 0.0,
      "step": 11430
    },
    {
      "epoch": 0.16174412193018423,
      "grad_norm": 0.0,
      "learning_rate": 8.384114012639795e-06,
      "loss": 0.0007,
      "step": 11440
    },
    {
      "epoch": 0.16188550665215118,
      "grad_norm": 0.0,
      "learning_rate": 8.382700165420126e-06,
      "loss": 0.0,
      "step": 11450
    },
    {
      "epoch": 0.1620268913741181,
      "grad_norm": 0.0,
      "learning_rate": 8.381286318200455e-06,
      "loss": 0.0,
      "step": 11460
    },
    {
      "epoch": 0.16216827609608506,
      "grad_norm": 0.0,
      "learning_rate": 8.379872470980787e-06,
      "loss": 0.0809,
      "step": 11470
    },
    {
      "epoch": 0.162309660818052,
      "grad_norm": 9.59616719420353e-10,
      "learning_rate": 8.378458623761117e-06,
      "loss": 0.0,
      "step": 11480
    },
    {
      "epoch": 0.16245104554001893,
      "grad_norm": 2.1895549297332764,
      "learning_rate": 8.377044776541447e-06,
      "loss": 0.0262,
      "step": 11490
    },
    {
      "epoch": 0.1625924302619859,
      "grad_norm": 0.0,
      "learning_rate": 8.375630929321778e-06,
      "loss": 0.027,
      "step": 11500
    },
    {
      "epoch": 0.16273381498395284,
      "grad_norm": 0.0,
      "learning_rate": 8.374217082102109e-06,
      "loss": 0.0,
      "step": 11510
    },
    {
      "epoch": 0.16287519970591977,
      "grad_norm": 1.3461744856613223e-06,
      "learning_rate": 8.37280323488244e-06,
      "loss": 0.0,
      "step": 11520
    },
    {
      "epoch": 0.16301658442788672,
      "grad_norm": 0.00037584290839731693,
      "learning_rate": 8.371389387662769e-06,
      "loss": 0.0,
      "step": 11530
    },
    {
      "epoch": 0.16315796914985367,
      "grad_norm": 0.15481694042682648,
      "learning_rate": 8.3699755404431e-06,
      "loss": 0.0,
      "step": 11540
    },
    {
      "epoch": 0.16329935387182062,
      "grad_norm": 4.899234772892669e-05,
      "learning_rate": 8.368561693223431e-06,
      "loss": 0.0001,
      "step": 11550
    },
    {
      "epoch": 0.16344073859378755,
      "grad_norm": 0.0,
      "learning_rate": 8.367147846003762e-06,
      "loss": 0.0,
      "step": 11560
    },
    {
      "epoch": 0.1635821233157545,
      "grad_norm": 0.0,
      "learning_rate": 8.365733998784093e-06,
      "loss": 0.0,
      "step": 11570
    },
    {
      "epoch": 0.16372350803772145,
      "grad_norm": 1.0337376510349827e-09,
      "learning_rate": 8.364320151564422e-06,
      "loss": 0.0,
      "step": 11580
    },
    {
      "epoch": 0.16386489275968838,
      "grad_norm": 0.0,
      "learning_rate": 8.362906304344753e-06,
      "loss": 0.0004,
      "step": 11590
    },
    {
      "epoch": 0.16400627748165533,
      "grad_norm": 0.0,
      "learning_rate": 8.361492457125083e-06,
      "loss": 0.0,
      "step": 11600
    },
    {
      "epoch": 0.16414766220362229,
      "grad_norm": 3.620204370236024e-05,
      "learning_rate": 8.360078609905414e-06,
      "loss": 0.0,
      "step": 11610
    },
    {
      "epoch": 0.1642890469255892,
      "grad_norm": 0.004926975816488266,
      "learning_rate": 8.358664762685745e-06,
      "loss": 0.0038,
      "step": 11620
    },
    {
      "epoch": 0.16443043164755616,
      "grad_norm": 9.350727969703598e-10,
      "learning_rate": 8.357250915466076e-06,
      "loss": 0.0,
      "step": 11630
    },
    {
      "epoch": 0.16457181636952312,
      "grad_norm": 0.0,
      "learning_rate": 8.355837068246407e-06,
      "loss": 0.0,
      "step": 11640
    },
    {
      "epoch": 0.16471320109149004,
      "grad_norm": 0.0,
      "learning_rate": 8.354423221026736e-06,
      "loss": 0.012,
      "step": 11650
    },
    {
      "epoch": 0.164854585813457,
      "grad_norm": 1.117266151595686e-06,
      "learning_rate": 8.353009373807067e-06,
      "loss": 0.0,
      "step": 11660
    },
    {
      "epoch": 0.16499597053542395,
      "grad_norm": 0.0,
      "learning_rate": 8.351595526587398e-06,
      "loss": 0.0016,
      "step": 11670
    },
    {
      "epoch": 0.1651373552573909,
      "grad_norm": 0.0,
      "learning_rate": 8.350181679367729e-06,
      "loss": 0.0054,
      "step": 11680
    },
    {
      "epoch": 0.16527873997935782,
      "grad_norm": 0.0,
      "learning_rate": 8.34876783214806e-06,
      "loss": 0.7008,
      "step": 11690
    },
    {
      "epoch": 0.16542012470132478,
      "grad_norm": 0.0,
      "learning_rate": 8.347353984928389e-06,
      "loss": 0.0,
      "step": 11700
    },
    {
      "epoch": 0.16556150942329173,
      "grad_norm": 0.03117610514163971,
      "learning_rate": 8.34594013770872e-06,
      "loss": 0.0,
      "step": 11710
    },
    {
      "epoch": 0.16570289414525866,
      "grad_norm": 0.0,
      "learning_rate": 8.34452629048905e-06,
      "loss": 0.0,
      "step": 11720
    },
    {
      "epoch": 0.1658442788672256,
      "grad_norm": 0.0,
      "learning_rate": 8.343112443269381e-06,
      "loss": 0.0419,
      "step": 11730
    },
    {
      "epoch": 0.16598566358919256,
      "grad_norm": 2.2350726794684306e-05,
      "learning_rate": 8.341698596049712e-06,
      "loss": 0.0,
      "step": 11740
    },
    {
      "epoch": 0.1661270483111595,
      "grad_norm": 0.0,
      "learning_rate": 8.340284748830041e-06,
      "loss": 0.0,
      "step": 11750
    },
    {
      "epoch": 0.16626843303312644,
      "grad_norm": 0.0,
      "learning_rate": 8.338870901610374e-06,
      "loss": 0.0001,
      "step": 11760
    },
    {
      "epoch": 0.1664098177550934,
      "grad_norm": 8.735386813896184e-08,
      "learning_rate": 8.337457054390703e-06,
      "loss": 0.0,
      "step": 11770
    },
    {
      "epoch": 0.16655120247706032,
      "grad_norm": 0.0,
      "learning_rate": 8.336043207171034e-06,
      "loss": 0.0,
      "step": 11780
    },
    {
      "epoch": 0.16669258719902727,
      "grad_norm": 0.0,
      "learning_rate": 8.334629359951365e-06,
      "loss": 0.0003,
      "step": 11790
    },
    {
      "epoch": 0.16683397192099422,
      "grad_norm": 0.0,
      "learning_rate": 8.333215512731695e-06,
      "loss": 0.0,
      "step": 11800
    },
    {
      "epoch": 0.16697535664296115,
      "grad_norm": 2.0173556549707428e-05,
      "learning_rate": 8.331801665512026e-06,
      "loss": 0.0001,
      "step": 11810
    },
    {
      "epoch": 0.1671167413649281,
      "grad_norm": 1.4995960555097554e-05,
      "learning_rate": 8.330387818292356e-06,
      "loss": 0.0,
      "step": 11820
    },
    {
      "epoch": 0.16725812608689505,
      "grad_norm": 7.425883086398244e-05,
      "learning_rate": 8.328973971072686e-06,
      "loss": 0.0,
      "step": 11830
    },
    {
      "epoch": 0.167399510808862,
      "grad_norm": 0.000952282571233809,
      "learning_rate": 8.327560123853017e-06,
      "loss": 0.0001,
      "step": 11840
    },
    {
      "epoch": 0.16754089553082893,
      "grad_norm": 0.0,
      "learning_rate": 8.326146276633348e-06,
      "loss": 0.0189,
      "step": 11850
    },
    {
      "epoch": 0.16768228025279588,
      "grad_norm": 0.0,
      "learning_rate": 8.324732429413677e-06,
      "loss": 0.0,
      "step": 11860
    },
    {
      "epoch": 0.16782366497476284,
      "grad_norm": 0.0,
      "learning_rate": 8.323318582194008e-06,
      "loss": 0.0,
      "step": 11870
    },
    {
      "epoch": 0.16796504969672976,
      "grad_norm": 17.79913330078125,
      "learning_rate": 8.32190473497434e-06,
      "loss": 0.0023,
      "step": 11880
    },
    {
      "epoch": 0.16810643441869672,
      "grad_norm": 0.0,
      "learning_rate": 8.32049088775467e-06,
      "loss": 0.0395,
      "step": 11890
    },
    {
      "epoch": 0.16824781914066367,
      "grad_norm": 4.44397336707425e-08,
      "learning_rate": 8.319077040535e-06,
      "loss": 0.0041,
      "step": 11900
    },
    {
      "epoch": 0.1683892038626306,
      "grad_norm": 0.0134517140686512,
      "learning_rate": 8.317663193315332e-06,
      "loss": 0.0,
      "step": 11910
    },
    {
      "epoch": 0.16853058858459755,
      "grad_norm": 0.0,
      "learning_rate": 8.316249346095662e-06,
      "loss": 0.0001,
      "step": 11920
    },
    {
      "epoch": 0.1686719733065645,
      "grad_norm": 8.969185394391843e-10,
      "learning_rate": 8.314835498875992e-06,
      "loss": 0.0141,
      "step": 11930
    },
    {
      "epoch": 0.16881335802853142,
      "grad_norm": 2.9253122193040326e-05,
      "learning_rate": 8.313421651656322e-06,
      "loss": 0.0,
      "step": 11940
    },
    {
      "epoch": 0.16895474275049838,
      "grad_norm": 9.430356495698788e-10,
      "learning_rate": 8.312007804436653e-06,
      "loss": 0.0,
      "step": 11950
    },
    {
      "epoch": 0.16909612747246533,
      "grad_norm": 0.25099024176597595,
      "learning_rate": 8.310593957216984e-06,
      "loss": 0.0001,
      "step": 11960
    },
    {
      "epoch": 0.16923751219443228,
      "grad_norm": 0.0,
      "learning_rate": 8.309180109997315e-06,
      "loss": 0.0497,
      "step": 11970
    },
    {
      "epoch": 0.1693788969163992,
      "grad_norm": 0.02599899098277092,
      "learning_rate": 8.307766262777644e-06,
      "loss": 0.0,
      "step": 11980
    },
    {
      "epoch": 0.16952028163836616,
      "grad_norm": 8.987071642430067e-10,
      "learning_rate": 8.306352415557975e-06,
      "loss": 0.001,
      "step": 11990
    },
    {
      "epoch": 0.1696616663603331,
      "grad_norm": 0.0,
      "learning_rate": 8.304938568338306e-06,
      "loss": 0.1201,
      "step": 12000
    },
    {
      "epoch": 0.16980305108230004,
      "grad_norm": 1.7693642803351395e-05,
      "learning_rate": 8.303524721118637e-06,
      "loss": 0.0,
      "step": 12010
    },
    {
      "epoch": 0.169944435804267,
      "grad_norm": 2.0109357833862305,
      "learning_rate": 8.302110873898968e-06,
      "loss": 0.0027,
      "step": 12020
    },
    {
      "epoch": 0.17008582052623394,
      "grad_norm": 0.0,
      "learning_rate": 8.300697026679297e-06,
      "loss": 0.0,
      "step": 12030
    },
    {
      "epoch": 0.17022720524820087,
      "grad_norm": 0.0,
      "learning_rate": 8.29928317945963e-06,
      "loss": 0.0,
      "step": 12040
    },
    {
      "epoch": 0.17036858997016782,
      "grad_norm": 0.0,
      "learning_rate": 8.297869332239958e-06,
      "loss": 0.0,
      "step": 12050
    },
    {
      "epoch": 0.17050997469213477,
      "grad_norm": 0.028291864320635796,
      "learning_rate": 8.29645548502029e-06,
      "loss": 0.0007,
      "step": 12060
    },
    {
      "epoch": 0.1706513594141017,
      "grad_norm": 7.587967729705269e-07,
      "learning_rate": 8.29504163780062e-06,
      "loss": 0.0,
      "step": 12070
    },
    {
      "epoch": 0.17079274413606865,
      "grad_norm": 0.0,
      "learning_rate": 8.293627790580951e-06,
      "loss": 0.0,
      "step": 12080
    },
    {
      "epoch": 0.1709341288580356,
      "grad_norm": 0.0,
      "learning_rate": 8.292213943361282e-06,
      "loss": 0.0,
      "step": 12090
    },
    {
      "epoch": 0.17107551358000256,
      "grad_norm": 0.0,
      "learning_rate": 8.290800096141611e-06,
      "loss": 0.0001,
      "step": 12100
    },
    {
      "epoch": 0.17121689830196948,
      "grad_norm": 7.696965553805057e-08,
      "learning_rate": 8.289386248921942e-06,
      "loss": 2.1078,
      "step": 12110
    },
    {
      "epoch": 0.17135828302393644,
      "grad_norm": 0.0,
      "learning_rate": 8.287972401702273e-06,
      "loss": 0.3727,
      "step": 12120
    },
    {
      "epoch": 0.1714996677459034,
      "grad_norm": 0.0,
      "learning_rate": 8.286558554482604e-06,
      "loss": 0.0,
      "step": 12130
    },
    {
      "epoch": 0.17164105246787031,
      "grad_norm": 9.437144399271347e-10,
      "learning_rate": 8.285144707262934e-06,
      "loss": 0.088,
      "step": 12140
    },
    {
      "epoch": 0.17178243718983727,
      "grad_norm": 0.0,
      "learning_rate": 8.283730860043264e-06,
      "loss": 0.0,
      "step": 12150
    },
    {
      "epoch": 0.17192382191180422,
      "grad_norm": 0.0,
      "learning_rate": 8.282317012823594e-06,
      "loss": 0.0,
      "step": 12160
    },
    {
      "epoch": 0.17206520663377114,
      "grad_norm": 0.0,
      "learning_rate": 8.280903165603925e-06,
      "loss": 0.0,
      "step": 12170
    },
    {
      "epoch": 0.1722065913557381,
      "grad_norm": 0.0,
      "learning_rate": 8.279489318384256e-06,
      "loss": 0.0,
      "step": 12180
    },
    {
      "epoch": 0.17234797607770505,
      "grad_norm": 0.0023877648636698723,
      "learning_rate": 8.278075471164585e-06,
      "loss": 0.0,
      "step": 12190
    },
    {
      "epoch": 0.17248936079967198,
      "grad_norm": 4.662229184759781e-05,
      "learning_rate": 8.276661623944918e-06,
      "loss": 0.0,
      "step": 12200
    },
    {
      "epoch": 0.17263074552163893,
      "grad_norm": 0.0,
      "learning_rate": 8.275247776725249e-06,
      "loss": 0.0913,
      "step": 12210
    },
    {
      "epoch": 0.17277213024360588,
      "grad_norm": 0.09373795241117477,
      "learning_rate": 8.273833929505578e-06,
      "loss": 0.0,
      "step": 12220
    },
    {
      "epoch": 0.1729135149655728,
      "grad_norm": 0.0,
      "learning_rate": 8.272420082285909e-06,
      "loss": 0.0454,
      "step": 12230
    },
    {
      "epoch": 0.17305489968753976,
      "grad_norm": 0.0,
      "learning_rate": 8.27100623506624e-06,
      "loss": 0.0,
      "step": 12240
    },
    {
      "epoch": 0.1731962844095067,
      "grad_norm": 7.563905091956258e-05,
      "learning_rate": 8.26959238784657e-06,
      "loss": 0.0,
      "step": 12250
    },
    {
      "epoch": 0.17333766913147367,
      "grad_norm": 1.4082638699619565e-06,
      "learning_rate": 8.268178540626901e-06,
      "loss": 0.0014,
      "step": 12260
    },
    {
      "epoch": 0.1734790538534406,
      "grad_norm": 0.0,
      "learning_rate": 8.26676469340723e-06,
      "loss": 0.0,
      "step": 12270
    },
    {
      "epoch": 0.17362043857540754,
      "grad_norm": 0.00018624673248268664,
      "learning_rate": 8.265350846187561e-06,
      "loss": 0.0,
      "step": 12280
    },
    {
      "epoch": 0.1737618232973745,
      "grad_norm": 9.601203032616468e-08,
      "learning_rate": 8.263936998967892e-06,
      "loss": 0.0001,
      "step": 12290
    },
    {
      "epoch": 0.17390320801934142,
      "grad_norm": 0.0,
      "learning_rate": 8.262523151748223e-06,
      "loss": 0.0,
      "step": 12300
    },
    {
      "epoch": 0.17404459274130837,
      "grad_norm": 0.6738077998161316,
      "learning_rate": 8.261109304528552e-06,
      "loss": 0.0074,
      "step": 12310
    },
    {
      "epoch": 0.17418597746327533,
      "grad_norm": 1.2543964658107143e-06,
      "learning_rate": 8.259695457308883e-06,
      "loss": 0.1208,
      "step": 12320
    },
    {
      "epoch": 0.17432736218524225,
      "grad_norm": 4.943180087479959e-09,
      "learning_rate": 8.258281610089216e-06,
      "loss": 0.0,
      "step": 12330
    },
    {
      "epoch": 0.1744687469072092,
      "grad_norm": 0.2080339789390564,
      "learning_rate": 8.256867762869545e-06,
      "loss": 0.0,
      "step": 12340
    },
    {
      "epoch": 0.17461013162917616,
      "grad_norm": 7.036044991082235e-08,
      "learning_rate": 8.255453915649876e-06,
      "loss": 0.0,
      "step": 12350
    },
    {
      "epoch": 0.17475151635114308,
      "grad_norm": 0.0,
      "learning_rate": 8.254040068430207e-06,
      "loss": 0.0,
      "step": 12360
    },
    {
      "epoch": 0.17489290107311004,
      "grad_norm": 6.61181376315767e-09,
      "learning_rate": 8.252626221210537e-06,
      "loss": 0.003,
      "step": 12370
    },
    {
      "epoch": 0.175034285795077,
      "grad_norm": 9.855897209476439e-10,
      "learning_rate": 8.251212373990867e-06,
      "loss": 0.0,
      "step": 12380
    },
    {
      "epoch": 0.17517567051704394,
      "grad_norm": 0.0,
      "learning_rate": 8.249798526771197e-06,
      "loss": 0.0003,
      "step": 12390
    },
    {
      "epoch": 0.17531705523901087,
      "grad_norm": 0.023492075502872467,
      "learning_rate": 8.248384679551528e-06,
      "loss": 0.0,
      "step": 12400
    },
    {
      "epoch": 0.17545843996097782,
      "grad_norm": 0.000367307715350762,
      "learning_rate": 8.246970832331859e-06,
      "loss": 0.0,
      "step": 12410
    },
    {
      "epoch": 0.17559982468294477,
      "grad_norm": 0.0032781537156552076,
      "learning_rate": 8.24555698511219e-06,
      "loss": 0.0001,
      "step": 12420
    },
    {
      "epoch": 0.1757412094049117,
      "grad_norm": 0.0,
      "learning_rate": 8.244143137892519e-06,
      "loss": 0.0001,
      "step": 12430
    },
    {
      "epoch": 0.17588259412687865,
      "grad_norm": 1.757716859174252e-06,
      "learning_rate": 8.24272929067285e-06,
      "loss": 0.3168,
      "step": 12440
    },
    {
      "epoch": 0.1760239788488456,
      "grad_norm": 0.0,
      "learning_rate": 8.24131544345318e-06,
      "loss": 0.0,
      "step": 12450
    },
    {
      "epoch": 0.17616536357081253,
      "grad_norm": 0.060821283608675,
      "learning_rate": 8.239901596233512e-06,
      "loss": 0.0001,
      "step": 12460
    },
    {
      "epoch": 0.17630674829277948,
      "grad_norm": 21.60634422302246,
      "learning_rate": 8.238487749013843e-06,
      "loss": 0.0035,
      "step": 12470
    },
    {
      "epoch": 0.17644813301474643,
      "grad_norm": 0.0,
      "learning_rate": 8.237073901794172e-06,
      "loss": 0.0,
      "step": 12480
    },
    {
      "epoch": 0.17658951773671336,
      "grad_norm": 0.0,
      "learning_rate": 8.235660054574504e-06,
      "loss": 0.0,
      "step": 12490
    },
    {
      "epoch": 0.1767309024586803,
      "grad_norm": 6.397416996151151e-07,
      "learning_rate": 8.234246207354833e-06,
      "loss": 0.0,
      "step": 12500
    },
    {
      "epoch": 0.17687228718064726,
      "grad_norm": 0.0,
      "learning_rate": 8.232832360135164e-06,
      "loss": 0.0001,
      "step": 12510
    },
    {
      "epoch": 0.17701367190261422,
      "grad_norm": 0.0004592011100612581,
      "learning_rate": 8.231418512915495e-06,
      "loss": 0.0,
      "step": 12520
    },
    {
      "epoch": 0.17715505662458114,
      "grad_norm": 0.0,
      "learning_rate": 8.230004665695826e-06,
      "loss": 0.0,
      "step": 12530
    },
    {
      "epoch": 0.1772964413465481,
      "grad_norm": 0.0,
      "learning_rate": 8.228590818476157e-06,
      "loss": 0.0,
      "step": 12540
    },
    {
      "epoch": 0.17743782606851505,
      "grad_norm": 0.0,
      "learning_rate": 8.227176971256486e-06,
      "loss": 0.0,
      "step": 12550
    },
    {
      "epoch": 0.17757921079048197,
      "grad_norm": 0.0,
      "learning_rate": 8.225763124036817e-06,
      "loss": 0.0066,
      "step": 12560
    },
    {
      "epoch": 0.17772059551244893,
      "grad_norm": 0.11962255835533142,
      "learning_rate": 8.224349276817148e-06,
      "loss": 0.0,
      "step": 12570
    },
    {
      "epoch": 0.17786198023441588,
      "grad_norm": 0.0,
      "learning_rate": 8.222935429597479e-06,
      "loss": 0.0,
      "step": 12580
    },
    {
      "epoch": 0.1780033649563828,
      "grad_norm": 0.0,
      "learning_rate": 8.22152158237781e-06,
      "loss": 0.1454,
      "step": 12590
    },
    {
      "epoch": 0.17814474967834976,
      "grad_norm": 0.0005606537451967597,
      "learning_rate": 8.220107735158139e-06,
      "loss": 0.0,
      "step": 12600
    },
    {
      "epoch": 0.1782861344003167,
      "grad_norm": 936.10791015625,
      "learning_rate": 8.218693887938471e-06,
      "loss": 1.1314,
      "step": 12610
    },
    {
      "epoch": 0.17842751912228363,
      "grad_norm": 0.0,
      "learning_rate": 8.2172800407188e-06,
      "loss": 0.0001,
      "step": 12620
    },
    {
      "epoch": 0.1785689038442506,
      "grad_norm": 0.0,
      "learning_rate": 8.215866193499131e-06,
      "loss": 0.0005,
      "step": 12630
    },
    {
      "epoch": 0.17871028856621754,
      "grad_norm": 0.0,
      "learning_rate": 8.214452346279462e-06,
      "loss": 0.0018,
      "step": 12640
    },
    {
      "epoch": 0.17885167328818447,
      "grad_norm": 0.0,
      "learning_rate": 8.213038499059793e-06,
      "loss": 0.0002,
      "step": 12650
    },
    {
      "epoch": 0.17899305801015142,
      "grad_norm": 8.525044559171135e-10,
      "learning_rate": 8.211624651840124e-06,
      "loss": 0.0,
      "step": 12660
    },
    {
      "epoch": 0.17913444273211837,
      "grad_norm": 0.0,
      "learning_rate": 8.210210804620453e-06,
      "loss": 0.0,
      "step": 12670
    },
    {
      "epoch": 0.17927582745408532,
      "grad_norm": 3.9131760597229004,
      "learning_rate": 8.208796957400784e-06,
      "loss": 0.0006,
      "step": 12680
    },
    {
      "epoch": 0.17941721217605225,
      "grad_norm": 0.0,
      "learning_rate": 8.207383110181115e-06,
      "loss": 0.0,
      "step": 12690
    },
    {
      "epoch": 0.1795585968980192,
      "grad_norm": 0.0,
      "learning_rate": 8.205969262961445e-06,
      "loss": 0.0089,
      "step": 12700
    },
    {
      "epoch": 0.17969998161998615,
      "grad_norm": 0.0,
      "learning_rate": 8.204555415741776e-06,
      "loss": 0.0,
      "step": 12710
    },
    {
      "epoch": 0.17984136634195308,
      "grad_norm": 0.0004097099299542606,
      "learning_rate": 8.203141568522105e-06,
      "loss": 0.0,
      "step": 12720
    },
    {
      "epoch": 0.17998275106392003,
      "grad_norm": 6.536467367368459e-08,
      "learning_rate": 8.201727721302436e-06,
      "loss": 0.0,
      "step": 12730
    },
    {
      "epoch": 0.18012413578588699,
      "grad_norm": 3.0486609148283605e-07,
      "learning_rate": 8.200313874082767e-06,
      "loss": 0.0,
      "step": 12740
    },
    {
      "epoch": 0.1802655205078539,
      "grad_norm": 0.0,
      "learning_rate": 8.198900026863098e-06,
      "loss": 0.0,
      "step": 12750
    },
    {
      "epoch": 0.18040690522982086,
      "grad_norm": 0.0,
      "learning_rate": 8.197486179643427e-06,
      "loss": 0.0,
      "step": 12760
    },
    {
      "epoch": 0.18054828995178782,
      "grad_norm": 6.395480056653469e-08,
      "learning_rate": 8.19607233242376e-06,
      "loss": 0.0,
      "step": 12770
    },
    {
      "epoch": 0.18068967467375474,
      "grad_norm": 0.0,
      "learning_rate": 8.19465848520409e-06,
      "loss": 0.8875,
      "step": 12780
    },
    {
      "epoch": 0.1808310593957217,
      "grad_norm": 0.0,
      "learning_rate": 8.19324463798442e-06,
      "loss": 0.0037,
      "step": 12790
    },
    {
      "epoch": 0.18097244411768865,
      "grad_norm": 4.084122338099405e-06,
      "learning_rate": 8.19183079076475e-06,
      "loss": 0.0,
      "step": 12800
    },
    {
      "epoch": 0.1811138288396556,
      "grad_norm": 0.0,
      "learning_rate": 8.190416943545081e-06,
      "loss": 0.0001,
      "step": 12810
    },
    {
      "epoch": 0.18125521356162252,
      "grad_norm": 0.0,
      "learning_rate": 8.189003096325412e-06,
      "loss": 0.1375,
      "step": 12820
    },
    {
      "epoch": 0.18139659828358948,
      "grad_norm": 4.188918296677002e-07,
      "learning_rate": 8.187589249105742e-06,
      "loss": 0.0001,
      "step": 12830
    },
    {
      "epoch": 0.18153798300555643,
      "grad_norm": 2.249469499560064e-09,
      "learning_rate": 8.186175401886072e-06,
      "loss": 0.0,
      "step": 12840
    },
    {
      "epoch": 0.18167936772752336,
      "grad_norm": 0.0,
      "learning_rate": 8.184761554666403e-06,
      "loss": 0.0,
      "step": 12850
    },
    {
      "epoch": 0.1818207524494903,
      "grad_norm": 0.011910339817404747,
      "learning_rate": 8.183347707446734e-06,
      "loss": 0.0,
      "step": 12860
    },
    {
      "epoch": 0.18196213717145726,
      "grad_norm": 4.802675903192721e-05,
      "learning_rate": 8.181933860227065e-06,
      "loss": 0.0005,
      "step": 12870
    },
    {
      "epoch": 0.1821035218934242,
      "grad_norm": 0.0,
      "learning_rate": 8.180520013007394e-06,
      "loss": 0.0009,
      "step": 12880
    },
    {
      "epoch": 0.18224490661539114,
      "grad_norm": 0.0,
      "learning_rate": 8.179106165787725e-06,
      "loss": 0.0002,
      "step": 12890
    },
    {
      "epoch": 0.1823862913373581,
      "grad_norm": 0.0,
      "learning_rate": 8.177692318568056e-06,
      "loss": 0.5973,
      "step": 12900
    },
    {
      "epoch": 0.18252767605932502,
      "grad_norm": 8.596648513048422e-07,
      "learning_rate": 8.176278471348387e-06,
      "loss": 0.0,
      "step": 12910
    },
    {
      "epoch": 0.18266906078129197,
      "grad_norm": 0.0,
      "learning_rate": 8.174864624128718e-06,
      "loss": 0.0,
      "step": 12920
    },
    {
      "epoch": 0.18281044550325892,
      "grad_norm": 3.038127260879264e-07,
      "learning_rate": 8.173450776909048e-06,
      "loss": 0.0926,
      "step": 12930
    },
    {
      "epoch": 0.18295183022522588,
      "grad_norm": 0.0,
      "learning_rate": 8.17203692968938e-06,
      "loss": 0.2072,
      "step": 12940
    },
    {
      "epoch": 0.1830932149471928,
      "grad_norm": 0.0,
      "learning_rate": 8.170623082469708e-06,
      "loss": 0.0,
      "step": 12950
    },
    {
      "epoch": 0.18323459966915975,
      "grad_norm": 0.0007655876688659191,
      "learning_rate": 8.16920923525004e-06,
      "loss": 0.0071,
      "step": 12960
    },
    {
      "epoch": 0.1833759843911267,
      "grad_norm": 855.03857421875,
      "learning_rate": 8.16779538803037e-06,
      "loss": 1.0078,
      "step": 12970
    },
    {
      "epoch": 0.18351736911309363,
      "grad_norm": 0.0,
      "learning_rate": 8.166381540810701e-06,
      "loss": 0.0031,
      "step": 12980
    },
    {
      "epoch": 0.18365875383506058,
      "grad_norm": 0.0,
      "learning_rate": 8.164967693591032e-06,
      "loss": 0.0005,
      "step": 12990
    },
    {
      "epoch": 0.18380013855702754,
      "grad_norm": 0.0,
      "learning_rate": 8.163553846371361e-06,
      "loss": 0.0011,
      "step": 13000
    },
    {
      "epoch": 0.18394152327899446,
      "grad_norm": 0.0,
      "learning_rate": 8.162139999151692e-06,
      "loss": 0.0181,
      "step": 13010
    },
    {
      "epoch": 0.18408290800096141,
      "grad_norm": 0.08995730429887772,
      "learning_rate": 8.160726151932023e-06,
      "loss": 0.0,
      "step": 13020
    },
    {
      "epoch": 0.18422429272292837,
      "grad_norm": 0.0,
      "learning_rate": 8.159312304712354e-06,
      "loss": 0.0006,
      "step": 13030
    },
    {
      "epoch": 0.1843656774448953,
      "grad_norm": 0.0,
      "learning_rate": 8.157898457492684e-06,
      "loss": 0.0,
      "step": 13040
    },
    {
      "epoch": 0.18450706216686225,
      "grad_norm": 0.0093761021271348,
      "learning_rate": 8.156484610273014e-06,
      "loss": 0.0001,
      "step": 13050
    },
    {
      "epoch": 0.1846484468888292,
      "grad_norm": 7.661139534320682e-05,
      "learning_rate": 8.155070763053346e-06,
      "loss": 0.0003,
      "step": 13060
    },
    {
      "epoch": 0.18478983161079612,
      "grad_norm": 7.833442623450537e-07,
      "learning_rate": 8.153656915833675e-06,
      "loss": 0.0,
      "step": 13070
    },
    {
      "epoch": 0.18493121633276308,
      "grad_norm": 1.7192593812942505,
      "learning_rate": 8.152243068614006e-06,
      "loss": 0.0004,
      "step": 13080
    },
    {
      "epoch": 0.18507260105473003,
      "grad_norm": 0.0,
      "learning_rate": 8.150829221394337e-06,
      "loss": 0.0,
      "step": 13090
    },
    {
      "epoch": 0.18521398577669698,
      "grad_norm": 0.8240106105804443,
      "learning_rate": 8.149415374174668e-06,
      "loss": 0.5459,
      "step": 13100
    },
    {
      "epoch": 0.1853553704986639,
      "grad_norm": 0.06860623508691788,
      "learning_rate": 8.148001526954999e-06,
      "loss": 0.0004,
      "step": 13110
    },
    {
      "epoch": 0.18549675522063086,
      "grad_norm": 0.0,
      "learning_rate": 8.146587679735328e-06,
      "loss": 0.0,
      "step": 13120
    },
    {
      "epoch": 0.1856381399425978,
      "grad_norm": 0.40525615215301514,
      "learning_rate": 8.145173832515659e-06,
      "loss": 0.0015,
      "step": 13130
    },
    {
      "epoch": 0.18577952466456474,
      "grad_norm": 118.06385803222656,
      "learning_rate": 8.14375998529599e-06,
      "loss": 0.0425,
      "step": 13140
    },
    {
      "epoch": 0.1859209093865317,
      "grad_norm": 0.0,
      "learning_rate": 8.14234613807632e-06,
      "loss": 0.0001,
      "step": 13150
    },
    {
      "epoch": 0.18606229410849864,
      "grad_norm": 0.00034804712049663067,
      "learning_rate": 8.14093229085665e-06,
      "loss": 0.0,
      "step": 13160
    },
    {
      "epoch": 0.18620367883046557,
      "grad_norm": 0.0,
      "learning_rate": 8.13951844363698e-06,
      "loss": 0.0,
      "step": 13170
    },
    {
      "epoch": 0.18634506355243252,
      "grad_norm": 0.0009972041007131338,
      "learning_rate": 8.138104596417313e-06,
      "loss": 0.0003,
      "step": 13180
    },
    {
      "epoch": 0.18648644827439947,
      "grad_norm": 0.0,
      "learning_rate": 8.136690749197642e-06,
      "loss": 0.0,
      "step": 13190
    },
    {
      "epoch": 0.1866278329963664,
      "grad_norm": 3.152325007249601e-05,
      "learning_rate": 8.135276901977973e-06,
      "loss": 0.0,
      "step": 13200
    },
    {
      "epoch": 0.18676921771833335,
      "grad_norm": 0.00669251149520278,
      "learning_rate": 8.133863054758304e-06,
      "loss": 0.0106,
      "step": 13210
    },
    {
      "epoch": 0.1869106024403003,
      "grad_norm": 3.2846816111486987e-07,
      "learning_rate": 8.132449207538635e-06,
      "loss": 0.0092,
      "step": 13220
    },
    {
      "epoch": 0.18705198716226726,
      "grad_norm": 0.11194054037332535,
      "learning_rate": 8.131035360318966e-06,
      "loss": 0.0,
      "step": 13230
    },
    {
      "epoch": 0.18719337188423418,
      "grad_norm": 1.126837196352426e-05,
      "learning_rate": 8.129621513099295e-06,
      "loss": 0.0,
      "step": 13240
    },
    {
      "epoch": 0.18733475660620114,
      "grad_norm": 1.7603806554689072e-05,
      "learning_rate": 8.128207665879626e-06,
      "loss": 0.0002,
      "step": 13250
    },
    {
      "epoch": 0.1874761413281681,
      "grad_norm": 0.12303480505943298,
      "learning_rate": 8.126793818659956e-06,
      "loss": 0.0002,
      "step": 13260
    },
    {
      "epoch": 0.18761752605013501,
      "grad_norm": 0.0,
      "learning_rate": 8.125379971440287e-06,
      "loss": 0.0,
      "step": 13270
    },
    {
      "epoch": 0.18775891077210197,
      "grad_norm": 2.1436839233501814e-05,
      "learning_rate": 8.123966124220617e-06,
      "loss": 0.0,
      "step": 13280
    },
    {
      "epoch": 0.18790029549406892,
      "grad_norm": 0.006340509746223688,
      "learning_rate": 8.122552277000947e-06,
      "loss": 0.0,
      "step": 13290
    },
    {
      "epoch": 0.18804168021603584,
      "grad_norm": 6.861700057983398,
      "learning_rate": 8.121138429781278e-06,
      "loss": 0.0011,
      "step": 13300
    },
    {
      "epoch": 0.1881830649380028,
      "grad_norm": 0.0,
      "learning_rate": 8.119724582561609e-06,
      "loss": 0.0001,
      "step": 13310
    },
    {
      "epoch": 0.18832444965996975,
      "grad_norm": 0.0,
      "learning_rate": 8.11831073534194e-06,
      "loss": 0.0,
      "step": 13320
    },
    {
      "epoch": 0.18846583438193668,
      "grad_norm": 1.0193533794833343e-09,
      "learning_rate": 8.116896888122269e-06,
      "loss": 0.0005,
      "step": 13330
    },
    {
      "epoch": 0.18860721910390363,
      "grad_norm": 0.011018290184438229,
      "learning_rate": 8.115483040902602e-06,
      "loss": 0.0058,
      "step": 13340
    },
    {
      "epoch": 0.18874860382587058,
      "grad_norm": 0.009165837429463863,
      "learning_rate": 8.11406919368293e-06,
      "loss": 0.0618,
      "step": 13350
    },
    {
      "epoch": 0.18888998854783753,
      "grad_norm": 0.0,
      "learning_rate": 8.112655346463262e-06,
      "loss": 0.0,
      "step": 13360
    },
    {
      "epoch": 0.18903137326980446,
      "grad_norm": 4.763144323405299e-10,
      "learning_rate": 8.111241499243593e-06,
      "loss": 0.0,
      "step": 13370
    },
    {
      "epoch": 0.1891727579917714,
      "grad_norm": 9.006541068501406e-10,
      "learning_rate": 8.109827652023923e-06,
      "loss": 0.0,
      "step": 13380
    },
    {
      "epoch": 0.18931414271373836,
      "grad_norm": 0.0008326731040142477,
      "learning_rate": 8.108413804804254e-06,
      "loss": 0.0,
      "step": 13390
    },
    {
      "epoch": 0.1894555274357053,
      "grad_norm": 0.00013501722423825413,
      "learning_rate": 8.106999957584583e-06,
      "loss": 0.0,
      "step": 13400
    },
    {
      "epoch": 0.18959691215767224,
      "grad_norm": 1.8067972673563304e-09,
      "learning_rate": 8.105586110364914e-06,
      "loss": 0.0,
      "step": 13410
    },
    {
      "epoch": 0.1897382968796392,
      "grad_norm": 0.0,
      "learning_rate": 8.104172263145245e-06,
      "loss": 0.0004,
      "step": 13420
    },
    {
      "epoch": 0.18987968160160612,
      "grad_norm": 4.567545630607128e-08,
      "learning_rate": 8.102758415925576e-06,
      "loss": 0.0,
      "step": 13430
    },
    {
      "epoch": 0.19002106632357307,
      "grad_norm": 0.0,
      "learning_rate": 8.101344568705907e-06,
      "loss": 0.0,
      "step": 13440
    },
    {
      "epoch": 0.19016245104554003,
      "grad_norm": 2.0103516578674316,
      "learning_rate": 8.099930721486236e-06,
      "loss": 0.0003,
      "step": 13450
    },
    {
      "epoch": 0.19030383576750695,
      "grad_norm": 2.9207259899521887e-07,
      "learning_rate": 8.098516874266567e-06,
      "loss": 0.0,
      "step": 13460
    },
    {
      "epoch": 0.1904452204894739,
      "grad_norm": 3.885817022819538e-06,
      "learning_rate": 8.097103027046898e-06,
      "loss": 0.3545,
      "step": 13470
    },
    {
      "epoch": 0.19058660521144086,
      "grad_norm": 0.0,
      "learning_rate": 8.095689179827229e-06,
      "loss": 0.0001,
      "step": 13480
    },
    {
      "epoch": 0.19072798993340778,
      "grad_norm": 1.3867494885744236e-07,
      "learning_rate": 8.09427533260756e-06,
      "loss": 0.0,
      "step": 13490
    },
    {
      "epoch": 0.19086937465537474,
      "grad_norm": 6.799352263442415e-07,
      "learning_rate": 8.09286148538789e-06,
      "loss": 0.0,
      "step": 13500
    },
    {
      "epoch": 0.1910107593773417,
      "grad_norm": 0.0,
      "learning_rate": 8.091447638168221e-06,
      "loss": 0.0012,
      "step": 13510
    },
    {
      "epoch": 0.19115214409930864,
      "grad_norm": 0.0,
      "learning_rate": 8.09003379094855e-06,
      "loss": 0.0,
      "step": 13520
    },
    {
      "epoch": 0.19129352882127557,
      "grad_norm": 0.007611885201185942,
      "learning_rate": 8.088619943728881e-06,
      "loss": 0.0,
      "step": 13530
    },
    {
      "epoch": 0.19143491354324252,
      "grad_norm": 0.0,
      "learning_rate": 8.087206096509212e-06,
      "loss": 0.0,
      "step": 13540
    },
    {
      "epoch": 0.19157629826520947,
      "grad_norm": 0.0,
      "learning_rate": 8.085792249289543e-06,
      "loss": 0.0,
      "step": 13550
    },
    {
      "epoch": 0.1917176829871764,
      "grad_norm": 0.0,
      "learning_rate": 8.084378402069874e-06,
      "loss": 0.0008,
      "step": 13560
    },
    {
      "epoch": 0.19185906770914335,
      "grad_norm": 0.0005957330577075481,
      "learning_rate": 8.082964554850203e-06,
      "loss": 0.0001,
      "step": 13570
    },
    {
      "epoch": 0.1920004524311103,
      "grad_norm": 0.00022731375065632164,
      "learning_rate": 8.081550707630534e-06,
      "loss": 0.0132,
      "step": 13580
    },
    {
      "epoch": 0.19214183715307723,
      "grad_norm": 0.0,
      "learning_rate": 8.080136860410865e-06,
      "loss": 0.0001,
      "step": 13590
    },
    {
      "epoch": 0.19228322187504418,
      "grad_norm": 0.0,
      "learning_rate": 8.078723013191195e-06,
      "loss": 0.0804,
      "step": 13600
    },
    {
      "epoch": 0.19242460659701113,
      "grad_norm": 6.731197288445401e-08,
      "learning_rate": 8.077309165971525e-06,
      "loss": 0.0,
      "step": 13610
    },
    {
      "epoch": 0.19256599131897806,
      "grad_norm": 0.0,
      "learning_rate": 8.075895318751855e-06,
      "loss": 0.0474,
      "step": 13620
    },
    {
      "epoch": 0.192707376040945,
      "grad_norm": 0.0,
      "learning_rate": 8.074481471532188e-06,
      "loss": 0.0678,
      "step": 13630
    },
    {
      "epoch": 0.19284876076291196,
      "grad_norm": 0.0,
      "learning_rate": 8.073067624312517e-06,
      "loss": 0.0,
      "step": 13640
    },
    {
      "epoch": 0.19299014548487892,
      "grad_norm": 0.12267220765352249,
      "learning_rate": 8.071653777092848e-06,
      "loss": 0.0,
      "step": 13650
    },
    {
      "epoch": 0.19313153020684584,
      "grad_norm": 0.0,
      "learning_rate": 8.070239929873179e-06,
      "loss": 0.0,
      "step": 13660
    },
    {
      "epoch": 0.1932729149288128,
      "grad_norm": 0.0,
      "learning_rate": 8.06882608265351e-06,
      "loss": 0.0,
      "step": 13670
    },
    {
      "epoch": 0.19341429965077975,
      "grad_norm": 0.0,
      "learning_rate": 8.06741223543384e-06,
      "loss": 0.0,
      "step": 13680
    },
    {
      "epoch": 0.19355568437274667,
      "grad_norm": 0.0,
      "learning_rate": 8.06599838821417e-06,
      "loss": 0.0,
      "step": 13690
    },
    {
      "epoch": 0.19369706909471363,
      "grad_norm": 0.0,
      "learning_rate": 8.0645845409945e-06,
      "loss": 0.0,
      "step": 13700
    },
    {
      "epoch": 0.19383845381668058,
      "grad_norm": 0.006358843296766281,
      "learning_rate": 8.063170693774831e-06,
      "loss": 0.0,
      "step": 13710
    },
    {
      "epoch": 0.1939798385386475,
      "grad_norm": 0.00017460159142501652,
      "learning_rate": 8.061756846555162e-06,
      "loss": 0.4221,
      "step": 13720
    },
    {
      "epoch": 0.19412122326061446,
      "grad_norm": 0.0,
      "learning_rate": 8.060342999335491e-06,
      "loss": 0.0,
      "step": 13730
    },
    {
      "epoch": 0.1942626079825814,
      "grad_norm": 0.00013077628682367504,
      "learning_rate": 8.058929152115822e-06,
      "loss": 0.0,
      "step": 13740
    },
    {
      "epoch": 0.19440399270454833,
      "grad_norm": 0.0,
      "learning_rate": 8.057515304896155e-06,
      "loss": 0.0,
      "step": 13750
    },
    {
      "epoch": 0.1945453774265153,
      "grad_norm": 0.0,
      "learning_rate": 8.056101457676484e-06,
      "loss": 0.0,
      "step": 13760
    },
    {
      "epoch": 0.19468676214848224,
      "grad_norm": 0.0,
      "learning_rate": 8.054687610456815e-06,
      "loss": 0.1778,
      "step": 13770
    },
    {
      "epoch": 0.1948281468704492,
      "grad_norm": 3.133833414992182e-09,
      "learning_rate": 8.053273763237146e-06,
      "loss": 0.0924,
      "step": 13780
    },
    {
      "epoch": 0.19496953159241612,
      "grad_norm": 0.0,
      "learning_rate": 8.051859916017477e-06,
      "loss": 0.0,
      "step": 13790
    },
    {
      "epoch": 0.19511091631438307,
      "grad_norm": 0.0,
      "learning_rate": 8.050446068797806e-06,
      "loss": 0.0,
      "step": 13800
    },
    {
      "epoch": 0.19525230103635002,
      "grad_norm": 0.0,
      "learning_rate": 8.049032221578137e-06,
      "loss": 0.1597,
      "step": 13810
    },
    {
      "epoch": 0.19539368575831695,
      "grad_norm": 0.0,
      "learning_rate": 8.047618374358468e-06,
      "loss": 0.0003,
      "step": 13820
    },
    {
      "epoch": 0.1955350704802839,
      "grad_norm": 8.766481914790347e-05,
      "learning_rate": 8.046204527138798e-06,
      "loss": 0.0018,
      "step": 13830
    },
    {
      "epoch": 0.19567645520225085,
      "grad_norm": 1.9561859517125413e-05,
      "learning_rate": 8.04479067991913e-06,
      "loss": 0.0,
      "step": 13840
    },
    {
      "epoch": 0.19581783992421778,
      "grad_norm": 4.570603062692413e-10,
      "learning_rate": 8.043376832699458e-06,
      "loss": 0.0,
      "step": 13850
    },
    {
      "epoch": 0.19595922464618473,
      "grad_norm": 0.0004735244147013873,
      "learning_rate": 8.04196298547979e-06,
      "loss": 0.0,
      "step": 13860
    },
    {
      "epoch": 0.19610060936815168,
      "grad_norm": 9.886751417553796e-10,
      "learning_rate": 8.04054913826012e-06,
      "loss": 0.0,
      "step": 13870
    },
    {
      "epoch": 0.1962419940901186,
      "grad_norm": 0.0,
      "learning_rate": 8.039135291040451e-06,
      "loss": 0.0,
      "step": 13880
    },
    {
      "epoch": 0.19638337881208556,
      "grad_norm": 0.0,
      "learning_rate": 8.037721443820782e-06,
      "loss": 0.0,
      "step": 13890
    },
    {
      "epoch": 0.19652476353405252,
      "grad_norm": 0.07328192889690399,
      "learning_rate": 8.036307596601111e-06,
      "loss": 0.0418,
      "step": 13900
    },
    {
      "epoch": 0.19666614825601944,
      "grad_norm": 0.0,
      "learning_rate": 8.034893749381444e-06,
      "loss": 0.0009,
      "step": 13910
    },
    {
      "epoch": 0.1968075329779864,
      "grad_norm": 0.0,
      "learning_rate": 8.033479902161773e-06,
      "loss": 0.0002,
      "step": 13920
    },
    {
      "epoch": 0.19694891769995335,
      "grad_norm": 0.0,
      "learning_rate": 8.032066054942104e-06,
      "loss": 0.0,
      "step": 13930
    },
    {
      "epoch": 0.1970903024219203,
      "grad_norm": 0.0,
      "learning_rate": 8.030652207722434e-06,
      "loss": 0.5166,
      "step": 13940
    },
    {
      "epoch": 0.19723168714388722,
      "grad_norm": 0.0,
      "learning_rate": 8.029238360502765e-06,
      "loss": 0.0,
      "step": 13950
    },
    {
      "epoch": 0.19737307186585418,
      "grad_norm": 0.0,
      "learning_rate": 8.027824513283096e-06,
      "loss": 0.0,
      "step": 13960
    },
    {
      "epoch": 0.19751445658782113,
      "grad_norm": 0.0,
      "learning_rate": 8.026552050785393e-06,
      "loss": 0.293,
      "step": 13970
    },
    {
      "epoch": 0.19765584130978806,
      "grad_norm": 0.0,
      "learning_rate": 8.025138203565723e-06,
      "loss": 0.0001,
      "step": 13980
    },
    {
      "epoch": 0.197797226031755,
      "grad_norm": 0.0,
      "learning_rate": 8.023724356346054e-06,
      "loss": 0.0,
      "step": 13990
    },
    {
      "epoch": 0.19793861075372196,
      "grad_norm": 3.563415305052331e-07,
      "learning_rate": 8.022310509126383e-06,
      "loss": 0.0,
      "step": 14000
    },
    {
      "epoch": 0.19807999547568889,
      "grad_norm": 0.0,
      "learning_rate": 8.020896661906716e-06,
      "loss": 0.0,
      "step": 14010
    },
    {
      "epoch": 0.19822138019765584,
      "grad_norm": 700.89794921875,
      "learning_rate": 8.019482814687045e-06,
      "loss": 0.115,
      "step": 14020
    },
    {
      "epoch": 0.1983627649196228,
      "grad_norm": 6.124090194702148,
      "learning_rate": 8.018068967467376e-06,
      "loss": 0.0071,
      "step": 14030
    },
    {
      "epoch": 0.19850414964158972,
      "grad_norm": 0.11164742708206177,
      "learning_rate": 8.016655120247707e-06,
      "loss": 0.0003,
      "step": 14040
    },
    {
      "epoch": 0.19864553436355667,
      "grad_norm": 0.0010586567223072052,
      "learning_rate": 8.015241273028038e-06,
      "loss": 0.0,
      "step": 14050
    },
    {
      "epoch": 0.19878691908552362,
      "grad_norm": 0.0,
      "learning_rate": 8.013827425808369e-06,
      "loss": 0.0,
      "step": 14060
    },
    {
      "epoch": 0.19892830380749058,
      "grad_norm": 1.0161681984754978e-07,
      "learning_rate": 8.012413578588698e-06,
      "loss": 0.0033,
      "step": 14070
    },
    {
      "epoch": 0.1990696885294575,
      "grad_norm": 0.02188796177506447,
      "learning_rate": 8.010999731369029e-06,
      "loss": 0.0,
      "step": 14080
    },
    {
      "epoch": 0.19921107325142445,
      "grad_norm": 1.5311732568079606e-05,
      "learning_rate": 8.00958588414936e-06,
      "loss": 0.1701,
      "step": 14090
    },
    {
      "epoch": 0.1993524579733914,
      "grad_norm": 9.757832319934323e-10,
      "learning_rate": 8.00817203692969e-06,
      "loss": 0.0,
      "step": 14100
    },
    {
      "epoch": 0.19949384269535833,
      "grad_norm": 0.0,
      "learning_rate": 8.006758189710021e-06,
      "loss": 0.0,
      "step": 14110
    },
    {
      "epoch": 0.19963522741732528,
      "grad_norm": 0.0,
      "learning_rate": 8.00534434249035e-06,
      "loss": 0.0009,
      "step": 14120
    },
    {
      "epoch": 0.19977661213929224,
      "grad_norm": 438.103515625,
      "learning_rate": 8.003930495270683e-06,
      "loss": 0.9047,
      "step": 14130
    },
    {
      "epoch": 0.19991799686125916,
      "grad_norm": 0.0,
      "learning_rate": 8.002516648051012e-06,
      "loss": 0.0,
      "step": 14140
    },
    {
      "epoch": 0.20005938158322611,
      "grad_norm": 0.0,
      "learning_rate": 8.001102800831343e-06,
      "loss": 0.0001,
      "step": 14150
    },
    {
      "epoch": 0.20020076630519307,
      "grad_norm": 0.0,
      "learning_rate": 7.999688953611672e-06,
      "loss": 0.0,
      "step": 14160
    },
    {
      "epoch": 0.20034215102716,
      "grad_norm": 6.9016339310223884e-09,
      "learning_rate": 7.998275106392005e-06,
      "loss": 0.0,
      "step": 14170
    },
    {
      "epoch": 0.20048353574912695,
      "grad_norm": 0.0023233932442963123,
      "learning_rate": 7.996861259172335e-06,
      "loss": 0.0,
      "step": 14180
    },
    {
      "epoch": 0.2006249204710939,
      "grad_norm": 0.0,
      "learning_rate": 7.995447411952665e-06,
      "loss": 0.0,
      "step": 14190
    },
    {
      "epoch": 0.20076630519306085,
      "grad_norm": 651.2219848632812,
      "learning_rate": 7.994033564732995e-06,
      "loss": 0.893,
      "step": 14200
    },
    {
      "epoch": 0.20090768991502778,
      "grad_norm": 0.0,
      "learning_rate": 7.992619717513326e-06,
      "loss": 0.0001,
      "step": 14210
    },
    {
      "epoch": 0.20104907463699473,
      "grad_norm": 0.0,
      "learning_rate": 7.991205870293657e-06,
      "loss": 0.0008,
      "step": 14220
    },
    {
      "epoch": 0.20119045935896168,
      "grad_norm": 2.368493915128056e-05,
      "learning_rate": 7.989792023073988e-06,
      "loss": 0.0,
      "step": 14230
    },
    {
      "epoch": 0.2013318440809286,
      "grad_norm": 0.0,
      "learning_rate": 7.988378175854317e-06,
      "loss": 0.0056,
      "step": 14240
    },
    {
      "epoch": 0.20147322880289556,
      "grad_norm": 0.0,
      "learning_rate": 7.986964328634648e-06,
      "loss": 0.0,
      "step": 14250
    },
    {
      "epoch": 0.2016146135248625,
      "grad_norm": 0.1227315366268158,
      "learning_rate": 7.985550481414979e-06,
      "loss": 0.0,
      "step": 14260
    },
    {
      "epoch": 0.20175599824682944,
      "grad_norm": 0.0,
      "learning_rate": 7.98413663419531e-06,
      "loss": 0.0,
      "step": 14270
    },
    {
      "epoch": 0.2018973829687964,
      "grad_norm": 0.0,
      "learning_rate": 7.982722786975639e-06,
      "loss": 0.0,
      "step": 14280
    },
    {
      "epoch": 0.20203876769076334,
      "grad_norm": 0.0001248226617462933,
      "learning_rate": 7.981308939755971e-06,
      "loss": 0.0,
      "step": 14290
    },
    {
      "epoch": 0.20218015241273027,
      "grad_norm": 0.0,
      "learning_rate": 7.979895092536302e-06,
      "loss": 0.0,
      "step": 14300
    },
    {
      "epoch": 0.20232153713469722,
      "grad_norm": 1.1038158618248417e-06,
      "learning_rate": 7.978481245316631e-06,
      "loss": 0.0,
      "step": 14310
    },
    {
      "epoch": 0.20246292185666417,
      "grad_norm": 0.0,
      "learning_rate": 7.977067398096962e-06,
      "loss": 0.0043,
      "step": 14320
    },
    {
      "epoch": 0.2026043065786311,
      "grad_norm": 0.000483812706079334,
      "learning_rate": 7.975653550877293e-06,
      "loss": 0.2916,
      "step": 14330
    },
    {
      "epoch": 0.20274569130059805,
      "grad_norm": 0.0,
      "learning_rate": 7.974239703657624e-06,
      "loss": 0.0016,
      "step": 14340
    },
    {
      "epoch": 0.202887076022565,
      "grad_norm": 8.887479907571105e-07,
      "learning_rate": 7.972825856437953e-06,
      "loss": 0.0,
      "step": 14350
    },
    {
      "epoch": 0.20302846074453196,
      "grad_norm": 2.6853487725020386e-05,
      "learning_rate": 7.971412009218284e-06,
      "loss": 0.0,
      "step": 14360
    },
    {
      "epoch": 0.20316984546649888,
      "grad_norm": 0.0,
      "learning_rate": 7.969998161998615e-06,
      "loss": 0.0001,
      "step": 14370
    },
    {
      "epoch": 0.20331123018846584,
      "grad_norm": 0.0,
      "learning_rate": 7.968584314778946e-06,
      "loss": 0.0,
      "step": 14380
    },
    {
      "epoch": 0.2034526149104328,
      "grad_norm": 0.0036679022014141083,
      "learning_rate": 7.967170467559277e-06,
      "loss": 0.0,
      "step": 14390
    },
    {
      "epoch": 0.2035939996323997,
      "grad_norm": 0.0,
      "learning_rate": 7.965756620339606e-06,
      "loss": 0.0057,
      "step": 14400
    },
    {
      "epoch": 0.20373538435436667,
      "grad_norm": 0.0,
      "learning_rate": 7.964342773119937e-06,
      "loss": 0.0006,
      "step": 14410
    },
    {
      "epoch": 0.20387676907633362,
      "grad_norm": 0.0,
      "learning_rate": 7.962928925900268e-06,
      "loss": 0.0,
      "step": 14420
    },
    {
      "epoch": 0.20401815379830054,
      "grad_norm": 1.2161302720414824e-07,
      "learning_rate": 7.961515078680598e-06,
      "loss": 0.0001,
      "step": 14430
    },
    {
      "epoch": 0.2041595385202675,
      "grad_norm": 0.0,
      "learning_rate": 7.96010123146093e-06,
      "loss": 0.0,
      "step": 14440
    },
    {
      "epoch": 0.20430092324223445,
      "grad_norm": 0.6046896576881409,
      "learning_rate": 7.95868738424126e-06,
      "loss": 0.0001,
      "step": 14450
    },
    {
      "epoch": 0.20444230796420138,
      "grad_norm": 0.0,
      "learning_rate": 7.957273537021591e-06,
      "loss": 0.0009,
      "step": 14460
    },
    {
      "epoch": 0.20458369268616833,
      "grad_norm": 0.020249253138899803,
      "learning_rate": 7.95585968980192e-06,
      "loss": 0.0147,
      "step": 14470
    },
    {
      "epoch": 0.20472507740813528,
      "grad_norm": 9.450385550735518e-05,
      "learning_rate": 7.954445842582251e-06,
      "loss": 0.088,
      "step": 14480
    },
    {
      "epoch": 0.20486646213010223,
      "grad_norm": 0.0,
      "learning_rate": 7.953031995362582e-06,
      "loss": 0.0,
      "step": 14490
    },
    {
      "epoch": 0.20500784685206916,
      "grad_norm": 5.787521786260186e-06,
      "learning_rate": 7.951618148142913e-06,
      "loss": 0.1955,
      "step": 14500
    },
    {
      "epoch": 0.2051492315740361,
      "grad_norm": 0.00065419333986938,
      "learning_rate": 7.950204300923244e-06,
      "loss": 0.0,
      "step": 14510
    },
    {
      "epoch": 0.20529061629600306,
      "grad_norm": 0.0,
      "learning_rate": 7.948790453703573e-06,
      "loss": 0.0001,
      "step": 14520
    },
    {
      "epoch": 0.20543200101797,
      "grad_norm": 326.3916320800781,
      "learning_rate": 7.947376606483904e-06,
      "loss": 0.0433,
      "step": 14530
    },
    {
      "epoch": 0.20557338573993694,
      "grad_norm": 0.0,
      "learning_rate": 7.945962759264234e-06,
      "loss": 0.027,
      "step": 14540
    },
    {
      "epoch": 0.2057147704619039,
      "grad_norm": 0.0,
      "learning_rate": 7.944548912044565e-06,
      "loss": 0.0003,
      "step": 14550
    },
    {
      "epoch": 0.20585615518387082,
      "grad_norm": 1.932786226272583,
      "learning_rate": 7.943135064824896e-06,
      "loss": 0.0007,
      "step": 14560
    },
    {
      "epoch": 0.20599753990583777,
      "grad_norm": 0.0,
      "learning_rate": 7.941721217605225e-06,
      "loss": 0.0001,
      "step": 14570
    },
    {
      "epoch": 0.20613892462780473,
      "grad_norm": 1.8328416562241046e-09,
      "learning_rate": 7.940307370385558e-06,
      "loss": 0.0,
      "step": 14580
    },
    {
      "epoch": 0.20628030934977165,
      "grad_norm": 0.0,
      "learning_rate": 7.938893523165887e-06,
      "loss": 0.0357,
      "step": 14590
    },
    {
      "epoch": 0.2064216940717386,
      "grad_norm": 1.2726569309506885e-07,
      "learning_rate": 7.937479675946218e-06,
      "loss": 0.0001,
      "step": 14600
    },
    {
      "epoch": 0.20656307879370556,
      "grad_norm": 0.0,
      "learning_rate": 7.936065828726549e-06,
      "loss": 0.0026,
      "step": 14610
    },
    {
      "epoch": 0.2067044635156725,
      "grad_norm": 0.0,
      "learning_rate": 7.93465198150688e-06,
      "loss": 0.0,
      "step": 14620
    },
    {
      "epoch": 0.20684584823763943,
      "grad_norm": 9.565452874227276e-10,
      "learning_rate": 7.93323813428721e-06,
      "loss": 0.0,
      "step": 14630
    },
    {
      "epoch": 0.2069872329596064,
      "grad_norm": 6.200631474939655e-08,
      "learning_rate": 7.93182428706754e-06,
      "loss": 0.0,
      "step": 14640
    },
    {
      "epoch": 0.20712861768157334,
      "grad_norm": 0.0,
      "learning_rate": 7.93041043984787e-06,
      "loss": 0.0001,
      "step": 14650
    },
    {
      "epoch": 0.20727000240354027,
      "grad_norm": 0.0,
      "learning_rate": 7.928996592628201e-06,
      "loss": 0.0036,
      "step": 14660
    },
    {
      "epoch": 0.20741138712550722,
      "grad_norm": 9.614486694335938,
      "learning_rate": 7.927582745408532e-06,
      "loss": 0.0025,
      "step": 14670
    },
    {
      "epoch": 0.20755277184747417,
      "grad_norm": 0.0,
      "learning_rate": 7.926168898188861e-06,
      "loss": 0.0,
      "step": 14680
    },
    {
      "epoch": 0.2076941565694411,
      "grad_norm": 0.0,
      "learning_rate": 7.924755050969192e-06,
      "loss": 0.0,
      "step": 14690
    },
    {
      "epoch": 0.20783554129140805,
      "grad_norm": 0.12843330204486847,
      "learning_rate": 7.923341203749523e-06,
      "loss": 0.0,
      "step": 14700
    },
    {
      "epoch": 0.207976926013375,
      "grad_norm": 5.271654046623553e-08,
      "learning_rate": 7.921927356529854e-06,
      "loss": 0.0,
      "step": 14710
    },
    {
      "epoch": 0.20811831073534193,
      "grad_norm": 1.652938863117015e-06,
      "learning_rate": 7.920513509310185e-06,
      "loss": 0.0001,
      "step": 14720
    },
    {
      "epoch": 0.20825969545730888,
      "grad_norm": 0.03714042529463768,
      "learning_rate": 7.919099662090514e-06,
      "loss": 0.0,
      "step": 14730
    },
    {
      "epoch": 0.20840108017927583,
      "grad_norm": 1.0125278393502413e-09,
      "learning_rate": 7.917685814870846e-06,
      "loss": 0.0,
      "step": 14740
    },
    {
      "epoch": 0.20854246490124279,
      "grad_norm": 0.0,
      "learning_rate": 7.916271967651177e-06,
      "loss": 0.0135,
      "step": 14750
    },
    {
      "epoch": 0.2086838496232097,
      "grad_norm": 0.0,
      "learning_rate": 7.914858120431506e-06,
      "loss": 0.0002,
      "step": 14760
    },
    {
      "epoch": 0.20882523434517666,
      "grad_norm": 0.0,
      "learning_rate": 7.913444273211837e-06,
      "loss": 0.0,
      "step": 14770
    },
    {
      "epoch": 0.20896661906714362,
      "grad_norm": 0.0,
      "learning_rate": 7.912030425992168e-06,
      "loss": 0.0,
      "step": 14780
    },
    {
      "epoch": 0.20910800378911054,
      "grad_norm": 6.504414341179654e-05,
      "learning_rate": 7.910616578772499e-06,
      "loss": 0.0,
      "step": 14790
    },
    {
      "epoch": 0.2092493885110775,
      "grad_norm": 0.0,
      "learning_rate": 7.909202731552828e-06,
      "loss": 0.0,
      "step": 14800
    },
    {
      "epoch": 0.20939077323304445,
      "grad_norm": 0.0,
      "learning_rate": 7.907788884333159e-06,
      "loss": 0.0,
      "step": 14810
    },
    {
      "epoch": 0.20953215795501137,
      "grad_norm": 1.0340350854676217e-05,
      "learning_rate": 7.90637503711349e-06,
      "loss": 0.0,
      "step": 14820
    },
    {
      "epoch": 0.20967354267697833,
      "grad_norm": 0.0,
      "learning_rate": 7.90496118989382e-06,
      "loss": 0.684,
      "step": 14830
    },
    {
      "epoch": 0.20981492739894528,
      "grad_norm": 0.0,
      "learning_rate": 7.903547342674152e-06,
      "loss": 0.0,
      "step": 14840
    },
    {
      "epoch": 0.2099563121209122,
      "grad_norm": 0.0,
      "learning_rate": 7.90213349545448e-06,
      "loss": 0.0,
      "step": 14850
    },
    {
      "epoch": 0.21009769684287916,
      "grad_norm": 0.0,
      "learning_rate": 7.900719648234813e-06,
      "loss": 0.0001,
      "step": 14860
    },
    {
      "epoch": 0.2102390815648461,
      "grad_norm": 5.105671405792236,
      "learning_rate": 7.899305801015143e-06,
      "loss": 0.0009,
      "step": 14870
    },
    {
      "epoch": 0.21038046628681303,
      "grad_norm": 0.0,
      "learning_rate": 7.897891953795473e-06,
      "loss": 0.0293,
      "step": 14880
    },
    {
      "epoch": 0.21052185100878,
      "grad_norm": 0.0,
      "learning_rate": 7.896478106575804e-06,
      "loss": 0.1548,
      "step": 14890
    },
    {
      "epoch": 0.21066323573074694,
      "grad_norm": 0.0,
      "learning_rate": 7.895064259356135e-06,
      "loss": 0.0,
      "step": 14900
    },
    {
      "epoch": 0.2108046204527139,
      "grad_norm": 1.40136179993533e-07,
      "learning_rate": 7.893650412136466e-06,
      "loss": 0.0008,
      "step": 14910
    },
    {
      "epoch": 0.21094600517468082,
      "grad_norm": 0.0,
      "learning_rate": 7.892236564916795e-06,
      "loss": 0.0,
      "step": 14920
    },
    {
      "epoch": 0.21108738989664777,
      "grad_norm": 0.0,
      "learning_rate": 7.890822717697126e-06,
      "loss": 0.0003,
      "step": 14930
    },
    {
      "epoch": 0.21122877461861472,
      "grad_norm": 0.0,
      "learning_rate": 7.889408870477457e-06,
      "loss": 0.8375,
      "step": 14940
    },
    {
      "epoch": 0.21137015934058165,
      "grad_norm": 0.0,
      "learning_rate": 7.887995023257788e-06,
      "loss": 0.0,
      "step": 14950
    },
    {
      "epoch": 0.2115115440625486,
      "grad_norm": 0.0,
      "learning_rate": 7.886581176038119e-06,
      "loss": 0.0,
      "step": 14960
    },
    {
      "epoch": 0.21165292878451555,
      "grad_norm": 6.710956235167487e-09,
      "learning_rate": 7.885167328818448e-06,
      "loss": 0.0,
      "step": 14970
    },
    {
      "epoch": 0.21179431350648248,
      "grad_norm": 0.0,
      "learning_rate": 7.883753481598779e-06,
      "loss": 0.0006,
      "step": 14980
    },
    {
      "epoch": 0.21193569822844943,
      "grad_norm": 0.0,
      "learning_rate": 7.88233963437911e-06,
      "loss": 0.0,
      "step": 14990
    },
    {
      "epoch": 0.21207708295041638,
      "grad_norm": 0.0,
      "learning_rate": 7.88092578715944e-06,
      "loss": 0.8228,
      "step": 15000
    },
    {
      "epoch": 0.2122184676723833,
      "grad_norm": 9.342995326733217e-05,
      "learning_rate": 7.879511939939771e-06,
      "loss": 0.0,
      "step": 15010
    },
    {
      "epoch": 0.21235985239435026,
      "grad_norm": 1.0015273055330454e-09,
      "learning_rate": 7.878098092720102e-06,
      "loss": 0.0,
      "step": 15020
    },
    {
      "epoch": 0.21250123711631722,
      "grad_norm": 0.0,
      "learning_rate": 7.876684245500433e-06,
      "loss": 0.0001,
      "step": 15030
    },
    {
      "epoch": 0.21264262183828417,
      "grad_norm": 0.0,
      "learning_rate": 7.875270398280762e-06,
      "loss": 0.0,
      "step": 15040
    },
    {
      "epoch": 0.2127840065602511,
      "grad_norm": 0.0032139727845788,
      "learning_rate": 7.873856551061093e-06,
      "loss": 0.0,
      "step": 15050
    },
    {
      "epoch": 0.21292539128221805,
      "grad_norm": 0.0,
      "learning_rate": 7.872442703841424e-06,
      "loss": 0.4297,
      "step": 15060
    },
    {
      "epoch": 0.213066776004185,
      "grad_norm": 1.3008137941360474,
      "learning_rate": 7.871028856621755e-06,
      "loss": 0.368,
      "step": 15070
    },
    {
      "epoch": 0.21320816072615192,
      "grad_norm": 1.3885233784094453e-05,
      "learning_rate": 7.869615009402085e-06,
      "loss": 0.0377,
      "step": 15080
    },
    {
      "epoch": 0.21334954544811888,
      "grad_norm": 2.4150132958311588e-05,
      "learning_rate": 7.868201162182415e-06,
      "loss": 0.0346,
      "step": 15090
    },
    {
      "epoch": 0.21349093017008583,
      "grad_norm": 0.16601355373859406,
      "learning_rate": 7.866787314962745e-06,
      "loss": 0.0,
      "step": 15100
    },
    {
      "epoch": 0.21363231489205275,
      "grad_norm": 0.0,
      "learning_rate": 7.865373467743076e-06,
      "loss": 0.0,
      "step": 15110
    },
    {
      "epoch": 0.2137736996140197,
      "grad_norm": 0.0,
      "learning_rate": 7.863959620523407e-06,
      "loss": 0.0,
      "step": 15120
    },
    {
      "epoch": 0.21391508433598666,
      "grad_norm": 0.0,
      "learning_rate": 7.862545773303736e-06,
      "loss": 0.0,
      "step": 15130
    },
    {
      "epoch": 0.21405646905795359,
      "grad_norm": 173.95474243164062,
      "learning_rate": 7.861131926084067e-06,
      "loss": 0.2893,
      "step": 15140
    },
    {
      "epoch": 0.21419785377992054,
      "grad_norm": 0.0002379270299570635,
      "learning_rate": 7.8597180788644e-06,
      "loss": 0.0,
      "step": 15150
    },
    {
      "epoch": 0.2143392385018875,
      "grad_norm": 0.0,
      "learning_rate": 7.858304231644729e-06,
      "loss": 0.0,
      "step": 15160
    },
    {
      "epoch": 0.21448062322385444,
      "grad_norm": 0.0,
      "learning_rate": 7.85689038442506e-06,
      "loss": 0.0017,
      "step": 15170
    },
    {
      "epoch": 0.21462200794582137,
      "grad_norm": 0.0,
      "learning_rate": 7.85547653720539e-06,
      "loss": 0.0,
      "step": 15180
    },
    {
      "epoch": 0.21476339266778832,
      "grad_norm": 0.00010856680455617607,
      "learning_rate": 7.854062689985721e-06,
      "loss": 0.0001,
      "step": 15190
    },
    {
      "epoch": 0.21490477738975527,
      "grad_norm": 5.179504825036929e-09,
      "learning_rate": 7.852648842766052e-06,
      "loss": 0.0,
      "step": 15200
    },
    {
      "epoch": 0.2150461621117222,
      "grad_norm": 0.0,
      "learning_rate": 7.851234995546381e-06,
      "loss": 0.0074,
      "step": 15210
    },
    {
      "epoch": 0.21518754683368915,
      "grad_norm": 0.0,
      "learning_rate": 7.849821148326712e-06,
      "loss": 0.0008,
      "step": 15220
    },
    {
      "epoch": 0.2153289315556561,
      "grad_norm": 0.0,
      "learning_rate": 7.848407301107043e-06,
      "loss": 0.0,
      "step": 15230
    },
    {
      "epoch": 0.21547031627762303,
      "grad_norm": 0.0,
      "learning_rate": 7.846993453887374e-06,
      "loss": 0.0,
      "step": 15240
    },
    {
      "epoch": 0.21561170099958998,
      "grad_norm": 0.012158541940152645,
      "learning_rate": 7.845579606667703e-06,
      "loss": 0.0,
      "step": 15250
    },
    {
      "epoch": 0.21575308572155694,
      "grad_norm": 0.0,
      "learning_rate": 7.844165759448034e-06,
      "loss": 0.0,
      "step": 15260
    },
    {
      "epoch": 0.21589447044352386,
      "grad_norm": 1.79984635906294e-05,
      "learning_rate": 7.842751912228365e-06,
      "loss": 0.0001,
      "step": 15270
    },
    {
      "epoch": 0.21603585516549081,
      "grad_norm": 0.0,
      "learning_rate": 7.841338065008696e-06,
      "loss": 0.0,
      "step": 15280
    },
    {
      "epoch": 0.21617723988745777,
      "grad_norm": 0.31921544671058655,
      "learning_rate": 7.839924217789027e-06,
      "loss": 0.0,
      "step": 15290
    },
    {
      "epoch": 0.2163186246094247,
      "grad_norm": 0.00037113812868483365,
      "learning_rate": 7.838510370569356e-06,
      "loss": 0.0009,
      "step": 15300
    },
    {
      "epoch": 0.21646000933139165,
      "grad_norm": 818.7623901367188,
      "learning_rate": 7.837096523349688e-06,
      "loss": 0.3896,
      "step": 15310
    },
    {
      "epoch": 0.2166013940533586,
      "grad_norm": 0.0,
      "learning_rate": 7.835682676130017e-06,
      "loss": 0.0,
      "step": 15320
    },
    {
      "epoch": 0.21674277877532555,
      "grad_norm": 0.0,
      "learning_rate": 7.834268828910348e-06,
      "loss": 0.0,
      "step": 15330
    },
    {
      "epoch": 0.21688416349729248,
      "grad_norm": 0.0,
      "learning_rate": 7.83285498169068e-06,
      "loss": 0.0,
      "step": 15340
    },
    {
      "epoch": 0.21702554821925943,
      "grad_norm": 2.0836594104766846,
      "learning_rate": 7.83144113447101e-06,
      "loss": 0.0005,
      "step": 15350
    },
    {
      "epoch": 0.21716693294122638,
      "grad_norm": 0.0,
      "learning_rate": 7.830027287251341e-06,
      "loss": 0.1526,
      "step": 15360
    },
    {
      "epoch": 0.2173083176631933,
      "grad_norm": 0.00036569838994182646,
      "learning_rate": 7.82861344003167e-06,
      "loss": 0.0,
      "step": 15370
    },
    {
      "epoch": 0.21744970238516026,
      "grad_norm": 0.0,
      "learning_rate": 7.827199592812001e-06,
      "loss": 0.0003,
      "step": 15380
    },
    {
      "epoch": 0.2175910871071272,
      "grad_norm": 0.0,
      "learning_rate": 7.825785745592332e-06,
      "loss": 0.0,
      "step": 15390
    },
    {
      "epoch": 0.21773247182909414,
      "grad_norm": 2.0097002561669797e-05,
      "learning_rate": 7.824371898372663e-06,
      "loss": 0.0094,
      "step": 15400
    },
    {
      "epoch": 0.2178738565510611,
      "grad_norm": 0.0,
      "learning_rate": 7.822958051152994e-06,
      "loss": 0.0182,
      "step": 15410
    },
    {
      "epoch": 0.21801524127302804,
      "grad_norm": 0.0,
      "learning_rate": 7.821544203933323e-06,
      "loss": 0.0,
      "step": 15420
    },
    {
      "epoch": 0.21815662599499497,
      "grad_norm": 0.009611339308321476,
      "learning_rate": 7.820130356713655e-06,
      "loss": 0.0281,
      "step": 15430
    },
    {
      "epoch": 0.21829801071696192,
      "grad_norm": 1.519382635706279e-06,
      "learning_rate": 7.818716509493984e-06,
      "loss": 0.0,
      "step": 15440
    },
    {
      "epoch": 0.21843939543892887,
      "grad_norm": 0.0,
      "learning_rate": 7.817302662274315e-06,
      "loss": 0.0,
      "step": 15450
    },
    {
      "epoch": 0.21858078016089583,
      "grad_norm": 7.896234819781967e-06,
      "learning_rate": 7.815888815054646e-06,
      "loss": 0.0,
      "step": 15460
    },
    {
      "epoch": 0.21872216488286275,
      "grad_norm": 0.0,
      "learning_rate": 7.814474967834977e-06,
      "loss": 0.0018,
      "step": 15470
    },
    {
      "epoch": 0.2188635496048297,
      "grad_norm": 8.268356759799644e-05,
      "learning_rate": 7.813061120615308e-06,
      "loss": 0.0,
      "step": 15480
    },
    {
      "epoch": 0.21900493432679666,
      "grad_norm": 6.328043422598739e-09,
      "learning_rate": 7.811647273395637e-06,
      "loss": 0.0003,
      "step": 15490
    },
    {
      "epoch": 0.21914631904876358,
      "grad_norm": 0.0,
      "learning_rate": 7.810233426175968e-06,
      "loss": 0.0003,
      "step": 15500
    },
    {
      "epoch": 0.21928770377073054,
      "grad_norm": 3.023546923941467e-08,
      "learning_rate": 7.808819578956299e-06,
      "loss": 0.0,
      "step": 15510
    },
    {
      "epoch": 0.2194290884926975,
      "grad_norm": 0.0011232788674533367,
      "learning_rate": 7.80740573173663e-06,
      "loss": 0.0012,
      "step": 15520
    },
    {
      "epoch": 0.2195704732146644,
      "grad_norm": 1.460618841520045e-05,
      "learning_rate": 7.80599188451696e-06,
      "loss": 0.0021,
      "step": 15530
    },
    {
      "epoch": 0.21971185793663137,
      "grad_norm": 8.374451863346621e-05,
      "learning_rate": 7.80457803729729e-06,
      "loss": 0.0,
      "step": 15540
    },
    {
      "epoch": 0.21985324265859832,
      "grad_norm": 0.01859309896826744,
      "learning_rate": 7.80316419007762e-06,
      "loss": 0.0007,
      "step": 15550
    },
    {
      "epoch": 0.21999462738056524,
      "grad_norm": 4.056178113387432e-07,
      "learning_rate": 7.801750342857951e-06,
      "loss": 0.0,
      "step": 15560
    },
    {
      "epoch": 0.2201360121025322,
      "grad_norm": 0.0,
      "learning_rate": 7.800336495638282e-06,
      "loss": 0.1926,
      "step": 15570
    },
    {
      "epoch": 0.22027739682449915,
      "grad_norm": 9.831071656662971e-05,
      "learning_rate": 7.798922648418611e-06,
      "loss": 0.08,
      "step": 15580
    },
    {
      "epoch": 0.2204187815464661,
      "grad_norm": 0.00392174581065774,
      "learning_rate": 7.797508801198944e-06,
      "loss": 0.2436,
      "step": 15590
    },
    {
      "epoch": 0.22056016626843303,
      "grad_norm": 8.707856125056423e-08,
      "learning_rate": 7.796094953979275e-06,
      "loss": 0.0468,
      "step": 15600
    },
    {
      "epoch": 0.22070155099039998,
      "grad_norm": 0.0002962327853310853,
      "learning_rate": 7.794681106759604e-06,
      "loss": 0.6207,
      "step": 15610
    },
    {
      "epoch": 0.22084293571236693,
      "grad_norm": 7.161142434597423e-08,
      "learning_rate": 7.793267259539935e-06,
      "loss": 0.0,
      "step": 15620
    },
    {
      "epoch": 0.22098432043433386,
      "grad_norm": 0.001210339949466288,
      "learning_rate": 7.791853412320266e-06,
      "loss": 0.0853,
      "step": 15630
    },
    {
      "epoch": 0.2211257051563008,
      "grad_norm": 0.00046000437578186393,
      "learning_rate": 7.790439565100596e-06,
      "loss": 0.2365,
      "step": 15640
    },
    {
      "epoch": 0.22126708987826776,
      "grad_norm": 1.2262564268894494e-05,
      "learning_rate": 7.789025717880926e-06,
      "loss": 0.0741,
      "step": 15650
    },
    {
      "epoch": 0.2214084746002347,
      "grad_norm": 0.0007660693372599781,
      "learning_rate": 7.787611870661256e-06,
      "loss": 0.0006,
      "step": 15660
    },
    {
      "epoch": 0.22154985932220164,
      "grad_norm": 1.5083750213307212e-06,
      "learning_rate": 7.786198023441587e-06,
      "loss": 0.0,
      "step": 15670
    },
    {
      "epoch": 0.2216912440441686,
      "grad_norm": 0.00018248667765874416,
      "learning_rate": 7.784784176221918e-06,
      "loss": 0.0007,
      "step": 15680
    },
    {
      "epoch": 0.22183262876613552,
      "grad_norm": 1.3807598975290603e-07,
      "learning_rate": 7.783370329002249e-06,
      "loss": 0.0001,
      "step": 15690
    },
    {
      "epoch": 0.22197401348810247,
      "grad_norm": 6.134424666015548e-07,
      "learning_rate": 7.781956481782578e-06,
      "loss": 0.0,
      "step": 15700
    },
    {
      "epoch": 0.22211539821006943,
      "grad_norm": 0.0,
      "learning_rate": 7.780542634562909e-06,
      "loss": 0.0,
      "step": 15710
    },
    {
      "epoch": 0.22225678293203635,
      "grad_norm": 1.967760621113257e-09,
      "learning_rate": 7.779128787343242e-06,
      "loss": 0.0,
      "step": 15720
    },
    {
      "epoch": 0.2223981676540033,
      "grad_norm": 0.0,
      "learning_rate": 7.77771494012357e-06,
      "loss": 0.0,
      "step": 15730
    },
    {
      "epoch": 0.22253955237597026,
      "grad_norm": 0.00017304519133176655,
      "learning_rate": 7.776301092903902e-06,
      "loss": 0.0,
      "step": 15740
    },
    {
      "epoch": 0.2226809370979372,
      "grad_norm": 0.0,
      "learning_rate": 7.774887245684232e-06,
      "loss": 0.0,
      "step": 15750
    },
    {
      "epoch": 0.22282232181990413,
      "grad_norm": 0.0,
      "learning_rate": 7.773473398464563e-06,
      "loss": 0.0,
      "step": 15760
    },
    {
      "epoch": 0.2229637065418711,
      "grad_norm": 0.0,
      "learning_rate": 7.772059551244892e-06,
      "loss": 0.0002,
      "step": 15770
    },
    {
      "epoch": 0.22310509126383804,
      "grad_norm": 2.8602373731700936e-06,
      "learning_rate": 7.770645704025223e-06,
      "loss": 0.0,
      "step": 15780
    },
    {
      "epoch": 0.22324647598580497,
      "grad_norm": 0.0010616828221827745,
      "learning_rate": 7.769231856805554e-06,
      "loss": 0.0,
      "step": 15790
    },
    {
      "epoch": 0.22338786070777192,
      "grad_norm": 0.01217032503336668,
      "learning_rate": 7.767818009585885e-06,
      "loss": 0.0,
      "step": 15800
    },
    {
      "epoch": 0.22352924542973887,
      "grad_norm": 7.354527770075947e-05,
      "learning_rate": 7.766404162366216e-06,
      "loss": 0.6051,
      "step": 15810
    },
    {
      "epoch": 0.2236706301517058,
      "grad_norm": 0.0,
      "learning_rate": 7.764990315146545e-06,
      "loss": 0.0,
      "step": 15820
    },
    {
      "epoch": 0.22381201487367275,
      "grad_norm": 0.0,
      "learning_rate": 7.763576467926876e-06,
      "loss": 0.0,
      "step": 15830
    },
    {
      "epoch": 0.2239533995956397,
      "grad_norm": 0.0,
      "learning_rate": 7.762162620707207e-06,
      "loss": 0.0023,
      "step": 15840
    },
    {
      "epoch": 0.22409478431760663,
      "grad_norm": 0.0,
      "learning_rate": 7.760748773487538e-06,
      "loss": 0.0023,
      "step": 15850
    },
    {
      "epoch": 0.22423616903957358,
      "grad_norm": 0.0,
      "learning_rate": 7.759334926267868e-06,
      "loss": 0.0,
      "step": 15860
    },
    {
      "epoch": 0.22437755376154053,
      "grad_norm": 0.00019745112513191998,
      "learning_rate": 7.757921079048198e-06,
      "loss": 0.1745,
      "step": 15870
    },
    {
      "epoch": 0.22451893848350749,
      "grad_norm": 33.07639694213867,
      "learning_rate": 7.75650723182853e-06,
      "loss": 0.0119,
      "step": 15880
    },
    {
      "epoch": 0.2246603232054744,
      "grad_norm": 0.8473035097122192,
      "learning_rate": 7.75509338460886e-06,
      "loss": 0.0687,
      "step": 15890
    },
    {
      "epoch": 0.22480170792744136,
      "grad_norm": 4.012678971321293e-07,
      "learning_rate": 7.75367953738919e-06,
      "loss": 0.0,
      "step": 15900
    },
    {
      "epoch": 0.22494309264940832,
      "grad_norm": 0.0,
      "learning_rate": 7.752265690169521e-06,
      "loss": 0.0,
      "step": 15910
    },
    {
      "epoch": 0.22508447737137524,
      "grad_norm": 7.858971002860926e-06,
      "learning_rate": 7.750851842949852e-06,
      "loss": 0.5008,
      "step": 15920
    },
    {
      "epoch": 0.2252258620933422,
      "grad_norm": 0.009890151210129261,
      "learning_rate": 7.749437995730183e-06,
      "loss": 0.0,
      "step": 15930
    },
    {
      "epoch": 0.22536724681530915,
      "grad_norm": 0.019279371947050095,
      "learning_rate": 7.748024148510512e-06,
      "loss": 0.0033,
      "step": 15940
    },
    {
      "epoch": 0.22550863153727607,
      "grad_norm": 0.0,
      "learning_rate": 7.746610301290843e-06,
      "loss": 0.0004,
      "step": 15950
    },
    {
      "epoch": 0.22565001625924302,
      "grad_norm": 384.0064392089844,
      "learning_rate": 7.745196454071174e-06,
      "loss": 0.21,
      "step": 15960
    },
    {
      "epoch": 0.22579140098120998,
      "grad_norm": 0.0,
      "learning_rate": 7.743782606851505e-06,
      "loss": 0.0058,
      "step": 15970
    },
    {
      "epoch": 0.2259327857031769,
      "grad_norm": 1.5828705102194363e-07,
      "learning_rate": 7.742368759631835e-06,
      "loss": 0.0,
      "step": 15980
    },
    {
      "epoch": 0.22607417042514386,
      "grad_norm": 2.3995486117200926e-05,
      "learning_rate": 7.740954912412165e-06,
      "loss": 0.0,
      "step": 15990
    },
    {
      "epoch": 0.2262155551471108,
      "grad_norm": 0.0,
      "learning_rate": 7.739541065192497e-06,
      "loss": 0.0,
      "step": 16000
    },
    {
      "epoch": 0.22635693986907776,
      "grad_norm": 6.950689315795898,
      "learning_rate": 7.738127217972826e-06,
      "loss": 0.004,
      "step": 16010
    },
    {
      "epoch": 0.2264983245910447,
      "grad_norm": 0.0,
      "learning_rate": 7.736713370753157e-06,
      "loss": 0.0007,
      "step": 16020
    },
    {
      "epoch": 0.22663970931301164,
      "grad_norm": 0.0,
      "learning_rate": 7.735299523533486e-06,
      "loss": 0.2449,
      "step": 16030
    },
    {
      "epoch": 0.2267810940349786,
      "grad_norm": 0.0,
      "learning_rate": 7.733885676313819e-06,
      "loss": 0.0,
      "step": 16040
    },
    {
      "epoch": 0.22692247875694552,
      "grad_norm": 0.0,
      "learning_rate": 7.73247182909415e-06,
      "loss": 0.0,
      "step": 16050
    },
    {
      "epoch": 0.22706386347891247,
      "grad_norm": 0.0,
      "learning_rate": 7.731057981874479e-06,
      "loss": 0.0001,
      "step": 16060
    },
    {
      "epoch": 0.22720524820087942,
      "grad_norm": 6.059000952518545e-05,
      "learning_rate": 7.72964413465481e-06,
      "loss": 0.0,
      "step": 16070
    },
    {
      "epoch": 0.22734663292284635,
      "grad_norm": 7.676234235987067e-05,
      "learning_rate": 7.72823028743514e-06,
      "loss": 0.0,
      "step": 16080
    },
    {
      "epoch": 0.2274880176448133,
      "grad_norm": 0.0,
      "learning_rate": 7.726816440215471e-06,
      "loss": 0.0001,
      "step": 16090
    },
    {
      "epoch": 0.22762940236678025,
      "grad_norm": 0.0,
      "learning_rate": 7.7254025929958e-06,
      "loss": 0.0,
      "step": 16100
    },
    {
      "epoch": 0.22777078708874718,
      "grad_norm": 5.408024250641574e-09,
      "learning_rate": 7.723988745776131e-06,
      "loss": 0.0021,
      "step": 16110
    },
    {
      "epoch": 0.22791217181071413,
      "grad_norm": 0.8439585566520691,
      "learning_rate": 7.722574898556462e-06,
      "loss": 0.0002,
      "step": 16120
    },
    {
      "epoch": 0.22805355653268108,
      "grad_norm": 0.0010487654944881797,
      "learning_rate": 7.721161051336793e-06,
      "loss": 0.0103,
      "step": 16130
    },
    {
      "epoch": 0.228194941254648,
      "grad_norm": 0.0,
      "learning_rate": 7.719747204117124e-06,
      "loss": 0.0001,
      "step": 16140
    },
    {
      "epoch": 0.22833632597661496,
      "grad_norm": 4.31095215081001e-10,
      "learning_rate": 7.718333356897453e-06,
      "loss": 0.0003,
      "step": 16150
    },
    {
      "epoch": 0.22847771069858192,
      "grad_norm": 2.4146247223910677e-09,
      "learning_rate": 7.716919509677786e-06,
      "loss": 0.0,
      "step": 16160
    },
    {
      "epoch": 0.22861909542054887,
      "grad_norm": 4.660934971312258e-10,
      "learning_rate": 7.715505662458117e-06,
      "loss": 0.0,
      "step": 16170
    },
    {
      "epoch": 0.2287604801425158,
      "grad_norm": 0.0,
      "learning_rate": 7.714091815238446e-06,
      "loss": 0.0,
      "step": 16180
    },
    {
      "epoch": 0.22890186486448275,
      "grad_norm": 0.0,
      "learning_rate": 7.712677968018777e-06,
      "loss": 0.0,
      "step": 16190
    },
    {
      "epoch": 0.2290432495864497,
      "grad_norm": 3.363616372098477e-07,
      "learning_rate": 7.711264120799107e-06,
      "loss": 0.0174,
      "step": 16200
    },
    {
      "epoch": 0.22918463430841662,
      "grad_norm": 9.51656723022461,
      "learning_rate": 7.709850273579438e-06,
      "loss": 0.0014,
      "step": 16210
    },
    {
      "epoch": 0.22932601903038358,
      "grad_norm": 0.0,
      "learning_rate": 7.708436426359767e-06,
      "loss": 0.0,
      "step": 16220
    },
    {
      "epoch": 0.22946740375235053,
      "grad_norm": 0.005272713955491781,
      "learning_rate": 7.707022579140098e-06,
      "loss": 0.0,
      "step": 16230
    },
    {
      "epoch": 0.22960878847431745,
      "grad_norm": 2.716358267207397e-07,
      "learning_rate": 7.70560873192043e-06,
      "loss": 0.0,
      "step": 16240
    },
    {
      "epoch": 0.2297501731962844,
      "grad_norm": 0.00013388549268711358,
      "learning_rate": 7.70419488470076e-06,
      "loss": 0.0,
      "step": 16250
    },
    {
      "epoch": 0.22989155791825136,
      "grad_norm": 0.0,
      "learning_rate": 7.702781037481091e-06,
      "loss": 0.0008,
      "step": 16260
    },
    {
      "epoch": 0.23003294264021829,
      "grad_norm": 0.48999086022377014,
      "learning_rate": 7.70136719026142e-06,
      "loss": 0.0001,
      "step": 16270
    },
    {
      "epoch": 0.23017432736218524,
      "grad_norm": 0.0,
      "learning_rate": 7.699953343041751e-06,
      "loss": 0.0,
      "step": 16280
    },
    {
      "epoch": 0.2303157120841522,
      "grad_norm": 4.787600316191742e-10,
      "learning_rate": 7.698539495822082e-06,
      "loss": 0.0053,
      "step": 16290
    },
    {
      "epoch": 0.23045709680611914,
      "grad_norm": 0.0,
      "learning_rate": 7.697125648602413e-06,
      "loss": 0.0001,
      "step": 16300
    },
    {
      "epoch": 0.23059848152808607,
      "grad_norm": 0.0,
      "learning_rate": 7.695711801382743e-06,
      "loss": 0.0,
      "step": 16310
    },
    {
      "epoch": 0.23073986625005302,
      "grad_norm": 0.0,
      "learning_rate": 7.694297954163074e-06,
      "loss": 0.0001,
      "step": 16320
    },
    {
      "epoch": 0.23088125097201997,
      "grad_norm": 0.0,
      "learning_rate": 7.692884106943405e-06,
      "loss": 0.0,
      "step": 16330
    },
    {
      "epoch": 0.2310226356939869,
      "grad_norm": 4.936647091113855e-09,
      "learning_rate": 7.691470259723734e-06,
      "loss": 0.0,
      "step": 16340
    },
    {
      "epoch": 0.23116402041595385,
      "grad_norm": 0.0,
      "learning_rate": 7.690056412504065e-06,
      "loss": 0.0,
      "step": 16350
    },
    {
      "epoch": 0.2313054051379208,
      "grad_norm": 0.0,
      "learning_rate": 7.688642565284396e-06,
      "loss": 0.0,
      "step": 16360
    },
    {
      "epoch": 0.23144678985988773,
      "grad_norm": 0.0,
      "learning_rate": 7.687228718064727e-06,
      "loss": 0.0,
      "step": 16370
    },
    {
      "epoch": 0.23158817458185468,
      "grad_norm": 0.0,
      "learning_rate": 7.685814870845058e-06,
      "loss": 0.0,
      "step": 16380
    },
    {
      "epoch": 0.23172955930382164,
      "grad_norm": 0.0,
      "learning_rate": 7.684401023625387e-06,
      "loss": 0.0002,
      "step": 16390
    },
    {
      "epoch": 0.23187094402578856,
      "grad_norm": 0.0,
      "learning_rate": 7.682987176405718e-06,
      "loss": 0.0,
      "step": 16400
    },
    {
      "epoch": 0.23201232874775551,
      "grad_norm": 0.0003527377266436815,
      "learning_rate": 7.681573329186049e-06,
      "loss": 0.0001,
      "step": 16410
    },
    {
      "epoch": 0.23215371346972247,
      "grad_norm": 1.527001404610928e-05,
      "learning_rate": 7.68015948196638e-06,
      "loss": 0.0,
      "step": 16420
    },
    {
      "epoch": 0.23229509819168942,
      "grad_norm": 6.716198186040856e-06,
      "learning_rate": 7.67874563474671e-06,
      "loss": 0.0,
      "step": 16430
    },
    {
      "epoch": 0.23243648291365634,
      "grad_norm": 0.009650299325585365,
      "learning_rate": 7.67733178752704e-06,
      "loss": 0.0014,
      "step": 16440
    },
    {
      "epoch": 0.2325778676356233,
      "grad_norm": 2.692001714876824e-07,
      "learning_rate": 7.675917940307372e-06,
      "loss": 0.0152,
      "step": 16450
    },
    {
      "epoch": 0.23271925235759025,
      "grad_norm": 0.008912213146686554,
      "learning_rate": 7.674504093087701e-06,
      "loss": 0.0,
      "step": 16460
    },
    {
      "epoch": 0.23286063707955718,
      "grad_norm": 0.0,
      "learning_rate": 7.673090245868032e-06,
      "loss": 0.0,
      "step": 16470
    },
    {
      "epoch": 0.23300202180152413,
      "grad_norm": 0.0,
      "learning_rate": 7.671676398648363e-06,
      "loss": 0.0001,
      "step": 16480
    },
    {
      "epoch": 0.23314340652349108,
      "grad_norm": 0.0,
      "learning_rate": 7.670262551428694e-06,
      "loss": 0.0,
      "step": 16490
    },
    {
      "epoch": 0.233284791245458,
      "grad_norm": 0.0,
      "learning_rate": 7.668848704209025e-06,
      "loss": 0.0,
      "step": 16500
    },
    {
      "epoch": 0.23342617596742496,
      "grad_norm": 4.820749381906353e-05,
      "learning_rate": 7.667434856989354e-06,
      "loss": 0.022,
      "step": 16510
    },
    {
      "epoch": 0.2335675606893919,
      "grad_norm": 0.0,
      "learning_rate": 7.666021009769685e-06,
      "loss": 0.0,
      "step": 16520
    },
    {
      "epoch": 0.23370894541135884,
      "grad_norm": 0.017151763662695885,
      "learning_rate": 7.664607162550016e-06,
      "loss": 0.0,
      "step": 16530
    },
    {
      "epoch": 0.2338503301333258,
      "grad_norm": 0.0,
      "learning_rate": 7.663193315330346e-06,
      "loss": 0.0024,
      "step": 16540
    },
    {
      "epoch": 0.23399171485529274,
      "grad_norm": 9.263195765996102e-10,
      "learning_rate": 7.661779468110676e-06,
      "loss": 0.0019,
      "step": 16550
    },
    {
      "epoch": 0.23413309957725967,
      "grad_norm": 0.0,
      "learning_rate": 7.660365620891006e-06,
      "loss": 0.0,
      "step": 16560
    },
    {
      "epoch": 0.23427448429922662,
      "grad_norm": 1.4724590755577083e-06,
      "learning_rate": 7.658951773671339e-06,
      "loss": 0.0873,
      "step": 16570
    },
    {
      "epoch": 0.23441586902119357,
      "grad_norm": 2.3020900741244077e-09,
      "learning_rate": 7.657537926451668e-06,
      "loss": 0.0002,
      "step": 16580
    },
    {
      "epoch": 0.23455725374316053,
      "grad_norm": 0.0,
      "learning_rate": 7.656124079231999e-06,
      "loss": 0.0,
      "step": 16590
    },
    {
      "epoch": 0.23469863846512745,
      "grad_norm": 0.03874632343649864,
      "learning_rate": 7.654710232012328e-06,
      "loss": 0.0,
      "step": 16600
    },
    {
      "epoch": 0.2348400231870944,
      "grad_norm": 0.0,
      "learning_rate": 7.65329638479266e-06,
      "loss": 0.0,
      "step": 16610
    },
    {
      "epoch": 0.23498140790906136,
      "grad_norm": 0.0,
      "learning_rate": 7.651882537572992e-06,
      "loss": 0.0002,
      "step": 16620
    },
    {
      "epoch": 0.23512279263102828,
      "grad_norm": 247.10252380371094,
      "learning_rate": 7.65046869035332e-06,
      "loss": 0.075,
      "step": 16630
    },
    {
      "epoch": 0.23526417735299524,
      "grad_norm": 0.0,
      "learning_rate": 7.649054843133652e-06,
      "loss": 0.0001,
      "step": 16640
    },
    {
      "epoch": 0.2354055620749622,
      "grad_norm": 0.014272580854594707,
      "learning_rate": 7.647640995913982e-06,
      "loss": 0.0004,
      "step": 16650
    },
    {
      "epoch": 0.2355469467969291,
      "grad_norm": 0.0,
      "learning_rate": 7.646227148694313e-06,
      "loss": 0.0,
      "step": 16660
    },
    {
      "epoch": 0.23568833151889607,
      "grad_norm": 0.0,
      "learning_rate": 7.644813301474642e-06,
      "loss": 0.0074,
      "step": 16670
    },
    {
      "epoch": 0.23582971624086302,
      "grad_norm": 1.1708904821716715e-05,
      "learning_rate": 7.643399454254973e-06,
      "loss": 0.0028,
      "step": 16680
    },
    {
      "epoch": 0.23597110096282994,
      "grad_norm": 5.820841586512415e-09,
      "learning_rate": 7.641985607035304e-06,
      "loss": 0.0003,
      "step": 16690
    },
    {
      "epoch": 0.2361124856847969,
      "grad_norm": 0.0,
      "learning_rate": 7.640571759815635e-06,
      "loss": 0.0,
      "step": 16700
    },
    {
      "epoch": 0.23625387040676385,
      "grad_norm": 1.265174388885498,
      "learning_rate": 7.639157912595966e-06,
      "loss": 0.0405,
      "step": 16710
    },
    {
      "epoch": 0.2363952551287308,
      "grad_norm": 0.0,
      "learning_rate": 7.637744065376295e-06,
      "loss": 0.0356,
      "step": 16720
    },
    {
      "epoch": 0.23653663985069773,
      "grad_norm": 9.08695563239803e-10,
      "learning_rate": 7.636330218156628e-06,
      "loss": 0.0001,
      "step": 16730
    },
    {
      "epoch": 0.23667802457266468,
      "grad_norm": 0.0,
      "learning_rate": 7.634916370936957e-06,
      "loss": 0.0008,
      "step": 16740
    },
    {
      "epoch": 0.23681940929463163,
      "grad_norm": 0.0,
      "learning_rate": 7.633502523717288e-06,
      "loss": 0.0156,
      "step": 16750
    },
    {
      "epoch": 0.23696079401659856,
      "grad_norm": 6.073784606996924e-05,
      "learning_rate": 7.632088676497618e-06,
      "loss": 0.0,
      "step": 16760
    },
    {
      "epoch": 0.2371021787385655,
      "grad_norm": 9.642614895710722e-05,
      "learning_rate": 7.63067482927795e-06,
      "loss": 0.0,
      "step": 16770
    },
    {
      "epoch": 0.23724356346053246,
      "grad_norm": 3.797436193053727e-06,
      "learning_rate": 7.629260982058279e-06,
      "loss": 0.0,
      "step": 16780
    },
    {
      "epoch": 0.2373849481824994,
      "grad_norm": 2.525410491216462e-05,
      "learning_rate": 7.627847134838609e-06,
      "loss": 0.0,
      "step": 16790
    },
    {
      "epoch": 0.23752633290446634,
      "grad_norm": 8.439864905085415e-05,
      "learning_rate": 7.626433287618941e-06,
      "loss": 0.0,
      "step": 16800
    },
    {
      "epoch": 0.2376677176264333,
      "grad_norm": 4.739465657621622e-05,
      "learning_rate": 7.62501944039927e-06,
      "loss": 0.0017,
      "step": 16810
    },
    {
      "epoch": 0.23780910234840022,
      "grad_norm": 9.467141515173694e-10,
      "learning_rate": 7.623605593179602e-06,
      "loss": 0.0713,
      "step": 16820
    },
    {
      "epoch": 0.23795048707036717,
      "grad_norm": 0.02165275067090988,
      "learning_rate": 7.622191745959933e-06,
      "loss": 0.0004,
      "step": 16830
    },
    {
      "epoch": 0.23809187179233413,
      "grad_norm": 0.0,
      "learning_rate": 7.620777898740263e-06,
      "loss": 0.0,
      "step": 16840
    },
    {
      "epoch": 0.23823325651430108,
      "grad_norm": 0.0,
      "learning_rate": 7.619364051520594e-06,
      "loss": 0.0001,
      "step": 16850
    },
    {
      "epoch": 0.238374641236268,
      "grad_norm": 0.0,
      "learning_rate": 7.617950204300924e-06,
      "loss": 0.0322,
      "step": 16860
    },
    {
      "epoch": 0.23851602595823496,
      "grad_norm": 2.5425032745829412e-08,
      "learning_rate": 7.6165363570812545e-06,
      "loss": 0.0001,
      "step": 16870
    },
    {
      "epoch": 0.2386574106802019,
      "grad_norm": 0.00029088868177495897,
      "learning_rate": 7.615122509861585e-06,
      "loss": 0.0,
      "step": 16880
    },
    {
      "epoch": 0.23879879540216883,
      "grad_norm": 0.0,
      "learning_rate": 7.613708662641915e-06,
      "loss": 0.0,
      "step": 16890
    },
    {
      "epoch": 0.2389401801241358,
      "grad_norm": 0.08469953387975693,
      "learning_rate": 7.612294815422246e-06,
      "loss": 0.0,
      "step": 16900
    },
    {
      "epoch": 0.23908156484610274,
      "grad_norm": 0.0,
      "learning_rate": 7.610880968202576e-06,
      "loss": 0.0001,
      "step": 16910
    },
    {
      "epoch": 0.23922294956806966,
      "grad_norm": 0.009634886868298054,
      "learning_rate": 7.609467120982907e-06,
      "loss": 0.0,
      "step": 16920
    },
    {
      "epoch": 0.23936433429003662,
      "grad_norm": 0.0,
      "learning_rate": 7.608053273763237e-06,
      "loss": 0.0001,
      "step": 16930
    },
    {
      "epoch": 0.23950571901200357,
      "grad_norm": 0.0,
      "learning_rate": 7.606639426543568e-06,
      "loss": 0.0,
      "step": 16940
    },
    {
      "epoch": 0.2396471037339705,
      "grad_norm": 0.0,
      "learning_rate": 7.6052255793239e-06,
      "loss": 0.0,
      "step": 16950
    },
    {
      "epoch": 0.23978848845593745,
      "grad_norm": 0.0,
      "learning_rate": 7.60381173210423e-06,
      "loss": 0.0131,
      "step": 16960
    },
    {
      "epoch": 0.2399298731779044,
      "grad_norm": 0.0,
      "learning_rate": 7.6023978848845605e-06,
      "loss": 0.4652,
      "step": 16970
    },
    {
      "epoch": 0.24007125789987133,
      "grad_norm": 0.0,
      "learning_rate": 7.6009840376648905e-06,
      "loss": 1.4234,
      "step": 16980
    },
    {
      "epoch": 0.24021264262183828,
      "grad_norm": 0.0,
      "learning_rate": 7.599570190445221e-06,
      "loss": 0.0207,
      "step": 16990
    },
    {
      "epoch": 0.24035402734380523,
      "grad_norm": 0.0,
      "learning_rate": 7.598156343225551e-06,
      "loss": 0.0001,
      "step": 17000
    },
    {
      "epoch": 0.24049541206577218,
      "grad_norm": 1.1756892126513208e-09,
      "learning_rate": 7.596742496005882e-06,
      "loss": 0.0,
      "step": 17010
    },
    {
      "epoch": 0.2406367967877391,
      "grad_norm": 0.0,
      "learning_rate": 7.595328648786213e-06,
      "loss": 0.003,
      "step": 17020
    },
    {
      "epoch": 0.24077818150970606,
      "grad_norm": 0.0,
      "learning_rate": 7.593914801566543e-06,
      "loss": 0.0016,
      "step": 17030
    },
    {
      "epoch": 0.24091956623167302,
      "grad_norm": 0.0,
      "learning_rate": 7.592500954346874e-06,
      "loss": 0.0001,
      "step": 17040
    },
    {
      "epoch": 0.24106095095363994,
      "grad_norm": 0.0,
      "learning_rate": 7.591087107127204e-06,
      "loss": 0.0,
      "step": 17050
    },
    {
      "epoch": 0.2412023356756069,
      "grad_norm": 4.2952663648065936e-07,
      "learning_rate": 7.589673259907535e-06,
      "loss": 0.0,
      "step": 17060
    },
    {
      "epoch": 0.24134372039757385,
      "grad_norm": 0.0015041049337014556,
      "learning_rate": 7.588259412687865e-06,
      "loss": 0.0,
      "step": 17070
    },
    {
      "epoch": 0.24148510511954077,
      "grad_norm": 0.001684025744907558,
      "learning_rate": 7.586845565468196e-06,
      "loss": 0.0,
      "step": 17080
    },
    {
      "epoch": 0.24162648984150772,
      "grad_norm": 0.0022822702303528786,
      "learning_rate": 7.585431718248527e-06,
      "loss": 0.0,
      "step": 17090
    },
    {
      "epoch": 0.24176787456347468,
      "grad_norm": 0.0,
      "learning_rate": 7.5840178710288574e-06,
      "loss": 0.0,
      "step": 17100
    },
    {
      "epoch": 0.2419092592854416,
      "grad_norm": 7.1166155066748615e-06,
      "learning_rate": 7.582604023809188e-06,
      "loss": 0.0,
      "step": 17110
    },
    {
      "epoch": 0.24205064400740856,
      "grad_norm": 0.0,
      "learning_rate": 7.581190176589518e-06,
      "loss": 0.003,
      "step": 17120
    },
    {
      "epoch": 0.2421920287293755,
      "grad_norm": 0.023437492549419403,
      "learning_rate": 7.579776329369849e-06,
      "loss": 0.0,
      "step": 17130
    },
    {
      "epoch": 0.24233341345134246,
      "grad_norm": 2.780109571176581e-05,
      "learning_rate": 7.57836248215018e-06,
      "loss": 0.1137,
      "step": 17140
    },
    {
      "epoch": 0.24247479817330939,
      "grad_norm": 4.7632301658495635e-08,
      "learning_rate": 7.57694863493051e-06,
      "loss": 0.0005,
      "step": 17150
    },
    {
      "epoch": 0.24261618289527634,
      "grad_norm": 0.0006923701148480177,
      "learning_rate": 7.575534787710841e-06,
      "loss": 0.0,
      "step": 17160
    },
    {
      "epoch": 0.2427575676172433,
      "grad_norm": 0.0,
      "learning_rate": 7.574120940491171e-06,
      "loss": 0.5295,
      "step": 17170
    },
    {
      "epoch": 0.24289895233921022,
      "grad_norm": 0.0,
      "learning_rate": 7.572707093271502e-06,
      "loss": 0.0,
      "step": 17180
    },
    {
      "epoch": 0.24304033706117717,
      "grad_norm": 2.791028919091332e-07,
      "learning_rate": 7.571293246051832e-06,
      "loss": 0.0,
      "step": 17190
    },
    {
      "epoch": 0.24318172178314412,
      "grad_norm": 0.0,
      "learning_rate": 7.569879398832163e-06,
      "loss": 0.0,
      "step": 17200
    },
    {
      "epoch": 0.24332310650511105,
      "grad_norm": 1.8889709796354737e-09,
      "learning_rate": 7.5684655516124935e-06,
      "loss": 0.0002,
      "step": 17210
    },
    {
      "epoch": 0.243464491227078,
      "grad_norm": 5.543108272831887e-05,
      "learning_rate": 7.5670517043928235e-06,
      "loss": 0.0,
      "step": 17220
    },
    {
      "epoch": 0.24360587594904495,
      "grad_norm": 0.0,
      "learning_rate": 7.565637857173155e-06,
      "loss": 0.0031,
      "step": 17230
    },
    {
      "epoch": 0.24374726067101188,
      "grad_norm": 56.55934143066406,
      "learning_rate": 7.564224009953484e-06,
      "loss": 0.0056,
      "step": 17240
    },
    {
      "epoch": 0.24388864539297883,
      "grad_norm": 8.853492181515321e-05,
      "learning_rate": 7.562810162733816e-06,
      "loss": 0.0001,
      "step": 17250
    },
    {
      "epoch": 0.24403003011494578,
      "grad_norm": 0.003909654915332794,
      "learning_rate": 7.561396315514146e-06,
      "loss": 0.0,
      "step": 17260
    },
    {
      "epoch": 0.24417141483691274,
      "grad_norm": 4.4435549684251896e-10,
      "learning_rate": 7.559982468294477e-06,
      "loss": 0.0,
      "step": 17270
    },
    {
      "epoch": 0.24431279955887966,
      "grad_norm": 0.07372479140758514,
      "learning_rate": 7.558568621074808e-06,
      "loss": 0.0001,
      "step": 17280
    },
    {
      "epoch": 0.24445418428084661,
      "grad_norm": 493.7487487792969,
      "learning_rate": 7.557154773855138e-06,
      "loss": 0.347,
      "step": 17290
    },
    {
      "epoch": 0.24459556900281357,
      "grad_norm": 0.0,
      "learning_rate": 7.555740926635469e-06,
      "loss": 0.0,
      "step": 17300
    },
    {
      "epoch": 0.2447369537247805,
      "grad_norm": 0.0,
      "learning_rate": 7.554327079415799e-06,
      "loss": 0.2517,
      "step": 17310
    },
    {
      "epoch": 0.24487833844674745,
      "grad_norm": 0.0,
      "learning_rate": 7.5529132321961295e-06,
      "loss": 0.0002,
      "step": 17320
    },
    {
      "epoch": 0.2450197231687144,
      "grad_norm": 0.0,
      "learning_rate": 7.5514993849764595e-06,
      "loss": 0.0004,
      "step": 17330
    },
    {
      "epoch": 0.24516110789068132,
      "grad_norm": 0.0,
      "learning_rate": 7.55008553775679e-06,
      "loss": 0.0,
      "step": 17340
    },
    {
      "epoch": 0.24530249261264828,
      "grad_norm": 0.0007993951439857483,
      "learning_rate": 7.548671690537121e-06,
      "loss": 0.0,
      "step": 17350
    },
    {
      "epoch": 0.24544387733461523,
      "grad_norm": 0.0,
      "learning_rate": 7.547257843317451e-06,
      "loss": 0.0,
      "step": 17360
    },
    {
      "epoch": 0.24558526205658215,
      "grad_norm": 6.759959433111362e-06,
      "learning_rate": 7.545843996097783e-06,
      "loss": 0.0,
      "step": 17370
    },
    {
      "epoch": 0.2457266467785491,
      "grad_norm": 0.0,
      "learning_rate": 7.544430148878112e-06,
      "loss": 0.0,
      "step": 17380
    },
    {
      "epoch": 0.24586803150051606,
      "grad_norm": 1.3261726508062566e-06,
      "learning_rate": 7.543016301658444e-06,
      "loss": 0.0,
      "step": 17390
    },
    {
      "epoch": 0.24600941622248299,
      "grad_norm": 1.9772748947143555,
      "learning_rate": 7.541602454438775e-06,
      "loss": 0.0018,
      "step": 17400
    },
    {
      "epoch": 0.24615080094444994,
      "grad_norm": 1.797180402718368e-06,
      "learning_rate": 7.540188607219105e-06,
      "loss": 0.0025,
      "step": 17410
    },
    {
      "epoch": 0.2462921856664169,
      "grad_norm": 0.0,
      "learning_rate": 7.5387747599994355e-06,
      "loss": 0.0,
      "step": 17420
    },
    {
      "epoch": 0.24643357038838384,
      "grad_norm": 0.0001141618995461613,
      "learning_rate": 7.5373609127797655e-06,
      "loss": 0.0,
      "step": 17430
    },
    {
      "epoch": 0.24657495511035077,
      "grad_norm": 0.0,
      "learning_rate": 7.535947065560096e-06,
      "loss": 0.0001,
      "step": 17440
    },
    {
      "epoch": 0.24671633983231772,
      "grad_norm": 0.0,
      "learning_rate": 7.534533218340426e-06,
      "loss": 0.0,
      "step": 17450
    },
    {
      "epoch": 0.24685772455428467,
      "grad_norm": 0.0,
      "learning_rate": 7.533119371120757e-06,
      "loss": 0.0,
      "step": 17460
    },
    {
      "epoch": 0.2469991092762516,
      "grad_norm": 1.4866112962863554e-07,
      "learning_rate": 7.531705523901088e-06,
      "loss": 0.0,
      "step": 17470
    },
    {
      "epoch": 0.24714049399821855,
      "grad_norm": 1.3350400251965766e-07,
      "learning_rate": 7.530291676681418e-06,
      "loss": 0.0,
      "step": 17480
    },
    {
      "epoch": 0.2472818787201855,
      "grad_norm": 0.0,
      "learning_rate": 7.528877829461749e-06,
      "loss": 0.0,
      "step": 17490
    },
    {
      "epoch": 0.24742326344215243,
      "grad_norm": 0.0,
      "learning_rate": 7.527463982242079e-06,
      "loss": 0.0003,
      "step": 17500
    },
    {
      "epoch": 0.24756464816411938,
      "grad_norm": 0.0,
      "learning_rate": 7.52605013502241e-06,
      "loss": 0.0,
      "step": 17510
    },
    {
      "epoch": 0.24770603288608634,
      "grad_norm": 0.0,
      "learning_rate": 7.52463628780274e-06,
      "loss": 0.0,
      "step": 17520
    },
    {
      "epoch": 0.24784741760805326,
      "grad_norm": 0.0,
      "learning_rate": 7.5232224405830715e-06,
      "loss": 0.0,
      "step": 17530
    },
    {
      "epoch": 0.2479888023300202,
      "grad_norm": 0.00011001788516296074,
      "learning_rate": 7.521808593363402e-06,
      "loss": 0.0,
      "step": 17540
    },
    {
      "epoch": 0.24813018705198717,
      "grad_norm": 0.0,
      "learning_rate": 7.520394746143732e-06,
      "loss": 0.0,
      "step": 17550
    },
    {
      "epoch": 0.24827157177395412,
      "grad_norm": 0.0,
      "learning_rate": 7.518980898924063e-06,
      "loss": 0.0,
      "step": 17560
    },
    {
      "epoch": 0.24841295649592104,
      "grad_norm": 0.0,
      "learning_rate": 7.517567051704393e-06,
      "loss": 0.0022,
      "step": 17570
    },
    {
      "epoch": 0.248554341217888,
      "grad_norm": 0.0,
      "learning_rate": 7.516153204484724e-06,
      "loss": 0.0,
      "step": 17580
    },
    {
      "epoch": 0.24869572593985495,
      "grad_norm": 2.132098416041117e-05,
      "learning_rate": 7.514739357265055e-06,
      "loss": 0.0004,
      "step": 17590
    },
    {
      "epoch": 0.24883711066182188,
      "grad_norm": 0.0,
      "learning_rate": 7.513325510045385e-06,
      "loss": 0.0342,
      "step": 17600
    },
    {
      "epoch": 0.24897849538378883,
      "grad_norm": 3.554272893779853e-07,
      "learning_rate": 7.511911662825716e-06,
      "loss": 0.0,
      "step": 17610
    },
    {
      "epoch": 0.24911988010575578,
      "grad_norm": 0.0,
      "learning_rate": 7.510497815606046e-06,
      "loss": 0.0065,
      "step": 17620
    },
    {
      "epoch": 0.2492612648277227,
      "grad_norm": 8.51981712912675e-06,
      "learning_rate": 7.509083968386377e-06,
      "loss": 0.2812,
      "step": 17630
    },
    {
      "epoch": 0.24940264954968966,
      "grad_norm": 0.0,
      "learning_rate": 7.507670121166707e-06,
      "loss": 0.0,
      "step": 17640
    },
    {
      "epoch": 0.2495440342716566,
      "grad_norm": 0.0,
      "learning_rate": 7.506256273947038e-06,
      "loss": 0.0003,
      "step": 17650
    },
    {
      "epoch": 0.24968541899362354,
      "grad_norm": 0.0,
      "learning_rate": 7.504842426727369e-06,
      "loss": 0.0,
      "step": 17660
    },
    {
      "epoch": 0.2498268037155905,
      "grad_norm": 0.0,
      "learning_rate": 7.5034285795076985e-06,
      "loss": 0.0001,
      "step": 17670
    },
    {
      "epoch": 0.24996818843755744,
      "grad_norm": 0.0,
      "learning_rate": 7.50201473228803e-06,
      "loss": 0.0561,
      "step": 17680
    },
    {
      "epoch": 0.25010957315952437,
      "grad_norm": 0.0,
      "learning_rate": 7.50060088506836e-06,
      "loss": 0.0001,
      "step": 17690
    },
    {
      "epoch": 0.2502509578814913,
      "grad_norm": 0.0009557771845720708,
      "learning_rate": 7.499187037848691e-06,
      "loss": 0.0,
      "step": 17700
    },
    {
      "epoch": 0.2503923426034583,
      "grad_norm": 2.8926028239339985e-09,
      "learning_rate": 7.497773190629021e-06,
      "loss": 0.0,
      "step": 17710
    },
    {
      "epoch": 0.2505337273254252,
      "grad_norm": 5.9540877117569835e-08,
      "learning_rate": 7.496359343409352e-06,
      "loss": 0.0,
      "step": 17720
    },
    {
      "epoch": 0.2506751120473922,
      "grad_norm": 0.0,
      "learning_rate": 7.494945496189683e-06,
      "loss": 0.0,
      "step": 17730
    },
    {
      "epoch": 0.2508164967693591,
      "grad_norm": 1.0548131257337445e-07,
      "learning_rate": 7.493531648970013e-06,
      "loss": 0.0001,
      "step": 17740
    },
    {
      "epoch": 0.25095788149132603,
      "grad_norm": 0.0,
      "learning_rate": 7.492117801750344e-06,
      "loss": 0.0007,
      "step": 17750
    },
    {
      "epoch": 0.251099266213293,
      "grad_norm": 0.0,
      "learning_rate": 7.490703954530674e-06,
      "loss": 0.0001,
      "step": 17760
    },
    {
      "epoch": 0.25124065093525993,
      "grad_norm": 0.0,
      "learning_rate": 7.4892901073110045e-06,
      "loss": 0.0,
      "step": 17770
    },
    {
      "epoch": 0.2513820356572269,
      "grad_norm": 42.631980895996094,
      "learning_rate": 7.4878762600913345e-06,
      "loss": 0.0111,
      "step": 17780
    },
    {
      "epoch": 0.25152342037919384,
      "grad_norm": 1.3048841118745713e-08,
      "learning_rate": 7.486462412871665e-06,
      "loss": 0.1116,
      "step": 17790
    },
    {
      "epoch": 0.2516648051011608,
      "grad_norm": 9.134288325718387e-10,
      "learning_rate": 7.485048565651997e-06,
      "loss": 0.0,
      "step": 17800
    },
    {
      "epoch": 0.2518061898231277,
      "grad_norm": 0.0,
      "learning_rate": 7.483634718432326e-06,
      "loss": 0.0,
      "step": 17810
    },
    {
      "epoch": 0.25194757454509464,
      "grad_norm": 2.100367879620535e-08,
      "learning_rate": 7.482220871212658e-06,
      "loss": 0.0,
      "step": 17820
    },
    {
      "epoch": 0.2520889592670616,
      "grad_norm": 0.0,
      "learning_rate": 7.480807023992988e-06,
      "loss": 0.0,
      "step": 17830
    },
    {
      "epoch": 0.25223034398902855,
      "grad_norm": 1.0857368124561617e-06,
      "learning_rate": 7.479393176773319e-06,
      "loss": 0.0651,
      "step": 17840
    },
    {
      "epoch": 0.2523717287109955,
      "grad_norm": 0.0,
      "learning_rate": 7.47797932955365e-06,
      "loss": 0.0,
      "step": 17850
    },
    {
      "epoch": 0.25251311343296245,
      "grad_norm": 0.0,
      "learning_rate": 7.47656548233398e-06,
      "loss": 0.1526,
      "step": 17860
    },
    {
      "epoch": 0.25265449815492935,
      "grad_norm": 0.0026789435651153326,
      "learning_rate": 7.4751516351143105e-06,
      "loss": 0.0,
      "step": 17870
    },
    {
      "epoch": 0.2527958828768963,
      "grad_norm": 0.0,
      "learning_rate": 7.4737377878946405e-06,
      "loss": 0.0028,
      "step": 17880
    },
    {
      "epoch": 0.25293726759886326,
      "grad_norm": 1.113403698127513e-07,
      "learning_rate": 7.472323940674971e-06,
      "loss": 0.0044,
      "step": 17890
    },
    {
      "epoch": 0.2530786523208302,
      "grad_norm": 0.0,
      "learning_rate": 7.470910093455301e-06,
      "loss": 0.0,
      "step": 17900
    },
    {
      "epoch": 0.25322003704279716,
      "grad_norm": 0.0,
      "learning_rate": 7.469496246235632e-06,
      "loss": 0.0,
      "step": 17910
    },
    {
      "epoch": 0.2533614217647641,
      "grad_norm": 0.0,
      "learning_rate": 7.468082399015963e-06,
      "loss": 0.0,
      "step": 17920
    },
    {
      "epoch": 0.25350280648673107,
      "grad_norm": 0.0,
      "learning_rate": 7.466668551796293e-06,
      "loss": 0.0002,
      "step": 17930
    },
    {
      "epoch": 0.25364419120869797,
      "grad_norm": 0.0,
      "learning_rate": 7.465254704576624e-06,
      "loss": 0.0,
      "step": 17940
    },
    {
      "epoch": 0.2537855759306649,
      "grad_norm": 0.0,
      "learning_rate": 7.463840857356954e-06,
      "loss": 0.0,
      "step": 17950
    },
    {
      "epoch": 0.2539269606526319,
      "grad_norm": 0.0022349869832396507,
      "learning_rate": 7.462427010137286e-06,
      "loss": 0.1253,
      "step": 17960
    },
    {
      "epoch": 0.2540683453745988,
      "grad_norm": 0.0013959917705506086,
      "learning_rate": 7.461013162917615e-06,
      "loss": 0.0,
      "step": 17970
    },
    {
      "epoch": 0.2542097300965658,
      "grad_norm": 0.0,
      "learning_rate": 7.4595993156979465e-06,
      "loss": 0.0,
      "step": 17980
    },
    {
      "epoch": 0.25435111481853273,
      "grad_norm": 0.00361010804772377,
      "learning_rate": 7.458185468478277e-06,
      "loss": 0.0894,
      "step": 17990
    },
    {
      "epoch": 0.25449249954049963,
      "grad_norm": 6.27872731584489e-09,
      "learning_rate": 7.456771621258607e-06,
      "loss": 0.0,
      "step": 18000
    },
    {
      "epoch": 0.2546338842624666,
      "grad_norm": 0.0,
      "learning_rate": 7.455357774038938e-06,
      "loss": 0.06,
      "step": 18010
    },
    {
      "epoch": 0.25477526898443353,
      "grad_norm": 7.977178739793089e-08,
      "learning_rate": 7.453943926819268e-06,
      "loss": 0.0,
      "step": 18020
    },
    {
      "epoch": 0.2549166537064005,
      "grad_norm": 0.012951911427080631,
      "learning_rate": 7.4526714643215655e-06,
      "loss": 0.1011,
      "step": 18030
    },
    {
      "epoch": 0.25505803842836744,
      "grad_norm": 6.5936922588605285e-09,
      "learning_rate": 7.451257617101897e-06,
      "loss": 0.0,
      "step": 18040
    },
    {
      "epoch": 0.2551994231503344,
      "grad_norm": 0.0,
      "learning_rate": 7.449843769882226e-06,
      "loss": 0.0,
      "step": 18050
    },
    {
      "epoch": 0.25534080787230135,
      "grad_norm": 0.0,
      "learning_rate": 7.448429922662558e-06,
      "loss": 0.0,
      "step": 18060
    },
    {
      "epoch": 0.25548219259426824,
      "grad_norm": 0.0,
      "learning_rate": 7.447016075442888e-06,
      "loss": 0.0,
      "step": 18070
    },
    {
      "epoch": 0.2556235773162352,
      "grad_norm": 0.0005468088202178478,
      "learning_rate": 7.445602228223219e-06,
      "loss": 0.0,
      "step": 18080
    },
    {
      "epoch": 0.25576496203820215,
      "grad_norm": 4.083522853193244e-09,
      "learning_rate": 7.44418838100355e-06,
      "loss": 0.0,
      "step": 18090
    },
    {
      "epoch": 0.2559063467601691,
      "grad_norm": 1.7411153748980723e-05,
      "learning_rate": 7.44277453378388e-06,
      "loss": 0.0003,
      "step": 18100
    },
    {
      "epoch": 0.25604773148213605,
      "grad_norm": 3.8285432424345345e-07,
      "learning_rate": 7.441360686564211e-06,
      "loss": 0.0093,
      "step": 18110
    },
    {
      "epoch": 0.256189116204103,
      "grad_norm": 0.00022356545377988368,
      "learning_rate": 7.439946839344541e-06,
      "loss": 0.0,
      "step": 18120
    },
    {
      "epoch": 0.2563305009260699,
      "grad_norm": 0.0006985213840380311,
      "learning_rate": 7.4385329921248716e-06,
      "loss": 0.0,
      "step": 18130
    },
    {
      "epoch": 0.25647188564803686,
      "grad_norm": 4.797543473600285e-10,
      "learning_rate": 7.437119144905202e-06,
      "loss": 0.0231,
      "step": 18140
    },
    {
      "epoch": 0.2566132703700038,
      "grad_norm": 0.0,
      "learning_rate": 7.435705297685532e-06,
      "loss": 0.0,
      "step": 18150
    },
    {
      "epoch": 0.25675465509197076,
      "grad_norm": 0.0033964463509619236,
      "learning_rate": 7.434291450465863e-06,
      "loss": 0.0,
      "step": 18160
    },
    {
      "epoch": 0.2568960398139377,
      "grad_norm": 0.0,
      "learning_rate": 7.432877603246193e-06,
      "loss": 0.0,
      "step": 18170
    },
    {
      "epoch": 0.25703742453590467,
      "grad_norm": 0.0,
      "learning_rate": 7.431463756026525e-06,
      "loss": 0.0,
      "step": 18180
    },
    {
      "epoch": 0.2571788092578716,
      "grad_norm": 0.013641740195453167,
      "learning_rate": 7.430049908806854e-06,
      "loss": 0.0001,
      "step": 18190
    },
    {
      "epoch": 0.2573201939798385,
      "grad_norm": 2.964885048584165e-08,
      "learning_rate": 7.428636061587186e-06,
      "loss": 0.0,
      "step": 18200
    },
    {
      "epoch": 0.25746157870180547,
      "grad_norm": 0.001508136629126966,
      "learning_rate": 7.427222214367517e-06,
      "loss": 0.0,
      "step": 18210
    },
    {
      "epoch": 0.2576029634237724,
      "grad_norm": 4.470716703508515e-07,
      "learning_rate": 7.425808367147847e-06,
      "loss": 0.0048,
      "step": 18220
    },
    {
      "epoch": 0.2577443481457394,
      "grad_norm": 5.777415212904202e-10,
      "learning_rate": 7.424394519928178e-06,
      "loss": 0.0001,
      "step": 18230
    },
    {
      "epoch": 0.25788573286770633,
      "grad_norm": 0.00015170076221693307,
      "learning_rate": 7.422980672708508e-06,
      "loss": 0.0183,
      "step": 18240
    },
    {
      "epoch": 0.2580271175896733,
      "grad_norm": 0.003475755453109741,
      "learning_rate": 7.4215668254888384e-06,
      "loss": 0.0,
      "step": 18250
    },
    {
      "epoch": 0.2581685023116402,
      "grad_norm": 0.0,
      "learning_rate": 7.4201529782691685e-06,
      "loss": 0.0,
      "step": 18260
    },
    {
      "epoch": 0.25830988703360713,
      "grad_norm": 0.00031518543255515397,
      "learning_rate": 7.418739131049499e-06,
      "loss": 0.0,
      "step": 18270
    },
    {
      "epoch": 0.2584512717555741,
      "grad_norm": 0.0,
      "learning_rate": 7.41732528382983e-06,
      "loss": 0.0,
      "step": 18280
    },
    {
      "epoch": 0.25859265647754104,
      "grad_norm": 4.530704131866514e-07,
      "learning_rate": 7.41591143661016e-06,
      "loss": 0.0053,
      "step": 18290
    },
    {
      "epoch": 0.258734041199508,
      "grad_norm": 8.28564734547399e-05,
      "learning_rate": 7.414497589390491e-06,
      "loss": 0.001,
      "step": 18300
    },
    {
      "epoch": 0.25887542592147494,
      "grad_norm": 0.0020703719928860664,
      "learning_rate": 7.413083742170821e-06,
      "loss": 0.0001,
      "step": 18310
    },
    {
      "epoch": 0.2590168106434419,
      "grad_norm": 6.764614823850934e-08,
      "learning_rate": 7.411669894951152e-06,
      "loss": 0.0,
      "step": 18320
    },
    {
      "epoch": 0.2591581953654088,
      "grad_norm": 2.710294877417141e-09,
      "learning_rate": 7.410256047731482e-06,
      "loss": 0.0171,
      "step": 18330
    },
    {
      "epoch": 0.25929958008737575,
      "grad_norm": 0.003246122971177101,
      "learning_rate": 7.408842200511814e-06,
      "loss": 0.0011,
      "step": 18340
    },
    {
      "epoch": 0.2594409648093427,
      "grad_norm": 0.0,
      "learning_rate": 7.4074283532921445e-06,
      "loss": 0.0,
      "step": 18350
    },
    {
      "epoch": 0.25958234953130965,
      "grad_norm": 0.0,
      "learning_rate": 7.4060145060724745e-06,
      "loss": 0.0019,
      "step": 18360
    },
    {
      "epoch": 0.2597237342532766,
      "grad_norm": 9.419285925105214e-05,
      "learning_rate": 7.404600658852805e-06,
      "loss": 0.0,
      "step": 18370
    },
    {
      "epoch": 0.25986511897524356,
      "grad_norm": 0.0,
      "learning_rate": 7.403186811633135e-06,
      "loss": 0.0002,
      "step": 18380
    },
    {
      "epoch": 0.26000650369721046,
      "grad_norm": 11.070841789245605,
      "learning_rate": 7.401772964413466e-06,
      "loss": 0.0011,
      "step": 18390
    },
    {
      "epoch": 0.2601478884191774,
      "grad_norm": 0.0,
      "learning_rate": 7.400359117193797e-06,
      "loss": 0.0331,
      "step": 18400
    },
    {
      "epoch": 0.26028927314114436,
      "grad_norm": 0.0,
      "learning_rate": 7.398945269974127e-06,
      "loss": 0.0001,
      "step": 18410
    },
    {
      "epoch": 0.2604306578631113,
      "grad_norm": 2.0453303495315822e-09,
      "learning_rate": 7.397531422754458e-06,
      "loss": 0.0,
      "step": 18420
    },
    {
      "epoch": 0.26057204258507827,
      "grad_norm": 0.0,
      "learning_rate": 7.396117575534788e-06,
      "loss": 0.0,
      "step": 18430
    },
    {
      "epoch": 0.2607134273070452,
      "grad_norm": 0.0,
      "learning_rate": 7.394703728315119e-06,
      "loss": 0.0,
      "step": 18440
    },
    {
      "epoch": 0.2608548120290121,
      "grad_norm": 0.0,
      "learning_rate": 7.393289881095449e-06,
      "loss": 0.0,
      "step": 18450
    },
    {
      "epoch": 0.26099619675097907,
      "grad_norm": 1.2820138181268703e-05,
      "learning_rate": 7.39187603387578e-06,
      "loss": 0.0252,
      "step": 18460
    },
    {
      "epoch": 0.261137581472946,
      "grad_norm": 0.016001038253307343,
      "learning_rate": 7.390462186656111e-06,
      "loss": 0.0,
      "step": 18470
    },
    {
      "epoch": 0.261278966194913,
      "grad_norm": 0.0,
      "learning_rate": 7.389048339436441e-06,
      "loss": 0.0003,
      "step": 18480
    },
    {
      "epoch": 0.26142035091687993,
      "grad_norm": 0.0,
      "learning_rate": 7.387634492216772e-06,
      "loss": 0.0,
      "step": 18490
    },
    {
      "epoch": 0.2615617356388469,
      "grad_norm": 8.662727850605734e-06,
      "learning_rate": 7.386220644997102e-06,
      "loss": 0.0049,
      "step": 18500
    },
    {
      "epoch": 0.26170312036081383,
      "grad_norm": 0.017574599012732506,
      "learning_rate": 7.384806797777433e-06,
      "loss": 0.0002,
      "step": 18510
    },
    {
      "epoch": 0.26184450508278073,
      "grad_norm": 2.262263536453247,
      "learning_rate": 7.383392950557763e-06,
      "loss": 0.0006,
      "step": 18520
    },
    {
      "epoch": 0.2619858898047477,
      "grad_norm": 0.0,
      "learning_rate": 7.381979103338094e-06,
      "loss": 0.0,
      "step": 18530
    },
    {
      "epoch": 0.26212727452671464,
      "grad_norm": 1.8335425920668058e-05,
      "learning_rate": 7.380565256118425e-06,
      "loss": 0.0,
      "step": 18540
    },
    {
      "epoch": 0.2622686592486816,
      "grad_norm": 0.0,
      "learning_rate": 7.379151408898755e-06,
      "loss": 0.0004,
      "step": 18550
    },
    {
      "epoch": 0.26241004397064854,
      "grad_norm": 2.931096787506249e-07,
      "learning_rate": 7.377737561679086e-06,
      "loss": 0.0,
      "step": 18560
    },
    {
      "epoch": 0.2625514286926155,
      "grad_norm": 4.6167193090695946e-07,
      "learning_rate": 7.376323714459416e-06,
      "loss": 0.0,
      "step": 18570
    },
    {
      "epoch": 0.2626928134145824,
      "grad_norm": 0.0,
      "learning_rate": 7.3749098672397465e-06,
      "loss": 0.0,
      "step": 18580
    },
    {
      "epoch": 0.26283419813654935,
      "grad_norm": 0.0,
      "learning_rate": 7.3734960200200766e-06,
      "loss": 0.0,
      "step": 18590
    },
    {
      "epoch": 0.2629755828585163,
      "grad_norm": 5.064321739745026e-10,
      "learning_rate": 7.372082172800407e-06,
      "loss": 0.0013,
      "step": 18600
    },
    {
      "epoch": 0.26311696758048325,
      "grad_norm": 5.374787974687933e-07,
      "learning_rate": 7.370668325580739e-06,
      "loss": 0.0,
      "step": 18610
    },
    {
      "epoch": 0.2632583523024502,
      "grad_norm": 0.032970231026411057,
      "learning_rate": 7.369254478361068e-06,
      "loss": 0.0,
      "step": 18620
    },
    {
      "epoch": 0.26339973702441716,
      "grad_norm": 0.0,
      "learning_rate": 7.3678406311414e-06,
      "loss": 0.0699,
      "step": 18630
    },
    {
      "epoch": 0.2635411217463841,
      "grad_norm": 2.6269810859957943e-06,
      "learning_rate": 7.36642678392173e-06,
      "loss": 0.0,
      "step": 18640
    },
    {
      "epoch": 0.263682506468351,
      "grad_norm": 3.6546197179632145e-07,
      "learning_rate": 7.365012936702061e-06,
      "loss": 0.0,
      "step": 18650
    },
    {
      "epoch": 0.26382389119031796,
      "grad_norm": 0.0,
      "learning_rate": 7.363599089482392e-06,
      "loss": 0.0003,
      "step": 18660
    },
    {
      "epoch": 0.2639652759122849,
      "grad_norm": 0.0,
      "learning_rate": 7.362185242262722e-06,
      "loss": 0.0,
      "step": 18670
    },
    {
      "epoch": 0.26410666063425187,
      "grad_norm": 0.0,
      "learning_rate": 7.3607713950430526e-06,
      "loss": 0.0003,
      "step": 18680
    },
    {
      "epoch": 0.2642480453562188,
      "grad_norm": 0.0,
      "learning_rate": 7.3593575478233826e-06,
      "loss": 0.0,
      "step": 18690
    },
    {
      "epoch": 0.26438943007818577,
      "grad_norm": 0.0005752091528847814,
      "learning_rate": 7.3579437006037134e-06,
      "loss": 0.0,
      "step": 18700
    },
    {
      "epoch": 0.26453081480015267,
      "grad_norm": 3.4711404168774607e-06,
      "learning_rate": 7.3565298533840434e-06,
      "loss": 0.0,
      "step": 18710
    },
    {
      "epoch": 0.2646721995221196,
      "grad_norm": 0.0,
      "learning_rate": 7.355116006164374e-06,
      "loss": 0.0,
      "step": 18720
    },
    {
      "epoch": 0.2648135842440866,
      "grad_norm": 0.0,
      "learning_rate": 7.353702158944705e-06,
      "loss": 0.0001,
      "step": 18730
    },
    {
      "epoch": 0.26495496896605353,
      "grad_norm": 1.0205200240420709e-08,
      "learning_rate": 7.352288311725035e-06,
      "loss": 0.0,
      "step": 18740
    },
    {
      "epoch": 0.2650963536880205,
      "grad_norm": 12.650232315063477,
      "learning_rate": 7.350874464505367e-06,
      "loss": 0.3343,
      "step": 18750
    },
    {
      "epoch": 0.26523773840998743,
      "grad_norm": 0.030889024958014488,
      "learning_rate": 7.349460617285696e-06,
      "loss": 0.4387,
      "step": 18760
    },
    {
      "epoch": 0.2653791231319544,
      "grad_norm": 0.0,
      "learning_rate": 7.348046770066028e-06,
      "loss": 0.0068,
      "step": 18770
    },
    {
      "epoch": 0.2655205078539213,
      "grad_norm": 0.0,
      "learning_rate": 7.346632922846357e-06,
      "loss": 0.0077,
      "step": 18780
    },
    {
      "epoch": 0.26566189257588824,
      "grad_norm": 0.047232117503881454,
      "learning_rate": 7.345219075626689e-06,
      "loss": 0.0,
      "step": 18790
    },
    {
      "epoch": 0.2658032772978552,
      "grad_norm": 0.0,
      "learning_rate": 7.3438052284070195e-06,
      "loss": 0.0062,
      "step": 18800
    },
    {
      "epoch": 0.26594466201982214,
      "grad_norm": 0.0,
      "learning_rate": 7.3423913811873495e-06,
      "loss": 0.0,
      "step": 18810
    },
    {
      "epoch": 0.2660860467417891,
      "grad_norm": 0.00010468849359313026,
      "learning_rate": 7.34097753396768e-06,
      "loss": 0.0,
      "step": 18820
    },
    {
      "epoch": 0.26622743146375605,
      "grad_norm": 0.0,
      "learning_rate": 7.33956368674801e-06,
      "loss": 0.0007,
      "step": 18830
    },
    {
      "epoch": 0.26636881618572295,
      "grad_norm": 0.0,
      "learning_rate": 7.338149839528341e-06,
      "loss": 0.0,
      "step": 18840
    },
    {
      "epoch": 0.2665102009076899,
      "grad_norm": 0.0,
      "learning_rate": 7.336735992308671e-06,
      "loss": 0.0,
      "step": 18850
    },
    {
      "epoch": 0.26665158562965685,
      "grad_norm": 0.0007888884865678847,
      "learning_rate": 7.335322145089002e-06,
      "loss": 0.001,
      "step": 18860
    },
    {
      "epoch": 0.2667929703516238,
      "grad_norm": 0.0,
      "learning_rate": 7.333908297869333e-06,
      "loss": 0.0,
      "step": 18870
    },
    {
      "epoch": 0.26693435507359076,
      "grad_norm": 0.0,
      "learning_rate": 7.332494450649663e-06,
      "loss": 0.0,
      "step": 18880
    },
    {
      "epoch": 0.2670757397955577,
      "grad_norm": 0.0,
      "learning_rate": 7.331080603429994e-06,
      "loss": 0.002,
      "step": 18890
    },
    {
      "epoch": 0.26721712451752466,
      "grad_norm": 0.0,
      "learning_rate": 7.329666756210324e-06,
      "loss": 0.0,
      "step": 18900
    },
    {
      "epoch": 0.26735850923949156,
      "grad_norm": 0.0,
      "learning_rate": 7.3282529089906555e-06,
      "loss": 0.0,
      "step": 18910
    },
    {
      "epoch": 0.2674998939614585,
      "grad_norm": 0.0,
      "learning_rate": 7.326839061770986e-06,
      "loss": 0.0,
      "step": 18920
    },
    {
      "epoch": 0.26764127868342547,
      "grad_norm": 0.0,
      "learning_rate": 7.325425214551316e-06,
      "loss": 0.0,
      "step": 18930
    },
    {
      "epoch": 0.2677826634053924,
      "grad_norm": 0.0,
      "learning_rate": 7.324011367331647e-06,
      "loss": 0.0,
      "step": 18940
    },
    {
      "epoch": 0.26792404812735937,
      "grad_norm": 2.3323609639192e-06,
      "learning_rate": 7.322597520111977e-06,
      "loss": 0.0,
      "step": 18950
    },
    {
      "epoch": 0.2680654328493263,
      "grad_norm": 0.0,
      "learning_rate": 7.321183672892308e-06,
      "loss": 0.0006,
      "step": 18960
    },
    {
      "epoch": 0.2682068175712932,
      "grad_norm": 0.00036181905306875706,
      "learning_rate": 7.319769825672638e-06,
      "loss": 0.1924,
      "step": 18970
    },
    {
      "epoch": 0.2683482022932602,
      "grad_norm": 0.0,
      "learning_rate": 7.318355978452969e-06,
      "loss": 0.0,
      "step": 18980
    },
    {
      "epoch": 0.2684895870152271,
      "grad_norm": 0.0,
      "learning_rate": 7.3169421312333e-06,
      "loss": 0.0,
      "step": 18990
    },
    {
      "epoch": 0.2686309717371941,
      "grad_norm": 0.016750194132328033,
      "learning_rate": 7.31552828401363e-06,
      "loss": 0.0017,
      "step": 19000
    },
    {
      "epoch": 0.26877235645916103,
      "grad_norm": 0.0,
      "learning_rate": 7.314114436793961e-06,
      "loss": 0.0007,
      "step": 19010
    },
    {
      "epoch": 0.268913741181128,
      "grad_norm": 0.0,
      "learning_rate": 7.312700589574291e-06,
      "loss": 0.0,
      "step": 19020
    },
    {
      "epoch": 0.26905512590309494,
      "grad_norm": 1.677651653153589e-06,
      "learning_rate": 7.3112867423546215e-06,
      "loss": 0.0021,
      "step": 19030
    },
    {
      "epoch": 0.26919651062506184,
      "grad_norm": 0.8356758952140808,
      "learning_rate": 7.3098728951349515e-06,
      "loss": 0.0001,
      "step": 19040
    },
    {
      "epoch": 0.2693378953470288,
      "grad_norm": 0.0,
      "learning_rate": 7.308459047915283e-06,
      "loss": 0.0,
      "step": 19050
    },
    {
      "epoch": 0.26947928006899574,
      "grad_norm": 0.0013115304755046964,
      "learning_rate": 7.307045200695614e-06,
      "loss": 0.1095,
      "step": 19060
    },
    {
      "epoch": 0.2696206647909627,
      "grad_norm": 8.67325600228952e-10,
      "learning_rate": 7.305631353475944e-06,
      "loss": 0.0,
      "step": 19070
    },
    {
      "epoch": 0.26976204951292965,
      "grad_norm": 0.022044822573661804,
      "learning_rate": 7.304217506256275e-06,
      "loss": 0.2465,
      "step": 19080
    },
    {
      "epoch": 0.2699034342348966,
      "grad_norm": 4.12808084487915,
      "learning_rate": 7.302803659036605e-06,
      "loss": 0.0006,
      "step": 19090
    },
    {
      "epoch": 0.2700448189568635,
      "grad_norm": 0.0,
      "learning_rate": 7.301389811816936e-06,
      "loss": 0.349,
      "step": 19100
    },
    {
      "epoch": 0.27018620367883045,
      "grad_norm": 0.0007214005454443395,
      "learning_rate": 7.299975964597267e-06,
      "loss": 0.0001,
      "step": 19110
    },
    {
      "epoch": 0.2703275884007974,
      "grad_norm": 0.0,
      "learning_rate": 7.298562117377597e-06,
      "loss": 0.0001,
      "step": 19120
    },
    {
      "epoch": 0.27046897312276436,
      "grad_norm": 2.160079002380371,
      "learning_rate": 7.2971482701579276e-06,
      "loss": 0.0003,
      "step": 19130
    },
    {
      "epoch": 0.2706103578447313,
      "grad_norm": 1.9711517095565796,
      "learning_rate": 7.2957344229382576e-06,
      "loss": 0.1227,
      "step": 19140
    },
    {
      "epoch": 0.27075174256669826,
      "grad_norm": 0.0,
      "learning_rate": 7.294320575718588e-06,
      "loss": 0.0001,
      "step": 19150
    },
    {
      "epoch": 0.2708931272886652,
      "grad_norm": 0.648871660232544,
      "learning_rate": 7.2929067284989184e-06,
      "loss": 0.0211,
      "step": 19160
    },
    {
      "epoch": 0.2710345120106321,
      "grad_norm": 0.0019722306169569492,
      "learning_rate": 7.291492881279249e-06,
      "loss": 0.0,
      "step": 19170
    },
    {
      "epoch": 0.27117589673259906,
      "grad_norm": 0.0,
      "learning_rate": 7.290079034059581e-06,
      "loss": 0.0001,
      "step": 19180
    },
    {
      "epoch": 0.271317281454566,
      "grad_norm": 0.0,
      "learning_rate": 7.28866518683991e-06,
      "loss": 0.0009,
      "step": 19190
    },
    {
      "epoch": 0.27145866617653297,
      "grad_norm": 0.0,
      "learning_rate": 7.287251339620242e-06,
      "loss": 0.0,
      "step": 19200
    },
    {
      "epoch": 0.2716000508984999,
      "grad_norm": 0.0,
      "learning_rate": 7.285837492400572e-06,
      "loss": 0.0001,
      "step": 19210
    },
    {
      "epoch": 0.2717414356204669,
      "grad_norm": 0.0,
      "learning_rate": 7.284423645180903e-06,
      "loss": 0.0073,
      "step": 19220
    },
    {
      "epoch": 0.2718828203424338,
      "grad_norm": 4.366849282178009e-08,
      "learning_rate": 7.283009797961233e-06,
      "loss": 0.0,
      "step": 19230
    },
    {
      "epoch": 0.2720242050644007,
      "grad_norm": 0.0,
      "learning_rate": 7.281595950741564e-06,
      "loss": 0.0,
      "step": 19240
    },
    {
      "epoch": 0.2721655897863677,
      "grad_norm": 3.881504653691081e-06,
      "learning_rate": 7.2801821035218944e-06,
      "loss": 0.0,
      "step": 19250
    },
    {
      "epoch": 0.27230697450833463,
      "grad_norm": 0.0,
      "learning_rate": 7.2787682563022245e-06,
      "loss": 0.0,
      "step": 19260
    },
    {
      "epoch": 0.2724483592303016,
      "grad_norm": 3.3117496514023514e-06,
      "learning_rate": 7.277354409082555e-06,
      "loss": 0.0,
      "step": 19270
    },
    {
      "epoch": 0.27258974395226854,
      "grad_norm": 1.8810507071975735e-06,
      "learning_rate": 7.275940561862885e-06,
      "loss": 0.0001,
      "step": 19280
    },
    {
      "epoch": 0.27273112867423543,
      "grad_norm": 0.0,
      "learning_rate": 7.274526714643216e-06,
      "loss": 0.0,
      "step": 19290
    },
    {
      "epoch": 0.2728725133962024,
      "grad_norm": 0.0,
      "learning_rate": 7.273112867423546e-06,
      "loss": 0.0,
      "step": 19300
    },
    {
      "epoch": 0.27301389811816934,
      "grad_norm": 1.295627356512341e-07,
      "learning_rate": 7.271699020203877e-06,
      "loss": 0.0,
      "step": 19310
    },
    {
      "epoch": 0.2731552828401363,
      "grad_norm": 0.0,
      "learning_rate": 7.270285172984209e-06,
      "loss": 0.0001,
      "step": 19320
    },
    {
      "epoch": 0.27329666756210325,
      "grad_norm": 2.4445521162874684e-08,
      "learning_rate": 7.268871325764538e-06,
      "loss": 0.0,
      "step": 19330
    },
    {
      "epoch": 0.2734380522840702,
      "grad_norm": 0.0,
      "learning_rate": 7.26745747854487e-06,
      "loss": 0.0001,
      "step": 19340
    },
    {
      "epoch": 0.27357943700603715,
      "grad_norm": 0.0,
      "learning_rate": 7.266043631325199e-06,
      "loss": 0.0006,
      "step": 19350
    },
    {
      "epoch": 0.27372082172800405,
      "grad_norm": 3.375949745532125e-05,
      "learning_rate": 7.2646297841055305e-06,
      "loss": 0.0,
      "step": 19360
    },
    {
      "epoch": 0.273862206449971,
      "grad_norm": 5.035494243799121e-10,
      "learning_rate": 7.263215936885861e-06,
      "loss": 0.0,
      "step": 19370
    },
    {
      "epoch": 0.27400359117193795,
      "grad_norm": 0.00019492171122692525,
      "learning_rate": 7.261802089666191e-06,
      "loss": 0.0,
      "step": 19380
    },
    {
      "epoch": 0.2741449758939049,
      "grad_norm": 0.0,
      "learning_rate": 7.260388242446522e-06,
      "loss": 0.0,
      "step": 19390
    },
    {
      "epoch": 0.27428636061587186,
      "grad_norm": 5.427960331871873e-06,
      "learning_rate": 7.258974395226852e-06,
      "loss": 0.0017,
      "step": 19400
    },
    {
      "epoch": 0.2744277453378388,
      "grad_norm": 0.0,
      "learning_rate": 7.257560548007183e-06,
      "loss": 0.0,
      "step": 19410
    },
    {
      "epoch": 0.2745691300598057,
      "grad_norm": 1.001112193144138e-09,
      "learning_rate": 7.256146700787513e-06,
      "loss": 0.0,
      "step": 19420
    },
    {
      "epoch": 0.27471051478177266,
      "grad_norm": 0.00024068793572951108,
      "learning_rate": 7.254732853567844e-06,
      "loss": 0.0,
      "step": 19430
    },
    {
      "epoch": 0.2748518995037396,
      "grad_norm": 0.0,
      "learning_rate": 7.253319006348175e-06,
      "loss": 0.0,
      "step": 19440
    },
    {
      "epoch": 0.27499328422570657,
      "grad_norm": 7.573939800262451,
      "learning_rate": 7.251905159128505e-06,
      "loss": 0.001,
      "step": 19450
    },
    {
      "epoch": 0.2751346689476735,
      "grad_norm": 0.0,
      "learning_rate": 7.250491311908836e-06,
      "loss": 0.0,
      "step": 19460
    },
    {
      "epoch": 0.2752760536696405,
      "grad_norm": 0.0,
      "learning_rate": 7.249077464689166e-06,
      "loss": 0.0001,
      "step": 19470
    },
    {
      "epoch": 0.2754174383916074,
      "grad_norm": 0.0,
      "learning_rate": 7.247663617469497e-06,
      "loss": 0.0,
      "step": 19480
    },
    {
      "epoch": 0.2755588231135743,
      "grad_norm": 0.0,
      "learning_rate": 7.2462497702498265e-06,
      "loss": 0.1676,
      "step": 19490
    },
    {
      "epoch": 0.2757002078355413,
      "grad_norm": 0.0,
      "learning_rate": 7.244835923030158e-06,
      "loss": 0.0,
      "step": 19500
    },
    {
      "epoch": 0.27584159255750823,
      "grad_norm": 0.00013164107804186642,
      "learning_rate": 7.243422075810489e-06,
      "loss": 0.0,
      "step": 19510
    },
    {
      "epoch": 0.2759829772794752,
      "grad_norm": 2.76420242251163e-09,
      "learning_rate": 7.242008228590819e-06,
      "loss": 0.0,
      "step": 19520
    },
    {
      "epoch": 0.27612436200144214,
      "grad_norm": 0.0,
      "learning_rate": 7.24059438137115e-06,
      "loss": 0.0048,
      "step": 19530
    },
    {
      "epoch": 0.2762657467234091,
      "grad_norm": 0.0,
      "learning_rate": 7.23918053415148e-06,
      "loss": 0.0216,
      "step": 19540
    },
    {
      "epoch": 0.276407131445376,
      "grad_norm": 5.232087914919248e-06,
      "learning_rate": 7.237766686931811e-06,
      "loss": 0.0,
      "step": 19550
    },
    {
      "epoch": 0.27654851616734294,
      "grad_norm": 6.815110828028992e-05,
      "learning_rate": 7.236352839712141e-06,
      "loss": 0.0,
      "step": 19560
    },
    {
      "epoch": 0.2766899008893099,
      "grad_norm": 2.1432821029065963e-07,
      "learning_rate": 7.234938992492472e-06,
      "loss": 0.0,
      "step": 19570
    },
    {
      "epoch": 0.27683128561127684,
      "grad_norm": 1.6055788876201404e-07,
      "learning_rate": 7.2335251452728025e-06,
      "loss": 0.0,
      "step": 19580
    },
    {
      "epoch": 0.2769726703332438,
      "grad_norm": 0.0,
      "learning_rate": 7.2321112980531326e-06,
      "loss": 0.0,
      "step": 19590
    },
    {
      "epoch": 0.27711405505521075,
      "grad_norm": 0.018740355968475342,
      "learning_rate": 7.230697450833463e-06,
      "loss": 0.0277,
      "step": 19600
    },
    {
      "epoch": 0.2772554397771777,
      "grad_norm": 6.012508824149165e-10,
      "learning_rate": 7.229283603613793e-06,
      "loss": 0.0011,
      "step": 19610
    },
    {
      "epoch": 0.2773968244991446,
      "grad_norm": 0.026651911437511444,
      "learning_rate": 7.227869756394124e-06,
      "loss": 0.0,
      "step": 19620
    },
    {
      "epoch": 0.27753820922111155,
      "grad_norm": 0.0,
      "learning_rate": 7.226455909174456e-06,
      "loss": 0.0,
      "step": 19630
    },
    {
      "epoch": 0.2776795939430785,
      "grad_norm": 1.2249242331563437e-07,
      "learning_rate": 7.225042061954786e-06,
      "loss": 0.0001,
      "step": 19640
    },
    {
      "epoch": 0.27782097866504546,
      "grad_norm": 2.407741249044193e-06,
      "learning_rate": 7.223628214735117e-06,
      "loss": 0.0,
      "step": 19650
    },
    {
      "epoch": 0.2779623633870124,
      "grad_norm": 2.8735733032226562,
      "learning_rate": 7.222214367515447e-06,
      "loss": 0.0005,
      "step": 19660
    },
    {
      "epoch": 0.27810374810897937,
      "grad_norm": 0.0,
      "learning_rate": 7.220800520295778e-06,
      "loss": 0.0,
      "step": 19670
    },
    {
      "epoch": 0.27824513283094626,
      "grad_norm": 0.0,
      "learning_rate": 7.219386673076108e-06,
      "loss": 0.0014,
      "step": 19680
    },
    {
      "epoch": 0.2783865175529132,
      "grad_norm": 0.031074225902557373,
      "learning_rate": 7.217972825856439e-06,
      "loss": 0.0,
      "step": 19690
    },
    {
      "epoch": 0.27852790227488017,
      "grad_norm": 0.0008860647794790566,
      "learning_rate": 7.2165589786367694e-06,
      "loss": 0.0,
      "step": 19700
    },
    {
      "epoch": 0.2786692869968471,
      "grad_norm": 0.0001637602981645614,
      "learning_rate": 7.2151451314170994e-06,
      "loss": 0.0,
      "step": 19710
    },
    {
      "epoch": 0.2788106717188141,
      "grad_norm": 0.0,
      "learning_rate": 7.21373128419743e-06,
      "loss": 0.0504,
      "step": 19720
    },
    {
      "epoch": 0.278952056440781,
      "grad_norm": 2.525979780898524e-08,
      "learning_rate": 7.21231743697776e-06,
      "loss": 0.0,
      "step": 19730
    },
    {
      "epoch": 0.279093441162748,
      "grad_norm": 0.0,
      "learning_rate": 7.210903589758091e-06,
      "loss": 0.0066,
      "step": 19740
    },
    {
      "epoch": 0.2792348258847149,
      "grad_norm": 0.0,
      "learning_rate": 7.209489742538421e-06,
      "loss": 0.0,
      "step": 19750
    },
    {
      "epoch": 0.27937621060668183,
      "grad_norm": 0.00011817516497103497,
      "learning_rate": 7.208075895318752e-06,
      "loss": 0.0,
      "step": 19760
    },
    {
      "epoch": 0.2795175953286488,
      "grad_norm": 0.0,
      "learning_rate": 7.206662048099084e-06,
      "loss": 0.0002,
      "step": 19770
    },
    {
      "epoch": 0.27965898005061574,
      "grad_norm": 0.0,
      "learning_rate": 7.205248200879414e-06,
      "loss": 0.0,
      "step": 19780
    },
    {
      "epoch": 0.2798003647725827,
      "grad_norm": 0.0,
      "learning_rate": 7.203834353659745e-06,
      "loss": 0.0003,
      "step": 19790
    },
    {
      "epoch": 0.27994174949454964,
      "grad_norm": 0.0,
      "learning_rate": 7.202420506440075e-06,
      "loss": 0.0,
      "step": 19800
    },
    {
      "epoch": 0.28008313421651654,
      "grad_norm": 2.5095722675323486,
      "learning_rate": 7.2010066592204055e-06,
      "loss": 0.0007,
      "step": 19810
    },
    {
      "epoch": 0.2802245189384835,
      "grad_norm": 0.12071364372968674,
      "learning_rate": 7.199592812000736e-06,
      "loss": 0.0004,
      "step": 19820
    },
    {
      "epoch": 0.28036590366045044,
      "grad_norm": 0.0,
      "learning_rate": 7.198178964781066e-06,
      "loss": 0.0,
      "step": 19830
    },
    {
      "epoch": 0.2805072883824174,
      "grad_norm": 0.0001015519883367233,
      "learning_rate": 7.196765117561397e-06,
      "loss": 0.0,
      "step": 19840
    },
    {
      "epoch": 0.28064867310438435,
      "grad_norm": 0.0,
      "learning_rate": 7.195351270341727e-06,
      "loss": 0.0421,
      "step": 19850
    },
    {
      "epoch": 0.2807900578263513,
      "grad_norm": 0.0,
      "learning_rate": 7.193937423122058e-06,
      "loss": 0.0,
      "step": 19860
    },
    {
      "epoch": 0.28093144254831826,
      "grad_norm": 0.00012060360313626006,
      "learning_rate": 7.192523575902388e-06,
      "loss": 0.0022,
      "step": 19870
    },
    {
      "epoch": 0.28107282727028515,
      "grad_norm": 0.0,
      "learning_rate": 7.191109728682719e-06,
      "loss": 0.0,
      "step": 19880
    },
    {
      "epoch": 0.2812142119922521,
      "grad_norm": 0.0016349200159311295,
      "learning_rate": 7.18969588146305e-06,
      "loss": 0.0372,
      "step": 19890
    },
    {
      "epoch": 0.28135559671421906,
      "grad_norm": 0.0,
      "learning_rate": 7.18828203424338e-06,
      "loss": 0.0003,
      "step": 19900
    },
    {
      "epoch": 0.281496981436186,
      "grad_norm": 0.0,
      "learning_rate": 7.1868681870237115e-06,
      "loss": 0.0,
      "step": 19910
    },
    {
      "epoch": 0.28163836615815296,
      "grad_norm": 0.0,
      "learning_rate": 7.185454339804041e-06,
      "loss": 0.0,
      "step": 19920
    },
    {
      "epoch": 0.2817797508801199,
      "grad_norm": 5.1331057875358965e-06,
      "learning_rate": 7.184040492584372e-06,
      "loss": 0.0,
      "step": 19930
    },
    {
      "epoch": 0.2819211356020868,
      "grad_norm": 0.24405677616596222,
      "learning_rate": 7.182626645364702e-06,
      "loss": 0.0,
      "step": 19940
    },
    {
      "epoch": 0.28206252032405377,
      "grad_norm": 0.0,
      "learning_rate": 7.181212798145033e-06,
      "loss": 0.0113,
      "step": 19950
    },
    {
      "epoch": 0.2822039050460207,
      "grad_norm": 7.114088429460708e-09,
      "learning_rate": 7.179798950925364e-06,
      "loss": 0.0,
      "step": 19960
    },
    {
      "epoch": 0.2823452897679877,
      "grad_norm": 0.0,
      "learning_rate": 7.178385103705694e-06,
      "loss": 0.0,
      "step": 19970
    },
    {
      "epoch": 0.2824866744899546,
      "grad_norm": 0.0,
      "learning_rate": 7.176971256486025e-06,
      "loss": 0.0,
      "step": 19980
    },
    {
      "epoch": 0.2826280592119216,
      "grad_norm": 0.0,
      "learning_rate": 7.175557409266355e-06,
      "loss": 0.0,
      "step": 19990
    },
    {
      "epoch": 0.28276944393388853,
      "grad_norm": 0.0,
      "learning_rate": 7.174143562046686e-06,
      "loss": 0.0,
      "step": 20000
    },
    {
      "epoch": 0.28291082865585543,
      "grad_norm": 0.0017176216933876276,
      "learning_rate": 7.172729714827016e-06,
      "loss": 0.0,
      "step": 20010
    },
    {
      "epoch": 0.2830522133778224,
      "grad_norm": 3.0805159667579574e-07,
      "learning_rate": 7.171315867607347e-06,
      "loss": 0.0,
      "step": 20020
    },
    {
      "epoch": 0.28319359809978933,
      "grad_norm": 155.67153930664062,
      "learning_rate": 7.1699020203876775e-06,
      "loss": 0.044,
      "step": 20030
    },
    {
      "epoch": 0.2833349828217563,
      "grad_norm": 0.0,
      "learning_rate": 7.1684881731680075e-06,
      "loss": 0.0,
      "step": 20040
    },
    {
      "epoch": 0.28347636754372324,
      "grad_norm": 0.0,
      "learning_rate": 7.167074325948339e-06,
      "loss": 0.0,
      "step": 20050
    },
    {
      "epoch": 0.2836177522656902,
      "grad_norm": 0.0,
      "learning_rate": 7.165660478728668e-06,
      "loss": 0.0,
      "step": 20060
    },
    {
      "epoch": 0.2837591369876571,
      "grad_norm": 1.877069522038255e-08,
      "learning_rate": 7.164246631509e-06,
      "loss": 0.0001,
      "step": 20070
    },
    {
      "epoch": 0.28390052170962404,
      "grad_norm": 0.0009841707069426775,
      "learning_rate": 7.162832784289331e-06,
      "loss": 0.0,
      "step": 20080
    },
    {
      "epoch": 0.284041906431591,
      "grad_norm": 0.0,
      "learning_rate": 7.161418937069661e-06,
      "loss": 0.0,
      "step": 20090
    },
    {
      "epoch": 0.28418329115355795,
      "grad_norm": 1.7402517795562744,
      "learning_rate": 7.160005089849992e-06,
      "loss": 0.0003,
      "step": 20100
    },
    {
      "epoch": 0.2843246758755249,
      "grad_norm": 0.0,
      "learning_rate": 7.158591242630322e-06,
      "loss": 0.0008,
      "step": 20110
    },
    {
      "epoch": 0.28446606059749185,
      "grad_norm": 0.0,
      "learning_rate": 7.157177395410653e-06,
      "loss": 0.0,
      "step": 20120
    },
    {
      "epoch": 0.28460744531945875,
      "grad_norm": 4.067517238581786e-06,
      "learning_rate": 7.155763548190983e-06,
      "loss": 0.0,
      "step": 20130
    },
    {
      "epoch": 0.2847488300414257,
      "grad_norm": 8.402948310504144e-07,
      "learning_rate": 7.1543497009713136e-06,
      "loss": 0.0,
      "step": 20140
    },
    {
      "epoch": 0.28489021476339266,
      "grad_norm": 0.0,
      "learning_rate": 7.152935853751644e-06,
      "loss": 0.0,
      "step": 20150
    },
    {
      "epoch": 0.2850315994853596,
      "grad_norm": 3.874403020631689e-09,
      "learning_rate": 7.1515220065319744e-06,
      "loss": 0.0,
      "step": 20160
    },
    {
      "epoch": 0.28517298420732656,
      "grad_norm": 0.0,
      "learning_rate": 7.150108159312305e-06,
      "loss": 0.0001,
      "step": 20170
    },
    {
      "epoch": 0.2853143689292935,
      "grad_norm": 0.0,
      "learning_rate": 7.148694312092635e-06,
      "loss": 0.0,
      "step": 20180
    },
    {
      "epoch": 0.28545575365126047,
      "grad_norm": 1.7396455831431012e-08,
      "learning_rate": 7.147280464872966e-06,
      "loss": 0.0045,
      "step": 20190
    },
    {
      "epoch": 0.28559713837322737,
      "grad_norm": 0.0,
      "learning_rate": 7.145866617653296e-06,
      "loss": 0.0,
      "step": 20200
    },
    {
      "epoch": 0.2857385230951943,
      "grad_norm": 0.0,
      "learning_rate": 7.144452770433628e-06,
      "loss": 0.0005,
      "step": 20210
    },
    {
      "epoch": 0.28587990781716127,
      "grad_norm": 0.0,
      "learning_rate": 7.143038923213959e-06,
      "loss": 0.0034,
      "step": 20220
    },
    {
      "epoch": 0.2860212925391282,
      "grad_norm": 6.520008173538372e-05,
      "learning_rate": 7.141625075994289e-06,
      "loss": 0.0,
      "step": 20230
    },
    {
      "epoch": 0.2861626772610952,
      "grad_norm": 0.0,
      "learning_rate": 7.14021122877462e-06,
      "loss": 0.0,
      "step": 20240
    },
    {
      "epoch": 0.28630406198306213,
      "grad_norm": 0.0,
      "learning_rate": 7.13879738155495e-06,
      "loss": 0.0,
      "step": 20250
    },
    {
      "epoch": 0.286445446705029,
      "grad_norm": 0.0,
      "learning_rate": 7.1373835343352805e-06,
      "loss": 0.064,
      "step": 20260
    },
    {
      "epoch": 0.286586831426996,
      "grad_norm": NaN,
      "learning_rate": 7.136111071837578e-06,
      "loss": 0.3115,
      "step": 20270
    },
    {
      "epoch": 0.28672821614896293,
      "grad_norm": 0.0,
      "learning_rate": 7.134697224617908e-06,
      "loss": 0.0,
      "step": 20280
    },
    {
      "epoch": 0.2868696008709299,
      "grad_norm": 0.0,
      "learning_rate": 7.1332833773982394e-06,
      "loss": 0.0108,
      "step": 20290
    },
    {
      "epoch": 0.28701098559289684,
      "grad_norm": 0.0,
      "learning_rate": 7.131869530178569e-06,
      "loss": 1.4407,
      "step": 20300
    },
    {
      "epoch": 0.2871523703148638,
      "grad_norm": 0.3096134066581726,
      "learning_rate": 7.1304556829589e-06,
      "loss": 0.0,
      "step": 20310
    },
    {
      "epoch": 0.28729375503683074,
      "grad_norm": 0.012005496770143509,
      "learning_rate": 7.129041835739231e-06,
      "loss": 0.0,
      "step": 20320
    },
    {
      "epoch": 0.28743513975879764,
      "grad_norm": 4.692111144066757e-10,
      "learning_rate": 7.127627988519561e-06,
      "loss": 0.0,
      "step": 20330
    },
    {
      "epoch": 0.2875765244807646,
      "grad_norm": 0.0,
      "learning_rate": 7.126214141299892e-06,
      "loss": 0.0,
      "step": 20340
    },
    {
      "epoch": 0.28771790920273155,
      "grad_norm": 2.1685774953539294e-08,
      "learning_rate": 7.124800294080222e-06,
      "loss": 0.0001,
      "step": 20350
    },
    {
      "epoch": 0.2878592939246985,
      "grad_norm": 30.409849166870117,
      "learning_rate": 7.123386446860553e-06,
      "loss": 0.0067,
      "step": 20360
    },
    {
      "epoch": 0.28800067864666545,
      "grad_norm": 0.0,
      "learning_rate": 7.121972599640884e-06,
      "loss": 0.0001,
      "step": 20370
    },
    {
      "epoch": 0.2881420633686324,
      "grad_norm": 0.0005188676877878606,
      "learning_rate": 7.120558752421214e-06,
      "loss": 0.0001,
      "step": 20380
    },
    {
      "epoch": 0.2882834480905993,
      "grad_norm": 2.209754601523173e-09,
      "learning_rate": 7.119144905201545e-06,
      "loss": 0.0738,
      "step": 20390
    },
    {
      "epoch": 0.28842483281256626,
      "grad_norm": 4.1171773546011536e-08,
      "learning_rate": 7.117731057981875e-06,
      "loss": 0.0001,
      "step": 20400
    },
    {
      "epoch": 0.2885662175345332,
      "grad_norm": 0.001095255953259766,
      "learning_rate": 7.1163172107622055e-06,
      "loss": 0.0,
      "step": 20410
    },
    {
      "epoch": 0.28870760225650016,
      "grad_norm": 0.0,
      "learning_rate": 7.1149033635425355e-06,
      "loss": 0.0,
      "step": 20420
    },
    {
      "epoch": 0.2888489869784671,
      "grad_norm": 0.0,
      "learning_rate": 7.113489516322867e-06,
      "loss": 0.0,
      "step": 20430
    },
    {
      "epoch": 0.28899037170043407,
      "grad_norm": 0.0,
      "learning_rate": 7.112075669103198e-06,
      "loss": 0.0,
      "step": 20440
    },
    {
      "epoch": 0.289131756422401,
      "grad_norm": 0.0,
      "learning_rate": 7.110661821883528e-06,
      "loss": 0.0,
      "step": 20450
    },
    {
      "epoch": 0.2892731411443679,
      "grad_norm": 0.0,
      "learning_rate": 7.109247974663859e-06,
      "loss": 0.0,
      "step": 20460
    },
    {
      "epoch": 0.28941452586633487,
      "grad_norm": 1.0261100413799795e-07,
      "learning_rate": 7.107834127444189e-06,
      "loss": 0.0,
      "step": 20470
    },
    {
      "epoch": 0.2895559105883018,
      "grad_norm": 0.0,
      "learning_rate": 7.10642028022452e-06,
      "loss": 0.0,
      "step": 20480
    },
    {
      "epoch": 0.2896972953102688,
      "grad_norm": 0.0,
      "learning_rate": 7.10500643300485e-06,
      "loss": 0.0,
      "step": 20490
    },
    {
      "epoch": 0.28983868003223573,
      "grad_norm": 0.0,
      "learning_rate": 7.103592585785181e-06,
      "loss": 0.8219,
      "step": 20500
    },
    {
      "epoch": 0.2899800647542027,
      "grad_norm": 2.3005533478226425e-07,
      "learning_rate": 7.1021787385655115e-06,
      "loss": 0.0,
      "step": 20510
    },
    {
      "epoch": 0.2901214494761696,
      "grad_norm": 0.0,
      "learning_rate": 7.1007648913458415e-06,
      "loss": 0.0001,
      "step": 20520
    },
    {
      "epoch": 0.29026283419813653,
      "grad_norm": 0.0,
      "learning_rate": 7.099351044126172e-06,
      "loss": 0.0004,
      "step": 20530
    },
    {
      "epoch": 0.2904042189201035,
      "grad_norm": 1.0352010360037411e-09,
      "learning_rate": 7.097937196906502e-06,
      "loss": 0.2003,
      "step": 20540
    },
    {
      "epoch": 0.29054560364207044,
      "grad_norm": 0.0,
      "learning_rate": 7.096523349686833e-06,
      "loss": 0.0413,
      "step": 20550
    },
    {
      "epoch": 0.2906869883640374,
      "grad_norm": 0.0,
      "learning_rate": 7.095109502467163e-06,
      "loss": 0.0,
      "step": 20560
    },
    {
      "epoch": 0.29082837308600434,
      "grad_norm": 0.01241308357566595,
      "learning_rate": 7.093695655247494e-06,
      "loss": 0.0001,
      "step": 20570
    },
    {
      "epoch": 0.2909697578079713,
      "grad_norm": 0.0,
      "learning_rate": 7.092281808027826e-06,
      "loss": 0.0001,
      "step": 20580
    },
    {
      "epoch": 0.2911111425299382,
      "grad_norm": 0.0,
      "learning_rate": 7.090867960808156e-06,
      "loss": 0.002,
      "step": 20590
    },
    {
      "epoch": 0.29125252725190515,
      "grad_norm": 4.235701234200917e-10,
      "learning_rate": 7.089454113588487e-06,
      "loss": 0.0001,
      "step": 20600
    },
    {
      "epoch": 0.2913939119738721,
      "grad_norm": 0.0,
      "learning_rate": 7.088040266368817e-06,
      "loss": 0.0,
      "step": 20610
    },
    {
      "epoch": 0.29153529669583905,
      "grad_norm": 3.370668650859443e-07,
      "learning_rate": 7.0866264191491475e-06,
      "loss": 0.0,
      "step": 20620
    },
    {
      "epoch": 0.291676681417806,
      "grad_norm": 3.9657438719586935e-06,
      "learning_rate": 7.085212571929478e-06,
      "loss": 0.0001,
      "step": 20630
    },
    {
      "epoch": 0.29181806613977296,
      "grad_norm": 27.798961639404297,
      "learning_rate": 7.083798724709808e-06,
      "loss": 0.0029,
      "step": 20640
    },
    {
      "epoch": 0.29195945086173986,
      "grad_norm": 0.0,
      "learning_rate": 7.082384877490139e-06,
      "loss": 0.0008,
      "step": 20650
    },
    {
      "epoch": 0.2921008355837068,
      "grad_norm": 0.0002531248319428414,
      "learning_rate": 7.080971030270469e-06,
      "loss": 0.1345,
      "step": 20660
    },
    {
      "epoch": 0.29224222030567376,
      "grad_norm": 0.533421516418457,
      "learning_rate": 7.0795571830508e-06,
      "loss": 0.0001,
      "step": 20670
    },
    {
      "epoch": 0.2923836050276407,
      "grad_norm": 4.538920350682929e-10,
      "learning_rate": 7.07814333583113e-06,
      "loss": 0.0,
      "step": 20680
    },
    {
      "epoch": 0.29252498974960767,
      "grad_norm": 0.011160247027873993,
      "learning_rate": 7.076729488611461e-06,
      "loss": 0.0,
      "step": 20690
    },
    {
      "epoch": 0.2926663744715746,
      "grad_norm": 0.0,
      "learning_rate": 7.075315641391793e-06,
      "loss": 0.0,
      "step": 20700
    },
    {
      "epoch": 0.2928077591935416,
      "grad_norm": 8.825881377561018e-05,
      "learning_rate": 7.073901794172122e-06,
      "loss": 0.0,
      "step": 20710
    },
    {
      "epoch": 0.29294914391550847,
      "grad_norm": 0.0020233604591339827,
      "learning_rate": 7.0724879469524536e-06,
      "loss": 0.0,
      "step": 20720
    },
    {
      "epoch": 0.2930905286374754,
      "grad_norm": 0.0,
      "learning_rate": 7.071074099732783e-06,
      "loss": 0.0,
      "step": 20730
    },
    {
      "epoch": 0.2932319133594424,
      "grad_norm": 0.0,
      "learning_rate": 7.069660252513114e-06,
      "loss": 0.0,
      "step": 20740
    },
    {
      "epoch": 0.29337329808140933,
      "grad_norm": 1.1811200906208796e-08,
      "learning_rate": 7.0682464052934444e-06,
      "loss": 0.0,
      "step": 20750
    },
    {
      "epoch": 0.2935146828033763,
      "grad_norm": 0.4819936156272888,
      "learning_rate": 7.066832558073775e-06,
      "loss": 0.0001,
      "step": 20760
    },
    {
      "epoch": 0.29365606752534323,
      "grad_norm": 0.0,
      "learning_rate": 7.065418710854106e-06,
      "loss": 0.1224,
      "step": 20770
    },
    {
      "epoch": 0.29379745224731013,
      "grad_norm": 0.0,
      "learning_rate": 7.064004863634436e-06,
      "loss": 0.0047,
      "step": 20780
    },
    {
      "epoch": 0.2939388369692771,
      "grad_norm": 0.0,
      "learning_rate": 7.062591016414767e-06,
      "loss": 0.0045,
      "step": 20790
    },
    {
      "epoch": 0.29408022169124404,
      "grad_norm": 0.0,
      "learning_rate": 7.061177169195097e-06,
      "loss": 0.0299,
      "step": 20800
    },
    {
      "epoch": 0.294221606413211,
      "grad_norm": 4.102105890524399e-07,
      "learning_rate": 7.059763321975428e-06,
      "loss": 0.0001,
      "step": 20810
    },
    {
      "epoch": 0.29436299113517794,
      "grad_norm": 3.9243118976628466e-08,
      "learning_rate": 7.058349474755758e-06,
      "loss": 0.0005,
      "step": 20820
    },
    {
      "epoch": 0.2945043758571449,
      "grad_norm": 0.0,
      "learning_rate": 7.056935627536089e-06,
      "loss": 0.0,
      "step": 20830
    },
    {
      "epoch": 0.29464576057911185,
      "grad_norm": 9.701336400880223e-10,
      "learning_rate": 7.05552178031642e-06,
      "loss": 0.0,
      "step": 20840
    },
    {
      "epoch": 0.29478714530107875,
      "grad_norm": 0.0,
      "learning_rate": 7.05410793309675e-06,
      "loss": 0.0,
      "step": 20850
    },
    {
      "epoch": 0.2949285300230457,
      "grad_norm": 0.0,
      "learning_rate": 7.052694085877081e-06,
      "loss": 0.0,
      "step": 20860
    },
    {
      "epoch": 0.29506991474501265,
      "grad_norm": 0.0,
      "learning_rate": 7.0512802386574105e-06,
      "loss": 0.0,
      "step": 20870
    },
    {
      "epoch": 0.2952112994669796,
      "grad_norm": 3.1071689932105073e-07,
      "learning_rate": 7.049866391437742e-06,
      "loss": 0.0057,
      "step": 20880
    },
    {
      "epoch": 0.29535268418894656,
      "grad_norm": 0.0,
      "learning_rate": 7.048452544218073e-06,
      "loss": 0.0397,
      "step": 20890
    },
    {
      "epoch": 0.2954940689109135,
      "grad_norm": 0.0,
      "learning_rate": 7.047038696998403e-06,
      "loss": 0.0002,
      "step": 20900
    },
    {
      "epoch": 0.2956354536328804,
      "grad_norm": 0.001842650817707181,
      "learning_rate": 7.045624849778734e-06,
      "loss": 0.0,
      "step": 20910
    },
    {
      "epoch": 0.29577683835484736,
      "grad_norm": 0.0,
      "learning_rate": 7.044211002559064e-06,
      "loss": 0.0001,
      "step": 20920
    },
    {
      "epoch": 0.2959182230768143,
      "grad_norm": 5.9578049160791124e-09,
      "learning_rate": 7.042797155339395e-06,
      "loss": 0.0061,
      "step": 20930
    },
    {
      "epoch": 0.29605960779878127,
      "grad_norm": 4.721582014255432e-10,
      "learning_rate": 7.041383308119725e-06,
      "loss": 0.0059,
      "step": 20940
    },
    {
      "epoch": 0.2962009925207482,
      "grad_norm": 0.006349506787955761,
      "learning_rate": 7.039969460900056e-06,
      "loss": 0.0,
      "step": 20950
    },
    {
      "epoch": 0.29634237724271517,
      "grad_norm": 5.944539793745207e-07,
      "learning_rate": 7.0385556136803865e-06,
      "loss": 0.0,
      "step": 20960
    },
    {
      "epoch": 0.29648376196468207,
      "grad_norm": 9.315570537182793e-07,
      "learning_rate": 7.0371417664607165e-06,
      "loss": 0.0,
      "step": 20970
    },
    {
      "epoch": 0.296625146686649,
      "grad_norm": 0.0,
      "learning_rate": 7.035727919241047e-06,
      "loss": 0.0,
      "step": 20980
    },
    {
      "epoch": 0.296766531408616,
      "grad_norm": 0.2328932285308838,
      "learning_rate": 7.034314072021377e-06,
      "loss": 0.0008,
      "step": 20990
    },
    {
      "epoch": 0.2969079161305829,
      "grad_norm": 4.179552888672333e-06,
      "learning_rate": 7.032900224801709e-06,
      "loss": 0.0093,
      "step": 21000
    },
    {
      "epoch": 0.2970493008525499,
      "grad_norm": 0.0,
      "learning_rate": 7.031486377582038e-06,
      "loss": 0.0001,
      "step": 21010
    },
    {
      "epoch": 0.29719068557451683,
      "grad_norm": 4.872438008618474e-10,
      "learning_rate": 7.03007253036237e-06,
      "loss": 0.0,
      "step": 21020
    },
    {
      "epoch": 0.2973320702964838,
      "grad_norm": 0.0,
      "learning_rate": 7.028658683142701e-06,
      "loss": 0.0,
      "step": 21030
    },
    {
      "epoch": 0.2974734550184507,
      "grad_norm": 6.824045328812645e-08,
      "learning_rate": 7.027244835923031e-06,
      "loss": 0.0817,
      "step": 21040
    },
    {
      "epoch": 0.29761483974041764,
      "grad_norm": 0.0,
      "learning_rate": 7.025830988703362e-06,
      "loss": 0.0451,
      "step": 21050
    },
    {
      "epoch": 0.2977562244623846,
      "grad_norm": 0.0005038263625465333,
      "learning_rate": 7.024417141483692e-06,
      "loss": 0.0,
      "step": 21060
    },
    {
      "epoch": 0.29789760918435154,
      "grad_norm": 0.0,
      "learning_rate": 7.0230032942640225e-06,
      "loss": 0.0003,
      "step": 21070
    },
    {
      "epoch": 0.2980389939063185,
      "grad_norm": 0.0,
      "learning_rate": 7.0215894470443525e-06,
      "loss": 0.0023,
      "step": 21080
    },
    {
      "epoch": 0.29818037862828545,
      "grad_norm": 23.34463119506836,
      "learning_rate": 7.020175599824683e-06,
      "loss": 0.0045,
      "step": 21090
    },
    {
      "epoch": 0.29832176335025234,
      "grad_norm": 172.94744873046875,
      "learning_rate": 7.018761752605014e-06,
      "loss": 0.0442,
      "step": 21100
    },
    {
      "epoch": 0.2984631480722193,
      "grad_norm": 0.0,
      "learning_rate": 7.017347905385344e-06,
      "loss": 0.0,
      "step": 21110
    },
    {
      "epoch": 0.29860453279418625,
      "grad_norm": 0.0,
      "learning_rate": 7.015934058165675e-06,
      "loss": 0.0,
      "step": 21120
    },
    {
      "epoch": 0.2987459175161532,
      "grad_norm": 0.0,
      "learning_rate": 7.014520210946005e-06,
      "loss": 0.0,
      "step": 21130
    },
    {
      "epoch": 0.29888730223812016,
      "grad_norm": 0.3816421926021576,
      "learning_rate": 7.013106363726336e-06,
      "loss": 0.0001,
      "step": 21140
    },
    {
      "epoch": 0.2990286869600871,
      "grad_norm": 9.593681170372292e-05,
      "learning_rate": 7.011692516506668e-06,
      "loss": 0.0,
      "step": 21150
    },
    {
      "epoch": 0.29917007168205406,
      "grad_norm": 0.0,
      "learning_rate": 7.010278669286998e-06,
      "loss": 0.0034,
      "step": 21160
    },
    {
      "epoch": 0.29931145640402096,
      "grad_norm": 0.0,
      "learning_rate": 7.0088648220673285e-06,
      "loss": 0.0,
      "step": 21170
    },
    {
      "epoch": 0.2994528411259879,
      "grad_norm": 0.0,
      "learning_rate": 7.0074509748476586e-06,
      "loss": 0.0,
      "step": 21180
    },
    {
      "epoch": 0.29959422584795486,
      "grad_norm": 1.2575299912498394e-08,
      "learning_rate": 7.006037127627989e-06,
      "loss": 0.0089,
      "step": 21190
    },
    {
      "epoch": 0.2997356105699218,
      "grad_norm": 0.0,
      "learning_rate": 7.004623280408319e-06,
      "loss": 0.0,
      "step": 21200
    },
    {
      "epoch": 0.29987699529188877,
      "grad_norm": 4.691771415821222e-10,
      "learning_rate": 7.00320943318865e-06,
      "loss": 0.0,
      "step": 21210
    },
    {
      "epoch": 0.3000183800138557,
      "grad_norm": 4.4874592930455037e-10,
      "learning_rate": 7.001795585968981e-06,
      "loss": 0.0,
      "step": 21220
    },
    {
      "epoch": 0.3001597647358226,
      "grad_norm": 0.0001500092912465334,
      "learning_rate": 7.000381738749311e-06,
      "loss": 0.0,
      "step": 21230
    },
    {
      "epoch": 0.3003011494577896,
      "grad_norm": 1.3435293340080534e-06,
      "learning_rate": 6.998967891529642e-06,
      "loss": 0.0413,
      "step": 21240
    },
    {
      "epoch": 0.3004425341797565,
      "grad_norm": 0.0,
      "learning_rate": 6.997554044309972e-06,
      "loss": 0.0,
      "step": 21250
    },
    {
      "epoch": 0.3005839189017235,
      "grad_norm": 0.0,
      "learning_rate": 6.996140197090303e-06,
      "loss": 0.0,
      "step": 21260
    },
    {
      "epoch": 0.30072530362369043,
      "grad_norm": 0.024822061881422997,
      "learning_rate": 6.994726349870633e-06,
      "loss": 0.0204,
      "step": 21270
    },
    {
      "epoch": 0.3008666883456574,
      "grad_norm": 0.0,
      "learning_rate": 6.993312502650964e-06,
      "loss": 0.0,
      "step": 21280
    },
    {
      "epoch": 0.30100807306762434,
      "grad_norm": 0.0013269687769934535,
      "learning_rate": 6.9918986554312954e-06,
      "loss": 0.0,
      "step": 21290
    },
    {
      "epoch": 0.30114945778959124,
      "grad_norm": 0.0,
      "learning_rate": 6.990484808211625e-06,
      "loss": 0.0039,
      "step": 21300
    },
    {
      "epoch": 0.3012908425115582,
      "grad_norm": 3.2877164812816773e-06,
      "learning_rate": 6.989070960991956e-06,
      "loss": 0.001,
      "step": 21310
    },
    {
      "epoch": 0.30143222723352514,
      "grad_norm": 50.9969596862793,
      "learning_rate": 6.987657113772286e-06,
      "loss": 0.0167,
      "step": 21320
    },
    {
      "epoch": 0.3015736119554921,
      "grad_norm": 0.0,
      "learning_rate": 6.986243266552617e-06,
      "loss": 0.0,
      "step": 21330
    },
    {
      "epoch": 0.30171499667745905,
      "grad_norm": 6.4701458768468e-07,
      "learning_rate": 6.984829419332948e-06,
      "loss": 0.0001,
      "step": 21340
    },
    {
      "epoch": 0.301856381399426,
      "grad_norm": 0.0,
      "learning_rate": 6.983415572113278e-06,
      "loss": 0.0222,
      "step": 21350
    },
    {
      "epoch": 0.3019977661213929,
      "grad_norm": 0.0,
      "learning_rate": 6.982001724893609e-06,
      "loss": 0.0001,
      "step": 21360
    },
    {
      "epoch": 0.30213915084335985,
      "grad_norm": 2.9752311725417258e-09,
      "learning_rate": 6.980587877673939e-06,
      "loss": 0.0,
      "step": 21370
    },
    {
      "epoch": 0.3022805355653268,
      "grad_norm": 0.0,
      "learning_rate": 6.97917403045427e-06,
      "loss": 0.0,
      "step": 21380
    },
    {
      "epoch": 0.30242192028729376,
      "grad_norm": 0.0,
      "learning_rate": 6.9777601832346e-06,
      "loss": 0.0387,
      "step": 21390
    },
    {
      "epoch": 0.3025633050092607,
      "grad_norm": 7.366599561464682e-07,
      "learning_rate": 6.976346336014931e-06,
      "loss": 0.0011,
      "step": 21400
    },
    {
      "epoch": 0.30270468973122766,
      "grad_norm": 0.0,
      "learning_rate": 6.9749324887952615e-06,
      "loss": 0.0011,
      "step": 21410
    },
    {
      "epoch": 0.3028460744531946,
      "grad_norm": 0.04564426466822624,
      "learning_rate": 6.9735186415755915e-06,
      "loss": 0.0,
      "step": 21420
    },
    {
      "epoch": 0.3029874591751615,
      "grad_norm": 0.0,
      "learning_rate": 6.972104794355923e-06,
      "loss": 0.0,
      "step": 21430
    },
    {
      "epoch": 0.30312884389712846,
      "grad_norm": 0.0,
      "learning_rate": 6.970690947136252e-06,
      "loss": 0.0,
      "step": 21440
    },
    {
      "epoch": 0.3032702286190954,
      "grad_norm": 0.0,
      "learning_rate": 6.969277099916584e-06,
      "loss": 0.0,
      "step": 21450
    },
    {
      "epoch": 0.30341161334106237,
      "grad_norm": 0.0,
      "learning_rate": 6.967863252696914e-06,
      "loss": 0.0099,
      "step": 21460
    },
    {
      "epoch": 0.3035529980630293,
      "grad_norm": 0.09334143251180649,
      "learning_rate": 6.966449405477245e-06,
      "loss": 0.0,
      "step": 21470
    },
    {
      "epoch": 0.3036943827849963,
      "grad_norm": 0.0,
      "learning_rate": 6.965035558257576e-06,
      "loss": 0.0207,
      "step": 21480
    },
    {
      "epoch": 0.3038357675069632,
      "grad_norm": 0.0,
      "learning_rate": 6.963621711037906e-06,
      "loss": 0.0007,
      "step": 21490
    },
    {
      "epoch": 0.3039771522289301,
      "grad_norm": 2.9197288313298486e-05,
      "learning_rate": 6.962207863818237e-06,
      "loss": 0.0001,
      "step": 21500
    },
    {
      "epoch": 0.3041185369508971,
      "grad_norm": 0.0,
      "learning_rate": 6.960794016598567e-06,
      "loss": 0.0,
      "step": 21510
    },
    {
      "epoch": 0.30425992167286403,
      "grad_norm": 0.0,
      "learning_rate": 6.9593801693788975e-06,
      "loss": 0.0,
      "step": 21520
    },
    {
      "epoch": 0.304401306394831,
      "grad_norm": 0.00017858739010989666,
      "learning_rate": 6.9579663221592275e-06,
      "loss": 0.0,
      "step": 21530
    },
    {
      "epoch": 0.30454269111679794,
      "grad_norm": 0.0,
      "learning_rate": 6.956552474939558e-06,
      "loss": 0.0,
      "step": 21540
    },
    {
      "epoch": 0.3046840758387649,
      "grad_norm": 3.459684133529663,
      "learning_rate": 6.955138627719889e-06,
      "loss": 0.0502,
      "step": 21550
    },
    {
      "epoch": 0.3048254605607318,
      "grad_norm": 9.975699424743652,
      "learning_rate": 6.953724780500219e-06,
      "loss": 0.001,
      "step": 21560
    },
    {
      "epoch": 0.30496684528269874,
      "grad_norm": 0.0,
      "learning_rate": 6.95231093328055e-06,
      "loss": 0.0,
      "step": 21570
    },
    {
      "epoch": 0.3051082300046657,
      "grad_norm": 0.0,
      "learning_rate": 6.95089708606088e-06,
      "loss": 0.0001,
      "step": 21580
    },
    {
      "epoch": 0.30524961472663265,
      "grad_norm": 0.0,
      "learning_rate": 6.949483238841212e-06,
      "loss": 0.0,
      "step": 21590
    },
    {
      "epoch": 0.3053909994485996,
      "grad_norm": 0.0,
      "learning_rate": 6.948069391621543e-06,
      "loss": 0.0,
      "step": 21600
    },
    {
      "epoch": 0.30553238417056655,
      "grad_norm": 8.25595680709057e-09,
      "learning_rate": 6.946655544401873e-06,
      "loss": 0.0,
      "step": 21610
    },
    {
      "epoch": 0.30567376889253345,
      "grad_norm": 1.5505277133343043e-06,
      "learning_rate": 6.9452416971822035e-06,
      "loss": 0.0001,
      "step": 21620
    },
    {
      "epoch": 0.3058151536145004,
      "grad_norm": 0.0,
      "learning_rate": 6.9438278499625335e-06,
      "loss": 0.0034,
      "step": 21630
    },
    {
      "epoch": 0.30595653833646735,
      "grad_norm": 0.16156938672065735,
      "learning_rate": 6.942414002742864e-06,
      "loss": 0.0,
      "step": 21640
    },
    {
      "epoch": 0.3060979230584343,
      "grad_norm": 6.769121227989672e-06,
      "learning_rate": 6.941000155523194e-06,
      "loss": 0.0022,
      "step": 21650
    },
    {
      "epoch": 0.30623930778040126,
      "grad_norm": 0.0,
      "learning_rate": 6.939586308303525e-06,
      "loss": 0.0,
      "step": 21660
    },
    {
      "epoch": 0.3063806925023682,
      "grad_norm": 0.0011591468937695026,
      "learning_rate": 6.938172461083856e-06,
      "loss": 0.0,
      "step": 21670
    },
    {
      "epoch": 0.30652207722433517,
      "grad_norm": 0.006514933425933123,
      "learning_rate": 6.936758613864186e-06,
      "loss": 0.0,
      "step": 21680
    },
    {
      "epoch": 0.30666346194630206,
      "grad_norm": 0.0,
      "learning_rate": 6.935344766644517e-06,
      "loss": 0.0,
      "step": 21690
    },
    {
      "epoch": 0.306804846668269,
      "grad_norm": 0.0,
      "learning_rate": 6.933930919424847e-06,
      "loss": 0.0001,
      "step": 21700
    },
    {
      "epoch": 0.30694623139023597,
      "grad_norm": 0.0,
      "learning_rate": 6.932517072205178e-06,
      "loss": 0.0,
      "step": 21710
    },
    {
      "epoch": 0.3070876161122029,
      "grad_norm": 0.0,
      "learning_rate": 6.931103224985508e-06,
      "loss": 0.0,
      "step": 21720
    },
    {
      "epoch": 0.3072290008341699,
      "grad_norm": 9.657858754508197e-05,
      "learning_rate": 6.9296893777658396e-06,
      "loss": 0.0,
      "step": 21730
    },
    {
      "epoch": 0.3073703855561368,
      "grad_norm": 0.010599229484796524,
      "learning_rate": 6.92827553054617e-06,
      "loss": 0.0,
      "step": 21740
    },
    {
      "epoch": 0.3075117702781037,
      "grad_norm": 0.0,
      "learning_rate": 6.9268616833265004e-06,
      "loss": 0.0,
      "step": 21750
    },
    {
      "epoch": 0.3076531550000707,
      "grad_norm": 8.429868469939095e-10,
      "learning_rate": 6.925447836106831e-06,
      "loss": 0.0,
      "step": 21760
    },
    {
      "epoch": 0.30779453972203763,
      "grad_norm": 0.0,
      "learning_rate": 6.924033988887161e-06,
      "loss": 0.0,
      "step": 21770
    },
    {
      "epoch": 0.3079359244440046,
      "grad_norm": 0.0,
      "learning_rate": 6.922620141667492e-06,
      "loss": 0.0,
      "step": 21780
    },
    {
      "epoch": 0.30807730916597154,
      "grad_norm": 1.6750647091612336e-08,
      "learning_rate": 6.921206294447822e-06,
      "loss": 0.5238,
      "step": 21790
    },
    {
      "epoch": 0.3082186938879385,
      "grad_norm": 0.0,
      "learning_rate": 6.919792447228153e-06,
      "loss": 0.0,
      "step": 21800
    },
    {
      "epoch": 0.3083600786099054,
      "grad_norm": 0.0,
      "learning_rate": 6.918378600008484e-06,
      "loss": 0.0,
      "step": 21810
    },
    {
      "epoch": 0.30850146333187234,
      "grad_norm": 2.5746779441833496,
      "learning_rate": 6.916964752788814e-06,
      "loss": 0.0006,
      "step": 21820
    },
    {
      "epoch": 0.3086428480538393,
      "grad_norm": 0.0,
      "learning_rate": 6.915550905569145e-06,
      "loss": 0.0,
      "step": 21830
    },
    {
      "epoch": 0.30878423277580624,
      "grad_norm": 2.2926424207980745e-05,
      "learning_rate": 6.914137058349475e-06,
      "loss": 0.0,
      "step": 21840
    },
    {
      "epoch": 0.3089256174977732,
      "grad_norm": 0.0,
      "learning_rate": 6.912723211129806e-06,
      "loss": 0.0,
      "step": 21850
    },
    {
      "epoch": 0.30906700221974015,
      "grad_norm": 0.0013245906447991729,
      "learning_rate": 6.911309363910137e-06,
      "loss": 0.0049,
      "step": 21860
    },
    {
      "epoch": 0.3092083869417071,
      "grad_norm": 0.0,
      "learning_rate": 6.9098955166904665e-06,
      "loss": 0.0,
      "step": 21870
    },
    {
      "epoch": 0.309349771663674,
      "grad_norm": 12.635022163391113,
      "learning_rate": 6.908481669470798e-06,
      "loss": 0.0014,
      "step": 21880
    },
    {
      "epoch": 0.30949115638564095,
      "grad_norm": 0.0,
      "learning_rate": 6.907067822251128e-06,
      "loss": 0.0002,
      "step": 21890
    },
    {
      "epoch": 0.3096325411076079,
      "grad_norm": 0.00010148410365218297,
      "learning_rate": 6.905653975031459e-06,
      "loss": 0.0,
      "step": 21900
    },
    {
      "epoch": 0.30977392582957486,
      "grad_norm": 4.918297991096665e-10,
      "learning_rate": 6.904240127811789e-06,
      "loss": 0.0,
      "step": 21910
    },
    {
      "epoch": 0.3099153105515418,
      "grad_norm": 0.006662094034254551,
      "learning_rate": 6.90282628059212e-06,
      "loss": 0.0001,
      "step": 21920
    },
    {
      "epoch": 0.31005669527350876,
      "grad_norm": 3.76292383919008e-08,
      "learning_rate": 6.901412433372451e-06,
      "loss": 0.0003,
      "step": 21930
    },
    {
      "epoch": 0.31019807999547566,
      "grad_norm": 0.0,
      "learning_rate": 6.899998586152781e-06,
      "loss": 0.006,
      "step": 21940
    },
    {
      "epoch": 0.3103394647174426,
      "grad_norm": 536.436767578125,
      "learning_rate": 6.898584738933112e-06,
      "loss": 0.163,
      "step": 21950
    },
    {
      "epoch": 0.31048084943940957,
      "grad_norm": 0.0,
      "learning_rate": 6.897170891713442e-06,
      "loss": 0.0,
      "step": 21960
    },
    {
      "epoch": 0.3106222341613765,
      "grad_norm": 0.0,
      "learning_rate": 6.8957570444937725e-06,
      "loss": 0.0,
      "step": 21970
    },
    {
      "epoch": 0.3107636188833435,
      "grad_norm": 0.000209673642530106,
      "learning_rate": 6.8943431972741025e-06,
      "loss": 0.0397,
      "step": 21980
    },
    {
      "epoch": 0.3109050036053104,
      "grad_norm": 0.0,
      "learning_rate": 6.892929350054433e-06,
      "loss": 0.0,
      "step": 21990
    },
    {
      "epoch": 0.3110463883272774,
      "grad_norm": 0.0,
      "learning_rate": 6.891515502834765e-06,
      "loss": 0.0,
      "step": 22000
    },
    {
      "epoch": 0.3111877730492443,
      "grad_norm": 0.0,
      "learning_rate": 6.890101655615094e-06,
      "loss": 0.0002,
      "step": 22010
    },
    {
      "epoch": 0.31132915777121123,
      "grad_norm": 2.5488713362165072e-08,
      "learning_rate": 6.888687808395426e-06,
      "loss": 0.0,
      "step": 22020
    },
    {
      "epoch": 0.3114705424931782,
      "grad_norm": 0.0,
      "learning_rate": 6.887273961175756e-06,
      "loss": 0.0,
      "step": 22030
    },
    {
      "epoch": 0.31161192721514513,
      "grad_norm": 0.0,
      "learning_rate": 6.885860113956087e-06,
      "loss": 0.1406,
      "step": 22040
    },
    {
      "epoch": 0.3117533119371121,
      "grad_norm": 4.907583228686008e-10,
      "learning_rate": 6.884446266736417e-06,
      "loss": 0.0,
      "step": 22050
    },
    {
      "epoch": 0.31189469665907904,
      "grad_norm": 0.0,
      "learning_rate": 6.883032419516748e-06,
      "loss": 0.0,
      "step": 22060
    },
    {
      "epoch": 0.31203608138104594,
      "grad_norm": 1.3759927242062986e-06,
      "learning_rate": 6.8816185722970785e-06,
      "loss": 0.0,
      "step": 22070
    },
    {
      "epoch": 0.3121774661030129,
      "grad_norm": 0.0,
      "learning_rate": 6.8802047250774085e-06,
      "loss": 0.0001,
      "step": 22080
    },
    {
      "epoch": 0.31231885082497984,
      "grad_norm": 0.0,
      "learning_rate": 6.878790877857739e-06,
      "loss": 0.0,
      "step": 22090
    },
    {
      "epoch": 0.3124602355469468,
      "grad_norm": 0.005042793694883585,
      "learning_rate": 6.877377030638069e-06,
      "loss": 0.0056,
      "step": 22100
    },
    {
      "epoch": 0.31260162026891375,
      "grad_norm": 0.0,
      "learning_rate": 6.8759631834184e-06,
      "loss": 0.0,
      "step": 22110
    },
    {
      "epoch": 0.3127430049908807,
      "grad_norm": 0.0,
      "learning_rate": 6.874549336198731e-06,
      "loss": 0.003,
      "step": 22120
    },
    {
      "epoch": 0.31288438971284765,
      "grad_norm": 0.0,
      "learning_rate": 6.873135488979061e-06,
      "loss": 0.001,
      "step": 22130
    },
    {
      "epoch": 0.31302577443481455,
      "grad_norm": 2.7381349809729727e-06,
      "learning_rate": 6.871721641759392e-06,
      "loss": 0.0,
      "step": 22140
    },
    {
      "epoch": 0.3131671591567815,
      "grad_norm": 0.0,
      "learning_rate": 6.870307794539722e-06,
      "loss": 0.0,
      "step": 22150
    },
    {
      "epoch": 0.31330854387874846,
      "grad_norm": 4.722036650584016e-10,
      "learning_rate": 6.868893947320054e-06,
      "loss": 0.0017,
      "step": 22160
    },
    {
      "epoch": 0.3134499286007154,
      "grad_norm": 0.0,
      "learning_rate": 6.867480100100383e-06,
      "loss": 0.0,
      "step": 22170
    },
    {
      "epoch": 0.31359131332268236,
      "grad_norm": 0.007136649917811155,
      "learning_rate": 6.8660662528807146e-06,
      "loss": 0.0,
      "step": 22180
    },
    {
      "epoch": 0.3137326980446493,
      "grad_norm": 2.2297773361206055,
      "learning_rate": 6.864652405661045e-06,
      "loss": 0.0003,
      "step": 22190
    },
    {
      "epoch": 0.3138740827666162,
      "grad_norm": 0.00014856761845294386,
      "learning_rate": 6.863238558441375e-06,
      "loss": 0.0,
      "step": 22200
    },
    {
      "epoch": 0.31401546748858317,
      "grad_norm": 0.038358986377716064,
      "learning_rate": 6.861824711221706e-06,
      "loss": 0.0,
      "step": 22210
    },
    {
      "epoch": 0.3141568522105501,
      "grad_norm": 2.9788800759433798e-08,
      "learning_rate": 6.860410864002036e-06,
      "loss": 0.0,
      "step": 22220
    },
    {
      "epoch": 0.31429823693251707,
      "grad_norm": 8.74077699108966e-10,
      "learning_rate": 6.858997016782367e-06,
      "loss": 0.0,
      "step": 22230
    },
    {
      "epoch": 0.314439621654484,
      "grad_norm": 0.0,
      "learning_rate": 6.857583169562697e-06,
      "loss": 1.1727,
      "step": 22240
    },
    {
      "epoch": 0.314581006376451,
      "grad_norm": 0.0,
      "learning_rate": 6.856169322343028e-06,
      "loss": 0.0,
      "step": 22250
    },
    {
      "epoch": 0.31472239109841793,
      "grad_norm": 0.0,
      "learning_rate": 6.854755475123359e-06,
      "loss": 0.7892,
      "step": 22260
    },
    {
      "epoch": 0.31486377582038483,
      "grad_norm": 0.0,
      "learning_rate": 6.853341627903689e-06,
      "loss": 0.0028,
      "step": 22270
    },
    {
      "epoch": 0.3150051605423518,
      "grad_norm": 0.0,
      "learning_rate": 6.85192778068402e-06,
      "loss": 0.0,
      "step": 22280
    },
    {
      "epoch": 0.31514654526431873,
      "grad_norm": 0.0,
      "learning_rate": 6.85051393346435e-06,
      "loss": 0.0,
      "step": 22290
    },
    {
      "epoch": 0.3152879299862857,
      "grad_norm": 9.170330304186791e-05,
      "learning_rate": 6.8491000862446814e-06,
      "loss": 0.0299,
      "step": 22300
    },
    {
      "epoch": 0.31542931470825264,
      "grad_norm": 120.33863067626953,
      "learning_rate": 6.847686239025012e-06,
      "loss": 0.0204,
      "step": 22310
    },
    {
      "epoch": 0.3155706994302196,
      "grad_norm": 0.0,
      "learning_rate": 6.846272391805342e-06,
      "loss": 0.0,
      "step": 22320
    },
    {
      "epoch": 0.3157120841521865,
      "grad_norm": 0.0,
      "learning_rate": 6.844858544585673e-06,
      "loss": 0.0,
      "step": 22330
    },
    {
      "epoch": 0.31585346887415344,
      "grad_norm": 5.508628873940324e-06,
      "learning_rate": 6.843444697366003e-06,
      "loss": 0.0,
      "step": 22340
    },
    {
      "epoch": 0.3159948535961204,
      "grad_norm": 2.3987339892173054e-10,
      "learning_rate": 6.842030850146334e-06,
      "loss": 0.0,
      "step": 22350
    },
    {
      "epoch": 0.31613623831808735,
      "grad_norm": 0.0,
      "learning_rate": 6.840617002926664e-06,
      "loss": 0.0012,
      "step": 22360
    },
    {
      "epoch": 0.3162776230400543,
      "grad_norm": 0.0,
      "learning_rate": 6.839203155706995e-06,
      "loss": 0.0,
      "step": 22370
    },
    {
      "epoch": 0.31641900776202125,
      "grad_norm": 1.7959575416170992e-05,
      "learning_rate": 6.837789308487326e-06,
      "loss": 0.0,
      "step": 22380
    },
    {
      "epoch": 0.3165603924839882,
      "grad_norm": 0.5647758841514587,
      "learning_rate": 6.836375461267656e-06,
      "loss": 0.0001,
      "step": 22390
    },
    {
      "epoch": 0.3167017772059551,
      "grad_norm": 5.157738414141022e-08,
      "learning_rate": 6.834961614047987e-06,
      "loss": 0.0,
      "step": 22400
    },
    {
      "epoch": 0.31684316192792206,
      "grad_norm": 2.989961558341747e-06,
      "learning_rate": 6.833547766828317e-06,
      "loss": 0.0011,
      "step": 22410
    },
    {
      "epoch": 0.316984546649889,
      "grad_norm": 0.6259296536445618,
      "learning_rate": 6.8321339196086475e-06,
      "loss": 0.0001,
      "step": 22420
    },
    {
      "epoch": 0.31712593137185596,
      "grad_norm": 0.0,
      "learning_rate": 6.8307200723889775e-06,
      "loss": 0.0,
      "step": 22430
    },
    {
      "epoch": 0.3172673160938229,
      "grad_norm": 0.0,
      "learning_rate": 6.829306225169308e-06,
      "loss": 0.0125,
      "step": 22440
    },
    {
      "epoch": 0.31740870081578987,
      "grad_norm": 0.0006778196548111737,
      "learning_rate": 6.82789237794964e-06,
      "loss": 0.0,
      "step": 22450
    },
    {
      "epoch": 0.31755008553775677,
      "grad_norm": 0.0,
      "learning_rate": 6.82647853072997e-06,
      "loss": 0.0,
      "step": 22460
    },
    {
      "epoch": 0.3176914702597237,
      "grad_norm": 0.0,
      "learning_rate": 6.825064683510301e-06,
      "loss": 0.0,
      "step": 22470
    },
    {
      "epoch": 0.31783285498169067,
      "grad_norm": 0.0,
      "learning_rate": 6.823650836290631e-06,
      "loss": 0.0,
      "step": 22480
    },
    {
      "epoch": 0.3179742397036576,
      "grad_norm": 9.301991667598486e-06,
      "learning_rate": 6.822236989070962e-06,
      "loss": 0.0,
      "step": 22490
    },
    {
      "epoch": 0.3181156244256246,
      "grad_norm": 0.0,
      "learning_rate": 6.820823141851292e-06,
      "loss": 0.0036,
      "step": 22500
    },
    {
      "epoch": 0.31825700914759153,
      "grad_norm": 7.624311137988116e-08,
      "learning_rate": 6.819409294631623e-06,
      "loss": 0.0,
      "step": 22510
    },
    {
      "epoch": 0.3183983938695585,
      "grad_norm": 0.00046698556980118155,
      "learning_rate": 6.8179954474119535e-06,
      "loss": 0.0,
      "step": 22520
    },
    {
      "epoch": 0.3185397785915254,
      "grad_norm": 2.6684111276154e-07,
      "learning_rate": 6.8165816001922835e-06,
      "loss": 0.0002,
      "step": 22530
    },
    {
      "epoch": 0.31868116331349233,
      "grad_norm": 3.0152229157920374e-08,
      "learning_rate": 6.815167752972614e-06,
      "loss": 0.0001,
      "step": 22540
    },
    {
      "epoch": 0.3188225480354593,
      "grad_norm": 0.0005577296833507717,
      "learning_rate": 6.813753905752944e-06,
      "loss": 0.0,
      "step": 22550
    },
    {
      "epoch": 0.31896393275742624,
      "grad_norm": 0.0,
      "learning_rate": 6.812340058533275e-06,
      "loss": 0.0,
      "step": 22560
    },
    {
      "epoch": 0.3191053174793932,
      "grad_norm": 0.0,
      "learning_rate": 6.810926211313607e-06,
      "loss": 0.0,
      "step": 22570
    },
    {
      "epoch": 0.31924670220136014,
      "grad_norm": 0.006110397167503834,
      "learning_rate": 6.809512364093936e-06,
      "loss": 0.0001,
      "step": 22580
    },
    {
      "epoch": 0.31938808692332704,
      "grad_norm": 0.0,
      "learning_rate": 6.808098516874268e-06,
      "loss": 0.0,
      "step": 22590
    },
    {
      "epoch": 0.319529471645294,
      "grad_norm": 0.0,
      "learning_rate": 6.806684669654598e-06,
      "loss": 0.0,
      "step": 22600
    },
    {
      "epoch": 0.31967085636726095,
      "grad_norm": 0.0,
      "learning_rate": 6.805270822434929e-06,
      "loss": 0.0,
      "step": 22610
    },
    {
      "epoch": 0.3198122410892279,
      "grad_norm": 0.0,
      "learning_rate": 6.803998359937226e-06,
      "loss": 0.7105,
      "step": 22620
    },
    {
      "epoch": 0.31995362581119485,
      "grad_norm": 0.0,
      "learning_rate": 6.802584512717556e-06,
      "loss": 0.0,
      "step": 22630
    },
    {
      "epoch": 0.3200950105331618,
      "grad_norm": 0.0,
      "learning_rate": 6.801170665497887e-06,
      "loss": 0.001,
      "step": 22640
    },
    {
      "epoch": 0.32023639525512876,
      "grad_norm": 0.0016553614987060428,
      "learning_rate": 6.799756818278217e-06,
      "loss": 0.0,
      "step": 22650
    },
    {
      "epoch": 0.32037777997709566,
      "grad_norm": 0.0025775700341910124,
      "learning_rate": 6.798342971058548e-06,
      "loss": 0.0,
      "step": 22660
    },
    {
      "epoch": 0.3205191646990626,
      "grad_norm": 0.0,
      "learning_rate": 6.796929123838879e-06,
      "loss": 0.0,
      "step": 22670
    },
    {
      "epoch": 0.32066054942102956,
      "grad_norm": 0.0,
      "learning_rate": 6.7955152766192085e-06,
      "loss": 0.0012,
      "step": 22680
    },
    {
      "epoch": 0.3208019341429965,
      "grad_norm": 0.0003143669164273888,
      "learning_rate": 6.79410142939954e-06,
      "loss": 0.0,
      "step": 22690
    },
    {
      "epoch": 0.32094331886496347,
      "grad_norm": 0.0,
      "learning_rate": 6.79268758217987e-06,
      "loss": 0.0031,
      "step": 22700
    },
    {
      "epoch": 0.3210847035869304,
      "grad_norm": 7.757282733917236,
      "learning_rate": 6.791273734960201e-06,
      "loss": 0.0029,
      "step": 22710
    },
    {
      "epoch": 0.3212260883088973,
      "grad_norm": 0.0,
      "learning_rate": 6.789859887740531e-06,
      "loss": 0.0,
      "step": 22720
    },
    {
      "epoch": 0.32136747303086427,
      "grad_norm": 7.541866943938658e-05,
      "learning_rate": 6.788446040520862e-06,
      "loss": 0.0,
      "step": 22730
    },
    {
      "epoch": 0.3215088577528312,
      "grad_norm": 0.0,
      "learning_rate": 6.787032193301193e-06,
      "loss": 0.0045,
      "step": 22740
    },
    {
      "epoch": 0.3216502424747982,
      "grad_norm": 0.03341659903526306,
      "learning_rate": 6.785618346081523e-06,
      "loss": 0.0,
      "step": 22750
    },
    {
      "epoch": 0.32179162719676513,
      "grad_norm": 0.0,
      "learning_rate": 6.784204498861854e-06,
      "loss": 0.0,
      "step": 22760
    },
    {
      "epoch": 0.3219330119187321,
      "grad_norm": 6.961276000083672e-08,
      "learning_rate": 6.782790651642184e-06,
      "loss": 0.0209,
      "step": 22770
    },
    {
      "epoch": 0.322074396640699,
      "grad_norm": 0.0,
      "learning_rate": 6.7813768044225146e-06,
      "loss": 0.0,
      "step": 22780
    },
    {
      "epoch": 0.32221578136266593,
      "grad_norm": 3.0251008865889162e-05,
      "learning_rate": 6.7799629572028446e-06,
      "loss": 0.0,
      "step": 22790
    },
    {
      "epoch": 0.3223571660846329,
      "grad_norm": 2.4446296720270766e-06,
      "learning_rate": 6.7785491099831754e-06,
      "loss": 0.0,
      "step": 22800
    },
    {
      "epoch": 0.32249855080659984,
      "grad_norm": 1.032343321938356e-09,
      "learning_rate": 6.777135262763507e-06,
      "loss": 0.0,
      "step": 22810
    },
    {
      "epoch": 0.3226399355285668,
      "grad_norm": 0.0,
      "learning_rate": 6.775721415543836e-06,
      "loss": 0.0,
      "step": 22820
    },
    {
      "epoch": 0.32278132025053374,
      "grad_norm": 0.013872246257960796,
      "learning_rate": 6.774307568324168e-06,
      "loss": 0.0147,
      "step": 22830
    },
    {
      "epoch": 0.3229227049725007,
      "grad_norm": 0.0,
      "learning_rate": 6.772893721104498e-06,
      "loss": 0.0,
      "step": 22840
    },
    {
      "epoch": 0.3230640896944676,
      "grad_norm": 0.0,
      "learning_rate": 6.771479873884829e-06,
      "loss": 0.041,
      "step": 22850
    },
    {
      "epoch": 0.32320547441643455,
      "grad_norm": 0.0002236179425381124,
      "learning_rate": 6.77006602666516e-06,
      "loss": 0.0,
      "step": 22860
    },
    {
      "epoch": 0.3233468591384015,
      "grad_norm": 0.052373237907886505,
      "learning_rate": 6.76865217944549e-06,
      "loss": 0.0021,
      "step": 22870
    },
    {
      "epoch": 0.32348824386036845,
      "grad_norm": 1.3655107068188954e-05,
      "learning_rate": 6.767238332225821e-06,
      "loss": 0.7111,
      "step": 22880
    },
    {
      "epoch": 0.3236296285823354,
      "grad_norm": 0.0,
      "learning_rate": 6.765824485006151e-06,
      "loss": 0.0,
      "step": 22890
    },
    {
      "epoch": 0.32377101330430236,
      "grad_norm": 0.0,
      "learning_rate": 6.7644106377864814e-06,
      "loss": 0.0,
      "step": 22900
    },
    {
      "epoch": 0.32391239802626925,
      "grad_norm": 0.17884545028209686,
      "learning_rate": 6.7629967905668115e-06,
      "loss": 0.0,
      "step": 22910
    },
    {
      "epoch": 0.3240537827482362,
      "grad_norm": 0.00239564455114305,
      "learning_rate": 6.761582943347142e-06,
      "loss": 0.0002,
      "step": 22920
    },
    {
      "epoch": 0.32419516747020316,
      "grad_norm": 0.0,
      "learning_rate": 6.760169096127473e-06,
      "loss": 0.0,
      "step": 22930
    },
    {
      "epoch": 0.3243365521921701,
      "grad_norm": 0.0024199169129133224,
      "learning_rate": 6.758755248907803e-06,
      "loss": 0.6703,
      "step": 22940
    },
    {
      "epoch": 0.32447793691413707,
      "grad_norm": 0.0,
      "learning_rate": 6.757341401688135e-06,
      "loss": 0.0001,
      "step": 22950
    },
    {
      "epoch": 0.324619321636104,
      "grad_norm": 0.0,
      "learning_rate": 6.755927554468464e-06,
      "loss": 0.0,
      "step": 22960
    },
    {
      "epoch": 0.32476070635807097,
      "grad_norm": 0.0,
      "learning_rate": 6.754513707248796e-06,
      "loss": 0.008,
      "step": 22970
    },
    {
      "epoch": 0.32490209108003787,
      "grad_norm": 0.0,
      "learning_rate": 6.753099860029125e-06,
      "loss": 0.0002,
      "step": 22980
    },
    {
      "epoch": 0.3250434758020048,
      "grad_norm": 405.6604309082031,
      "learning_rate": 6.751686012809457e-06,
      "loss": 0.1054,
      "step": 22990
    },
    {
      "epoch": 0.3251848605239718,
      "grad_norm": 0.00013209827011451125,
      "learning_rate": 6.7502721655897875e-06,
      "loss": 0.0005,
      "step": 23000
    },
    {
      "epoch": 0.3253262452459387,
      "grad_norm": 1.946953886999836e-07,
      "learning_rate": 6.7488583183701175e-06,
      "loss": 0.0058,
      "step": 23010
    },
    {
      "epoch": 0.3254676299679057,
      "grad_norm": 0.00032528667361475527,
      "learning_rate": 6.747444471150448e-06,
      "loss": 0.5793,
      "step": 23020
    },
    {
      "epoch": 0.32560901468987263,
      "grad_norm": 0.00020065282296855003,
      "learning_rate": 6.746030623930778e-06,
      "loss": 0.0,
      "step": 23030
    },
    {
      "epoch": 0.32575039941183953,
      "grad_norm": 4.369329326436855e-05,
      "learning_rate": 6.744616776711109e-06,
      "loss": 0.0001,
      "step": 23040
    },
    {
      "epoch": 0.3258917841338065,
      "grad_norm": 0.0,
      "learning_rate": 6.743202929491439e-06,
      "loss": 0.0,
      "step": 23050
    },
    {
      "epoch": 0.32603316885577344,
      "grad_norm": 1.0755422152897154e-07,
      "learning_rate": 6.74178908227177e-06,
      "loss": 0.0017,
      "step": 23060
    },
    {
      "epoch": 0.3261745535777404,
      "grad_norm": 0.0,
      "learning_rate": 6.740375235052101e-06,
      "loss": 0.0044,
      "step": 23070
    },
    {
      "epoch": 0.32631593829970734,
      "grad_norm": 5.216883591785404e-10,
      "learning_rate": 6.738961387832431e-06,
      "loss": 0.0,
      "step": 23080
    },
    {
      "epoch": 0.3264573230216743,
      "grad_norm": 1.1617937161645386e-05,
      "learning_rate": 6.737547540612762e-06,
      "loss": 0.0006,
      "step": 23090
    },
    {
      "epoch": 0.32659870774364125,
      "grad_norm": 0.27382147312164307,
      "learning_rate": 6.736133693393092e-06,
      "loss": 0.0,
      "step": 23100
    },
    {
      "epoch": 0.32674009246560815,
      "grad_norm": 0.06291769444942474,
      "learning_rate": 6.7347198461734235e-06,
      "loss": 0.0,
      "step": 23110
    },
    {
      "epoch": 0.3268814771875751,
      "grad_norm": 0.0,
      "learning_rate": 6.733305998953754e-06,
      "loss": 0.0,
      "step": 23120
    },
    {
      "epoch": 0.32702286190954205,
      "grad_norm": 0.0,
      "learning_rate": 6.731892151734084e-06,
      "loss": 0.0,
      "step": 23130
    },
    {
      "epoch": 0.327164246631509,
      "grad_norm": 0.0,
      "learning_rate": 6.730478304514415e-06,
      "loss": 0.0,
      "step": 23140
    },
    {
      "epoch": 0.32730563135347596,
      "grad_norm": 0.0,
      "learning_rate": 6.729064457294745e-06,
      "loss": 0.0004,
      "step": 23150
    },
    {
      "epoch": 0.3274470160754429,
      "grad_norm": 0.0,
      "learning_rate": 6.727650610075076e-06,
      "loss": 0.0047,
      "step": 23160
    },
    {
      "epoch": 0.3275884007974098,
      "grad_norm": 0.0,
      "learning_rate": 6.726236762855406e-06,
      "loss": 0.0001,
      "step": 23170
    },
    {
      "epoch": 0.32772978551937676,
      "grad_norm": 1.8442833606968634e-05,
      "learning_rate": 6.724822915635737e-06,
      "loss": 0.0039,
      "step": 23180
    },
    {
      "epoch": 0.3278711702413437,
      "grad_norm": 0.0,
      "learning_rate": 6.723409068416068e-06,
      "loss": 0.0002,
      "step": 23190
    },
    {
      "epoch": 0.32801255496331067,
      "grad_norm": 0.0,
      "learning_rate": 6.721995221196398e-06,
      "loss": 0.0,
      "step": 23200
    },
    {
      "epoch": 0.3281539396852776,
      "grad_norm": 0.006314144469797611,
      "learning_rate": 6.720581373976729e-06,
      "loss": 0.0001,
      "step": 23210
    },
    {
      "epoch": 0.32829532440724457,
      "grad_norm": 0.0002029902971116826,
      "learning_rate": 6.719167526757059e-06,
      "loss": 0.0,
      "step": 23220
    },
    {
      "epoch": 0.3284367091292115,
      "grad_norm": 0.003402900882065296,
      "learning_rate": 6.7177536795373895e-06,
      "loss": 0.0,
      "step": 23230
    },
    {
      "epoch": 0.3285780938511784,
      "grad_norm": 0.0,
      "learning_rate": 6.7163398323177196e-06,
      "loss": 0.0,
      "step": 23240
    },
    {
      "epoch": 0.3287194785731454,
      "grad_norm": 6.795073659304762e-08,
      "learning_rate": 6.71492598509805e-06,
      "loss": 0.0,
      "step": 23250
    },
    {
      "epoch": 0.3288608632951123,
      "grad_norm": 8.964084372564685e-06,
      "learning_rate": 6.713512137878382e-06,
      "loss": 0.0,
      "step": 23260
    },
    {
      "epoch": 0.3290022480170793,
      "grad_norm": 0.0,
      "learning_rate": 6.712098290658712e-06,
      "loss": 0.0,
      "step": 23270
    },
    {
      "epoch": 0.32914363273904623,
      "grad_norm": 2.5209265004377812e-05,
      "learning_rate": 6.710684443439043e-06,
      "loss": 0.0,
      "step": 23280
    },
    {
      "epoch": 0.3292850174610132,
      "grad_norm": 0.07287447154521942,
      "learning_rate": 6.709270596219373e-06,
      "loss": 0.0,
      "step": 23290
    },
    {
      "epoch": 0.3294264021829801,
      "grad_norm": 0.0,
      "learning_rate": 6.707856748999704e-06,
      "loss": 0.0,
      "step": 23300
    },
    {
      "epoch": 0.32956778690494704,
      "grad_norm": 0.0,
      "learning_rate": 6.706442901780034e-06,
      "loss": 0.0,
      "step": 23310
    },
    {
      "epoch": 0.329709171626914,
      "grad_norm": 0.0,
      "learning_rate": 6.705029054560365e-06,
      "loss": 0.0,
      "step": 23320
    },
    {
      "epoch": 0.32985055634888094,
      "grad_norm": 0.0,
      "learning_rate": 6.7036152073406956e-06,
      "loss": 0.0,
      "step": 23330
    },
    {
      "epoch": 0.3299919410708479,
      "grad_norm": 0.001593248569406569,
      "learning_rate": 6.702201360121026e-06,
      "loss": 0.0,
      "step": 23340
    },
    {
      "epoch": 0.33013332579281485,
      "grad_norm": 0.0708465650677681,
      "learning_rate": 6.7007875129013564e-06,
      "loss": 0.0009,
      "step": 23350
    },
    {
      "epoch": 0.3302747105147818,
      "grad_norm": 1.2200059629208226e-08,
      "learning_rate": 6.6993736656816864e-06,
      "loss": 0.0,
      "step": 23360
    },
    {
      "epoch": 0.3304160952367487,
      "grad_norm": 0.0,
      "learning_rate": 6.697959818462017e-06,
      "loss": 0.0,
      "step": 23370
    },
    {
      "epoch": 0.33055747995871565,
      "grad_norm": 0.0,
      "learning_rate": 6.696545971242349e-06,
      "loss": 0.0,
      "step": 23380
    },
    {
      "epoch": 0.3306988646806826,
      "grad_norm": 0.0,
      "learning_rate": 6.695132124022678e-06,
      "loss": 0.3004,
      "step": 23390
    },
    {
      "epoch": 0.33084024940264956,
      "grad_norm": 0.0,
      "learning_rate": 6.69371827680301e-06,
      "loss": 0.0137,
      "step": 23400
    },
    {
      "epoch": 0.3309816341246165,
      "grad_norm": 0.0,
      "learning_rate": 6.69230442958334e-06,
      "loss": 0.2717,
      "step": 23410
    },
    {
      "epoch": 0.33112301884658346,
      "grad_norm": 0.0,
      "learning_rate": 6.690890582363671e-06,
      "loss": 0.0006,
      "step": 23420
    },
    {
      "epoch": 0.33126440356855036,
      "grad_norm": 0.0,
      "learning_rate": 6.689476735144001e-06,
      "loss": 0.0001,
      "step": 23430
    },
    {
      "epoch": 0.3314057882905173,
      "grad_norm": 0.0,
      "learning_rate": 6.688062887924332e-06,
      "loss": 0.0016,
      "step": 23440
    },
    {
      "epoch": 0.33154717301248426,
      "grad_norm": 0.0,
      "learning_rate": 6.6866490407046625e-06,
      "loss": 0.0,
      "step": 23450
    },
    {
      "epoch": 0.3316885577344512,
      "grad_norm": 7.091159659466939e-06,
      "learning_rate": 6.6852351934849925e-06,
      "loss": 0.0,
      "step": 23460
    },
    {
      "epoch": 0.33182994245641817,
      "grad_norm": 7.45391218970326e-07,
      "learning_rate": 6.683821346265323e-06,
      "loss": 0.0,
      "step": 23470
    },
    {
      "epoch": 0.3319713271783851,
      "grad_norm": 0.0,
      "learning_rate": 6.682407499045653e-06,
      "loss": 0.0,
      "step": 23480
    },
    {
      "epoch": 0.3321127119003521,
      "grad_norm": 0.0,
      "learning_rate": 6.680993651825984e-06,
      "loss": 0.0047,
      "step": 23490
    },
    {
      "epoch": 0.332254096622319,
      "grad_norm": 0.0,
      "learning_rate": 6.679579804606314e-06,
      "loss": 0.0028,
      "step": 23500
    },
    {
      "epoch": 0.3323954813442859,
      "grad_norm": 0.0,
      "learning_rate": 6.678165957386645e-06,
      "loss": 0.0002,
      "step": 23510
    },
    {
      "epoch": 0.3325368660662529,
      "grad_norm": 6.323474372038618e-05,
      "learning_rate": 6.676752110166976e-06,
      "loss": 0.0002,
      "step": 23520
    },
    {
      "epoch": 0.33267825078821983,
      "grad_norm": 0.1869805008172989,
      "learning_rate": 6.675338262947306e-06,
      "loss": 0.0,
      "step": 23530
    },
    {
      "epoch": 0.3328196355101868,
      "grad_norm": 7.85689906024345e-07,
      "learning_rate": 6.673924415727638e-06,
      "loss": 0.0,
      "step": 23540
    },
    {
      "epoch": 0.33296102023215374,
      "grad_norm": 6.167760346897921e-08,
      "learning_rate": 6.672510568507967e-06,
      "loss": 0.0934,
      "step": 23550
    },
    {
      "epoch": 0.33310240495412063,
      "grad_norm": 8.802681986708194e-05,
      "learning_rate": 6.6710967212882985e-06,
      "loss": 0.0,
      "step": 23560
    },
    {
      "epoch": 0.3332437896760876,
      "grad_norm": 5.082761163066607e-06,
      "learning_rate": 6.669682874068629e-06,
      "loss": 0.0,
      "step": 23570
    },
    {
      "epoch": 0.33338517439805454,
      "grad_norm": 0.0,
      "learning_rate": 6.668269026848959e-06,
      "loss": 0.0,
      "step": 23580
    },
    {
      "epoch": 0.3335265591200215,
      "grad_norm": 0.0,
      "learning_rate": 6.66685517962929e-06,
      "loss": 0.0097,
      "step": 23590
    },
    {
      "epoch": 0.33366794384198845,
      "grad_norm": 0.0,
      "learning_rate": 6.66544133240962e-06,
      "loss": 0.0,
      "step": 23600
    },
    {
      "epoch": 0.3338093285639554,
      "grad_norm": 0.0,
      "learning_rate": 6.664027485189951e-06,
      "loss": 0.0018,
      "step": 23610
    },
    {
      "epoch": 0.3339507132859223,
      "grad_norm": 0.025788042694330215,
      "learning_rate": 6.662613637970281e-06,
      "loss": 0.0,
      "step": 23620
    },
    {
      "epoch": 0.33409209800788925,
      "grad_norm": 3.084696231780981e-07,
      "learning_rate": 6.661199790750612e-06,
      "loss": 0.0,
      "step": 23630
    },
    {
      "epoch": 0.3342334827298562,
      "grad_norm": 0.0,
      "learning_rate": 6.659785943530943e-06,
      "loss": 0.0,
      "step": 23640
    },
    {
      "epoch": 0.33437486745182315,
      "grad_norm": 0.02054157294332981,
      "learning_rate": 6.658372096311273e-06,
      "loss": 0.0007,
      "step": 23650
    },
    {
      "epoch": 0.3345162521737901,
      "grad_norm": 0.0004670349298976362,
      "learning_rate": 6.656958249091604e-06,
      "loss": 0.0,
      "step": 23660
    },
    {
      "epoch": 0.33465763689575706,
      "grad_norm": 1.1486704920571356e-07,
      "learning_rate": 6.655544401871934e-06,
      "loss": 0.0001,
      "step": 23670
    },
    {
      "epoch": 0.334799021617724,
      "grad_norm": 0.0,
      "learning_rate": 6.654130554652265e-06,
      "loss": 0.0006,
      "step": 23680
    },
    {
      "epoch": 0.3349404063396909,
      "grad_norm": 6.496736526489258,
      "learning_rate": 6.6527167074325945e-06,
      "loss": 0.0011,
      "step": 23690
    },
    {
      "epoch": 0.33508179106165786,
      "grad_norm": 0.0,
      "learning_rate": 6.651302860212926e-06,
      "loss": 0.0,
      "step": 23700
    },
    {
      "epoch": 0.3352231757836248,
      "grad_norm": 0.0,
      "learning_rate": 6.649889012993257e-06,
      "loss": 0.0,
      "step": 23710
    },
    {
      "epoch": 0.33536456050559177,
      "grad_norm": 0.0,
      "learning_rate": 6.648475165773587e-06,
      "loss": 0.0022,
      "step": 23720
    },
    {
      "epoch": 0.3355059452275587,
      "grad_norm": 1.9492238934049055e-08,
      "learning_rate": 6.647061318553918e-06,
      "loss": 0.0,
      "step": 23730
    },
    {
      "epoch": 0.3356473299495257,
      "grad_norm": 3.021239081135718e-08,
      "learning_rate": 6.645647471334248e-06,
      "loss": 0.0,
      "step": 23740
    },
    {
      "epoch": 0.33578871467149257,
      "grad_norm": 0.0014214710099622607,
      "learning_rate": 6.644233624114579e-06,
      "loss": 0.0,
      "step": 23750
    },
    {
      "epoch": 0.3359300993934595,
      "grad_norm": 0.0,
      "learning_rate": 6.642819776894909e-06,
      "loss": 0.0,
      "step": 23760
    },
    {
      "epoch": 0.3360714841154265,
      "grad_norm": 6.299166125245392e-05,
      "learning_rate": 6.64140592967524e-06,
      "loss": 0.0,
      "step": 23770
    },
    {
      "epoch": 0.33621286883739343,
      "grad_norm": 0.012249110266566277,
      "learning_rate": 6.6399920824555706e-06,
      "loss": 0.0,
      "step": 23780
    },
    {
      "epoch": 0.3363542535593604,
      "grad_norm": 0.0008823423413559794,
      "learning_rate": 6.6385782352359006e-06,
      "loss": 0.1995,
      "step": 23790
    },
    {
      "epoch": 0.33649563828132734,
      "grad_norm": 0.0014889877056702971,
      "learning_rate": 6.6371643880162314e-06,
      "loss": 0.0,
      "step": 23800
    },
    {
      "epoch": 0.3366370230032943,
      "grad_norm": 0.06459177285432816,
      "learning_rate": 6.6357505407965614e-06,
      "loss": 0.0,
      "step": 23810
    },
    {
      "epoch": 0.3367784077252612,
      "grad_norm": 1.639478068682365e-05,
      "learning_rate": 6.634336693576892e-06,
      "loss": 0.0071,
      "step": 23820
    },
    {
      "epoch": 0.33691979244722814,
      "grad_norm": 0.0,
      "learning_rate": 6.632922846357224e-06,
      "loss": 0.0,
      "step": 23830
    },
    {
      "epoch": 0.3370611771691951,
      "grad_norm": 45.959415435791016,
      "learning_rate": 6.631508999137554e-06,
      "loss": 0.0127,
      "step": 23840
    },
    {
      "epoch": 0.33720256189116204,
      "grad_norm": 0.0,
      "learning_rate": 6.630095151917885e-06,
      "loss": 0.0,
      "step": 23850
    },
    {
      "epoch": 0.337343946613129,
      "grad_norm": 0.0,
      "learning_rate": 6.628681304698215e-06,
      "loss": 0.0,
      "step": 23860
    },
    {
      "epoch": 0.33748533133509595,
      "grad_norm": 0.0,
      "learning_rate": 6.627267457478546e-06,
      "loss": 0.0156,
      "step": 23870
    },
    {
      "epoch": 0.33762671605706285,
      "grad_norm": 0.0,
      "learning_rate": 6.625853610258876e-06,
      "loss": 0.0132,
      "step": 23880
    },
    {
      "epoch": 0.3377681007790298,
      "grad_norm": 0.11704470962285995,
      "learning_rate": 6.624439763039207e-06,
      "loss": 0.0,
      "step": 23890
    },
    {
      "epoch": 0.33790948550099675,
      "grad_norm": 3.690584890136961e-06,
      "learning_rate": 6.6230259158195374e-06,
      "loss": 0.0,
      "step": 23900
    },
    {
      "epoch": 0.3380508702229637,
      "grad_norm": 0.0,
      "learning_rate": 6.6216120685998675e-06,
      "loss": 0.0,
      "step": 23910
    },
    {
      "epoch": 0.33819225494493066,
      "grad_norm": 0.0,
      "learning_rate": 6.620198221380198e-06,
      "loss": 0.0044,
      "step": 23920
    },
    {
      "epoch": 0.3383336396668976,
      "grad_norm": 4.202532636554679e-06,
      "learning_rate": 6.618784374160528e-06,
      "loss": 0.0,
      "step": 23930
    },
    {
      "epoch": 0.33847502438886456,
      "grad_norm": 0.0,
      "learning_rate": 6.617370526940859e-06,
      "loss": 0.0002,
      "step": 23940
    },
    {
      "epoch": 0.33861640911083146,
      "grad_norm": 0.0,
      "learning_rate": 6.615956679721189e-06,
      "loss": 0.0,
      "step": 23950
    },
    {
      "epoch": 0.3387577938327984,
      "grad_norm": 0.0,
      "learning_rate": 6.61454283250152e-06,
      "loss": 0.0011,
      "step": 23960
    },
    {
      "epoch": 0.33889917855476537,
      "grad_norm": 0.0,
      "learning_rate": 6.613128985281852e-06,
      "loss": 0.0,
      "step": 23970
    },
    {
      "epoch": 0.3390405632767323,
      "grad_norm": 3.3809104934334755e-05,
      "learning_rate": 6.611715138062182e-06,
      "loss": 0.0225,
      "step": 23980
    },
    {
      "epoch": 0.3391819479986993,
      "grad_norm": 5.903702060550131e-08,
      "learning_rate": 6.610301290842513e-06,
      "loss": 0.0,
      "step": 23990
    },
    {
      "epoch": 0.3393233327206662,
      "grad_norm": 337.6971740722656,
      "learning_rate": 6.608887443622843e-06,
      "loss": 0.0701,
      "step": 24000
    },
    {
      "epoch": 0.3394647174426331,
      "grad_norm": 0.0,
      "learning_rate": 6.6074735964031735e-06,
      "loss": 0.0001,
      "step": 24010
    },
    {
      "epoch": 0.3396061021646001,
      "grad_norm": 0.0,
      "learning_rate": 6.6060597491835035e-06,
      "loss": 0.0001,
      "step": 24020
    },
    {
      "epoch": 0.33974748688656703,
      "grad_norm": 0.0,
      "learning_rate": 6.604645901963834e-06,
      "loss": 0.0,
      "step": 24030
    },
    {
      "epoch": 0.339888871608534,
      "grad_norm": 0.0002192558895330876,
      "learning_rate": 6.603232054744165e-06,
      "loss": 0.0,
      "step": 24040
    },
    {
      "epoch": 0.34003025633050094,
      "grad_norm": 0.0,
      "learning_rate": 6.601818207524495e-06,
      "loss": 0.0113,
      "step": 24050
    },
    {
      "epoch": 0.3401716410524679,
      "grad_norm": 0.0,
      "learning_rate": 6.600404360304826e-06,
      "loss": 0.1234,
      "step": 24060
    },
    {
      "epoch": 0.34031302577443484,
      "grad_norm": 0.0,
      "learning_rate": 6.598990513085156e-06,
      "loss": 0.0001,
      "step": 24070
    },
    {
      "epoch": 0.34045441049640174,
      "grad_norm": 0.5791575312614441,
      "learning_rate": 6.597576665865487e-06,
      "loss": 0.0002,
      "step": 24080
    },
    {
      "epoch": 0.3405957952183687,
      "grad_norm": 0.0,
      "learning_rate": 6.596162818645818e-06,
      "loss": 0.0,
      "step": 24090
    },
    {
      "epoch": 0.34073717994033564,
      "grad_norm": 0.0,
      "learning_rate": 6.594748971426148e-06,
      "loss": 0.0,
      "step": 24100
    },
    {
      "epoch": 0.3408785646623026,
      "grad_norm": 5.4417622408209354e-08,
      "learning_rate": 6.5933351242064795e-06,
      "loss": 0.0,
      "step": 24110
    },
    {
      "epoch": 0.34101994938426955,
      "grad_norm": 0.11263063549995422,
      "learning_rate": 6.591921276986809e-06,
      "loss": 0.0003,
      "step": 24120
    },
    {
      "epoch": 0.3411613341062365,
      "grad_norm": 0.0,
      "learning_rate": 6.59050742976714e-06,
      "loss": 0.0072,
      "step": 24130
    },
    {
      "epoch": 0.3413027188282034,
      "grad_norm": 0.00853163655847311,
      "learning_rate": 6.58909358254747e-06,
      "loss": 0.0002,
      "step": 24140
    },
    {
      "epoch": 0.34144410355017035,
      "grad_norm": 0.0,
      "learning_rate": 6.587679735327801e-06,
      "loss": 0.0,
      "step": 24150
    },
    {
      "epoch": 0.3415854882721373,
      "grad_norm": 0.0,
      "learning_rate": 6.586265888108132e-06,
      "loss": 0.0,
      "step": 24160
    },
    {
      "epoch": 0.34172687299410426,
      "grad_norm": 5.458611895825527e-10,
      "learning_rate": 6.584852040888462e-06,
      "loss": 0.0,
      "step": 24170
    },
    {
      "epoch": 0.3418682577160712,
      "grad_norm": 1.0040367669716943e-05,
      "learning_rate": 6.583438193668793e-06,
      "loss": 0.0003,
      "step": 24180
    },
    {
      "epoch": 0.34200964243803816,
      "grad_norm": 0.00021252315491437912,
      "learning_rate": 6.582024346449123e-06,
      "loss": 0.3768,
      "step": 24190
    },
    {
      "epoch": 0.3421510271600051,
      "grad_norm": 0.0,
      "learning_rate": 6.580610499229454e-06,
      "loss": 0.0001,
      "step": 24200
    },
    {
      "epoch": 0.342292411881972,
      "grad_norm": 0.0,
      "learning_rate": 6.579196652009784e-06,
      "loss": 0.0,
      "step": 24210
    },
    {
      "epoch": 0.34243379660393897,
      "grad_norm": 0.06635252386331558,
      "learning_rate": 6.577782804790115e-06,
      "loss": 0.0199,
      "step": 24220
    },
    {
      "epoch": 0.3425751813259059,
      "grad_norm": 0.0,
      "learning_rate": 6.5763689575704455e-06,
      "loss": 1.075,
      "step": 24230
    },
    {
      "epoch": 0.3427165660478729,
      "grad_norm": 0.060891203582286835,
      "learning_rate": 6.5749551103507756e-06,
      "loss": 0.0,
      "step": 24240
    },
    {
      "epoch": 0.3428579507698398,
      "grad_norm": 47.08693313598633,
      "learning_rate": 6.573541263131107e-06,
      "loss": 0.018,
      "step": 24250
    },
    {
      "epoch": 0.3429993354918068,
      "grad_norm": 0.0,
      "learning_rate": 6.5721274159114364e-06,
      "loss": 0.0036,
      "step": 24260
    },
    {
      "epoch": 0.3431407202137737,
      "grad_norm": 0.0,
      "learning_rate": 6.570713568691768e-06,
      "loss": 0.0,
      "step": 24270
    },
    {
      "epoch": 0.34328210493574063,
      "grad_norm": 0.0,
      "learning_rate": 6.569299721472097e-06,
      "loss": 0.0,
      "step": 24280
    },
    {
      "epoch": 0.3434234896577076,
      "grad_norm": 0.026406817138195038,
      "learning_rate": 6.567885874252429e-06,
      "loss": 0.0,
      "step": 24290
    },
    {
      "epoch": 0.34356487437967453,
      "grad_norm": 0.0,
      "learning_rate": 6.56647202703276e-06,
      "loss": 0.0139,
      "step": 24300
    },
    {
      "epoch": 0.3437062591016415,
      "grad_norm": 0.0,
      "learning_rate": 6.56505817981309e-06,
      "loss": 0.0,
      "step": 24310
    },
    {
      "epoch": 0.34384764382360844,
      "grad_norm": 0.005711978301405907,
      "learning_rate": 6.563644332593421e-06,
      "loss": 0.0,
      "step": 24320
    },
    {
      "epoch": 0.3439890285455754,
      "grad_norm": 0.0005349503480829298,
      "learning_rate": 6.562230485373751e-06,
      "loss": 0.0,
      "step": 24330
    },
    {
      "epoch": 0.3441304132675423,
      "grad_norm": 0.019354533404111862,
      "learning_rate": 6.560816638154082e-06,
      "loss": 0.0,
      "step": 24340
    },
    {
      "epoch": 0.34427179798950924,
      "grad_norm": 0.0,
      "learning_rate": 6.5594027909344124e-06,
      "loss": 0.1904,
      "step": 24350
    },
    {
      "epoch": 0.3444131827114762,
      "grad_norm": 0.00010112932068295777,
      "learning_rate": 6.5579889437147424e-06,
      "loss": 0.0,
      "step": 24360
    },
    {
      "epoch": 0.34455456743344315,
      "grad_norm": 0.0,
      "learning_rate": 6.556575096495073e-06,
      "loss": 0.0,
      "step": 24370
    },
    {
      "epoch": 0.3446959521554101,
      "grad_norm": 0.0,
      "learning_rate": 6.555161249275403e-06,
      "loss": 0.0006,
      "step": 24380
    },
    {
      "epoch": 0.34483733687737705,
      "grad_norm": 0.0,
      "learning_rate": 6.553747402055734e-06,
      "loss": 0.0059,
      "step": 24390
    },
    {
      "epoch": 0.34497872159934395,
      "grad_norm": 0.03449573740363121,
      "learning_rate": 6.552333554836064e-06,
      "loss": 0.0001,
      "step": 24400
    },
    {
      "epoch": 0.3451201063213109,
      "grad_norm": 4.672752424994542e-07,
      "learning_rate": 6.550919707616396e-06,
      "loss": 0.0016,
      "step": 24410
    },
    {
      "epoch": 0.34526149104327786,
      "grad_norm": 0.0018448916962370276,
      "learning_rate": 6.549505860396727e-06,
      "loss": 0.0,
      "step": 24420
    },
    {
      "epoch": 0.3454028757652448,
      "grad_norm": 0.0,
      "learning_rate": 6.548092013177057e-06,
      "loss": 0.0001,
      "step": 24430
    },
    {
      "epoch": 0.34554426048721176,
      "grad_norm": 0.0,
      "learning_rate": 6.546678165957388e-06,
      "loss": 0.0,
      "step": 24440
    },
    {
      "epoch": 0.3456856452091787,
      "grad_norm": 0.0,
      "learning_rate": 6.545264318737718e-06,
      "loss": 0.0,
      "step": 24450
    },
    {
      "epoch": 0.3458270299311456,
      "grad_norm": 0.002108534798026085,
      "learning_rate": 6.5438504715180485e-06,
      "loss": 0.0,
      "step": 24460
    },
    {
      "epoch": 0.34596841465311257,
      "grad_norm": 0.0,
      "learning_rate": 6.5424366242983785e-06,
      "loss": 0.8141,
      "step": 24470
    },
    {
      "epoch": 0.3461097993750795,
      "grad_norm": 0.0,
      "learning_rate": 6.541022777078709e-06,
      "loss": 0.0,
      "step": 24480
    },
    {
      "epoch": 0.34625118409704647,
      "grad_norm": 0.0,
      "learning_rate": 6.53960892985904e-06,
      "loss": 0.0064,
      "step": 24490
    },
    {
      "epoch": 0.3463925688190134,
      "grad_norm": 0.0,
      "learning_rate": 6.53819508263937e-06,
      "loss": 0.0921,
      "step": 24500
    },
    {
      "epoch": 0.3465339535409804,
      "grad_norm": 1.0122405313950367e-07,
      "learning_rate": 6.536781235419701e-06,
      "loss": 0.0,
      "step": 24510
    },
    {
      "epoch": 0.34667533826294733,
      "grad_norm": 0.0,
      "learning_rate": 6.535367388200031e-06,
      "loss": 0.0092,
      "step": 24520
    },
    {
      "epoch": 0.3468167229849142,
      "grad_norm": 0.0,
      "learning_rate": 6.533953540980362e-06,
      "loss": 0.0,
      "step": 24530
    },
    {
      "epoch": 0.3469581077068812,
      "grad_norm": 0.0,
      "learning_rate": 6.532539693760694e-06,
      "loss": 0.0034,
      "step": 24540
    },
    {
      "epoch": 0.34709949242884813,
      "grad_norm": 3.175860570081568e-07,
      "learning_rate": 6.531125846541024e-06,
      "loss": 0.0,
      "step": 24550
    },
    {
      "epoch": 0.3472408771508151,
      "grad_norm": 0.0,
      "learning_rate": 6.5297119993213545e-06,
      "loss": 0.6953,
      "step": 24560
    },
    {
      "epoch": 0.34738226187278204,
      "grad_norm": 0.0,
      "learning_rate": 6.5282981521016845e-06,
      "loss": 0.0,
      "step": 24570
    },
    {
      "epoch": 0.347523646594749,
      "grad_norm": 0.0,
      "learning_rate": 6.526884304882015e-06,
      "loss": 0.149,
      "step": 24580
    },
    {
      "epoch": 0.3476650313167159,
      "grad_norm": 0.0,
      "learning_rate": 6.525470457662345e-06,
      "loss": 0.0,
      "step": 24590
    },
    {
      "epoch": 0.34780641603868284,
      "grad_norm": 0.3189041316509247,
      "learning_rate": 6.524056610442676e-06,
      "loss": 0.0001,
      "step": 24600
    },
    {
      "epoch": 0.3479478007606498,
      "grad_norm": 0.020992513746023178,
      "learning_rate": 6.522642763223007e-06,
      "loss": 0.0,
      "step": 24610
    },
    {
      "epoch": 0.34808918548261675,
      "grad_norm": 0.0,
      "learning_rate": 6.521228916003337e-06,
      "loss": 0.0,
      "step": 24620
    },
    {
      "epoch": 0.3482305702045837,
      "grad_norm": 0.0,
      "learning_rate": 6.519815068783668e-06,
      "loss": 0.0,
      "step": 24630
    },
    {
      "epoch": 0.34837195492655065,
      "grad_norm": 6.545218145248555e-09,
      "learning_rate": 6.518401221563998e-06,
      "loss": 0.0001,
      "step": 24640
    },
    {
      "epoch": 0.3485133396485176,
      "grad_norm": 8.026245268411003e-07,
      "learning_rate": 6.516987374344329e-06,
      "loss": 0.0,
      "step": 24650
    },
    {
      "epoch": 0.3486547243704845,
      "grad_norm": 0.032301582396030426,
      "learning_rate": 6.515573527124659e-06,
      "loss": 0.0002,
      "step": 24660
    },
    {
      "epoch": 0.34879610909245146,
      "grad_norm": 0.0,
      "learning_rate": 6.51415967990499e-06,
      "loss": 0.0,
      "step": 24670
    },
    {
      "epoch": 0.3489374938144184,
      "grad_norm": 0.0,
      "learning_rate": 6.512745832685321e-06,
      "loss": 0.0,
      "step": 24680
    },
    {
      "epoch": 0.34907887853638536,
      "grad_norm": 2.794809574879764e-07,
      "learning_rate": 6.5113319854656505e-06,
      "loss": 0.0001,
      "step": 24690
    },
    {
      "epoch": 0.3492202632583523,
      "grad_norm": 0.0036798708606511354,
      "learning_rate": 6.510059522967949e-06,
      "loss": 0.8481,
      "step": 24700
    },
    {
      "epoch": 0.34936164798031927,
      "grad_norm": 0.0,
      "learning_rate": 6.5086456757482795e-06,
      "loss": 0.001,
      "step": 24710
    },
    {
      "epoch": 0.34950303270228616,
      "grad_norm": 0.0,
      "learning_rate": 6.5072318285286095e-06,
      "loss": 0.0006,
      "step": 24720
    },
    {
      "epoch": 0.3496444174242531,
      "grad_norm": 0.0,
      "learning_rate": 6.50581798130894e-06,
      "loss": 0.0,
      "step": 24730
    },
    {
      "epoch": 0.34978580214622007,
      "grad_norm": 0.0,
      "learning_rate": 6.50440413408927e-06,
      "loss": 0.0,
      "step": 24740
    },
    {
      "epoch": 0.349927186868187,
      "grad_norm": 0.0,
      "learning_rate": 6.502990286869601e-06,
      "loss": 0.0,
      "step": 24750
    },
    {
      "epoch": 0.350068571590154,
      "grad_norm": 0.0,
      "learning_rate": 6.501576439649931e-06,
      "loss": 0.0002,
      "step": 24760
    },
    {
      "epoch": 0.35020995631212093,
      "grad_norm": 0.0,
      "learning_rate": 6.500162592430262e-06,
      "loss": 0.0092,
      "step": 24770
    },
    {
      "epoch": 0.3503513410340879,
      "grad_norm": 2.602628774184268e-05,
      "learning_rate": 6.498748745210594e-06,
      "loss": 0.0,
      "step": 24780
    },
    {
      "epoch": 0.3504927257560548,
      "grad_norm": 26.188358306884766,
      "learning_rate": 6.497334897990924e-06,
      "loss": 0.006,
      "step": 24790
    },
    {
      "epoch": 0.35063411047802173,
      "grad_norm": 0.00032009556889533997,
      "learning_rate": 6.495921050771255e-06,
      "loss": 0.0,
      "step": 24800
    },
    {
      "epoch": 0.3507754951999887,
      "grad_norm": 0.0,
      "learning_rate": 6.494507203551585e-06,
      "loss": 0.0,
      "step": 24810
    },
    {
      "epoch": 0.35091687992195564,
      "grad_norm": 4.4700229628880095e-10,
      "learning_rate": 6.4930933563319155e-06,
      "loss": 0.0002,
      "step": 24820
    },
    {
      "epoch": 0.3510582646439226,
      "grad_norm": 0.0,
      "learning_rate": 6.4916795091122456e-06,
      "loss": 0.0001,
      "step": 24830
    },
    {
      "epoch": 0.35119964936588954,
      "grad_norm": 8.498043462168425e-05,
      "learning_rate": 6.490265661892576e-06,
      "loss": 0.0,
      "step": 24840
    },
    {
      "epoch": 0.35134103408785644,
      "grad_norm": 0.0606556162238121,
      "learning_rate": 6.488851814672907e-06,
      "loss": 0.0009,
      "step": 24850
    },
    {
      "epoch": 0.3514824188098234,
      "grad_norm": 0.0016254197107627988,
      "learning_rate": 6.487437967453237e-06,
      "loss": 0.0003,
      "step": 24860
    },
    {
      "epoch": 0.35162380353179035,
      "grad_norm": 0.0,
      "learning_rate": 6.486024120233568e-06,
      "loss": 0.0274,
      "step": 24870
    },
    {
      "epoch": 0.3517651882537573,
      "grad_norm": 0.007412867620587349,
      "learning_rate": 6.484610273013898e-06,
      "loss": 0.0,
      "step": 24880
    },
    {
      "epoch": 0.35190657297572425,
      "grad_norm": 0.0,
      "learning_rate": 6.483196425794229e-06,
      "loss": 0.0,
      "step": 24890
    },
    {
      "epoch": 0.3520479576976912,
      "grad_norm": 0.0,
      "learning_rate": 6.481782578574561e-06,
      "loss": 0.0,
      "step": 24900
    },
    {
      "epoch": 0.35218934241965816,
      "grad_norm": 2.907835006713867,
      "learning_rate": 6.48036873135489e-06,
      "loss": 0.0005,
      "step": 24910
    },
    {
      "epoch": 0.35233072714162506,
      "grad_norm": 0.0,
      "learning_rate": 6.4789548841352216e-06,
      "loss": 0.0007,
      "step": 24920
    },
    {
      "epoch": 0.352472111863592,
      "grad_norm": 0.04210911691188812,
      "learning_rate": 6.477541036915551e-06,
      "loss": 0.0113,
      "step": 24930
    },
    {
      "epoch": 0.35261349658555896,
      "grad_norm": 0.0,
      "learning_rate": 6.4761271896958824e-06,
      "loss": 0.0008,
      "step": 24940
    },
    {
      "epoch": 0.3527548813075259,
      "grad_norm": 0.07303735613822937,
      "learning_rate": 6.4747133424762124e-06,
      "loss": 0.0011,
      "step": 24950
    },
    {
      "epoch": 0.35289626602949287,
      "grad_norm": 0.0,
      "learning_rate": 6.473299495256543e-06,
      "loss": 0.0,
      "step": 24960
    },
    {
      "epoch": 0.3530376507514598,
      "grad_norm": 7.215125208404061e-08,
      "learning_rate": 6.471885648036874e-06,
      "loss": 0.0,
      "step": 24970
    },
    {
      "epoch": 0.3531790354734267,
      "grad_norm": 0.0,
      "learning_rate": 6.470471800817204e-06,
      "loss": 0.0,
      "step": 24980
    },
    {
      "epoch": 0.35332042019539367,
      "grad_norm": 0.0,
      "learning_rate": 6.469057953597535e-06,
      "loss": 0.0,
      "step": 24990
    },
    {
      "epoch": 0.3534618049173606,
      "grad_norm": 0.0,
      "learning_rate": 6.467644106377865e-06,
      "loss": 0.0,
      "step": 25000
    },
    {
      "epoch": 0.3536031896393276,
      "grad_norm": 4.885861715209217e-10,
      "learning_rate": 6.466230259158196e-06,
      "loss": 0.0,
      "step": 25010
    },
    {
      "epoch": 0.35374457436129453,
      "grad_norm": 0.0,
      "learning_rate": 6.464816411938526e-06,
      "loss": 0.0,
      "step": 25020
    },
    {
      "epoch": 0.3538859590832615,
      "grad_norm": 0.0,
      "learning_rate": 6.463402564718857e-06,
      "loss": 0.0,
      "step": 25030
    },
    {
      "epoch": 0.35402734380522843,
      "grad_norm": 0.0,
      "learning_rate": 6.461988717499188e-06,
      "loss": 0.0001,
      "step": 25040
    },
    {
      "epoch": 0.35416872852719533,
      "grad_norm": 0.0,
      "learning_rate": 6.460574870279518e-06,
      "loss": 0.0123,
      "step": 25050
    },
    {
      "epoch": 0.3543101132491623,
      "grad_norm": 0.0,
      "learning_rate": 6.459161023059849e-06,
      "loss": 0.4211,
      "step": 25060
    },
    {
      "epoch": 0.35445149797112924,
      "grad_norm": 0.0,
      "learning_rate": 6.4577471758401785e-06,
      "loss": 0.0006,
      "step": 25070
    },
    {
      "epoch": 0.3545928826930962,
      "grad_norm": 0.0,
      "learning_rate": 6.45633332862051e-06,
      "loss": 0.0,
      "step": 25080
    },
    {
      "epoch": 0.35473426741506314,
      "grad_norm": 0.00013334042159840465,
      "learning_rate": 6.454919481400841e-06,
      "loss": 0.0,
      "step": 25090
    },
    {
      "epoch": 0.3548756521370301,
      "grad_norm": 0.00047323975013569,
      "learning_rate": 6.453505634181171e-06,
      "loss": 0.0,
      "step": 25100
    },
    {
      "epoch": 0.355017036858997,
      "grad_norm": 709.3038330078125,
      "learning_rate": 6.452091786961502e-06,
      "loss": 0.4396,
      "step": 25110
    },
    {
      "epoch": 0.35515842158096395,
      "grad_norm": 0.0,
      "learning_rate": 6.450677939741832e-06,
      "loss": 0.0005,
      "step": 25120
    },
    {
      "epoch": 0.3552998063029309,
      "grad_norm": 0.0,
      "learning_rate": 6.449264092522163e-06,
      "loss": 0.0,
      "step": 25130
    },
    {
      "epoch": 0.35544119102489785,
      "grad_norm": 0.007971511222422123,
      "learning_rate": 6.447850245302493e-06,
      "loss": 0.0,
      "step": 25140
    },
    {
      "epoch": 0.3555825757468648,
      "grad_norm": 0.0,
      "learning_rate": 6.446436398082824e-06,
      "loss": 0.006,
      "step": 25150
    },
    {
      "epoch": 0.35572396046883176,
      "grad_norm": 0.0,
      "learning_rate": 6.4450225508631545e-06,
      "loss": 0.0,
      "step": 25160
    },
    {
      "epoch": 0.3558653451907987,
      "grad_norm": 8.720893451830136e-10,
      "learning_rate": 6.4436087036434845e-06,
      "loss": 0.0213,
      "step": 25170
    },
    {
      "epoch": 0.3560067299127656,
      "grad_norm": 0.0,
      "learning_rate": 6.442194856423815e-06,
      "loss": 0.0008,
      "step": 25180
    },
    {
      "epoch": 0.35614811463473256,
      "grad_norm": 0.0,
      "learning_rate": 6.440781009204145e-06,
      "loss": 0.0,
      "step": 25190
    },
    {
      "epoch": 0.3562894993566995,
      "grad_norm": 0.0,
      "learning_rate": 6.439367161984476e-06,
      "loss": 0.0,
      "step": 25200
    },
    {
      "epoch": 0.35643088407866647,
      "grad_norm": 7.70989790908061e-05,
      "learning_rate": 6.437953314764806e-06,
      "loss": 0.0,
      "step": 25210
    },
    {
      "epoch": 0.3565722688006334,
      "grad_norm": 0.0,
      "learning_rate": 6.436539467545138e-06,
      "loss": 0.0088,
      "step": 25220
    },
    {
      "epoch": 0.35671365352260037,
      "grad_norm": 0.0,
      "learning_rate": 6.435125620325469e-06,
      "loss": 0.0167,
      "step": 25230
    },
    {
      "epoch": 0.35685503824456727,
      "grad_norm": 0.0,
      "learning_rate": 6.433711773105799e-06,
      "loss": 0.0,
      "step": 25240
    },
    {
      "epoch": 0.3569964229665342,
      "grad_norm": 0.0,
      "learning_rate": 6.43229792588613e-06,
      "loss": 0.0,
      "step": 25250
    },
    {
      "epoch": 0.3571378076885012,
      "grad_norm": 3.7034469642094336e-06,
      "learning_rate": 6.43088407866646e-06,
      "loss": 0.0121,
      "step": 25260
    },
    {
      "epoch": 0.3572791924104681,
      "grad_norm": 4.1856363708348e-08,
      "learning_rate": 6.4294702314467905e-06,
      "loss": 0.0007,
      "step": 25270
    },
    {
      "epoch": 0.3574205771324351,
      "grad_norm": 9.147516077945284e-10,
      "learning_rate": 6.4280563842271205e-06,
      "loss": 0.0,
      "step": 25280
    },
    {
      "epoch": 0.35756196185440203,
      "grad_norm": 0.024559009820222855,
      "learning_rate": 6.426642537007451e-06,
      "loss": 0.0,
      "step": 25290
    },
    {
      "epoch": 0.35770334657636893,
      "grad_norm": 0.0,
      "learning_rate": 6.425228689787782e-06,
      "loss": 0.0131,
      "step": 25300
    },
    {
      "epoch": 0.3578447312983359,
      "grad_norm": 9.823027946609386e-10,
      "learning_rate": 6.423814842568112e-06,
      "loss": 0.0,
      "step": 25310
    },
    {
      "epoch": 0.35798611602030284,
      "grad_norm": 3.1790312959856237e-07,
      "learning_rate": 6.422400995348443e-06,
      "loss": 0.0,
      "step": 25320
    },
    {
      "epoch": 0.3581275007422698,
      "grad_norm": 597.3960571289062,
      "learning_rate": 6.420987148128773e-06,
      "loss": 0.0808,
      "step": 25330
    },
    {
      "epoch": 0.35826888546423674,
      "grad_norm": 0.0,
      "learning_rate": 6.419573300909104e-06,
      "loss": 0.0001,
      "step": 25340
    },
    {
      "epoch": 0.3584102701862037,
      "grad_norm": 0.0,
      "learning_rate": 6.418159453689436e-06,
      "loss": 0.0,
      "step": 25350
    },
    {
      "epoch": 0.35855165490817065,
      "grad_norm": 0.0,
      "learning_rate": 6.416745606469766e-06,
      "loss": 0.0001,
      "step": 25360
    },
    {
      "epoch": 0.35869303963013754,
      "grad_norm": 0.0,
      "learning_rate": 6.4153317592500966e-06,
      "loss": 0.0009,
      "step": 25370
    },
    {
      "epoch": 0.3588344243521045,
      "grad_norm": 0.0,
      "learning_rate": 6.4139179120304266e-06,
      "loss": 0.0002,
      "step": 25380
    },
    {
      "epoch": 0.35897580907407145,
      "grad_norm": 0.0,
      "learning_rate": 6.4125040648107574e-06,
      "loss": 0.0002,
      "step": 25390
    },
    {
      "epoch": 0.3591171937960384,
      "grad_norm": 1.3610708265332505e-05,
      "learning_rate": 6.4110902175910874e-06,
      "loss": 0.0018,
      "step": 25400
    },
    {
      "epoch": 0.35925857851800536,
      "grad_norm": 0.0003211074217688292,
      "learning_rate": 6.409676370371418e-06,
      "loss": 0.0,
      "step": 25410
    },
    {
      "epoch": 0.3593999632399723,
      "grad_norm": 0.0,
      "learning_rate": 6.408262523151749e-06,
      "loss": 0.0,
      "step": 25420
    },
    {
      "epoch": 0.3595413479619392,
      "grad_norm": 0.0,
      "learning_rate": 6.406848675932079e-06,
      "loss": 0.0028,
      "step": 25430
    },
    {
      "epoch": 0.35968273268390616,
      "grad_norm": 18.24318504333496,
      "learning_rate": 6.40543482871241e-06,
      "loss": 0.0028,
      "step": 25440
    },
    {
      "epoch": 0.3598241174058731,
      "grad_norm": 0.0,
      "learning_rate": 6.40402098149274e-06,
      "loss": 0.0,
      "step": 25450
    },
    {
      "epoch": 0.35996550212784006,
      "grad_norm": 0.0,
      "learning_rate": 6.402607134273071e-06,
      "loss": 0.0,
      "step": 25460
    },
    {
      "epoch": 0.360106886849807,
      "grad_norm": 0.0,
      "learning_rate": 6.401193287053401e-06,
      "loss": 0.3863,
      "step": 25470
    },
    {
      "epoch": 0.36024827157177397,
      "grad_norm": 0.0,
      "learning_rate": 6.399779439833732e-06,
      "loss": 0.0316,
      "step": 25480
    },
    {
      "epoch": 0.3603896562937409,
      "grad_norm": 2.1589479729300365e-05,
      "learning_rate": 6.3983655926140634e-06,
      "loss": 0.0,
      "step": 25490
    },
    {
      "epoch": 0.3605310410157078,
      "grad_norm": 0.0,
      "learning_rate": 6.39709313011636e-06,
      "loss": 0.3286,
      "step": 25500
    },
    {
      "epoch": 0.3606724257376748,
      "grad_norm": 71.11280822753906,
      "learning_rate": 6.395679282896691e-06,
      "loss": 0.0018,
      "step": 25510
    },
    {
      "epoch": 0.3608138104596417,
      "grad_norm": 0.011547685600817204,
      "learning_rate": 6.3942654356770216e-06,
      "loss": 0.0,
      "step": 25520
    },
    {
      "epoch": 0.3609551951816087,
      "grad_norm": 0.0,
      "learning_rate": 6.392851588457352e-06,
      "loss": 0.0001,
      "step": 25530
    },
    {
      "epoch": 0.36109657990357563,
      "grad_norm": 0.0,
      "learning_rate": 6.3914377412376824e-06,
      "loss": 0.0,
      "step": 25540
    },
    {
      "epoch": 0.3612379646255426,
      "grad_norm": 7.872540663811378e-06,
      "learning_rate": 6.3900238940180125e-06,
      "loss": 0.0001,
      "step": 25550
    },
    {
      "epoch": 0.3613793493475095,
      "grad_norm": 0.0,
      "learning_rate": 6.388610046798343e-06,
      "loss": 0.0,
      "step": 25560
    },
    {
      "epoch": 0.36152073406947643,
      "grad_norm": 9.369584859086899e-07,
      "learning_rate": 6.387196199578673e-06,
      "loss": 0.6176,
      "step": 25570
    },
    {
      "epoch": 0.3616621187914434,
      "grad_norm": 15.503994941711426,
      "learning_rate": 6.385782352359004e-06,
      "loss": 0.0035,
      "step": 25580
    },
    {
      "epoch": 0.36180350351341034,
      "grad_norm": 1.3572040188591927e-05,
      "learning_rate": 6.384368505139336e-06,
      "loss": 0.0,
      "step": 25590
    },
    {
      "epoch": 0.3619448882353773,
      "grad_norm": 1.1797681054304121e-07,
      "learning_rate": 6.382954657919666e-06,
      "loss": 0.0307,
      "step": 25600
    },
    {
      "epoch": 0.36208627295734425,
      "grad_norm": 1.9731027123270906e-07,
      "learning_rate": 6.381540810699997e-06,
      "loss": 0.0,
      "step": 25610
    },
    {
      "epoch": 0.3622276576793112,
      "grad_norm": 1.896857559913201e-09,
      "learning_rate": 6.380126963480327e-06,
      "loss": 0.1201,
      "step": 25620
    },
    {
      "epoch": 0.3623690424012781,
      "grad_norm": 0.0,
      "learning_rate": 6.378713116260658e-06,
      "loss": 0.0004,
      "step": 25630
    },
    {
      "epoch": 0.36251042712324505,
      "grad_norm": 1.5953669390000869e-06,
      "learning_rate": 6.3772992690409885e-06,
      "loss": 0.0,
      "step": 25640
    },
    {
      "epoch": 0.362651811845212,
      "grad_norm": 5.703493118286133,
      "learning_rate": 6.3758854218213185e-06,
      "loss": 0.0008,
      "step": 25650
    },
    {
      "epoch": 0.36279319656717895,
      "grad_norm": 0.0,
      "learning_rate": 6.374471574601649e-06,
      "loss": 0.0,
      "step": 25660
    },
    {
      "epoch": 0.3629345812891459,
      "grad_norm": 0.0,
      "learning_rate": 6.373057727381979e-06,
      "loss": 0.0001,
      "step": 25670
    },
    {
      "epoch": 0.36307596601111286,
      "grad_norm": 0.0,
      "learning_rate": 6.37164388016231e-06,
      "loss": 0.0006,
      "step": 25680
    },
    {
      "epoch": 0.36321735073307976,
      "grad_norm": 0.0,
      "learning_rate": 6.37023003294264e-06,
      "loss": 0.0156,
      "step": 25690
    },
    {
      "epoch": 0.3633587354550467,
      "grad_norm": 68.4031982421875,
      "learning_rate": 6.368816185722971e-06,
      "loss": 0.0162,
      "step": 25700
    },
    {
      "epoch": 0.36350012017701366,
      "grad_norm": 5.8223249652655795e-06,
      "learning_rate": 6.367543723225269e-06,
      "loss": 0.0303,
      "step": 25710
    },
    {
      "epoch": 0.3636415048989806,
      "grad_norm": 0.0,
      "learning_rate": 6.366129876005599e-06,
      "loss": 0.0207,
      "step": 25720
    },
    {
      "epoch": 0.36378288962094757,
      "grad_norm": 0.0,
      "learning_rate": 6.36471602878593e-06,
      "loss": 0.0001,
      "step": 25730
    },
    {
      "epoch": 0.3639242743429145,
      "grad_norm": 3.230453876312822e-05,
      "learning_rate": 6.363302181566261e-06,
      "loss": 0.0,
      "step": 25740
    },
    {
      "epoch": 0.3640656590648815,
      "grad_norm": 0.0,
      "learning_rate": 6.361888334346591e-06,
      "loss": 0.0,
      "step": 25750
    },
    {
      "epoch": 0.3642070437868484,
      "grad_norm": 0.0,
      "learning_rate": 6.360474487126922e-06,
      "loss": 0.641,
      "step": 25760
    },
    {
      "epoch": 0.3643484285088153,
      "grad_norm": 0.0,
      "learning_rate": 6.359060639907252e-06,
      "loss": 0.0003,
      "step": 25770
    },
    {
      "epoch": 0.3644898132307823,
      "grad_norm": 0.0,
      "learning_rate": 6.357646792687583e-06,
      "loss": 0.0,
      "step": 25780
    },
    {
      "epoch": 0.36463119795274923,
      "grad_norm": 0.00021032449149060994,
      "learning_rate": 6.356232945467913e-06,
      "loss": 0.0,
      "step": 25790
    },
    {
      "epoch": 0.3647725826747162,
      "grad_norm": 0.0,
      "learning_rate": 6.3548190982482435e-06,
      "loss": 0.0,
      "step": 25800
    },
    {
      "epoch": 0.36491396739668314,
      "grad_norm": 2.5714054572745226e-05,
      "learning_rate": 6.353405251028575e-06,
      "loss": 0.0,
      "step": 25810
    },
    {
      "epoch": 0.36505535211865003,
      "grad_norm": 2.128634696418885e-05,
      "learning_rate": 6.351991403808904e-06,
      "loss": 0.0,
      "step": 25820
    },
    {
      "epoch": 0.365196736840617,
      "grad_norm": 0.0,
      "learning_rate": 6.350577556589236e-06,
      "loss": 0.0,
      "step": 25830
    },
    {
      "epoch": 0.36533812156258394,
      "grad_norm": 0.0,
      "learning_rate": 6.349163709369566e-06,
      "loss": 0.0,
      "step": 25840
    },
    {
      "epoch": 0.3654795062845509,
      "grad_norm": 0.0,
      "learning_rate": 6.347749862149897e-06,
      "loss": 0.0035,
      "step": 25850
    },
    {
      "epoch": 0.36562089100651785,
      "grad_norm": 0.0014018273213878274,
      "learning_rate": 6.346336014930227e-06,
      "loss": 0.0644,
      "step": 25860
    },
    {
      "epoch": 0.3657622757284848,
      "grad_norm": Infinity,
      "learning_rate": 6.345204937154492e-06,
      "loss": 2.5226,
      "step": 25870
    },
    {
      "epoch": 0.36590366045045175,
      "grad_norm": 1805.8955078125,
      "learning_rate": 6.344073859378757e-06,
      "loss": 4.3525,
      "step": 25880
    },
    {
      "epoch": 0.36604504517241865,
      "grad_norm": 1.6051337297540158e-05,
      "learning_rate": 6.342801396881053e-06,
      "loss": 5.9345,
      "step": 25890
    },
    {
      "epoch": 0.3661864298943856,
      "grad_norm": 0.6561949849128723,
      "learning_rate": 6.341387549661384e-06,
      "loss": 5.4481,
      "step": 25900
    },
    {
      "epoch": 0.36632781461635255,
      "grad_norm": 2.0726037064378033e-07,
      "learning_rate": 6.339973702441715e-06,
      "loss": 0.1249,
      "step": 25910
    },
    {
      "epoch": 0.3664691993383195,
      "grad_norm": 0.0,
      "learning_rate": 6.338559855222045e-06,
      "loss": 0.1798,
      "step": 25920
    },
    {
      "epoch": 0.36661058406028646,
      "grad_norm": 1.7053840160369873,
      "learning_rate": 6.337146008002376e-06,
      "loss": 0.096,
      "step": 25930
    },
    {
      "epoch": 0.3667519687822534,
      "grad_norm": 0.0,
      "learning_rate": 6.335732160782706e-06,
      "loss": 0.6736,
      "step": 25940
    },
    {
      "epoch": 0.3668933535042203,
      "grad_norm": 78.41645050048828,
      "learning_rate": 6.334318313563037e-06,
      "loss": 0.0101,
      "step": 25950
    },
    {
      "epoch": 0.36703473822618726,
      "grad_norm": 0.0,
      "learning_rate": 6.332904466343367e-06,
      "loss": 0.7238,
      "step": 25960
    },
    {
      "epoch": 0.3671761229481542,
      "grad_norm": 0.38913947343826294,
      "learning_rate": 6.331490619123698e-06,
      "loss": 0.7628,
      "step": 25970
    },
    {
      "epoch": 0.36731750767012117,
      "grad_norm": 2.051710225714487e-06,
      "learning_rate": 6.330076771904029e-06,
      "loss": 0.2233,
      "step": 25980
    },
    {
      "epoch": 0.3674588923920881,
      "grad_norm": 0.0,
      "learning_rate": 6.328662924684359e-06,
      "loss": 1.6203,
      "step": 25990
    },
    {
      "epoch": 0.3676002771140551,
      "grad_norm": 1.4739557627763133e-05,
      "learning_rate": 6.32724907746469e-06,
      "loss": 0.0434,
      "step": 26000
    },
    {
      "epoch": 0.367741661836022,
      "grad_norm": 0.0,
      "learning_rate": 6.32583523024502e-06,
      "loss": 0.1026,
      "step": 26010
    },
    {
      "epoch": 0.3678830465579889,
      "grad_norm": 0.0,
      "learning_rate": 6.324421383025351e-06,
      "loss": 0.2401,
      "step": 26020
    },
    {
      "epoch": 0.3680244312799559,
      "grad_norm": 0.0,
      "learning_rate": 6.323007535805681e-06,
      "loss": 0.5146,
      "step": 26030
    },
    {
      "epoch": 0.36816581600192283,
      "grad_norm": 0.0,
      "learning_rate": 6.321593688586012e-06,
      "loss": 0.0018,
      "step": 26040
    },
    {
      "epoch": 0.3683072007238898,
      "grad_norm": 376.4212646484375,
      "learning_rate": 6.320179841366343e-06,
      "loss": 0.1611,
      "step": 26050
    },
    {
      "epoch": 0.36844858544585674,
      "grad_norm": 2.6031322704511695e-05,
      "learning_rate": 6.318765994146673e-06,
      "loss": 0.0,
      "step": 26060
    },
    {
      "epoch": 0.3685899701678237,
      "grad_norm": 344.38482666015625,
      "learning_rate": 6.3173521469270036e-06,
      "loss": 1.5797,
      "step": 26070
    },
    {
      "epoch": 0.3687313548897906,
      "grad_norm": 0.0,
      "learning_rate": 6.3159382997073336e-06,
      "loss": 0.0051,
      "step": 26080
    },
    {
      "epoch": 0.36887273961175754,
      "grad_norm": 0.0,
      "learning_rate": 6.3145244524876644e-06,
      "loss": 1.5321,
      "step": 26090
    },
    {
      "epoch": 0.3690141243337245,
      "grad_norm": 79.39460754394531,
      "learning_rate": 6.313110605267996e-06,
      "loss": 0.0357,
      "step": 26100
    },
    {
      "epoch": 0.36915550905569144,
      "grad_norm": 0.0,
      "learning_rate": 6.311696758048325e-06,
      "loss": 0.0004,
      "step": 26110
    },
    {
      "epoch": 0.3692968937776584,
      "grad_norm": 0.0,
      "learning_rate": 6.310282910828657e-06,
      "loss": 0.7169,
      "step": 26120
    },
    {
      "epoch": 0.36943827849962535,
      "grad_norm": 0.0,
      "learning_rate": 6.308869063608987e-06,
      "loss": 0.0003,
      "step": 26130
    },
    {
      "epoch": 0.36957966322159225,
      "grad_norm": 0.0,
      "learning_rate": 6.307455216389318e-06,
      "loss": 0.0004,
      "step": 26140
    },
    {
      "epoch": 0.3697210479435592,
      "grad_norm": 677.81298828125,
      "learning_rate": 6.306041369169648e-06,
      "loss": 0.6469,
      "step": 26150
    },
    {
      "epoch": 0.36986243266552615,
      "grad_norm": 2.176177440560423e-05,
      "learning_rate": 6.304627521949979e-06,
      "loss": 0.1779,
      "step": 26160
    },
    {
      "epoch": 0.3700038173874931,
      "grad_norm": 0.00038035077159292996,
      "learning_rate": 6.30321367473031e-06,
      "loss": 1.5574,
      "step": 26170
    },
    {
      "epoch": 0.37014520210946006,
      "grad_norm": 0.0,
      "learning_rate": 6.30179982751064e-06,
      "loss": 0.1689,
      "step": 26180
    },
    {
      "epoch": 0.370286586831427,
      "grad_norm": 0.0,
      "learning_rate": 6.3003859802909705e-06,
      "loss": 0.271,
      "step": 26190
    },
    {
      "epoch": 0.37042797155339396,
      "grad_norm": 0.0,
      "learning_rate": 6.2989721330713005e-06,
      "loss": 0.0026,
      "step": 26200
    },
    {
      "epoch": 0.37056935627536086,
      "grad_norm": 0.0,
      "learning_rate": 6.297558285851631e-06,
      "loss": 0.2642,
      "step": 26210
    },
    {
      "epoch": 0.3707107409973278,
      "grad_norm": 2.2063015421736054e-05,
      "learning_rate": 6.296144438631961e-06,
      "loss": 0.0021,
      "step": 26220
    },
    {
      "epoch": 0.37085212571929477,
      "grad_norm": 1.7833679914474487,
      "learning_rate": 6.294730591412292e-06,
      "loss": 1.3331,
      "step": 26230
    },
    {
      "epoch": 0.3709935104412617,
      "grad_norm": 564.844970703125,
      "learning_rate": 6.293316744192624e-06,
      "loss": 1.7892,
      "step": 26240
    },
    {
      "epoch": 0.3711348951632287,
      "grad_norm": 1.9764635794672358e-07,
      "learning_rate": 6.291902896972953e-06,
      "loss": 0.4621,
      "step": 26250
    },
    {
      "epoch": 0.3712762798851956,
      "grad_norm": 2.9579612601082772e-05,
      "learning_rate": 6.290489049753285e-06,
      "loss": 0.6594,
      "step": 26260
    },
    {
      "epoch": 0.3714176646071625,
      "grad_norm": 0.0,
      "learning_rate": 6.289075202533615e-06,
      "loss": 0.1264,
      "step": 26270
    },
    {
      "epoch": 0.3715590493291295,
      "grad_norm": 2.3619892597198486,
      "learning_rate": 6.287661355313946e-06,
      "loss": 1.1255,
      "step": 26280
    },
    {
      "epoch": 0.37170043405109643,
      "grad_norm": 1.221980437549064e-05,
      "learning_rate": 6.286247508094276e-06,
      "loss": 0.2436,
      "step": 26290
    },
    {
      "epoch": 0.3718418187730634,
      "grad_norm": 0.0,
      "learning_rate": 6.2848336608746065e-06,
      "loss": 0.0672,
      "step": 26300
    },
    {
      "epoch": 0.37198320349503033,
      "grad_norm": 405.5007019042969,
      "learning_rate": 6.283419813654937e-06,
      "loss": 0.6814,
      "step": 26310
    },
    {
      "epoch": 0.3721245882169973,
      "grad_norm": 0.0,
      "learning_rate": 6.282005966435267e-06,
      "loss": 0.9999,
      "step": 26320
    },
    {
      "epoch": 0.37226597293896424,
      "grad_norm": 3.92767333984375,
      "learning_rate": 6.280592119215598e-06,
      "loss": 1.0887,
      "step": 26330
    },
    {
      "epoch": 0.37240735766093114,
      "grad_norm": 0.0,
      "learning_rate": 6.279178271995928e-06,
      "loss": 0.1336,
      "step": 26340
    },
    {
      "epoch": 0.3725487423828981,
      "grad_norm": 0.0,
      "learning_rate": 6.277764424776259e-06,
      "loss": 0.6221,
      "step": 26350
    },
    {
      "epoch": 0.37269012710486504,
      "grad_norm": 0.0,
      "learning_rate": 6.27635057755659e-06,
      "loss": 0.0003,
      "step": 26360
    },
    {
      "epoch": 0.372831511826832,
      "grad_norm": 0.0,
      "learning_rate": 6.27493673033692e-06,
      "loss": 0.3053,
      "step": 26370
    },
    {
      "epoch": 0.37297289654879895,
      "grad_norm": 0.0,
      "learning_rate": 6.273522883117251e-06,
      "loss": 0.5586,
      "step": 26380
    },
    {
      "epoch": 0.3731142812707659,
      "grad_norm": 0.0,
      "learning_rate": 6.272109035897581e-06,
      "loss": 0.5526,
      "step": 26390
    },
    {
      "epoch": 0.3732556659927328,
      "grad_norm": 1.7028065919876099,
      "learning_rate": 6.2706951886779125e-06,
      "loss": 0.9464,
      "step": 26400
    },
    {
      "epoch": 0.37339705071469975,
      "grad_norm": 0.0,
      "learning_rate": 6.269281341458242e-06,
      "loss": 0.0756,
      "step": 26410
    },
    {
      "epoch": 0.3735384354366667,
      "grad_norm": 0.0,
      "learning_rate": 6.267867494238573e-06,
      "loss": 0.0216,
      "step": 26420
    },
    {
      "epoch": 0.37367982015863366,
      "grad_norm": 0.0,
      "learning_rate": 6.266453647018904e-06,
      "loss": 0.0009,
      "step": 26430
    },
    {
      "epoch": 0.3738212048806006,
      "grad_norm": 0.0,
      "learning_rate": 6.265039799799234e-06,
      "loss": 0.3095,
      "step": 26440
    },
    {
      "epoch": 0.37396258960256756,
      "grad_norm": 0.0,
      "learning_rate": 6.263625952579565e-06,
      "loss": 1.828,
      "step": 26450
    },
    {
      "epoch": 0.3741039743245345,
      "grad_norm": 5.087509632110596,
      "learning_rate": 6.262212105359895e-06,
      "loss": 0.6247,
      "step": 26460
    },
    {
      "epoch": 0.3742453590465014,
      "grad_norm": 0.00600783433765173,
      "learning_rate": 6.260798258140226e-06,
      "loss": 0.0,
      "step": 26470
    },
    {
      "epoch": 0.37438674376846837,
      "grad_norm": 0.005263954866677523,
      "learning_rate": 6.259384410920556e-06,
      "loss": 0.3221,
      "step": 26480
    },
    {
      "epoch": 0.3745281284904353,
      "grad_norm": 1.1368722141469334e-07,
      "learning_rate": 6.257970563700887e-06,
      "loss": 0.1457,
      "step": 26490
    },
    {
      "epoch": 0.37466951321240227,
      "grad_norm": 0.0,
      "learning_rate": 6.256556716481218e-06,
      "loss": 1.5762,
      "step": 26500
    },
    {
      "epoch": 0.3748108979343692,
      "grad_norm": 0.0,
      "learning_rate": 6.255142869261548e-06,
      "loss": 0.014,
      "step": 26510
    },
    {
      "epoch": 0.3749522826563362,
      "grad_norm": 4.773322359596932e-08,
      "learning_rate": 6.2537290220418786e-06,
      "loss": 2.1679,
      "step": 26520
    },
    {
      "epoch": 0.3750936673783031,
      "grad_norm": 0.0,
      "learning_rate": 6.2523151748222086e-06,
      "loss": 0.6019,
      "step": 26530
    },
    {
      "epoch": 0.37523505210027003,
      "grad_norm": 0.0,
      "learning_rate": 6.25090132760254e-06,
      "loss": 0.0407,
      "step": 26540
    },
    {
      "epoch": 0.375376436822237,
      "grad_norm": 0.0,
      "learning_rate": 6.2494874803828694e-06,
      "loss": 0.0,
      "step": 26550
    },
    {
      "epoch": 0.37551782154420393,
      "grad_norm": 0.0,
      "learning_rate": 6.248073633163201e-06,
      "loss": 0.0049,
      "step": 26560
    },
    {
      "epoch": 0.3756592062661709,
      "grad_norm": 0.0,
      "learning_rate": 6.246659785943532e-06,
      "loss": 0.4916,
      "step": 26570
    },
    {
      "epoch": 0.37580059098813784,
      "grad_norm": 0.0,
      "learning_rate": 6.245245938723862e-06,
      "loss": 0.0,
      "step": 26580
    },
    {
      "epoch": 0.3759419757101048,
      "grad_norm": 0.0,
      "learning_rate": 6.243832091504193e-06,
      "loss": 0.0634,
      "step": 26590
    },
    {
      "epoch": 0.3760833604320717,
      "grad_norm": 277.5834045410156,
      "learning_rate": 6.242418244284523e-06,
      "loss": 0.5011,
      "step": 26600
    },
    {
      "epoch": 0.37622474515403864,
      "grad_norm": 0.0,
      "learning_rate": 6.241004397064854e-06,
      "loss": 0.3433,
      "step": 26610
    },
    {
      "epoch": 0.3763661298760056,
      "grad_norm": 0.0,
      "learning_rate": 6.239590549845185e-06,
      "loss": 0.0,
      "step": 26620
    },
    {
      "epoch": 0.37650751459797255,
      "grad_norm": 0.0,
      "learning_rate": 6.238176702625515e-06,
      "loss": 0.0,
      "step": 26630
    },
    {
      "epoch": 0.3766488993199395,
      "grad_norm": 0.0,
      "learning_rate": 6.2367628554058454e-06,
      "loss": 0.0,
      "step": 26640
    },
    {
      "epoch": 0.37679028404190645,
      "grad_norm": 0.0,
      "learning_rate": 6.2353490081861755e-06,
      "loss": 0.4402,
      "step": 26650
    },
    {
      "epoch": 0.37693166876387335,
      "grad_norm": 0.5165342688560486,
      "learning_rate": 6.233935160966506e-06,
      "loss": 1.0887,
      "step": 26660
    },
    {
      "epoch": 0.3770730534858403,
      "grad_norm": 0.0,
      "learning_rate": 6.232521313746836e-06,
      "loss": 0.5902,
      "step": 26670
    },
    {
      "epoch": 0.37721443820780726,
      "grad_norm": 0.0,
      "learning_rate": 6.231107466527167e-06,
      "loss": 0.0003,
      "step": 26680
    },
    {
      "epoch": 0.3773558229297742,
      "grad_norm": 0.0,
      "learning_rate": 6.229693619307499e-06,
      "loss": 0.011,
      "step": 26690
    },
    {
      "epoch": 0.37749720765174116,
      "grad_norm": 0.0,
      "learning_rate": 6.228279772087829e-06,
      "loss": 0.0,
      "step": 26700
    },
    {
      "epoch": 0.3776385923737081,
      "grad_norm": 0.1432657539844513,
      "learning_rate": 6.22686592486816e-06,
      "loss": 0.0,
      "step": 26710
    },
    {
      "epoch": 0.37777997709567507,
      "grad_norm": 0.0,
      "learning_rate": 6.22545207764849e-06,
      "loss": 0.0002,
      "step": 26720
    },
    {
      "epoch": 0.37792136181764197,
      "grad_norm": 0.0,
      "learning_rate": 6.224038230428821e-06,
      "loss": 0.6564,
      "step": 26730
    },
    {
      "epoch": 0.3780627465396089,
      "grad_norm": 0.0008317759493365884,
      "learning_rate": 6.222624383209151e-06,
      "loss": 0.6555,
      "step": 26740
    },
    {
      "epoch": 0.37820413126157587,
      "grad_norm": 0.0,
      "learning_rate": 6.2212105359894815e-06,
      "loss": 0.0,
      "step": 26750
    },
    {
      "epoch": 0.3783455159835428,
      "grad_norm": 0.00012804924335796386,
      "learning_rate": 6.219796688769812e-06,
      "loss": 0.0857,
      "step": 26760
    },
    {
      "epoch": 0.3784869007055098,
      "grad_norm": 0.0,
      "learning_rate": 6.218382841550142e-06,
      "loss": 0.0075,
      "step": 26770
    },
    {
      "epoch": 0.37862828542747673,
      "grad_norm": 0.0009246919653378427,
      "learning_rate": 6.216968994330473e-06,
      "loss": 1.4195,
      "step": 26780
    },
    {
      "epoch": 0.3787696701494436,
      "grad_norm": 0.0,
      "learning_rate": 6.215555147110803e-06,
      "loss": 0.2593,
      "step": 26790
    },
    {
      "epoch": 0.3789110548714106,
      "grad_norm": 789.1742553710938,
      "learning_rate": 6.214141299891134e-06,
      "loss": 2.7903,
      "step": 26800
    },
    {
      "epoch": 0.37905243959337753,
      "grad_norm": 0.13268473744392395,
      "learning_rate": 6.212727452671466e-06,
      "loss": 0.2424,
      "step": 26810
    },
    {
      "epoch": 0.3791938243153445,
      "grad_norm": 28.110090255737305,
      "learning_rate": 6.211313605451795e-06,
      "loss": 0.0044,
      "step": 26820
    },
    {
      "epoch": 0.37933520903731144,
      "grad_norm": 0.0,
      "learning_rate": 6.209899758232127e-06,
      "loss": 1.172,
      "step": 26830
    },
    {
      "epoch": 0.3794765937592784,
      "grad_norm": 0.0,
      "learning_rate": 6.208485911012457e-06,
      "loss": 0.0,
      "step": 26840
    },
    {
      "epoch": 0.37961797848124534,
      "grad_norm": 0.0,
      "learning_rate": 6.2070720637927875e-06,
      "loss": 0.0096,
      "step": 26850
    },
    {
      "epoch": 0.37975936320321224,
      "grad_norm": 92.38627624511719,
      "learning_rate": 6.2056582165731175e-06,
      "loss": 0.0596,
      "step": 26860
    },
    {
      "epoch": 0.3799007479251792,
      "grad_norm": 0.0,
      "learning_rate": 6.204244369353448e-06,
      "loss": 0.0029,
      "step": 26870
    },
    {
      "epoch": 0.38004213264714615,
      "grad_norm": 0.0,
      "learning_rate": 6.202830522133779e-06,
      "loss": 0.0016,
      "step": 26880
    },
    {
      "epoch": 0.3801835173691131,
      "grad_norm": 0.0,
      "learning_rate": 6.201416674914109e-06,
      "loss": 0.0051,
      "step": 26890
    },
    {
      "epoch": 0.38032490209108005,
      "grad_norm": 0.0,
      "learning_rate": 6.20000282769444e-06,
      "loss": 0.0,
      "step": 26900
    },
    {
      "epoch": 0.380466286813047,
      "grad_norm": 0.0,
      "learning_rate": 6.19858898047477e-06,
      "loss": 0.834,
      "step": 26910
    },
    {
      "epoch": 0.3806076715350139,
      "grad_norm": 0.003516575787216425,
      "learning_rate": 6.197175133255101e-06,
      "loss": 0.6862,
      "step": 26920
    },
    {
      "epoch": 0.38074905625698086,
      "grad_norm": 0.0,
      "learning_rate": 6.195761286035431e-06,
      "loss": 0.3654,
      "step": 26930
    },
    {
      "epoch": 0.3808904409789478,
      "grad_norm": 0.0,
      "learning_rate": 6.194347438815762e-06,
      "loss": 0.0,
      "step": 26940
    },
    {
      "epoch": 0.38103182570091476,
      "grad_norm": 0.0,
      "learning_rate": 6.192933591596093e-06,
      "loss": 0.0097,
      "step": 26950
    },
    {
      "epoch": 0.3811732104228817,
      "grad_norm": 0.0,
      "learning_rate": 6.191519744376423e-06,
      "loss": 1.2821,
      "step": 26960
    },
    {
      "epoch": 0.38131459514484867,
      "grad_norm": 0.0,
      "learning_rate": 6.190105897156754e-06,
      "loss": 0.0,
      "step": 26970
    },
    {
      "epoch": 0.38145597986681556,
      "grad_norm": 0.0,
      "learning_rate": 6.1886920499370836e-06,
      "loss": 0.152,
      "step": 26980
    },
    {
      "epoch": 0.3815973645887825,
      "grad_norm": 0.0,
      "learning_rate": 6.187278202717415e-06,
      "loss": 1.6512,
      "step": 26990
    },
    {
      "epoch": 0.38173874931074947,
      "grad_norm": 0.0,
      "learning_rate": 6.185864355497745e-06,
      "loss": 0.0,
      "step": 27000
    },
    {
      "epoch": 0.3818801340327164,
      "grad_norm": 0.0,
      "learning_rate": 6.184450508278076e-06,
      "loss": 0.3815,
      "step": 27010
    },
    {
      "epoch": 0.3820215187546834,
      "grad_norm": 0.0,
      "learning_rate": 6.183036661058407e-06,
      "loss": 1.0437,
      "step": 27020
    },
    {
      "epoch": 0.38216290347665033,
      "grad_norm": 209.55690002441406,
      "learning_rate": 6.181622813838737e-06,
      "loss": 0.1085,
      "step": 27030
    },
    {
      "epoch": 0.3823042881986173,
      "grad_norm": 0.0,
      "learning_rate": 6.180208966619068e-06,
      "loss": 0.312,
      "step": 27040
    },
    {
      "epoch": 0.3824456729205842,
      "grad_norm": 0.0,
      "learning_rate": 6.178795119399398e-06,
      "loss": 0.0003,
      "step": 27050
    },
    {
      "epoch": 0.38258705764255113,
      "grad_norm": 0.0,
      "learning_rate": 6.177381272179729e-06,
      "loss": 0.2459,
      "step": 27060
    },
    {
      "epoch": 0.3827284423645181,
      "grad_norm": 0.0,
      "learning_rate": 6.1759674249600596e-06,
      "loss": 0.0,
      "step": 27070
    },
    {
      "epoch": 0.38286982708648504,
      "grad_norm": 0.0,
      "learning_rate": 6.17455357774039e-06,
      "loss": 0.3885,
      "step": 27080
    },
    {
      "epoch": 0.383011211808452,
      "grad_norm": 0.0,
      "learning_rate": 6.1731397305207204e-06,
      "loss": 0.0001,
      "step": 27090
    },
    {
      "epoch": 0.38315259653041894,
      "grad_norm": 0.0,
      "learning_rate": 6.1717258833010504e-06,
      "loss": 0.005,
      "step": 27100
    },
    {
      "epoch": 0.38329398125238584,
      "grad_norm": 171.72784423828125,
      "learning_rate": 6.170312036081382e-06,
      "loss": 0.1991,
      "step": 27110
    },
    {
      "epoch": 0.3834353659743528,
      "grad_norm": 7.756980895996094,
      "learning_rate": 6.168898188861711e-06,
      "loss": 0.0027,
      "step": 27120
    },
    {
      "epoch": 0.38357675069631975,
      "grad_norm": 8.043421257752925e-05,
      "learning_rate": 6.167484341642043e-06,
      "loss": 0.0083,
      "step": 27130
    },
    {
      "epoch": 0.3837181354182867,
      "grad_norm": 0.0,
      "learning_rate": 6.166070494422374e-06,
      "loss": 0.0402,
      "step": 27140
    },
    {
      "epoch": 0.38385952014025365,
      "grad_norm": 0.0,
      "learning_rate": 6.164656647202704e-06,
      "loss": 0.0,
      "step": 27150
    },
    {
      "epoch": 0.3840009048622206,
      "grad_norm": 0.0004216250963509083,
      "learning_rate": 6.163242799983035e-06,
      "loss": 0.0,
      "step": 27160
    },
    {
      "epoch": 0.38414228958418756,
      "grad_norm": 2.152288107026834e-05,
      "learning_rate": 6.161828952763365e-06,
      "loss": 1.1985,
      "step": 27170
    },
    {
      "epoch": 0.38428367430615445,
      "grad_norm": 0.0,
      "learning_rate": 6.160415105543696e-06,
      "loss": 0.0001,
      "step": 27180
    },
    {
      "epoch": 0.3844250590281214,
      "grad_norm": 10.571930885314941,
      "learning_rate": 6.159001258324026e-06,
      "loss": 0.9663,
      "step": 27190
    },
    {
      "epoch": 0.38456644375008836,
      "grad_norm": 0.46635881066322327,
      "learning_rate": 6.1575874111043565e-06,
      "loss": 0.1621,
      "step": 27200
    },
    {
      "epoch": 0.3847078284720553,
      "grad_norm": 0.0,
      "learning_rate": 6.156173563884687e-06,
      "loss": 1.1427,
      "step": 27210
    },
    {
      "epoch": 0.38484921319402227,
      "grad_norm": 0.0,
      "learning_rate": 6.154759716665017e-06,
      "loss": 0.0092,
      "step": 27220
    },
    {
      "epoch": 0.3849905979159892,
      "grad_norm": 0.0,
      "learning_rate": 6.153345869445348e-06,
      "loss": 0.2754,
      "step": 27230
    },
    {
      "epoch": 0.3851319826379561,
      "grad_norm": 0.0,
      "learning_rate": 6.151932022225678e-06,
      "loss": 0.5031,
      "step": 27240
    },
    {
      "epoch": 0.38527336735992307,
      "grad_norm": 0.0,
      "learning_rate": 6.150518175006009e-06,
      "loss": 0.876,
      "step": 27250
    },
    {
      "epoch": 0.38541475208189,
      "grad_norm": 0.0,
      "learning_rate": 6.149104327786339e-06,
      "loss": 0.6416,
      "step": 27260
    },
    {
      "epoch": 0.385556136803857,
      "grad_norm": 1.1376924646810949e-07,
      "learning_rate": 6.147690480566671e-06,
      "loss": 0.7037,
      "step": 27270
    },
    {
      "epoch": 0.3856975215258239,
      "grad_norm": 0.0,
      "learning_rate": 6.146276633347002e-06,
      "loss": 0.005,
      "step": 27280
    },
    {
      "epoch": 0.3858389062477909,
      "grad_norm": 0.0,
      "learning_rate": 6.144862786127332e-06,
      "loss": 0.0891,
      "step": 27290
    },
    {
      "epoch": 0.38598029096975783,
      "grad_norm": 4.385569095611572,
      "learning_rate": 6.1434489389076625e-06,
      "loss": 0.0008,
      "step": 27300
    },
    {
      "epoch": 0.38612167569172473,
      "grad_norm": 6.398317964340094e-08,
      "learning_rate": 6.1420350916879925e-06,
      "loss": 0.2768,
      "step": 27310
    },
    {
      "epoch": 0.3862630604136917,
      "grad_norm": 0.0,
      "learning_rate": 6.140621244468323e-06,
      "loss": 0.2916,
      "step": 27320
    },
    {
      "epoch": 0.38640444513565864,
      "grad_norm": 294.6524963378906,
      "learning_rate": 6.139207397248654e-06,
      "loss": 0.0717,
      "step": 27330
    },
    {
      "epoch": 0.3865458298576256,
      "grad_norm": 0.0,
      "learning_rate": 6.137793550028984e-06,
      "loss": 0.2709,
      "step": 27340
    },
    {
      "epoch": 0.38668721457959254,
      "grad_norm": 0.0,
      "learning_rate": 6.136379702809315e-06,
      "loss": 0.0001,
      "step": 27350
    },
    {
      "epoch": 0.3868285993015595,
      "grad_norm": 0.0,
      "learning_rate": 6.134965855589645e-06,
      "loss": 0.0001,
      "step": 27360
    },
    {
      "epoch": 0.3869699840235264,
      "grad_norm": 0.0,
      "learning_rate": 6.133552008369976e-06,
      "loss": 0.0,
      "step": 27370
    },
    {
      "epoch": 0.38711136874549334,
      "grad_norm": 0.0,
      "learning_rate": 6.132138161150306e-06,
      "loss": 0.0008,
      "step": 27380
    },
    {
      "epoch": 0.3872527534674603,
      "grad_norm": 313.241943359375,
      "learning_rate": 6.130724313930637e-06,
      "loss": 1.9621,
      "step": 27390
    },
    {
      "epoch": 0.38739413818942725,
      "grad_norm": 0.0,
      "learning_rate": 6.1293104667109685e-06,
      "loss": 0.0281,
      "step": 27400
    },
    {
      "epoch": 0.3875355229113942,
      "grad_norm": 0.0,
      "learning_rate": 6.1278966194912985e-06,
      "loss": 2.0641,
      "step": 27410
    },
    {
      "epoch": 0.38767690763336116,
      "grad_norm": 0.0,
      "learning_rate": 6.126482772271629e-06,
      "loss": 0.2414,
      "step": 27420
    },
    {
      "epoch": 0.3878182923553281,
      "grad_norm": 0.0,
      "learning_rate": 6.125068925051959e-06,
      "loss": 0.0062,
      "step": 27430
    },
    {
      "epoch": 0.387959677077295,
      "grad_norm": 0.0,
      "learning_rate": 6.12365507783229e-06,
      "loss": 0.2723,
      "step": 27440
    },
    {
      "epoch": 0.38810106179926196,
      "grad_norm": 0.0,
      "learning_rate": 6.12224123061262e-06,
      "loss": 0.0,
      "step": 27450
    },
    {
      "epoch": 0.3882424465212289,
      "grad_norm": 0.0,
      "learning_rate": 6.120827383392951e-06,
      "loss": 0.0283,
      "step": 27460
    },
    {
      "epoch": 0.38838383124319587,
      "grad_norm": 0.0,
      "learning_rate": 6.119413536173282e-06,
      "loss": 0.0074,
      "step": 27470
    },
    {
      "epoch": 0.3885252159651628,
      "grad_norm": 0.0,
      "learning_rate": 6.117999688953612e-06,
      "loss": 0.1292,
      "step": 27480
    },
    {
      "epoch": 0.38866660068712977,
      "grad_norm": 0.0,
      "learning_rate": 6.116585841733943e-06,
      "loss": 0.2536,
      "step": 27490
    },
    {
      "epoch": 0.38880798540909667,
      "grad_norm": 0.0,
      "learning_rate": 6.115171994514273e-06,
      "loss": 0.0,
      "step": 27500
    },
    {
      "epoch": 0.3889493701310636,
      "grad_norm": 7.644573543075239e-07,
      "learning_rate": 6.113758147294604e-06,
      "loss": 0.5566,
      "step": 27510
    },
    {
      "epoch": 0.3890907548530306,
      "grad_norm": 0.0,
      "learning_rate": 6.1123443000749346e-06,
      "loss": 0.0,
      "step": 27520
    },
    {
      "epoch": 0.3892321395749975,
      "grad_norm": 0.0,
      "learning_rate": 6.1109304528552646e-06,
      "loss": 0.0,
      "step": 27530
    },
    {
      "epoch": 0.3893735242969645,
      "grad_norm": 0.0,
      "learning_rate": 6.109516605635596e-06,
      "loss": 0.6621,
      "step": 27540
    },
    {
      "epoch": 0.38951490901893143,
      "grad_norm": 0.0,
      "learning_rate": 6.1081027584159254e-06,
      "loss": 0.0011,
      "step": 27550
    },
    {
      "epoch": 0.3896562937408984,
      "grad_norm": 0.0,
      "learning_rate": 6.106688911196257e-06,
      "loss": 0.1201,
      "step": 27560
    },
    {
      "epoch": 0.3897976784628653,
      "grad_norm": 0.0,
      "learning_rate": 6.105275063976587e-06,
      "loss": 0.0003,
      "step": 27570
    },
    {
      "epoch": 0.38993906318483224,
      "grad_norm": 12.432640075683594,
      "learning_rate": 6.103861216756918e-06,
      "loss": 0.6001,
      "step": 27580
    },
    {
      "epoch": 0.3900804479067992,
      "grad_norm": 0.0,
      "learning_rate": 6.102447369537249e-06,
      "loss": 0.8875,
      "step": 27590
    },
    {
      "epoch": 0.39022183262876614,
      "grad_norm": 0.0,
      "learning_rate": 6.101033522317579e-06,
      "loss": 0.0359,
      "step": 27600
    },
    {
      "epoch": 0.3903632173507331,
      "grad_norm": 0.0,
      "learning_rate": 6.09961967509791e-06,
      "loss": 0.6522,
      "step": 27610
    },
    {
      "epoch": 0.39050460207270005,
      "grad_norm": 0.0,
      "learning_rate": 6.09820582787824e-06,
      "loss": 0.201,
      "step": 27620
    },
    {
      "epoch": 0.39064598679466694,
      "grad_norm": 0.0006754971691407263,
      "learning_rate": 6.096791980658571e-06,
      "loss": 0.0,
      "step": 27630
    },
    {
      "epoch": 0.3907873715166339,
      "grad_norm": 5.1070959017351925e-08,
      "learning_rate": 6.095378133438901e-06,
      "loss": 0.0,
      "step": 27640
    },
    {
      "epoch": 0.39092875623860085,
      "grad_norm": 0.005227948538959026,
      "learning_rate": 6.0939642862192315e-06,
      "loss": 0.001,
      "step": 27650
    },
    {
      "epoch": 0.3910701409605678,
      "grad_norm": 0.0,
      "learning_rate": 6.092550438999562e-06,
      "loss": 0.1632,
      "step": 27660
    },
    {
      "epoch": 0.39121152568253476,
      "grad_norm": 0.0,
      "learning_rate": 6.091136591779892e-06,
      "loss": 0.0,
      "step": 27670
    },
    {
      "epoch": 0.3913529104045017,
      "grad_norm": 0.0,
      "learning_rate": 6.089722744560224e-06,
      "loss": 1.3489,
      "step": 27680
    },
    {
      "epoch": 0.39149429512646866,
      "grad_norm": 0.0,
      "learning_rate": 6.088308897340553e-06,
      "loss": 0.0,
      "step": 27690
    },
    {
      "epoch": 0.39163567984843556,
      "grad_norm": 0.0,
      "learning_rate": 6.086895050120885e-06,
      "loss": 2.1531,
      "step": 27700
    },
    {
      "epoch": 0.3917770645704025,
      "grad_norm": 0.0,
      "learning_rate": 6.085481202901214e-06,
      "loss": 0.0215,
      "step": 27710
    },
    {
      "epoch": 0.39191844929236946,
      "grad_norm": 0.012666872702538967,
      "learning_rate": 6.084067355681546e-06,
      "loss": 0.0005,
      "step": 27720
    },
    {
      "epoch": 0.3920598340143364,
      "grad_norm": 0.0,
      "learning_rate": 6.082653508461877e-06,
      "loss": 4.9312,
      "step": 27730
    },
    {
      "epoch": 0.39220121873630337,
      "grad_norm": 983.0516357421875,
      "learning_rate": 6.081239661242207e-06,
      "loss": 0.6532,
      "step": 27740
    },
    {
      "epoch": 0.3923426034582703,
      "grad_norm": 0.0,
      "learning_rate": 6.0798258140225375e-06,
      "loss": 0.0548,
      "step": 27750
    },
    {
      "epoch": 0.3924839881802372,
      "grad_norm": 0.01651010289788246,
      "learning_rate": 6.0784119668028675e-06,
      "loss": 0.0,
      "step": 27760
    },
    {
      "epoch": 0.3926253729022042,
      "grad_norm": 0.0,
      "learning_rate": 6.076998119583198e-06,
      "loss": 0.8941,
      "step": 27770
    },
    {
      "epoch": 0.3927667576241711,
      "grad_norm": 0.0,
      "learning_rate": 6.075584272363529e-06,
      "loss": 0.1369,
      "step": 27780
    },
    {
      "epoch": 0.3929081423461381,
      "grad_norm": 1.5382460355758667,
      "learning_rate": 6.074170425143859e-06,
      "loss": 0.0001,
      "step": 27790
    },
    {
      "epoch": 0.39304952706810503,
      "grad_norm": 6.979182103350468e-07,
      "learning_rate": 6.07275657792419e-06,
      "loss": 0.598,
      "step": 27800
    },
    {
      "epoch": 0.393190911790072,
      "grad_norm": 0.0,
      "learning_rate": 6.07134273070452e-06,
      "loss": 1.1291,
      "step": 27810
    },
    {
      "epoch": 0.3933322965120389,
      "grad_norm": 0.0,
      "learning_rate": 6.069928883484851e-06,
      "loss": 1.2406,
      "step": 27820
    },
    {
      "epoch": 0.39347368123400583,
      "grad_norm": 0.0,
      "learning_rate": 6.068515036265181e-06,
      "loss": 1.2185,
      "step": 27830
    },
    {
      "epoch": 0.3936150659559728,
      "grad_norm": 0.0,
      "learning_rate": 6.067101189045513e-06,
      "loss": 0.0,
      "step": 27840
    },
    {
      "epoch": 0.39375645067793974,
      "grad_norm": 817.6815795898438,
      "learning_rate": 6.0656873418258435e-06,
      "loss": 0.4667,
      "step": 27850
    },
    {
      "epoch": 0.3938978353999067,
      "grad_norm": 0.0005744485533796251,
      "learning_rate": 6.0642734946061735e-06,
      "loss": 0.0002,
      "step": 27860
    },
    {
      "epoch": 0.39403922012187365,
      "grad_norm": 0.0,
      "learning_rate": 6.062859647386504e-06,
      "loss": 1.3617,
      "step": 27870
    },
    {
      "epoch": 0.3941806048438406,
      "grad_norm": 0.0,
      "learning_rate": 6.061445800166834e-06,
      "loss": 0.0028,
      "step": 27880
    },
    {
      "epoch": 0.3943219895658075,
      "grad_norm": 0.0,
      "learning_rate": 6.060031952947165e-06,
      "loss": 0.0,
      "step": 27890
    },
    {
      "epoch": 0.39446337428777445,
      "grad_norm": 0.027473462745547295,
      "learning_rate": 6.058618105727495e-06,
      "loss": 0.0073,
      "step": 27900
    },
    {
      "epoch": 0.3946047590097414,
      "grad_norm": 0.0,
      "learning_rate": 6.057204258507826e-06,
      "loss": 1.2786,
      "step": 27910
    },
    {
      "epoch": 0.39474614373170835,
      "grad_norm": 0.0,
      "learning_rate": 6.055790411288157e-06,
      "loss": 0.0,
      "step": 27920
    },
    {
      "epoch": 0.3948875284536753,
      "grad_norm": 0.0,
      "learning_rate": 6.054376564068487e-06,
      "loss": 0.0049,
      "step": 27930
    },
    {
      "epoch": 0.39502891317564226,
      "grad_norm": 2.5845213968977987e-08,
      "learning_rate": 6.052962716848818e-06,
      "loss": 1.4217,
      "step": 27940
    },
    {
      "epoch": 0.39517029789760916,
      "grad_norm": 0.0,
      "learning_rate": 6.051548869629148e-06,
      "loss": 1.2623,
      "step": 27950
    },
    {
      "epoch": 0.3953116826195761,
      "grad_norm": 68.32649230957031,
      "learning_rate": 6.050135022409479e-06,
      "loss": 0.0392,
      "step": 27960
    },
    {
      "epoch": 0.39545306734154306,
      "grad_norm": 0.4508143961429596,
      "learning_rate": 6.048721175189809e-06,
      "loss": 0.2919,
      "step": 27970
    },
    {
      "epoch": 0.39559445206351,
      "grad_norm": 0.0,
      "learning_rate": 6.0473073279701396e-06,
      "loss": 0.0163,
      "step": 27980
    },
    {
      "epoch": 0.39573583678547697,
      "grad_norm": 0.0,
      "learning_rate": 6.045893480750471e-06,
      "loss": 0.0018,
      "step": 27990
    },
    {
      "epoch": 0.3958772215074439,
      "grad_norm": 0.0,
      "learning_rate": 6.044479633530801e-06,
      "loss": 0.3942,
      "step": 28000
    },
    {
      "epoch": 0.3960186062294109,
      "grad_norm": 0.0,
      "learning_rate": 6.043065786311132e-06,
      "loss": 0.0397,
      "step": 28010
    },
    {
      "epoch": 0.39615999095137777,
      "grad_norm": 0.0,
      "learning_rate": 6.041651939091462e-06,
      "loss": 0.2144,
      "step": 28020
    },
    {
      "epoch": 0.3963013756733447,
      "grad_norm": 0.0,
      "learning_rate": 6.040238091871793e-06,
      "loss": 0.0008,
      "step": 28030
    },
    {
      "epoch": 0.3964427603953117,
      "grad_norm": 3.844629191007698e-06,
      "learning_rate": 6.038824244652124e-06,
      "loss": 0.2422,
      "step": 28040
    },
    {
      "epoch": 0.39658414511727863,
      "grad_norm": 0.0,
      "learning_rate": 6.037410397432454e-06,
      "loss": 0.0797,
      "step": 28050
    },
    {
      "epoch": 0.3967255298392456,
      "grad_norm": 0.0,
      "learning_rate": 6.035996550212785e-06,
      "loss": 0.1238,
      "step": 28060
    },
    {
      "epoch": 0.39686691456121254,
      "grad_norm": 0.00011826851550722495,
      "learning_rate": 6.034582702993115e-06,
      "loss": 0.1382,
      "step": 28070
    },
    {
      "epoch": 0.39700829928317943,
      "grad_norm": 0.0,
      "learning_rate": 6.033168855773446e-06,
      "loss": 0.0593,
      "step": 28080
    },
    {
      "epoch": 0.3971496840051464,
      "grad_norm": 1.1062045359722106e-06,
      "learning_rate": 6.031755008553776e-06,
      "loss": 0.0881,
      "step": 28090
    },
    {
      "epoch": 0.39729106872711334,
      "grad_norm": 0.0,
      "learning_rate": 6.0303411613341064e-06,
      "loss": 0.4247,
      "step": 28100
    },
    {
      "epoch": 0.3974324534490803,
      "grad_norm": 0.0,
      "learning_rate": 6.028927314114438e-06,
      "loss": 0.0,
      "step": 28110
    },
    {
      "epoch": 0.39757383817104724,
      "grad_norm": 2.9959114300481815e-08,
      "learning_rate": 6.027513466894767e-06,
      "loss": 0.0001,
      "step": 28120
    },
    {
      "epoch": 0.3977152228930142,
      "grad_norm": 0.0,
      "learning_rate": 6.026099619675099e-06,
      "loss": 0.0048,
      "step": 28130
    },
    {
      "epoch": 0.39785660761498115,
      "grad_norm": 0.0,
      "learning_rate": 6.024685772455429e-06,
      "loss": 0.0096,
      "step": 28140
    },
    {
      "epoch": 0.39799799233694805,
      "grad_norm": 0.0,
      "learning_rate": 6.02327192523576e-06,
      "loss": 0.0067,
      "step": 28150
    },
    {
      "epoch": 0.398139377058915,
      "grad_norm": 0.0,
      "learning_rate": 6.02185807801609e-06,
      "loss": 1.3156,
      "step": 28160
    },
    {
      "epoch": 0.39828076178088195,
      "grad_norm": 0.0,
      "learning_rate": 6.020444230796421e-06,
      "loss": 0.0001,
      "step": 28170
    },
    {
      "epoch": 0.3984221465028489,
      "grad_norm": 2.777828740363475e-05,
      "learning_rate": 6.019030383576752e-06,
      "loss": 0.0005,
      "step": 28180
    },
    {
      "epoch": 0.39856353122481586,
      "grad_norm": 0.0,
      "learning_rate": 6.017616536357082e-06,
      "loss": 0.4406,
      "step": 28190
    },
    {
      "epoch": 0.3987049159467828,
      "grad_norm": 1.3108230234593066e-07,
      "learning_rate": 6.0162026891374125e-06,
      "loss": 0.6569,
      "step": 28200
    },
    {
      "epoch": 0.3988463006687497,
      "grad_norm": 0.0,
      "learning_rate": 6.0147888419177425e-06,
      "loss": 1.9648,
      "step": 28210
    },
    {
      "epoch": 0.39898768539071666,
      "grad_norm": 0.0,
      "learning_rate": 6.013374994698073e-06,
      "loss": 0.0489,
      "step": 28220
    },
    {
      "epoch": 0.3991290701126836,
      "grad_norm": 2.028915514529217e-05,
      "learning_rate": 6.011961147478403e-06,
      "loss": 1.2941,
      "step": 28230
    },
    {
      "epoch": 0.39927045483465057,
      "grad_norm": 0.00015990686370059848,
      "learning_rate": 6.010547300258734e-06,
      "loss": 0.0332,
      "step": 28240
    },
    {
      "epoch": 0.3994118395566175,
      "grad_norm": 0.0,
      "learning_rate": 6.009133453039065e-06,
      "loss": 0.1377,
      "step": 28250
    },
    {
      "epoch": 0.3995532242785845,
      "grad_norm": 0.0,
      "learning_rate": 6.007719605819395e-06,
      "loss": 1.3377,
      "step": 28260
    },
    {
      "epoch": 0.3996946090005514,
      "grad_norm": 0.0,
      "learning_rate": 6.006305758599727e-06,
      "loss": 0.4844,
      "step": 28270
    },
    {
      "epoch": 0.3998359937225183,
      "grad_norm": 0.0,
      "learning_rate": 6.004891911380056e-06,
      "loss": 1.7537,
      "step": 28280
    },
    {
      "epoch": 0.3999773784444853,
      "grad_norm": 0.0,
      "learning_rate": 6.003478064160388e-06,
      "loss": 0.0001,
      "step": 28290
    },
    {
      "epoch": 0.40011876316645223,
      "grad_norm": 0.0,
      "learning_rate": 6.0020642169407185e-06,
      "loss": 0.0042,
      "step": 28300
    },
    {
      "epoch": 0.4002601478884192,
      "grad_norm": 1.2562739849090576,
      "learning_rate": 6.0006503697210485e-06,
      "loss": 0.0519,
      "step": 28310
    },
    {
      "epoch": 0.40040153261038613,
      "grad_norm": 0.0,
      "learning_rate": 5.999236522501379e-06,
      "loss": 0.2458,
      "step": 28320
    },
    {
      "epoch": 0.4005429173323531,
      "grad_norm": 0.0,
      "learning_rate": 5.997822675281709e-06,
      "loss": 0.0101,
      "step": 28330
    },
    {
      "epoch": 0.40068430205432,
      "grad_norm": 0.0,
      "learning_rate": 5.99640882806204e-06,
      "loss": 0.0,
      "step": 28340
    },
    {
      "epoch": 0.40082568677628694,
      "grad_norm": 0.0,
      "learning_rate": 5.99499498084237e-06,
      "loss": 0.0003,
      "step": 28350
    },
    {
      "epoch": 0.4009670714982539,
      "grad_norm": 0.0,
      "learning_rate": 5.993581133622701e-06,
      "loss": 0.0413,
      "step": 28360
    },
    {
      "epoch": 0.40110845622022084,
      "grad_norm": 0.0,
      "learning_rate": 5.992167286403032e-06,
      "loss": 1.9876,
      "step": 28370
    },
    {
      "epoch": 0.4012498409421878,
      "grad_norm": 0.0,
      "learning_rate": 5.990753439183362e-06,
      "loss": 0.448,
      "step": 28380
    },
    {
      "epoch": 0.40139122566415475,
      "grad_norm": 38.444305419921875,
      "learning_rate": 5.989339591963693e-06,
      "loss": 0.0165,
      "step": 28390
    },
    {
      "epoch": 0.4015326103861217,
      "grad_norm": 380.70635986328125,
      "learning_rate": 5.987925744744023e-06,
      "loss": 0.4425,
      "step": 28400
    },
    {
      "epoch": 0.4016739951080886,
      "grad_norm": 0.0,
      "learning_rate": 5.9865118975243545e-06,
      "loss": 1.4083,
      "step": 28410
    },
    {
      "epoch": 0.40181537983005555,
      "grad_norm": 0.0,
      "learning_rate": 5.985098050304684e-06,
      "loss": 0.0,
      "step": 28420
    },
    {
      "epoch": 0.4019567645520225,
      "grad_norm": 583.844482421875,
      "learning_rate": 5.983684203085015e-06,
      "loss": 1.0648,
      "step": 28430
    },
    {
      "epoch": 0.40209814927398946,
      "grad_norm": 0.0,
      "learning_rate": 5.982270355865346e-06,
      "loss": 0.0001,
      "step": 28440
    },
    {
      "epoch": 0.4022395339959564,
      "grad_norm": 0.0,
      "learning_rate": 5.980856508645676e-06,
      "loss": 0.0,
      "step": 28450
    },
    {
      "epoch": 0.40238091871792336,
      "grad_norm": 0.0,
      "learning_rate": 5.979442661426007e-06,
      "loss": 0.1842,
      "step": 28460
    },
    {
      "epoch": 0.40252230343989026,
      "grad_norm": 0.0,
      "learning_rate": 5.978028814206337e-06,
      "loss": 0.0114,
      "step": 28470
    },
    {
      "epoch": 0.4026636881618572,
      "grad_norm": 0.0,
      "learning_rate": 5.976614966986668e-06,
      "loss": 0.0002,
      "step": 28480
    },
    {
      "epoch": 0.40280507288382417,
      "grad_norm": 832.0610961914062,
      "learning_rate": 5.975201119766999e-06,
      "loss": 0.2809,
      "step": 28490
    },
    {
      "epoch": 0.4029464576057911,
      "grad_norm": 0.0,
      "learning_rate": 5.973787272547329e-06,
      "loss": 0.0415,
      "step": 28500
    },
    {
      "epoch": 0.4030878423277581,
      "grad_norm": 0.0,
      "learning_rate": 5.97237342532766e-06,
      "loss": 0.6658,
      "step": 28510
    },
    {
      "epoch": 0.403229227049725,
      "grad_norm": 0.0,
      "learning_rate": 5.97095957810799e-06,
      "loss": 1.8793,
      "step": 28520
    },
    {
      "epoch": 0.403370611771692,
      "grad_norm": 0.0,
      "learning_rate": 5.9695457308883206e-06,
      "loss": 0.0006,
      "step": 28530
    },
    {
      "epoch": 0.4035119964936589,
      "grad_norm": 819.6660766601562,
      "learning_rate": 5.9681318836686506e-06,
      "loss": 0.4844,
      "step": 28540
    },
    {
      "epoch": 0.40365338121562583,
      "grad_norm": 4.414020077092573e-05,
      "learning_rate": 5.9667180364489814e-06,
      "loss": 0.0069,
      "step": 28550
    },
    {
      "epoch": 0.4037947659375928,
      "grad_norm": 0.0,
      "learning_rate": 5.965304189229313e-06,
      "loss": 0.0001,
      "step": 28560
    },
    {
      "epoch": 0.40393615065955973,
      "grad_norm": 780.0013427734375,
      "learning_rate": 5.963890342009643e-06,
      "loss": 1.0369,
      "step": 28570
    },
    {
      "epoch": 0.4040775353815267,
      "grad_norm": 518.1390991210938,
      "learning_rate": 5.962476494789974e-06,
      "loss": 0.2168,
      "step": 28580
    },
    {
      "epoch": 0.40421892010349364,
      "grad_norm": 6.684685707092285,
      "learning_rate": 5.961062647570304e-06,
      "loss": 1.2852,
      "step": 28590
    },
    {
      "epoch": 0.40436030482546054,
      "grad_norm": 160.00245666503906,
      "learning_rate": 5.959648800350635e-06,
      "loss": 0.8695,
      "step": 28600
    },
    {
      "epoch": 0.4045016895474275,
      "grad_norm": 1.1280154694759403e-06,
      "learning_rate": 5.958234953130965e-06,
      "loss": 0.004,
      "step": 28610
    },
    {
      "epoch": 0.40464307426939444,
      "grad_norm": 0.0,
      "learning_rate": 5.956821105911296e-06,
      "loss": 3.6665,
      "step": 28620
    },
    {
      "epoch": 0.4047844589913614,
      "grad_norm": 0.0,
      "learning_rate": 5.955407258691627e-06,
      "loss": 1.4207,
      "step": 28630
    },
    {
      "epoch": 0.40492584371332835,
      "grad_norm": 7.801887841196731e-05,
      "learning_rate": 5.953993411471957e-06,
      "loss": 0.0025,
      "step": 28640
    },
    {
      "epoch": 0.4050672284352953,
      "grad_norm": 3.1077590278982825e-08,
      "learning_rate": 5.9525795642522875e-06,
      "loss": 0.014,
      "step": 28650
    },
    {
      "epoch": 0.4052086131572622,
      "grad_norm": 0.0,
      "learning_rate": 5.9511657170326175e-06,
      "loss": 0.0879,
      "step": 28660
    },
    {
      "epoch": 0.40534999787922915,
      "grad_norm": 133.08009338378906,
      "learning_rate": 5.949751869812948e-06,
      "loss": 0.8462,
      "step": 28670
    },
    {
      "epoch": 0.4054913826011961,
      "grad_norm": 0.0,
      "learning_rate": 5.948338022593278e-06,
      "loss": 0.0004,
      "step": 28680
    },
    {
      "epoch": 0.40563276732316306,
      "grad_norm": 225.95416259765625,
      "learning_rate": 5.946924175373609e-06,
      "loss": 1.4498,
      "step": 28690
    },
    {
      "epoch": 0.40577415204513,
      "grad_norm": 0.0,
      "learning_rate": 5.945510328153941e-06,
      "loss": 0.4234,
      "step": 28700
    },
    {
      "epoch": 0.40591553676709696,
      "grad_norm": 0.0,
      "learning_rate": 5.944096480934271e-06,
      "loss": 0.0,
      "step": 28710
    },
    {
      "epoch": 0.4060569214890639,
      "grad_norm": 745.9388427734375,
      "learning_rate": 5.942682633714602e-06,
      "loss": 0.7102,
      "step": 28720
    },
    {
      "epoch": 0.4061983062110308,
      "grad_norm": 0.0,
      "learning_rate": 5.941268786494932e-06,
      "loss": 0.0,
      "step": 28730
    },
    {
      "epoch": 0.40633969093299777,
      "grad_norm": 67.38463592529297,
      "learning_rate": 5.939854939275263e-06,
      "loss": 0.0092,
      "step": 28740
    },
    {
      "epoch": 0.4064810756549647,
      "grad_norm": 0.0,
      "learning_rate": 5.9384410920555935e-06,
      "loss": 0.2036,
      "step": 28750
    },
    {
      "epoch": 0.40662246037693167,
      "grad_norm": 0.0,
      "learning_rate": 5.9370272448359235e-06,
      "loss": 0.0001,
      "step": 28760
    },
    {
      "epoch": 0.4067638450988986,
      "grad_norm": 0.0,
      "learning_rate": 5.935613397616254e-06,
      "loss": 0.014,
      "step": 28770
    },
    {
      "epoch": 0.4069052298208656,
      "grad_norm": 0.006942703388631344,
      "learning_rate": 5.934199550396584e-06,
      "loss": 0.01,
      "step": 28780
    },
    {
      "epoch": 0.4070466145428325,
      "grad_norm": 0.0,
      "learning_rate": 5.932785703176915e-06,
      "loss": 0.5585,
      "step": 28790
    },
    {
      "epoch": 0.4071879992647994,
      "grad_norm": 0.0,
      "learning_rate": 5.931371855957245e-06,
      "loss": 0.6989,
      "step": 28800
    },
    {
      "epoch": 0.4073293839867664,
      "grad_norm": 0.0,
      "learning_rate": 5.929958008737576e-06,
      "loss": 0.001,
      "step": 28810
    },
    {
      "epoch": 0.40747076870873333,
      "grad_norm": 0.0,
      "learning_rate": 5.928544161517907e-06,
      "loss": 0.0,
      "step": 28820
    },
    {
      "epoch": 0.4076121534307003,
      "grad_norm": 6.404433605666782e-08,
      "learning_rate": 5.927130314298237e-06,
      "loss": 0.8906,
      "step": 28830
    },
    {
      "epoch": 0.40775353815266724,
      "grad_norm": 0.0,
      "learning_rate": 5.925716467078569e-06,
      "loss": 0.3713,
      "step": 28840
    },
    {
      "epoch": 0.4078949228746342,
      "grad_norm": 0.0,
      "learning_rate": 5.924302619858898e-06,
      "loss": 0.5582,
      "step": 28850
    },
    {
      "epoch": 0.4080363075966011,
      "grad_norm": 0.0,
      "learning_rate": 5.9228887726392295e-06,
      "loss": 0.0825,
      "step": 28860
    },
    {
      "epoch": 0.40817769231856804,
      "grad_norm": 0.0,
      "learning_rate": 5.9214749254195595e-06,
      "loss": 0.0,
      "step": 28870
    },
    {
      "epoch": 0.408319077040535,
      "grad_norm": 0.04483996704220772,
      "learning_rate": 5.92006107819989e-06,
      "loss": 0.4079,
      "step": 28880
    },
    {
      "epoch": 0.40846046176250195,
      "grad_norm": 0.0,
      "learning_rate": 5.918647230980221e-06,
      "loss": 0.0917,
      "step": 28890
    },
    {
      "epoch": 0.4086018464844689,
      "grad_norm": 7.657625644696964e-08,
      "learning_rate": 5.917233383760551e-06,
      "loss": 1.3186,
      "step": 28900
    },
    {
      "epoch": 0.40874323120643585,
      "grad_norm": 0.14841189980506897,
      "learning_rate": 5.915819536540882e-06,
      "loss": 0.8716,
      "step": 28910
    },
    {
      "epoch": 0.40888461592840275,
      "grad_norm": 828.8787841796875,
      "learning_rate": 5.914405689321212e-06,
      "loss": 0.6598,
      "step": 28920
    },
    {
      "epoch": 0.4090260006503697,
      "grad_norm": 170.70884704589844,
      "learning_rate": 5.912991842101543e-06,
      "loss": 0.0261,
      "step": 28930
    },
    {
      "epoch": 0.40916738537233666,
      "grad_norm": 3.488387301331386e-05,
      "learning_rate": 5.911577994881873e-06,
      "loss": 0.0163,
      "step": 28940
    },
    {
      "epoch": 0.4093087700943036,
      "grad_norm": 0.0,
      "learning_rate": 5.910164147662204e-06,
      "loss": 0.0648,
      "step": 28950
    },
    {
      "epoch": 0.40945015481627056,
      "grad_norm": 0.0,
      "learning_rate": 5.908750300442535e-06,
      "loss": 0.0,
      "step": 28960
    },
    {
      "epoch": 0.4095915395382375,
      "grad_norm": 0.0,
      "learning_rate": 5.907336453222865e-06,
      "loss": 0.1387,
      "step": 28970
    },
    {
      "epoch": 0.40973292426020447,
      "grad_norm": 0.0003482167958281934,
      "learning_rate": 5.905922606003196e-06,
      "loss": 0.2278,
      "step": 28980
    },
    {
      "epoch": 0.40987430898217136,
      "grad_norm": 0.0,
      "learning_rate": 5.9045087587835256e-06,
      "loss": 0.0,
      "step": 28990
    },
    {
      "epoch": 0.4100156937041383,
      "grad_norm": 0.0,
      "learning_rate": 5.903094911563857e-06,
      "loss": 0.0,
      "step": 29000
    },
    {
      "epoch": 0.41015707842610527,
      "grad_norm": 0.0,
      "learning_rate": 5.901681064344188e-06,
      "loss": 0.3503,
      "step": 29010
    },
    {
      "epoch": 0.4102984631480722,
      "grad_norm": 0.9075237512588501,
      "learning_rate": 5.900267217124518e-06,
      "loss": 0.527,
      "step": 29020
    },
    {
      "epoch": 0.4104398478700392,
      "grad_norm": 1.564144492149353,
      "learning_rate": 5.898853369904849e-06,
      "loss": 0.0021,
      "step": 29030
    },
    {
      "epoch": 0.41058123259200613,
      "grad_norm": 0.0,
      "learning_rate": 5.897439522685179e-06,
      "loss": 0.0026,
      "step": 29040
    },
    {
      "epoch": 0.410722617313973,
      "grad_norm": 0.0,
      "learning_rate": 5.89602567546551e-06,
      "loss": 0.2152,
      "step": 29050
    },
    {
      "epoch": 0.41086400203594,
      "grad_norm": 0.00084006815450266,
      "learning_rate": 5.89461182824584e-06,
      "loss": 1.6875,
      "step": 29060
    },
    {
      "epoch": 0.41100538675790693,
      "grad_norm": 885.3824462890625,
      "learning_rate": 5.893197981026171e-06,
      "loss": 0.6009,
      "step": 29070
    },
    {
      "epoch": 0.4111467714798739,
      "grad_norm": 0.0,
      "learning_rate": 5.891784133806502e-06,
      "loss": 0.1808,
      "step": 29080
    },
    {
      "epoch": 0.41128815620184084,
      "grad_norm": 825.936279296875,
      "learning_rate": 5.890370286586832e-06,
      "loss": 1.9289,
      "step": 29090
    },
    {
      "epoch": 0.4114295409238078,
      "grad_norm": 70.25204467773438,
      "learning_rate": 5.8889564393671624e-06,
      "loss": 0.0069,
      "step": 29100
    },
    {
      "epoch": 0.41157092564577474,
      "grad_norm": 0.0,
      "learning_rate": 5.8875425921474925e-06,
      "loss": 0.5589,
      "step": 29110
    },
    {
      "epoch": 0.41171231036774164,
      "grad_norm": 0.0,
      "learning_rate": 5.886128744927823e-06,
      "loss": 1.1861,
      "step": 29120
    },
    {
      "epoch": 0.4118536950897086,
      "grad_norm": 0.0,
      "learning_rate": 5.884714897708153e-06,
      "loss": 0.6844,
      "step": 29130
    },
    {
      "epoch": 0.41199507981167555,
      "grad_norm": 0.0,
      "learning_rate": 5.883301050488485e-06,
      "loss": 1.3484,
      "step": 29140
    },
    {
      "epoch": 0.4121364645336425,
      "grad_norm": 0.0,
      "learning_rate": 5.881887203268816e-06,
      "loss": 0.0,
      "step": 29150
    },
    {
      "epoch": 0.41227784925560945,
      "grad_norm": 7.918738842010498,
      "learning_rate": 5.880473356049146e-06,
      "loss": 0.2195,
      "step": 29160
    },
    {
      "epoch": 0.4124192339775764,
      "grad_norm": 819.7987670898438,
      "learning_rate": 5.879059508829477e-06,
      "loss": 0.7151,
      "step": 29170
    },
    {
      "epoch": 0.4125606186995433,
      "grad_norm": 0.0,
      "learning_rate": 5.877645661609807e-06,
      "loss": 0.2239,
      "step": 29180
    },
    {
      "epoch": 0.41270200342151026,
      "grad_norm": 0.0,
      "learning_rate": 5.876231814390138e-06,
      "loss": 1.4805,
      "step": 29190
    },
    {
      "epoch": 0.4128433881434772,
      "grad_norm": 0.0,
      "learning_rate": 5.8748179671704685e-06,
      "loss": 0.4342,
      "step": 29200
    },
    {
      "epoch": 0.41298477286544416,
      "grad_norm": 0.006070623639971018,
      "learning_rate": 5.8734041199507985e-06,
      "loss": 0.3942,
      "step": 29210
    },
    {
      "epoch": 0.4131261575874111,
      "grad_norm": 0.0,
      "learning_rate": 5.871990272731129e-06,
      "loss": 0.0,
      "step": 29220
    },
    {
      "epoch": 0.41326754230937807,
      "grad_norm": 0.6653293967247009,
      "learning_rate": 5.870576425511459e-06,
      "loss": 0.0001,
      "step": 29230
    },
    {
      "epoch": 0.413408927031345,
      "grad_norm": 0.0,
      "learning_rate": 5.86916257829179e-06,
      "loss": 0.0018,
      "step": 29240
    },
    {
      "epoch": 0.4135503117533119,
      "grad_norm": 0.0,
      "learning_rate": 5.86774873107212e-06,
      "loss": 0.0,
      "step": 29250
    },
    {
      "epoch": 0.41369169647527887,
      "grad_norm": 0.0,
      "learning_rate": 5.866334883852451e-06,
      "loss": 0.1663,
      "step": 29260
    },
    {
      "epoch": 0.4138330811972458,
      "grad_norm": 0.0,
      "learning_rate": 5.864921036632783e-06,
      "loss": 0.0001,
      "step": 29270
    },
    {
      "epoch": 0.4139744659192128,
      "grad_norm": 0.0,
      "learning_rate": 5.863507189413113e-06,
      "loss": 0.0014,
      "step": 29280
    },
    {
      "epoch": 0.41411585064117973,
      "grad_norm": 0.0,
      "learning_rate": 5.862093342193444e-06,
      "loss": 0.9847,
      "step": 29290
    },
    {
      "epoch": 0.4142572353631467,
      "grad_norm": 0.0,
      "learning_rate": 5.860679494973774e-06,
      "loss": 0.0,
      "step": 29300
    },
    {
      "epoch": 0.4143986200851136,
      "grad_norm": 0.0,
      "learning_rate": 5.8592656477541045e-06,
      "loss": 0.5309,
      "step": 29310
    },
    {
      "epoch": 0.41454000480708053,
      "grad_norm": 0.0,
      "learning_rate": 5.8578518005344345e-06,
      "loss": 0.304,
      "step": 29320
    },
    {
      "epoch": 0.4146813895290475,
      "grad_norm": 0.0,
      "learning_rate": 5.856437953314765e-06,
      "loss": 0.1661,
      "step": 29330
    },
    {
      "epoch": 0.41482277425101444,
      "grad_norm": 1.8113323676516302e-06,
      "learning_rate": 5.855024106095096e-06,
      "loss": 0.0011,
      "step": 29340
    },
    {
      "epoch": 0.4149641589729814,
      "grad_norm": 0.0,
      "learning_rate": 5.853610258875426e-06,
      "loss": 0.0,
      "step": 29350
    },
    {
      "epoch": 0.41510554369494834,
      "grad_norm": 0.10812894254922867,
      "learning_rate": 5.852196411655757e-06,
      "loss": 0.0052,
      "step": 29360
    },
    {
      "epoch": 0.4152469284169153,
      "grad_norm": 0.05125533416867256,
      "learning_rate": 5.850782564436087e-06,
      "loss": 0.0,
      "step": 29370
    },
    {
      "epoch": 0.4153883131388822,
      "grad_norm": 0.0,
      "learning_rate": 5.849368717216418e-06,
      "loss": 0.0,
      "step": 29380
    },
    {
      "epoch": 0.41552969786084915,
      "grad_norm": 0.0,
      "learning_rate": 5.847954869996748e-06,
      "loss": 0.3807,
      "step": 29390
    },
    {
      "epoch": 0.4156710825828161,
      "grad_norm": 1191.9346923828125,
      "learning_rate": 5.846541022777079e-06,
      "loss": 0.5956,
      "step": 29400
    },
    {
      "epoch": 0.41581246730478305,
      "grad_norm": 0.0,
      "learning_rate": 5.8451271755574105e-06,
      "loss": 0.1321,
      "step": 29410
    },
    {
      "epoch": 0.41595385202675,
      "grad_norm": 3.99285888671875,
      "learning_rate": 5.84371332833774e-06,
      "loss": 0.124,
      "step": 29420
    },
    {
      "epoch": 0.41609523674871696,
      "grad_norm": 0.0,
      "learning_rate": 5.842299481118071e-06,
      "loss": 0.0168,
      "step": 29430
    },
    {
      "epoch": 0.41623662147068385,
      "grad_norm": 0.0,
      "learning_rate": 5.840885633898401e-06,
      "loss": 0.0,
      "step": 29440
    },
    {
      "epoch": 0.4163780061926508,
      "grad_norm": 0.0,
      "learning_rate": 5.839471786678732e-06,
      "loss": 0.0,
      "step": 29450
    },
    {
      "epoch": 0.41651939091461776,
      "grad_norm": 0.0,
      "learning_rate": 5.838057939459063e-06,
      "loss": 0.0,
      "step": 29460
    },
    {
      "epoch": 0.4166607756365847,
      "grad_norm": 965.9164428710938,
      "learning_rate": 5.836644092239393e-06,
      "loss": 1.2821,
      "step": 29470
    },
    {
      "epoch": 0.41680216035855167,
      "grad_norm": 0.758539617061615,
      "learning_rate": 5.835230245019724e-06,
      "loss": 0.5895,
      "step": 29480
    },
    {
      "epoch": 0.4169435450805186,
      "grad_norm": 0.0,
      "learning_rate": 5.833816397800054e-06,
      "loss": 0.9481,
      "step": 29490
    },
    {
      "epoch": 0.41708492980248557,
      "grad_norm": 0.0,
      "learning_rate": 5.832402550580385e-06,
      "loss": 0.0429,
      "step": 29500
    },
    {
      "epoch": 0.41722631452445247,
      "grad_norm": 0.0,
      "learning_rate": 5.830988703360715e-06,
      "loss": 0.0,
      "step": 29510
    },
    {
      "epoch": 0.4173676992464194,
      "grad_norm": 0.0,
      "learning_rate": 5.829574856141046e-06,
      "loss": 0.0002,
      "step": 29520
    },
    {
      "epoch": 0.4175090839683864,
      "grad_norm": 9.638549869350754e-08,
      "learning_rate": 5.8281610089213766e-06,
      "loss": 0.0259,
      "step": 29530
    },
    {
      "epoch": 0.4176504686903533,
      "grad_norm": 763.5450439453125,
      "learning_rate": 5.826747161701707e-06,
      "loss": 1.3289,
      "step": 29540
    },
    {
      "epoch": 0.4177918534123203,
      "grad_norm": 0.0,
      "learning_rate": 5.825333314482038e-06,
      "loss": 0.7938,
      "step": 29550
    },
    {
      "epoch": 0.41793323813428723,
      "grad_norm": 0.0,
      "learning_rate": 5.8239194672623674e-06,
      "loss": 0.0,
      "step": 29560
    },
    {
      "epoch": 0.41807462285625413,
      "grad_norm": 0.0,
      "learning_rate": 5.822505620042699e-06,
      "loss": 0.1044,
      "step": 29570
    },
    {
      "epoch": 0.4182160075782211,
      "grad_norm": 0.0,
      "learning_rate": 5.821091772823028e-06,
      "loss": 0.1137,
      "step": 29580
    },
    {
      "epoch": 0.41835739230018804,
      "grad_norm": 0.0,
      "learning_rate": 5.81967792560336e-06,
      "loss": 0.3637,
      "step": 29590
    },
    {
      "epoch": 0.418498777022155,
      "grad_norm": 866.7625732421875,
      "learning_rate": 5.818264078383691e-06,
      "loss": 1.6049,
      "step": 29600
    },
    {
      "epoch": 0.41864016174412194,
      "grad_norm": 0.0,
      "learning_rate": 5.816850231164021e-06,
      "loss": 0.0,
      "step": 29610
    },
    {
      "epoch": 0.4187815464660889,
      "grad_norm": 0.0,
      "learning_rate": 5.815436383944352e-06,
      "loss": 0.2636,
      "step": 29620
    },
    {
      "epoch": 0.4189229311880558,
      "grad_norm": 0.0,
      "learning_rate": 5.814022536724682e-06,
      "loss": 4.156,
      "step": 29630
    },
    {
      "epoch": 0.41906431591002274,
      "grad_norm": 8.0933256149292,
      "learning_rate": 5.812608689505013e-06,
      "loss": 0.2804,
      "step": 29640
    },
    {
      "epoch": 0.4192057006319897,
      "grad_norm": 2.3490385956392856e-06,
      "learning_rate": 5.811194842285343e-06,
      "loss": 0.0001,
      "step": 29650
    },
    {
      "epoch": 0.41934708535395665,
      "grad_norm": 0.009484075009822845,
      "learning_rate": 5.8097809950656735e-06,
      "loss": 1.3363,
      "step": 29660
    },
    {
      "epoch": 0.4194884700759236,
      "grad_norm": 284.4750061035156,
      "learning_rate": 5.808367147846004e-06,
      "loss": 0.2564,
      "step": 29670
    },
    {
      "epoch": 0.41962985479789056,
      "grad_norm": 0.0,
      "learning_rate": 5.806953300626334e-06,
      "loss": 0.0044,
      "step": 29680
    },
    {
      "epoch": 0.4197712395198575,
      "grad_norm": 0.0,
      "learning_rate": 5.805539453406665e-06,
      "loss": 0.0006,
      "step": 29690
    },
    {
      "epoch": 0.4199126242418244,
      "grad_norm": 0.029436349868774414,
      "learning_rate": 5.804125606186995e-06,
      "loss": 0.0002,
      "step": 29700
    },
    {
      "epoch": 0.42005400896379136,
      "grad_norm": 0.010565556585788727,
      "learning_rate": 5.802711758967327e-06,
      "loss": 0.1901,
      "step": 29710
    },
    {
      "epoch": 0.4201953936857583,
      "grad_norm": 0.0,
      "learning_rate": 5.801297911747658e-06,
      "loss": 0.1133,
      "step": 29720
    },
    {
      "epoch": 0.42033677840772526,
      "grad_norm": 0.0,
      "learning_rate": 5.799884064527988e-06,
      "loss": 0.0016,
      "step": 29730
    },
    {
      "epoch": 0.4204781631296922,
      "grad_norm": 0.0,
      "learning_rate": 5.798470217308319e-06,
      "loss": 0.6059,
      "step": 29740
    },
    {
      "epoch": 0.42061954785165917,
      "grad_norm": 0.0,
      "learning_rate": 5.797056370088649e-06,
      "loss": 0.6191,
      "step": 29750
    },
    {
      "epoch": 0.42076093257362607,
      "grad_norm": 2.9873728379925524e-08,
      "learning_rate": 5.7956425228689795e-06,
      "loss": 0.3766,
      "step": 29760
    },
    {
      "epoch": 0.420902317295593,
      "grad_norm": 0.0,
      "learning_rate": 5.7942286756493095e-06,
      "loss": 0.2931,
      "step": 29770
    },
    {
      "epoch": 0.42104370201756,
      "grad_norm": 0.0,
      "learning_rate": 5.79281482842964e-06,
      "loss": 0.2168,
      "step": 29780
    },
    {
      "epoch": 0.4211850867395269,
      "grad_norm": 0.0,
      "learning_rate": 5.791400981209971e-06,
      "loss": 0.9876,
      "step": 29790
    },
    {
      "epoch": 0.4213264714614939,
      "grad_norm": 0.0,
      "learning_rate": 5.789987133990301e-06,
      "loss": 0.0,
      "step": 29800
    },
    {
      "epoch": 0.42146785618346083,
      "grad_norm": 0.0,
      "learning_rate": 5.788573286770632e-06,
      "loss": 0.0003,
      "step": 29810
    },
    {
      "epoch": 0.4216092409054278,
      "grad_norm": 37.05244827270508,
      "learning_rate": 5.787159439550962e-06,
      "loss": 0.034,
      "step": 29820
    },
    {
      "epoch": 0.4217506256273947,
      "grad_norm": 37.91542434692383,
      "learning_rate": 5.785745592331293e-06,
      "loss": 3.5431,
      "step": 29830
    },
    {
      "epoch": 0.42189201034936163,
      "grad_norm": 0.0,
      "learning_rate": 5.784331745111623e-06,
      "loss": 0.0003,
      "step": 29840
    },
    {
      "epoch": 0.4220333950713286,
      "grad_norm": 0.010973389260470867,
      "learning_rate": 5.782917897891955e-06,
      "loss": 0.0003,
      "step": 29850
    },
    {
      "epoch": 0.42217477979329554,
      "grad_norm": 0.0,
      "learning_rate": 5.7815040506722855e-06,
      "loss": 0.0294,
      "step": 29860
    },
    {
      "epoch": 0.4223161645152625,
      "grad_norm": 0.0,
      "learning_rate": 5.7800902034526155e-06,
      "loss": 0.0002,
      "step": 29870
    },
    {
      "epoch": 0.42245754923722945,
      "grad_norm": 0.0,
      "learning_rate": 5.778676356232946e-06,
      "loss": 3.1205,
      "step": 29880
    },
    {
      "epoch": 0.42259893395919634,
      "grad_norm": 0.0,
      "learning_rate": 5.777262509013276e-06,
      "loss": 0.0,
      "step": 29890
    },
    {
      "epoch": 0.4227403186811633,
      "grad_norm": 12.983377456665039,
      "learning_rate": 5.775848661793607e-06,
      "loss": 0.0247,
      "step": 29900
    },
    {
      "epoch": 0.42288170340313025,
      "grad_norm": 0.0,
      "learning_rate": 5.774434814573937e-06,
      "loss": 0.0,
      "step": 29910
    },
    {
      "epoch": 0.4230230881250972,
      "grad_norm": 0.0,
      "learning_rate": 5.773020967354268e-06,
      "loss": 0.0001,
      "step": 29920
    },
    {
      "epoch": 0.42316447284706415,
      "grad_norm": 1.553365081008451e-08,
      "learning_rate": 5.771607120134599e-06,
      "loss": 0.0002,
      "step": 29930
    },
    {
      "epoch": 0.4233058575690311,
      "grad_norm": 3.901789114024723e-06,
      "learning_rate": 5.770193272914929e-06,
      "loss": 0.3762,
      "step": 29940
    },
    {
      "epoch": 0.42344724229099806,
      "grad_norm": 0.0,
      "learning_rate": 5.76877942569526e-06,
      "loss": 0.0646,
      "step": 29950
    },
    {
      "epoch": 0.42358862701296496,
      "grad_norm": 0.0,
      "learning_rate": 5.76736557847559e-06,
      "loss": 0.0,
      "step": 29960
    },
    {
      "epoch": 0.4237300117349319,
      "grad_norm": 2.3410846097249305e-06,
      "learning_rate": 5.765951731255921e-06,
      "loss": 0.1514,
      "step": 29970
    },
    {
      "epoch": 0.42387139645689886,
      "grad_norm": 0.0,
      "learning_rate": 5.764537884036252e-06,
      "loss": 0.0,
      "step": 29980
    },
    {
      "epoch": 0.4240127811788658,
      "grad_norm": 0.001385872601531446,
      "learning_rate": 5.7631240368165816e-06,
      "loss": 0.0,
      "step": 29990
    },
    {
      "epoch": 0.42415416590083277,
      "grad_norm": 0.0,
      "learning_rate": 5.761710189596913e-06,
      "loss": 0.0037,
      "step": 30000
    },
    {
      "epoch": 0.4242955506227997,
      "grad_norm": 0.0,
      "learning_rate": 5.760296342377243e-06,
      "loss": 0.4961,
      "step": 30010
    },
    {
      "epoch": 0.4244369353447666,
      "grad_norm": 3.3558262657606974e-06,
      "learning_rate": 5.758882495157574e-06,
      "loss": 0.8129,
      "step": 30020
    },
    {
      "epoch": 0.42457832006673357,
      "grad_norm": 0.0,
      "learning_rate": 5.757468647937904e-06,
      "loss": 1.6281,
      "step": 30030
    },
    {
      "epoch": 0.4247197047887005,
      "grad_norm": 0.0,
      "learning_rate": 5.756054800718235e-06,
      "loss": 0.0813,
      "step": 30040
    },
    {
      "epoch": 0.4248610895106675,
      "grad_norm": 0.3469892740249634,
      "learning_rate": 5.754640953498566e-06,
      "loss": 0.0058,
      "step": 30050
    },
    {
      "epoch": 0.42500247423263443,
      "grad_norm": 0.0,
      "learning_rate": 5.753227106278896e-06,
      "loss": 0.0081,
      "step": 30060
    },
    {
      "epoch": 0.4251438589546014,
      "grad_norm": 0.0,
      "learning_rate": 5.751813259059227e-06,
      "loss": 0.0,
      "step": 30070
    },
    {
      "epoch": 0.42528524367656834,
      "grad_norm": 5.760650765296305e-06,
      "learning_rate": 5.750399411839557e-06,
      "loss": 1.1764,
      "step": 30080
    },
    {
      "epoch": 0.42542662839853523,
      "grad_norm": 0.0,
      "learning_rate": 5.748985564619888e-06,
      "loss": 0.0101,
      "step": 30090
    },
    {
      "epoch": 0.4255680131205022,
      "grad_norm": 0.0,
      "learning_rate": 5.747571717400218e-06,
      "loss": 0.019,
      "step": 30100
    },
    {
      "epoch": 0.42570939784246914,
      "grad_norm": 0.0,
      "learning_rate": 5.7461578701805485e-06,
      "loss": 0.0002,
      "step": 30110
    },
    {
      "epoch": 0.4258507825644361,
      "grad_norm": 0.0,
      "learning_rate": 5.74474402296088e-06,
      "loss": 0.0004,
      "step": 30120
    },
    {
      "epoch": 0.42599216728640305,
      "grad_norm": 0.0,
      "learning_rate": 5.743330175741209e-06,
      "loss": 0.0145,
      "step": 30130
    },
    {
      "epoch": 0.42613355200837,
      "grad_norm": 7.35527282813564e-05,
      "learning_rate": 5.741916328521541e-06,
      "loss": 0.0157,
      "step": 30140
    },
    {
      "epoch": 0.4262749367303369,
      "grad_norm": 0.0,
      "learning_rate": 5.74050248130187e-06,
      "loss": 0.2866,
      "step": 30150
    },
    {
      "epoch": 0.42641632145230385,
      "grad_norm": 61.31375503540039,
      "learning_rate": 5.739088634082202e-06,
      "loss": 0.412,
      "step": 30160
    },
    {
      "epoch": 0.4265577061742708,
      "grad_norm": 0.0,
      "learning_rate": 5.737674786862533e-06,
      "loss": 0.0,
      "step": 30170
    },
    {
      "epoch": 0.42669909089623775,
      "grad_norm": 0.13383027911186218,
      "learning_rate": 5.736260939642863e-06,
      "loss": 0.0634,
      "step": 30180
    },
    {
      "epoch": 0.4268404756182047,
      "grad_norm": 0.0,
      "learning_rate": 5.734847092423194e-06,
      "loss": 0.0083,
      "step": 30190
    },
    {
      "epoch": 0.42698186034017166,
      "grad_norm": 0.0,
      "learning_rate": 5.733433245203524e-06,
      "loss": 0.0,
      "step": 30200
    },
    {
      "epoch": 0.4271232450621386,
      "grad_norm": 840.7017822265625,
      "learning_rate": 5.7320193979838545e-06,
      "loss": 0.6086,
      "step": 30210
    },
    {
      "epoch": 0.4272646297841055,
      "grad_norm": 0.0,
      "learning_rate": 5.7306055507641845e-06,
      "loss": 0.001,
      "step": 30220
    },
    {
      "epoch": 0.42740601450607246,
      "grad_norm": 0.0,
      "learning_rate": 5.729191703544515e-06,
      "loss": 0.3064,
      "step": 30230
    },
    {
      "epoch": 0.4275473992280394,
      "grad_norm": 0.0,
      "learning_rate": 5.727777856324846e-06,
      "loss": 0.3896,
      "step": 30240
    },
    {
      "epoch": 0.42768878395000637,
      "grad_norm": 0.002981561003252864,
      "learning_rate": 5.726364009105176e-06,
      "loss": 3.4859,
      "step": 30250
    },
    {
      "epoch": 0.4278301686719733,
      "grad_norm": 0.00352679705247283,
      "learning_rate": 5.724950161885507e-06,
      "loss": 1.0062,
      "step": 30260
    },
    {
      "epoch": 0.4279715533939403,
      "grad_norm": 20.9061222076416,
      "learning_rate": 5.723536314665837e-06,
      "loss": 0.0018,
      "step": 30270
    },
    {
      "epoch": 0.42811293811590717,
      "grad_norm": 0.33805790543556213,
      "learning_rate": 5.722122467446169e-06,
      "loss": 0.0318,
      "step": 30280
    },
    {
      "epoch": 0.4282543228378741,
      "grad_norm": 0.0,
      "learning_rate": 5.720708620226498e-06,
      "loss": 0.0,
      "step": 30290
    },
    {
      "epoch": 0.4283957075598411,
      "grad_norm": 60.920005798339844,
      "learning_rate": 5.71929477300683e-06,
      "loss": 1.2114,
      "step": 30300
    },
    {
      "epoch": 0.42853709228180803,
      "grad_norm": 0.0,
      "learning_rate": 5.7178809257871605e-06,
      "loss": 0.0785,
      "step": 30310
    },
    {
      "epoch": 0.428678477003775,
      "grad_norm": 0.0,
      "learning_rate": 5.7164670785674905e-06,
      "loss": 0.0051,
      "step": 30320
    },
    {
      "epoch": 0.42881986172574194,
      "grad_norm": 0.0015022627776488662,
      "learning_rate": 5.715053231347821e-06,
      "loss": 0.8938,
      "step": 30330
    },
    {
      "epoch": 0.4289612464477089,
      "grad_norm": 33.970157623291016,
      "learning_rate": 5.713639384128151e-06,
      "loss": 1.7763,
      "step": 30340
    },
    {
      "epoch": 0.4291026311696758,
      "grad_norm": 0.0,
      "learning_rate": 5.712225536908482e-06,
      "loss": 0.0,
      "step": 30350
    },
    {
      "epoch": 0.42924401589164274,
      "grad_norm": 0.015947062522172928,
      "learning_rate": 5.710811689688812e-06,
      "loss": 0.6223,
      "step": 30360
    },
    {
      "epoch": 0.4293854006136097,
      "grad_norm": 0.0,
      "learning_rate": 5.709397842469143e-06,
      "loss": 0.6299,
      "step": 30370
    },
    {
      "epoch": 0.42952678533557664,
      "grad_norm": 0.0,
      "learning_rate": 5.707983995249474e-06,
      "loss": 0.6792,
      "step": 30380
    },
    {
      "epoch": 0.4296681700575436,
      "grad_norm": 577.336181640625,
      "learning_rate": 5.706570148029804e-06,
      "loss": 0.3683,
      "step": 30390
    },
    {
      "epoch": 0.42980955477951055,
      "grad_norm": 0.0,
      "learning_rate": 5.705156300810135e-06,
      "loss": 0.1821,
      "step": 30400
    },
    {
      "epoch": 0.42995093950147745,
      "grad_norm": 777.7068481445312,
      "learning_rate": 5.703742453590465e-06,
      "loss": 0.9198,
      "step": 30410
    },
    {
      "epoch": 0.4300923242234444,
      "grad_norm": 0.0,
      "learning_rate": 5.702328606370796e-06,
      "loss": 0.0,
      "step": 30420
    },
    {
      "epoch": 0.43023370894541135,
      "grad_norm": 0.0,
      "learning_rate": 5.700914759151127e-06,
      "loss": 0.0,
      "step": 30430
    },
    {
      "epoch": 0.4303750936673783,
      "grad_norm": 0.0,
      "learning_rate": 5.699500911931457e-06,
      "loss": 0.0,
      "step": 30440
    },
    {
      "epoch": 0.43051647838934526,
      "grad_norm": 3.8227524782996625e-05,
      "learning_rate": 5.698087064711788e-06,
      "loss": 1.0057,
      "step": 30450
    },
    {
      "epoch": 0.4306578631113122,
      "grad_norm": 0.0,
      "learning_rate": 5.696673217492118e-06,
      "loss": 0.3273,
      "step": 30460
    },
    {
      "epoch": 0.4307992478332791,
      "grad_norm": 0.0,
      "learning_rate": 5.695259370272449e-06,
      "loss": 0.0003,
      "step": 30470
    },
    {
      "epoch": 0.43094063255524606,
      "grad_norm": 0.0,
      "learning_rate": 5.693845523052779e-06,
      "loss": 0.2652,
      "step": 30480
    },
    {
      "epoch": 0.431082017277213,
      "grad_norm": 3.111503701802576e-06,
      "learning_rate": 5.69243167583311e-06,
      "loss": 0.0002,
      "step": 30490
    },
    {
      "epoch": 0.43122340199917997,
      "grad_norm": 0.0,
      "learning_rate": 5.691017828613441e-06,
      "loss": 0.0559,
      "step": 30500
    },
    {
      "epoch": 0.4313647867211469,
      "grad_norm": 8.382195472717285,
      "learning_rate": 5.689603981393771e-06,
      "loss": 0.0289,
      "step": 30510
    },
    {
      "epoch": 0.4315061714431139,
      "grad_norm": 0.0,
      "learning_rate": 5.688190134174102e-06,
      "loss": 0.0084,
      "step": 30520
    },
    {
      "epoch": 0.4316475561650808,
      "grad_norm": 0.0,
      "learning_rate": 5.686776286954432e-06,
      "loss": 0.0469,
      "step": 30530
    },
    {
      "epoch": 0.4317889408870477,
      "grad_norm": 0.00011356471804901958,
      "learning_rate": 5.685362439734763e-06,
      "loss": 0.428,
      "step": 30540
    },
    {
      "epoch": 0.4319303256090147,
      "grad_norm": 18.799142837524414,
      "learning_rate": 5.683948592515093e-06,
      "loss": 0.0026,
      "step": 30550
    },
    {
      "epoch": 0.43207171033098163,
      "grad_norm": 0.0,
      "learning_rate": 5.6825347452954234e-06,
      "loss": 0.0026,
      "step": 30560
    },
    {
      "epoch": 0.4322130950529486,
      "grad_norm": 0.0,
      "learning_rate": 5.681120898075755e-06,
      "loss": 0.2814,
      "step": 30570
    },
    {
      "epoch": 0.43235447977491553,
      "grad_norm": 0.0,
      "learning_rate": 5.679707050856085e-06,
      "loss": 0.0002,
      "step": 30580
    },
    {
      "epoch": 0.4324958644968825,
      "grad_norm": 0.0,
      "learning_rate": 5.678293203636416e-06,
      "loss": 0.0221,
      "step": 30590
    },
    {
      "epoch": 0.4326372492188494,
      "grad_norm": 0.11526467651128769,
      "learning_rate": 5.676879356416746e-06,
      "loss": 0.0001,
      "step": 30600
    },
    {
      "epoch": 0.43277863394081634,
      "grad_norm": 0.0,
      "learning_rate": 5.675465509197077e-06,
      "loss": 0.3741,
      "step": 30610
    },
    {
      "epoch": 0.4329200186627833,
      "grad_norm": 0.0,
      "learning_rate": 5.674051661977407e-06,
      "loss": 0.1694,
      "step": 30620
    },
    {
      "epoch": 0.43306140338475024,
      "grad_norm": 5.925250661675818e-06,
      "learning_rate": 5.672637814757738e-06,
      "loss": 0.0,
      "step": 30630
    },
    {
      "epoch": 0.4332027881067172,
      "grad_norm": 851.9702758789062,
      "learning_rate": 5.671223967538069e-06,
      "loss": 0.9006,
      "step": 30640
    },
    {
      "epoch": 0.43334417282868415,
      "grad_norm": 0.0,
      "learning_rate": 5.669810120318399e-06,
      "loss": 0.4452,
      "step": 30650
    },
    {
      "epoch": 0.4334855575506511,
      "grad_norm": 0.0,
      "learning_rate": 5.6683962730987295e-06,
      "loss": 0.0021,
      "step": 30660
    },
    {
      "epoch": 0.433626942272618,
      "grad_norm": 0.00014278166054282337,
      "learning_rate": 5.6669824258790595e-06,
      "loss": 0.3757,
      "step": 30670
    },
    {
      "epoch": 0.43376832699458495,
      "grad_norm": 0.0,
      "learning_rate": 5.66556857865939e-06,
      "loss": 0.1297,
      "step": 30680
    },
    {
      "epoch": 0.4339097117165519,
      "grad_norm": 0.0,
      "learning_rate": 5.664154731439721e-06,
      "loss": 0.2141,
      "step": 30690
    },
    {
      "epoch": 0.43405109643851886,
      "grad_norm": 0.0,
      "learning_rate": 5.662740884220051e-06,
      "loss": 0.0131,
      "step": 30700
    },
    {
      "epoch": 0.4341924811604858,
      "grad_norm": 0.0,
      "learning_rate": 5.661327037000383e-06,
      "loss": 0.002,
      "step": 30710
    },
    {
      "epoch": 0.43433386588245276,
      "grad_norm": 0.0,
      "learning_rate": 5.659913189780712e-06,
      "loss": 0.0002,
      "step": 30720
    },
    {
      "epoch": 0.43447525060441966,
      "grad_norm": 0.0,
      "learning_rate": 5.658499342561044e-06,
      "loss": 0.0015,
      "step": 30730
    },
    {
      "epoch": 0.4346166353263866,
      "grad_norm": 0.0,
      "learning_rate": 5.657085495341374e-06,
      "loss": 0.0001,
      "step": 30740
    },
    {
      "epoch": 0.43475802004835357,
      "grad_norm": 0.0,
      "learning_rate": 5.655671648121705e-06,
      "loss": 0.0604,
      "step": 30750
    },
    {
      "epoch": 0.4348994047703205,
      "grad_norm": 0.0,
      "learning_rate": 5.6542578009020355e-06,
      "loss": 0.495,
      "step": 30760
    },
    {
      "epoch": 0.43504078949228747,
      "grad_norm": 0.08536089956760406,
      "learning_rate": 5.6528439536823655e-06,
      "loss": 0.0,
      "step": 30770
    },
    {
      "epoch": 0.4351821742142544,
      "grad_norm": 0.0,
      "learning_rate": 5.651430106462696e-06,
      "loss": 0.0,
      "step": 30780
    },
    {
      "epoch": 0.4353235589362214,
      "grad_norm": 0.0,
      "learning_rate": 5.650016259243026e-06,
      "loss": 0.268,
      "step": 30790
    },
    {
      "epoch": 0.4354649436581883,
      "grad_norm": 0.0,
      "learning_rate": 5.648602412023357e-06,
      "loss": 0.002,
      "step": 30800
    },
    {
      "epoch": 0.4356063283801552,
      "grad_norm": 0.0,
      "learning_rate": 5.647188564803687e-06,
      "loss": 0.0,
      "step": 30810
    },
    {
      "epoch": 0.4357477131021222,
      "grad_norm": 0.7454078197479248,
      "learning_rate": 5.645774717584018e-06,
      "loss": 0.1703,
      "step": 30820
    },
    {
      "epoch": 0.43588909782408913,
      "grad_norm": 0.0,
      "learning_rate": 5.644360870364349e-06,
      "loss": 0.0006,
      "step": 30830
    },
    {
      "epoch": 0.4360304825460561,
      "grad_norm": 0.0,
      "learning_rate": 5.642947023144679e-06,
      "loss": 0.0,
      "step": 30840
    },
    {
      "epoch": 0.43617186726802304,
      "grad_norm": 0.0,
      "learning_rate": 5.641533175925011e-06,
      "loss": 0.0005,
      "step": 30850
    },
    {
      "epoch": 0.43631325198998994,
      "grad_norm": 0.0,
      "learning_rate": 5.64011932870534e-06,
      "loss": 0.0,
      "step": 30860
    },
    {
      "epoch": 0.4364546367119569,
      "grad_norm": 0.0,
      "learning_rate": 5.6387054814856715e-06,
      "loss": 0.0,
      "step": 30870
    },
    {
      "epoch": 0.43659602143392384,
      "grad_norm": 0.0,
      "learning_rate": 5.6372916342660015e-06,
      "loss": 0.0002,
      "step": 30880
    },
    {
      "epoch": 0.4367374061558908,
      "grad_norm": 2.5574841799880232e-08,
      "learning_rate": 5.635877787046332e-06,
      "loss": 0.0,
      "step": 30890
    },
    {
      "epoch": 0.43687879087785775,
      "grad_norm": 0.0,
      "learning_rate": 5.634463939826663e-06,
      "loss": 0.0179,
      "step": 30900
    },
    {
      "epoch": 0.4370201755998247,
      "grad_norm": 0.0,
      "learning_rate": 5.633050092606993e-06,
      "loss": 0.0,
      "step": 30910
    },
    {
      "epoch": 0.43716156032179165,
      "grad_norm": 0.0,
      "learning_rate": 5.631636245387324e-06,
      "loss": 0.0,
      "step": 30920
    },
    {
      "epoch": 0.43730294504375855,
      "grad_norm": 0.0,
      "learning_rate": 5.630222398167654e-06,
      "loss": 0.0,
      "step": 30930
    },
    {
      "epoch": 0.4374443297657255,
      "grad_norm": 2.052870035171509,
      "learning_rate": 5.628808550947985e-06,
      "loss": 0.0006,
      "step": 30940
    },
    {
      "epoch": 0.43758571448769246,
      "grad_norm": 0.0,
      "learning_rate": 5.627394703728316e-06,
      "loss": 0.0,
      "step": 30950
    },
    {
      "epoch": 0.4377270992096594,
      "grad_norm": 0.0,
      "learning_rate": 5.625980856508646e-06,
      "loss": 0.0,
      "step": 30960
    },
    {
      "epoch": 0.43786848393162636,
      "grad_norm": 12.163866996765137,
      "learning_rate": 5.624567009288977e-06,
      "loss": 0.0016,
      "step": 30970
    },
    {
      "epoch": 0.4380098686535933,
      "grad_norm": 0.00174768827855587,
      "learning_rate": 5.623153162069307e-06,
      "loss": 0.0,
      "step": 30980
    },
    {
      "epoch": 0.4381512533755602,
      "grad_norm": 0.0,
      "learning_rate": 5.6217393148496376e-06,
      "loss": 0.0,
      "step": 30990
    },
    {
      "epoch": 0.43829263809752717,
      "grad_norm": 0.020458979532122612,
      "learning_rate": 5.6203254676299676e-06,
      "loss": 0.0194,
      "step": 31000
    },
    {
      "epoch": 0.4384340228194941,
      "grad_norm": 0.0,
      "learning_rate": 5.618911620410299e-06,
      "loss": 0.0001,
      "step": 31010
    },
    {
      "epoch": 0.43857540754146107,
      "grad_norm": 0.0,
      "learning_rate": 5.61749777319063e-06,
      "loss": 0.0,
      "step": 31020
    },
    {
      "epoch": 0.438716792263428,
      "grad_norm": 0.0014609458157792687,
      "learning_rate": 5.61608392597096e-06,
      "loss": 0.0,
      "step": 31030
    },
    {
      "epoch": 0.438858176985395,
      "grad_norm": 0.0,
      "learning_rate": 5.614670078751291e-06,
      "loss": 0.0,
      "step": 31040
    },
    {
      "epoch": 0.43899956170736193,
      "grad_norm": 0.0,
      "learning_rate": 5.613256231531621e-06,
      "loss": 0.0053,
      "step": 31050
    },
    {
      "epoch": 0.4391409464293288,
      "grad_norm": 0.0013977575581520796,
      "learning_rate": 5.611842384311952e-06,
      "loss": 0.0001,
      "step": 31060
    },
    {
      "epoch": 0.4392823311512958,
      "grad_norm": 5.384212272474542e-05,
      "learning_rate": 5.610428537092282e-06,
      "loss": 0.5855,
      "step": 31070
    },
    {
      "epoch": 0.43942371587326273,
      "grad_norm": 0.0,
      "learning_rate": 5.609014689872613e-06,
      "loss": 0.0005,
      "step": 31080
    },
    {
      "epoch": 0.4395651005952297,
      "grad_norm": 0.06867100298404694,
      "learning_rate": 5.607600842652944e-06,
      "loss": 0.1349,
      "step": 31090
    },
    {
      "epoch": 0.43970648531719664,
      "grad_norm": 8.251604413089808e-06,
      "learning_rate": 5.606186995433274e-06,
      "loss": 0.0,
      "step": 31100
    },
    {
      "epoch": 0.4398478700391636,
      "grad_norm": 0.0,
      "learning_rate": 5.6047731482136045e-06,
      "loss": 0.0,
      "step": 31110
    },
    {
      "epoch": 0.4399892547611305,
      "grad_norm": 0.0,
      "learning_rate": 5.6033593009939345e-06,
      "loss": 0.0,
      "step": 31120
    },
    {
      "epoch": 0.44013063948309744,
      "grad_norm": 0.0,
      "learning_rate": 5.601945453774265e-06,
      "loss": 0.0038,
      "step": 31130
    },
    {
      "epoch": 0.4402720242050644,
      "grad_norm": 315.183837890625,
      "learning_rate": 5.600531606554597e-06,
      "loss": 0.1116,
      "step": 31140
    },
    {
      "epoch": 0.44041340892703135,
      "grad_norm": 0.005894877947866917,
      "learning_rate": 5.599117759334927e-06,
      "loss": 0.0002,
      "step": 31150
    },
    {
      "epoch": 0.4405547936489983,
      "grad_norm": 0.0,
      "learning_rate": 5.597703912115258e-06,
      "loss": 0.0001,
      "step": 31160
    },
    {
      "epoch": 0.44069617837096525,
      "grad_norm": 0.0,
      "learning_rate": 5.596290064895588e-06,
      "loss": 0.0001,
      "step": 31170
    },
    {
      "epoch": 0.4408375630929322,
      "grad_norm": 0.0,
      "learning_rate": 5.594876217675919e-06,
      "loss": 0.0,
      "step": 31180
    },
    {
      "epoch": 0.4409789478148991,
      "grad_norm": 1.4047514262927052e-08,
      "learning_rate": 5.593462370456249e-06,
      "loss": 0.0,
      "step": 31190
    },
    {
      "epoch": 0.44112033253686606,
      "grad_norm": 17.166858673095703,
      "learning_rate": 5.59204852323658e-06,
      "loss": 0.0032,
      "step": 31200
    },
    {
      "epoch": 0.441261717258833,
      "grad_norm": 0.0,
      "learning_rate": 5.5906346760169105e-06,
      "loss": 0.0,
      "step": 31210
    },
    {
      "epoch": 0.44140310198079996,
      "grad_norm": 0.0,
      "learning_rate": 5.5892208287972405e-06,
      "loss": 0.0001,
      "step": 31220
    },
    {
      "epoch": 0.4415444867027669,
      "grad_norm": 0.0,
      "learning_rate": 5.587806981577571e-06,
      "loss": 0.0,
      "step": 31230
    },
    {
      "epoch": 0.44168587142473387,
      "grad_norm": 96.50838470458984,
      "learning_rate": 5.586393134357901e-06,
      "loss": 0.0129,
      "step": 31240
    },
    {
      "epoch": 0.44182725614670076,
      "grad_norm": 0.012016947381198406,
      "learning_rate": 5.584979287138232e-06,
      "loss": 0.0,
      "step": 31250
    },
    {
      "epoch": 0.4419686408686677,
      "grad_norm": 0.0,
      "learning_rate": 5.583565439918562e-06,
      "loss": 0.0,
      "step": 31260
    },
    {
      "epoch": 0.44211002559063467,
      "grad_norm": 0.0828559473156929,
      "learning_rate": 5.582151592698893e-06,
      "loss": 0.0,
      "step": 31270
    },
    {
      "epoch": 0.4422514103126016,
      "grad_norm": 0.0,
      "learning_rate": 5.580737745479225e-06,
      "loss": 0.0774,
      "step": 31280
    },
    {
      "epoch": 0.4423927950345686,
      "grad_norm": 0.0,
      "learning_rate": 5.579323898259554e-06,
      "loss": 0.0,
      "step": 31290
    },
    {
      "epoch": 0.44253417975653553,
      "grad_norm": 0.0,
      "learning_rate": 5.577910051039886e-06,
      "loss": 0.0003,
      "step": 31300
    },
    {
      "epoch": 0.4426755644785024,
      "grad_norm": 0.0,
      "learning_rate": 5.576496203820216e-06,
      "loss": 0.0045,
      "step": 31310
    },
    {
      "epoch": 0.4428169492004694,
      "grad_norm": 0.0,
      "learning_rate": 5.5750823566005465e-06,
      "loss": 0.0,
      "step": 31320
    },
    {
      "epoch": 0.44295833392243633,
      "grad_norm": 0.0,
      "learning_rate": 5.5736685093808765e-06,
      "loss": 0.0,
      "step": 31330
    },
    {
      "epoch": 0.4430997186444033,
      "grad_norm": 0.0,
      "learning_rate": 5.572254662161207e-06,
      "loss": 0.0036,
      "step": 31340
    },
    {
      "epoch": 0.44324110336637024,
      "grad_norm": 0.028796782717108727,
      "learning_rate": 5.570840814941538e-06,
      "loss": 0.0,
      "step": 31350
    },
    {
      "epoch": 0.4433824880883372,
      "grad_norm": 3.1274005323211895e-06,
      "learning_rate": 5.569426967721868e-06,
      "loss": 0.0001,
      "step": 31360
    },
    {
      "epoch": 0.44352387281030414,
      "grad_norm": 0.0,
      "learning_rate": 5.568013120502199e-06,
      "loss": 0.0,
      "step": 31370
    },
    {
      "epoch": 0.44366525753227104,
      "grad_norm": 0.0,
      "learning_rate": 5.566599273282529e-06,
      "loss": 0.0,
      "step": 31380
    },
    {
      "epoch": 0.443806642254238,
      "grad_norm": 0.0,
      "learning_rate": 5.56518542606286e-06,
      "loss": 0.0,
      "step": 31390
    },
    {
      "epoch": 0.44394802697620495,
      "grad_norm": 0.0,
      "learning_rate": 5.563771578843191e-06,
      "loss": 0.0003,
      "step": 31400
    },
    {
      "epoch": 0.4440894116981719,
      "grad_norm": 2.023126853600843e-06,
      "learning_rate": 5.562357731623521e-06,
      "loss": 0.0,
      "step": 31410
    },
    {
      "epoch": 0.44423079642013885,
      "grad_norm": 0.0,
      "learning_rate": 5.5609438844038525e-06,
      "loss": 0.1978,
      "step": 31420
    },
    {
      "epoch": 0.4443721811421058,
      "grad_norm": 0.0,
      "learning_rate": 5.559530037184182e-06,
      "loss": 0.0,
      "step": 31430
    },
    {
      "epoch": 0.4445135658640727,
      "grad_norm": 1.5934281449858645e-08,
      "learning_rate": 5.558116189964513e-06,
      "loss": 0.0,
      "step": 31440
    },
    {
      "epoch": 0.44465495058603965,
      "grad_norm": 0.0,
      "learning_rate": 5.556702342744843e-06,
      "loss": 0.0,
      "step": 31450
    },
    {
      "epoch": 0.4447963353080066,
      "grad_norm": 0.0,
      "learning_rate": 5.555288495525174e-06,
      "loss": 0.0,
      "step": 31460
    },
    {
      "epoch": 0.44493772002997356,
      "grad_norm": 178.60618591308594,
      "learning_rate": 5.553874648305505e-06,
      "loss": 0.051,
      "step": 31470
    },
    {
      "epoch": 0.4450791047519405,
      "grad_norm": 0.274539977312088,
      "learning_rate": 5.552460801085835e-06,
      "loss": 0.0,
      "step": 31480
    },
    {
      "epoch": 0.44522048947390747,
      "grad_norm": 0.0,
      "learning_rate": 5.551046953866166e-06,
      "loss": 0.0,
      "step": 31490
    },
    {
      "epoch": 0.4453618741958744,
      "grad_norm": 0.0,
      "learning_rate": 5.549633106646496e-06,
      "loss": 0.0018,
      "step": 31500
    },
    {
      "epoch": 0.4455032589178413,
      "grad_norm": 0.0,
      "learning_rate": 5.548219259426827e-06,
      "loss": 0.0,
      "step": 31510
    },
    {
      "epoch": 0.44564464363980827,
      "grad_norm": 0.0,
      "learning_rate": 5.546805412207157e-06,
      "loss": 0.013,
      "step": 31520
    },
    {
      "epoch": 0.4457860283617752,
      "grad_norm": 0.0,
      "learning_rate": 5.545391564987488e-06,
      "loss": 0.0003,
      "step": 31530
    },
    {
      "epoch": 0.4459274130837422,
      "grad_norm": 2.758803930191789e-06,
      "learning_rate": 5.543977717767819e-06,
      "loss": 0.0,
      "step": 31540
    },
    {
      "epoch": 0.4460687978057091,
      "grad_norm": 0.0,
      "learning_rate": 5.542563870548149e-06,
      "loss": 0.0,
      "step": 31550
    },
    {
      "epoch": 0.4462101825276761,
      "grad_norm": 0.0,
      "learning_rate": 5.5411500233284794e-06,
      "loss": 0.0,
      "step": 31560
    },
    {
      "epoch": 0.446351567249643,
      "grad_norm": 0.0,
      "learning_rate": 5.5397361761088095e-06,
      "loss": 0.0037,
      "step": 31570
    },
    {
      "epoch": 0.44649295197160993,
      "grad_norm": 0.0,
      "learning_rate": 5.538322328889141e-06,
      "loss": 0.0,
      "step": 31580
    },
    {
      "epoch": 0.4466343366935769,
      "grad_norm": 0.0,
      "learning_rate": 5.53690848166947e-06,
      "loss": 0.0,
      "step": 31590
    },
    {
      "epoch": 0.44677572141554384,
      "grad_norm": 0.0,
      "learning_rate": 5.535494634449802e-06,
      "loss": 0.0019,
      "step": 31600
    },
    {
      "epoch": 0.4469171061375108,
      "grad_norm": 1.420745920910349e-08,
      "learning_rate": 5.534080787230133e-06,
      "loss": 0.0,
      "step": 31610
    },
    {
      "epoch": 0.44705849085947774,
      "grad_norm": 0.0,
      "learning_rate": 5.532666940010463e-06,
      "loss": 0.0,
      "step": 31620
    },
    {
      "epoch": 0.4471998755814447,
      "grad_norm": 0.0,
      "learning_rate": 5.531253092790794e-06,
      "loss": 0.0,
      "step": 31630
    },
    {
      "epoch": 0.4473412603034116,
      "grad_norm": 0.02981310337781906,
      "learning_rate": 5.529839245571124e-06,
      "loss": 0.0725,
      "step": 31640
    },
    {
      "epoch": 0.44748264502537854,
      "grad_norm": 0.0,
      "learning_rate": 5.528425398351455e-06,
      "loss": 0.0,
      "step": 31650
    },
    {
      "epoch": 0.4476240297473455,
      "grad_norm": 2.4863791465759277,
      "learning_rate": 5.5270115511317855e-06,
      "loss": 0.001,
      "step": 31660
    },
    {
      "epoch": 0.44776541446931245,
      "grad_norm": 0.16629864275455475,
      "learning_rate": 5.5255977039121155e-06,
      "loss": 1.4969,
      "step": 31670
    },
    {
      "epoch": 0.4479067991912794,
      "grad_norm": 0.0,
      "learning_rate": 5.524183856692446e-06,
      "loss": 0.0335,
      "step": 31680
    },
    {
      "epoch": 0.44804818391324636,
      "grad_norm": 0.0,
      "learning_rate": 5.522770009472776e-06,
      "loss": 0.5473,
      "step": 31690
    },
    {
      "epoch": 0.44818956863521325,
      "grad_norm": 0.0,
      "learning_rate": 5.521356162253107e-06,
      "loss": 0.0,
      "step": 31700
    },
    {
      "epoch": 0.4483309533571802,
      "grad_norm": 0.0,
      "learning_rate": 5.519942315033437e-06,
      "loss": 0.0013,
      "step": 31710
    },
    {
      "epoch": 0.44847233807914716,
      "grad_norm": 1.912631432787748e-06,
      "learning_rate": 5.518528467813769e-06,
      "loss": 0.4424,
      "step": 31720
    },
    {
      "epoch": 0.4486137228011141,
      "grad_norm": 0.0,
      "learning_rate": 5.5171146205941e-06,
      "loss": 0.0001,
      "step": 31730
    },
    {
      "epoch": 0.44875510752308106,
      "grad_norm": 0.0,
      "learning_rate": 5.51570077337443e-06,
      "loss": 0.0,
      "step": 31740
    },
    {
      "epoch": 0.448896492245048,
      "grad_norm": 9.93836498260498,
      "learning_rate": 5.514286926154761e-06,
      "loss": 0.0028,
      "step": 31750
    },
    {
      "epoch": 0.44903787696701497,
      "grad_norm": 0.0,
      "learning_rate": 5.512873078935091e-06,
      "loss": 0.0062,
      "step": 31760
    },
    {
      "epoch": 0.44917926168898187,
      "grad_norm": 0.0,
      "learning_rate": 5.5114592317154215e-06,
      "loss": 0.0001,
      "step": 31770
    },
    {
      "epoch": 0.4493206464109488,
      "grad_norm": 0.0,
      "learning_rate": 5.5100453844957515e-06,
      "loss": 0.0,
      "step": 31780
    },
    {
      "epoch": 0.4494620311329158,
      "grad_norm": 0.0,
      "learning_rate": 5.508631537276082e-06,
      "loss": 0.0,
      "step": 31790
    },
    {
      "epoch": 0.4496034158548827,
      "grad_norm": 0.0,
      "learning_rate": 5.507217690056413e-06,
      "loss": 0.0,
      "step": 31800
    },
    {
      "epoch": 0.4497448005768497,
      "grad_norm": 0.0,
      "learning_rate": 5.505803842836743e-06,
      "loss": 0.0,
      "step": 31810
    },
    {
      "epoch": 0.44988618529881663,
      "grad_norm": 0.0,
      "learning_rate": 5.504389995617074e-06,
      "loss": 0.0,
      "step": 31820
    },
    {
      "epoch": 0.45002757002078353,
      "grad_norm": 0.0,
      "learning_rate": 5.502976148397404e-06,
      "loss": 0.0,
      "step": 31830
    },
    {
      "epoch": 0.4501689547427505,
      "grad_norm": 0.004471055697649717,
      "learning_rate": 5.501562301177735e-06,
      "loss": 0.0,
      "step": 31840
    },
    {
      "epoch": 0.45031033946471744,
      "grad_norm": 0.0,
      "learning_rate": 5.500148453958067e-06,
      "loss": 0.0002,
      "step": 31850
    },
    {
      "epoch": 0.4504517241866844,
      "grad_norm": 0.0,
      "learning_rate": 5.498734606738396e-06,
      "loss": 0.0,
      "step": 31860
    },
    {
      "epoch": 0.45059310890865134,
      "grad_norm": 2.6856472601366477e-08,
      "learning_rate": 5.4973207595187275e-06,
      "loss": 0.0,
      "step": 31870
    },
    {
      "epoch": 0.4507344936306183,
      "grad_norm": 0.0,
      "learning_rate": 5.4959069122990575e-06,
      "loss": 0.6308,
      "step": 31880
    },
    {
      "epoch": 0.45087587835258525,
      "grad_norm": 0.0,
      "learning_rate": 5.494493065079388e-06,
      "loss": 0.0009,
      "step": 31890
    },
    {
      "epoch": 0.45101726307455214,
      "grad_norm": 0.0,
      "learning_rate": 5.493079217859718e-06,
      "loss": 0.0001,
      "step": 31900
    },
    {
      "epoch": 0.4511586477965191,
      "grad_norm": 0.0003179685154464096,
      "learning_rate": 5.491665370640049e-06,
      "loss": 0.0002,
      "step": 31910
    },
    {
      "epoch": 0.45130003251848605,
      "grad_norm": 3.6541941881296225e-06,
      "learning_rate": 5.49025152342038e-06,
      "loss": 0.0,
      "step": 31920
    },
    {
      "epoch": 0.451441417240453,
      "grad_norm": 1.8942859014714486e-07,
      "learning_rate": 5.48883767620071e-06,
      "loss": 0.0158,
      "step": 31930
    },
    {
      "epoch": 0.45158280196241996,
      "grad_norm": 0.0,
      "learning_rate": 5.487423828981041e-06,
      "loss": 0.0,
      "step": 31940
    },
    {
      "epoch": 0.4517241866843869,
      "grad_norm": 0.0,
      "learning_rate": 5.486009981761371e-06,
      "loss": 0.0109,
      "step": 31950
    },
    {
      "epoch": 0.4518655714063538,
      "grad_norm": 0.0,
      "learning_rate": 5.484596134541702e-06,
      "loss": 0.0,
      "step": 31960
    },
    {
      "epoch": 0.45200695612832076,
      "grad_norm": 0.0,
      "learning_rate": 5.483182287322032e-06,
      "loss": 0.0,
      "step": 31970
    },
    {
      "epoch": 0.4521483408502877,
      "grad_norm": 0.0,
      "learning_rate": 5.481768440102363e-06,
      "loss": 0.0,
      "step": 31980
    },
    {
      "epoch": 0.45228972557225466,
      "grad_norm": 0.00633440725505352,
      "learning_rate": 5.480354592882694e-06,
      "loss": 0.0,
      "step": 31990
    },
    {
      "epoch": 0.4524311102942216,
      "grad_norm": 0.0,
      "learning_rate": 5.478940745663024e-06,
      "loss": 0.0,
      "step": 32000
    },
    {
      "epoch": 0.45257249501618857,
      "grad_norm": 0.0,
      "learning_rate": 5.477526898443355e-06,
      "loss": 0.0,
      "step": 32010
    },
    {
      "epoch": 0.4527138797381555,
      "grad_norm": 3.0155635613482445e-05,
      "learning_rate": 5.4761130512236844e-06,
      "loss": 0.0005,
      "step": 32020
    },
    {
      "epoch": 0.4528552644601224,
      "grad_norm": 0.0,
      "learning_rate": 5.474699204004016e-06,
      "loss": 0.0007,
      "step": 32030
    },
    {
      "epoch": 0.4529966491820894,
      "grad_norm": 0.0,
      "learning_rate": 5.473285356784346e-06,
      "loss": 0.0,
      "step": 32040
    },
    {
      "epoch": 0.4531380339040563,
      "grad_norm": 0.0,
      "learning_rate": 5.471871509564677e-06,
      "loss": 0.0,
      "step": 32050
    },
    {
      "epoch": 0.4532794186260233,
      "grad_norm": 0.0,
      "learning_rate": 5.470457662345008e-06,
      "loss": 0.0,
      "step": 32060
    },
    {
      "epoch": 0.45342080334799023,
      "grad_norm": 0.015447036363184452,
      "learning_rate": 5.469043815125338e-06,
      "loss": 0.0015,
      "step": 32070
    },
    {
      "epoch": 0.4535621880699572,
      "grad_norm": 0.0,
      "learning_rate": 5.467629967905669e-06,
      "loss": 0.0,
      "step": 32080
    },
    {
      "epoch": 0.4537035727919241,
      "grad_norm": 0.0,
      "learning_rate": 5.466216120685999e-06,
      "loss": 0.0,
      "step": 32090
    },
    {
      "epoch": 0.45384495751389103,
      "grad_norm": 0.115249402821064,
      "learning_rate": 5.46480227346633e-06,
      "loss": 0.0,
      "step": 32100
    },
    {
      "epoch": 0.453986342235858,
      "grad_norm": 0.0,
      "learning_rate": 5.4633884262466605e-06,
      "loss": 0.0,
      "step": 32110
    },
    {
      "epoch": 0.45412772695782494,
      "grad_norm": 0.0,
      "learning_rate": 5.4619745790269905e-06,
      "loss": 0.0002,
      "step": 32120
    },
    {
      "epoch": 0.4542691116797919,
      "grad_norm": 0.0,
      "learning_rate": 5.460560731807321e-06,
      "loss": 0.6938,
      "step": 32130
    },
    {
      "epoch": 0.45441049640175885,
      "grad_norm": 0.0,
      "learning_rate": 5.459146884587651e-06,
      "loss": 0.0001,
      "step": 32140
    },
    {
      "epoch": 0.45455188112372574,
      "grad_norm": 0.0006525324424728751,
      "learning_rate": 5.457733037367983e-06,
      "loss": 0.0,
      "step": 32150
    },
    {
      "epoch": 0.4546932658456927,
      "grad_norm": 1.1141456525365356e-05,
      "learning_rate": 5.456319190148312e-06,
      "loss": 0.018,
      "step": 32160
    },
    {
      "epoch": 0.45483465056765965,
      "grad_norm": 0.0,
      "learning_rate": 5.454905342928644e-06,
      "loss": 0.0002,
      "step": 32170
    },
    {
      "epoch": 0.4549760352896266,
      "grad_norm": 0.0,
      "learning_rate": 5.453491495708975e-06,
      "loss": 0.0,
      "step": 32180
    },
    {
      "epoch": 0.45511742001159355,
      "grad_norm": 0.0,
      "learning_rate": 5.452077648489305e-06,
      "loss": 0.3444,
      "step": 32190
    },
    {
      "epoch": 0.4552588047335605,
      "grad_norm": 109.9192886352539,
      "learning_rate": 5.450663801269636e-06,
      "loss": 0.0193,
      "step": 32200
    },
    {
      "epoch": 0.45540018945552746,
      "grad_norm": 4.902673254036927e-07,
      "learning_rate": 5.449249954049966e-06,
      "loss": 0.0,
      "step": 32210
    },
    {
      "epoch": 0.45554157417749436,
      "grad_norm": 0.0,
      "learning_rate": 5.4478361068302965e-06,
      "loss": 0.0,
      "step": 32220
    },
    {
      "epoch": 0.4556829588994613,
      "grad_norm": 0.09765512496232986,
      "learning_rate": 5.4464222596106265e-06,
      "loss": 0.0002,
      "step": 32230
    },
    {
      "epoch": 0.45582434362142826,
      "grad_norm": 0.0,
      "learning_rate": 5.445008412390957e-06,
      "loss": 0.0,
      "step": 32240
    },
    {
      "epoch": 0.4559657283433952,
      "grad_norm": 0.004761672578752041,
      "learning_rate": 5.443594565171288e-06,
      "loss": 0.0001,
      "step": 32250
    },
    {
      "epoch": 0.45610711306536217,
      "grad_norm": 0.0,
      "learning_rate": 5.442180717951618e-06,
      "loss": 0.0,
      "step": 32260
    },
    {
      "epoch": 0.4562484977873291,
      "grad_norm": 0.0,
      "learning_rate": 5.440766870731949e-06,
      "loss": 0.0576,
      "step": 32270
    },
    {
      "epoch": 0.456389882509296,
      "grad_norm": 0.0,
      "learning_rate": 5.439353023512279e-06,
      "loss": 0.0,
      "step": 32280
    },
    {
      "epoch": 0.45653126723126297,
      "grad_norm": 8.322464282173314e-07,
      "learning_rate": 5.43793917629261e-06,
      "loss": 0.0,
      "step": 32290
    },
    {
      "epoch": 0.4566726519532299,
      "grad_norm": 0.0,
      "learning_rate": 5.43652532907294e-06,
      "loss": 0.0,
      "step": 32300
    },
    {
      "epoch": 0.4568140366751969,
      "grad_norm": 0.0,
      "learning_rate": 5.435111481853272e-06,
      "loss": 0.0001,
      "step": 32310
    },
    {
      "epoch": 0.45695542139716383,
      "grad_norm": 0.2228931337594986,
      "learning_rate": 5.4336976346336025e-06,
      "loss": 0.0,
      "step": 32320
    },
    {
      "epoch": 0.4570968061191308,
      "grad_norm": 0.0,
      "learning_rate": 5.4322837874139325e-06,
      "loss": 0.0,
      "step": 32330
    },
    {
      "epoch": 0.45723819084109774,
      "grad_norm": 0.0,
      "learning_rate": 5.430869940194263e-06,
      "loss": 0.0,
      "step": 32340
    },
    {
      "epoch": 0.45737957556306463,
      "grad_norm": 0.0,
      "learning_rate": 5.429456092974593e-06,
      "loss": 0.0454,
      "step": 32350
    },
    {
      "epoch": 0.4575209602850316,
      "grad_norm": 0.0,
      "learning_rate": 5.428042245754924e-06,
      "loss": 0.0,
      "step": 32360
    },
    {
      "epoch": 0.45766234500699854,
      "grad_norm": 0.000785470416303724,
      "learning_rate": 5.426628398535255e-06,
      "loss": 0.0,
      "step": 32370
    },
    {
      "epoch": 0.4578037297289655,
      "grad_norm": 0.0,
      "learning_rate": 5.425214551315585e-06,
      "loss": 0.0,
      "step": 32380
    },
    {
      "epoch": 0.45794511445093244,
      "grad_norm": 5.2374637959928805e-08,
      "learning_rate": 5.423800704095916e-06,
      "loss": 0.0,
      "step": 32390
    },
    {
      "epoch": 0.4580864991728994,
      "grad_norm": 0.0,
      "learning_rate": 5.422386856876246e-06,
      "loss": 0.0,
      "step": 32400
    },
    {
      "epoch": 0.4582278838948663,
      "grad_norm": 0.0,
      "learning_rate": 5.420973009656577e-06,
      "loss": 0.0,
      "step": 32410
    },
    {
      "epoch": 0.45836926861683325,
      "grad_norm": 0.5673542022705078,
      "learning_rate": 5.419559162436907e-06,
      "loss": 0.0002,
      "step": 32420
    },
    {
      "epoch": 0.4585106533388002,
      "grad_norm": 0.0,
      "learning_rate": 5.418145315217238e-06,
      "loss": 0.0,
      "step": 32430
    },
    {
      "epoch": 0.45865203806076715,
      "grad_norm": 0.0,
      "learning_rate": 5.416731467997569e-06,
      "loss": 0.0,
      "step": 32440
    },
    {
      "epoch": 0.4587934227827341,
      "grad_norm": 0.0,
      "learning_rate": 5.415317620777899e-06,
      "loss": 0.0,
      "step": 32450
    },
    {
      "epoch": 0.45893480750470106,
      "grad_norm": 0.0,
      "learning_rate": 5.41390377355823e-06,
      "loss": 0.0,
      "step": 32460
    },
    {
      "epoch": 0.459076192226668,
      "grad_norm": 4.722916855826043e-06,
      "learning_rate": 5.41248992633856e-06,
      "loss": 0.0004,
      "step": 32470
    },
    {
      "epoch": 0.4592175769486349,
      "grad_norm": 0.0,
      "learning_rate": 5.411076079118891e-06,
      "loss": 0.0,
      "step": 32480
    },
    {
      "epoch": 0.45935896167060186,
      "grad_norm": 0.0,
      "learning_rate": 5.409662231899221e-06,
      "loss": 0.0322,
      "step": 32490
    },
    {
      "epoch": 0.4595003463925688,
      "grad_norm": 0.0033302297815680504,
      "learning_rate": 5.408248384679552e-06,
      "loss": 0.276,
      "step": 32500
    },
    {
      "epoch": 0.45964173111453577,
      "grad_norm": 0.0,
      "learning_rate": 5.406834537459883e-06,
      "loss": 0.0,
      "step": 32510
    },
    {
      "epoch": 0.4597831158365027,
      "grad_norm": 388.6514892578125,
      "learning_rate": 5.405420690240213e-06,
      "loss": 0.09,
      "step": 32520
    },
    {
      "epoch": 0.4599245005584697,
      "grad_norm": 0.0,
      "learning_rate": 5.404006843020544e-06,
      "loss": 0.0,
      "step": 32530
    },
    {
      "epoch": 0.46006588528043657,
      "grad_norm": 0.0,
      "learning_rate": 5.402592995800874e-06,
      "loss": 0.0006,
      "step": 32540
    },
    {
      "epoch": 0.4602072700024035,
      "grad_norm": 0.0,
      "learning_rate": 5.401179148581205e-06,
      "loss": 0.3409,
      "step": 32550
    },
    {
      "epoch": 0.4603486547243705,
      "grad_norm": 7.390647078864276e-05,
      "learning_rate": 5.399765301361535e-06,
      "loss": 0.0,
      "step": 32560
    },
    {
      "epoch": 0.46049003944633743,
      "grad_norm": 0.0,
      "learning_rate": 5.3983514541418655e-06,
      "loss": 0.0,
      "step": 32570
    },
    {
      "epoch": 0.4606314241683044,
      "grad_norm": 0.0,
      "learning_rate": 5.396937606922197e-06,
      "loss": 0.0046,
      "step": 32580
    },
    {
      "epoch": 0.46077280889027133,
      "grad_norm": 0.0,
      "learning_rate": 5.395523759702526e-06,
      "loss": 0.0,
      "step": 32590
    },
    {
      "epoch": 0.4609141936122383,
      "grad_norm": 0.0,
      "learning_rate": 5.394109912482858e-06,
      "loss": 0.0,
      "step": 32600
    },
    {
      "epoch": 0.4610555783342052,
      "grad_norm": 0.0,
      "learning_rate": 5.392696065263188e-06,
      "loss": 0.0006,
      "step": 32610
    },
    {
      "epoch": 0.46119696305617214,
      "grad_norm": 0.0,
      "learning_rate": 5.391282218043519e-06,
      "loss": 0.0,
      "step": 32620
    },
    {
      "epoch": 0.4613383477781391,
      "grad_norm": 0.0,
      "learning_rate": 5.38986837082385e-06,
      "loss": 0.0,
      "step": 32630
    },
    {
      "epoch": 0.46147973250010604,
      "grad_norm": 0.0,
      "learning_rate": 5.38845452360418e-06,
      "loss": 0.0001,
      "step": 32640
    },
    {
      "epoch": 0.461621117222073,
      "grad_norm": 0.0,
      "learning_rate": 5.387040676384511e-06,
      "loss": 0.0,
      "step": 32650
    },
    {
      "epoch": 0.46176250194403995,
      "grad_norm": 0.0004869940166827291,
      "learning_rate": 5.385626829164841e-06,
      "loss": 0.0,
      "step": 32660
    },
    {
      "epoch": 0.46190388666600685,
      "grad_norm": 0.0001001884083962068,
      "learning_rate": 5.3842129819451715e-06,
      "loss": 0.0,
      "step": 32670
    },
    {
      "epoch": 0.4620452713879738,
      "grad_norm": 0.0,
      "learning_rate": 5.3827991347255015e-06,
      "loss": 0.268,
      "step": 32680
    },
    {
      "epoch": 0.46218665610994075,
      "grad_norm": 0.07199425250291824,
      "learning_rate": 5.381385287505832e-06,
      "loss": 0.0,
      "step": 32690
    },
    {
      "epoch": 0.4623280408319077,
      "grad_norm": 0.0,
      "learning_rate": 5.379971440286163e-06,
      "loss": 0.0,
      "step": 32700
    },
    {
      "epoch": 0.46246942555387466,
      "grad_norm": 0.0,
      "learning_rate": 5.378557593066493e-06,
      "loss": 0.0,
      "step": 32710
    },
    {
      "epoch": 0.4626108102758416,
      "grad_norm": 0.0,
      "learning_rate": 5.377143745846825e-06,
      "loss": 0.0,
      "step": 32720
    },
    {
      "epoch": 0.46275219499780856,
      "grad_norm": 0.0,
      "learning_rate": 5.375729898627154e-06,
      "loss": 0.0,
      "step": 32730
    },
    {
      "epoch": 0.46289357971977546,
      "grad_norm": 0.0,
      "learning_rate": 5.374316051407486e-06,
      "loss": 0.0024,
      "step": 32740
    },
    {
      "epoch": 0.4630349644417424,
      "grad_norm": 0.0,
      "learning_rate": 5.372902204187816e-06,
      "loss": 0.0001,
      "step": 32750
    },
    {
      "epoch": 0.46317634916370937,
      "grad_norm": 0.0,
      "learning_rate": 5.371488356968147e-06,
      "loss": 0.0,
      "step": 32760
    },
    {
      "epoch": 0.4633177338856763,
      "grad_norm": 0.0,
      "learning_rate": 5.3700745097484775e-06,
      "loss": 0.0,
      "step": 32770
    },
    {
      "epoch": 0.4634591186076433,
      "grad_norm": 0.0,
      "learning_rate": 5.3686606625288075e-06,
      "loss": 0.0053,
      "step": 32780
    },
    {
      "epoch": 0.4636005033296102,
      "grad_norm": 0.0024108258076012135,
      "learning_rate": 5.367246815309138e-06,
      "loss": 0.0001,
      "step": 32790
    },
    {
      "epoch": 0.4637418880515771,
      "grad_norm": 0.0,
      "learning_rate": 5.365832968089468e-06,
      "loss": 0.0004,
      "step": 32800
    },
    {
      "epoch": 0.4638832727735441,
      "grad_norm": 6.4125306380447e-06,
      "learning_rate": 5.364419120869799e-06,
      "loss": 0.0001,
      "step": 32810
    },
    {
      "epoch": 0.46402465749551103,
      "grad_norm": 0.0,
      "learning_rate": 5.36300527365013e-06,
      "loss": 0.0002,
      "step": 32820
    },
    {
      "epoch": 0.464166042217478,
      "grad_norm": 0.0,
      "learning_rate": 5.36159142643046e-06,
      "loss": 0.0,
      "step": 32830
    },
    {
      "epoch": 0.46430742693944493,
      "grad_norm": 0.0,
      "learning_rate": 5.360177579210791e-06,
      "loss": 0.0,
      "step": 32840
    },
    {
      "epoch": 0.4644488116614119,
      "grad_norm": 0.0,
      "learning_rate": 5.358763731991121e-06,
      "loss": 0.0,
      "step": 32850
    },
    {
      "epoch": 0.46459019638337884,
      "grad_norm": 0.0,
      "learning_rate": 5.357349884771452e-06,
      "loss": 1.2487,
      "step": 32860
    },
    {
      "epoch": 0.46473158110534574,
      "grad_norm": 0.0,
      "learning_rate": 5.355936037551782e-06,
      "loss": 0.0,
      "step": 32870
    },
    {
      "epoch": 0.4648729658273127,
      "grad_norm": 0.0,
      "learning_rate": 5.3545221903321135e-06,
      "loss": 0.0,
      "step": 32880
    },
    {
      "epoch": 0.46501435054927964,
      "grad_norm": 0.0,
      "learning_rate": 5.353108343112444e-06,
      "loss": 0.0,
      "step": 32890
    },
    {
      "epoch": 0.4651557352712466,
      "grad_norm": 9.642462828196585e-05,
      "learning_rate": 5.351694495892774e-06,
      "loss": 0.0017,
      "step": 32900
    },
    {
      "epoch": 0.46529711999321355,
      "grad_norm": 0.0,
      "learning_rate": 5.350280648673105e-06,
      "loss": 0.0,
      "step": 32910
    },
    {
      "epoch": 0.4654385047151805,
      "grad_norm": 0.0,
      "learning_rate": 5.348866801453435e-06,
      "loss": 0.0,
      "step": 32920
    },
    {
      "epoch": 0.4655798894371474,
      "grad_norm": 11.199226379394531,
      "learning_rate": 5.347452954233766e-06,
      "loss": 0.025,
      "step": 32930
    },
    {
      "epoch": 0.46572127415911435,
      "grad_norm": 0.0,
      "learning_rate": 5.346039107014096e-06,
      "loss": 0.0066,
      "step": 32940
    },
    {
      "epoch": 0.4658626588810813,
      "grad_norm": 0.0,
      "learning_rate": 5.344625259794427e-06,
      "loss": 0.0,
      "step": 32950
    },
    {
      "epoch": 0.46600404360304826,
      "grad_norm": 0.0,
      "learning_rate": 5.343211412574758e-06,
      "loss": 0.0,
      "step": 32960
    },
    {
      "epoch": 0.4661454283250152,
      "grad_norm": 0.0,
      "learning_rate": 5.341797565355088e-06,
      "loss": 0.0,
      "step": 32970
    },
    {
      "epoch": 0.46628681304698216,
      "grad_norm": 0.03143073990941048,
      "learning_rate": 5.340383718135419e-06,
      "loss": 0.0,
      "step": 32980
    },
    {
      "epoch": 0.46642819776894906,
      "grad_norm": 0.04514511302113533,
      "learning_rate": 5.338969870915749e-06,
      "loss": 0.0011,
      "step": 32990
    },
    {
      "epoch": 0.466569582490916,
      "grad_norm": 0.0,
      "learning_rate": 5.33755602369608e-06,
      "loss": 0.0,
      "step": 33000
    },
    {
      "epoch": 0.46671096721288297,
      "grad_norm": 754.4208984375,
      "learning_rate": 5.33614217647641e-06,
      "loss": 1.144,
      "step": 33010
    },
    {
      "epoch": 0.4668523519348499,
      "grad_norm": 0.014079423621296883,
      "learning_rate": 5.334728329256741e-06,
      "loss": 0.3605,
      "step": 33020
    },
    {
      "epoch": 0.46699373665681687,
      "grad_norm": 0.0,
      "learning_rate": 5.333314482037072e-06,
      "loss": 0.1054,
      "step": 33030
    },
    {
      "epoch": 0.4671351213787838,
      "grad_norm": 0.0,
      "learning_rate": 5.331900634817402e-06,
      "loss": 0.0,
      "step": 33040
    },
    {
      "epoch": 0.4672765061007508,
      "grad_norm": 0.0,
      "learning_rate": 5.330486787597733e-06,
      "loss": 0.0,
      "step": 33050
    },
    {
      "epoch": 0.4674178908227177,
      "grad_norm": 0.0,
      "learning_rate": 5.329072940378063e-06,
      "loss": 0.0,
      "step": 33060
    },
    {
      "epoch": 0.4675592755446846,
      "grad_norm": 1.123670614333605e-07,
      "learning_rate": 5.327659093158394e-06,
      "loss": 0.0,
      "step": 33070
    },
    {
      "epoch": 0.4677006602666516,
      "grad_norm": 0.8260069489479065,
      "learning_rate": 5.326245245938725e-06,
      "loss": 0.0003,
      "step": 33080
    },
    {
      "epoch": 0.46784204498861853,
      "grad_norm": 0.0,
      "learning_rate": 5.324831398719055e-06,
      "loss": 0.0,
      "step": 33090
    },
    {
      "epoch": 0.4679834297105855,
      "grad_norm": 1.3413699865341187,
      "learning_rate": 5.323417551499386e-06,
      "loss": 0.0002,
      "step": 33100
    },
    {
      "epoch": 0.46812481443255244,
      "grad_norm": 0.0,
      "learning_rate": 5.322003704279716e-06,
      "loss": 0.0,
      "step": 33110
    },
    {
      "epoch": 0.46826619915451934,
      "grad_norm": 0.0,
      "learning_rate": 5.3205898570600465e-06,
      "loss": 0.0,
      "step": 33120
    },
    {
      "epoch": 0.4684075838764863,
      "grad_norm": 174.75094604492188,
      "learning_rate": 5.3191760098403765e-06,
      "loss": 0.0418,
      "step": 33130
    },
    {
      "epoch": 0.46854896859845324,
      "grad_norm": 0.0,
      "learning_rate": 5.317762162620707e-06,
      "loss": 0.0,
      "step": 33140
    },
    {
      "epoch": 0.4686903533204202,
      "grad_norm": 0.0,
      "learning_rate": 5.316348315401039e-06,
      "loss": 0.0,
      "step": 33150
    },
    {
      "epoch": 0.46883173804238715,
      "grad_norm": 0.0,
      "learning_rate": 5.314934468181368e-06,
      "loss": 0.0,
      "step": 33160
    },
    {
      "epoch": 0.4689731227643541,
      "grad_norm": 0.0,
      "learning_rate": 5.3135206209617e-06,
      "loss": 0.0,
      "step": 33170
    },
    {
      "epoch": 0.46911450748632105,
      "grad_norm": 0.0,
      "learning_rate": 5.31210677374203e-06,
      "loss": 0.0006,
      "step": 33180
    },
    {
      "epoch": 0.46925589220828795,
      "grad_norm": 0.0,
      "learning_rate": 5.310692926522361e-06,
      "loss": 0.0,
      "step": 33190
    },
    {
      "epoch": 0.4693972769302549,
      "grad_norm": 0.00023142783902585506,
      "learning_rate": 5.309279079302691e-06,
      "loss": 0.0,
      "step": 33200
    },
    {
      "epoch": 0.46953866165222186,
      "grad_norm": 0.0,
      "learning_rate": 5.307865232083022e-06,
      "loss": 0.0,
      "step": 33210
    },
    {
      "epoch": 0.4696800463741888,
      "grad_norm": 0.0,
      "learning_rate": 5.3064513848633525e-06,
      "loss": 0.0,
      "step": 33220
    },
    {
      "epoch": 0.46982143109615576,
      "grad_norm": 0.0,
      "learning_rate": 5.3050375376436825e-06,
      "loss": 0.0018,
      "step": 33230
    },
    {
      "epoch": 0.4699628158181227,
      "grad_norm": 0.0,
      "learning_rate": 5.303623690424013e-06,
      "loss": 0.0344,
      "step": 33240
    },
    {
      "epoch": 0.4701042005400896,
      "grad_norm": 0.0,
      "learning_rate": 5.302209843204343e-06,
      "loss": 0.0,
      "step": 33250
    },
    {
      "epoch": 0.47024558526205656,
      "grad_norm": 2.082613264065003e-07,
      "learning_rate": 5.300795995984674e-06,
      "loss": 0.0,
      "step": 33260
    },
    {
      "epoch": 0.4703869699840235,
      "grad_norm": 0.0,
      "learning_rate": 5.299382148765004e-06,
      "loss": 0.0,
      "step": 33270
    },
    {
      "epoch": 0.47052835470599047,
      "grad_norm": 0.0,
      "learning_rate": 5.297968301545335e-06,
      "loss": 0.0,
      "step": 33280
    },
    {
      "epoch": 0.4706697394279574,
      "grad_norm": 0.0,
      "learning_rate": 5.296554454325667e-06,
      "loss": 0.0,
      "step": 33290
    },
    {
      "epoch": 0.4708111241499244,
      "grad_norm": 0.0,
      "learning_rate": 5.295140607105996e-06,
      "loss": 0.0,
      "step": 33300
    },
    {
      "epoch": 0.47095250887189133,
      "grad_norm": 0.0,
      "learning_rate": 5.293726759886328e-06,
      "loss": 0.0,
      "step": 33310
    },
    {
      "epoch": 0.4710938935938582,
      "grad_norm": 0.0,
      "learning_rate": 5.292312912666658e-06,
      "loss": 0.0,
      "step": 33320
    },
    {
      "epoch": 0.4712352783158252,
      "grad_norm": 0.0,
      "learning_rate": 5.2908990654469885e-06,
      "loss": 0.0,
      "step": 33330
    },
    {
      "epoch": 0.47137666303779213,
      "grad_norm": 0.0,
      "learning_rate": 5.289485218227319e-06,
      "loss": 0.0,
      "step": 33340
    },
    {
      "epoch": 0.4715180477597591,
      "grad_norm": 0.0,
      "learning_rate": 5.288071371007649e-06,
      "loss": 0.0,
      "step": 33350
    },
    {
      "epoch": 0.47165943248172604,
      "grad_norm": 0.0036776193883270025,
      "learning_rate": 5.28665752378798e-06,
      "loss": 0.0,
      "step": 33360
    },
    {
      "epoch": 0.471800817203693,
      "grad_norm": 0.006765273865312338,
      "learning_rate": 5.28524367656831e-06,
      "loss": 0.0017,
      "step": 33370
    },
    {
      "epoch": 0.4719422019256599,
      "grad_norm": 0.0,
      "learning_rate": 5.283829829348641e-06,
      "loss": 0.4402,
      "step": 33380
    },
    {
      "epoch": 0.47208358664762684,
      "grad_norm": 0.0,
      "learning_rate": 5.282415982128971e-06,
      "loss": 0.0,
      "step": 33390
    },
    {
      "epoch": 0.4722249713695938,
      "grad_norm": 0.0,
      "learning_rate": 5.281002134909302e-06,
      "loss": 0.0,
      "step": 33400
    },
    {
      "epoch": 0.47236635609156075,
      "grad_norm": 9.77530944510363e-05,
      "learning_rate": 5.279588287689633e-06,
      "loss": 0.0,
      "step": 33410
    },
    {
      "epoch": 0.4725077408135277,
      "grad_norm": 0.0,
      "learning_rate": 5.278174440469963e-06,
      "loss": 0.0,
      "step": 33420
    },
    {
      "epoch": 0.47264912553549465,
      "grad_norm": 0.0,
      "learning_rate": 5.276760593250294e-06,
      "loss": 0.0,
      "step": 33430
    },
    {
      "epoch": 0.4727905102574616,
      "grad_norm": 0.0,
      "learning_rate": 5.275346746030624e-06,
      "loss": 0.0002,
      "step": 33440
    },
    {
      "epoch": 0.4729318949794285,
      "grad_norm": 0.02858731336891651,
      "learning_rate": 5.273932898810955e-06,
      "loss": 0.0034,
      "step": 33450
    },
    {
      "epoch": 0.47307327970139545,
      "grad_norm": 0.0016722199507057667,
      "learning_rate": 5.2725190515912846e-06,
      "loss": 0.0,
      "step": 33460
    },
    {
      "epoch": 0.4732146644233624,
      "grad_norm": 0.0,
      "learning_rate": 5.271105204371616e-06,
      "loss": 0.0,
      "step": 33470
    },
    {
      "epoch": 0.47335604914532936,
      "grad_norm": 0.0,
      "learning_rate": 5.269691357151947e-06,
      "loss": 0.0,
      "step": 33480
    },
    {
      "epoch": 0.4734974338672963,
      "grad_norm": 0.0,
      "learning_rate": 5.268277509932277e-06,
      "loss": 0.0,
      "step": 33490
    },
    {
      "epoch": 0.47363881858926327,
      "grad_norm": 0.0006835332023911178,
      "learning_rate": 5.266863662712608e-06,
      "loss": 0.0035,
      "step": 33500
    },
    {
      "epoch": 0.47378020331123016,
      "grad_norm": 0.0,
      "learning_rate": 5.265449815492938e-06,
      "loss": 0.0,
      "step": 33510
    },
    {
      "epoch": 0.4739215880331971,
      "grad_norm": 6.989022254943848,
      "learning_rate": 5.264035968273269e-06,
      "loss": 0.0944,
      "step": 33520
    },
    {
      "epoch": 0.47406297275516407,
      "grad_norm": 0.0,
      "learning_rate": 5.2626221210536e-06,
      "loss": 0.0,
      "step": 33530
    },
    {
      "epoch": 0.474204357477131,
      "grad_norm": 0.0,
      "learning_rate": 5.26120827383393e-06,
      "loss": 0.0,
      "step": 33540
    },
    {
      "epoch": 0.474345742199098,
      "grad_norm": 0.0,
      "learning_rate": 5.259794426614261e-06,
      "loss": 0.0,
      "step": 33550
    },
    {
      "epoch": 0.47448712692106493,
      "grad_norm": 5.3356106946012005e-06,
      "learning_rate": 5.258380579394591e-06,
      "loss": 0.0,
      "step": 33560
    },
    {
      "epoch": 0.4746285116430319,
      "grad_norm": 0.0,
      "learning_rate": 5.2569667321749215e-06,
      "loss": 0.0001,
      "step": 33570
    },
    {
      "epoch": 0.4747698963649988,
      "grad_norm": 0.0,
      "learning_rate": 5.2555528849552515e-06,
      "loss": 0.0,
      "step": 33580
    },
    {
      "epoch": 0.47491128108696573,
      "grad_norm": 1.302308128003915e-08,
      "learning_rate": 5.254139037735583e-06,
      "loss": 0.0005,
      "step": 33590
    },
    {
      "epoch": 0.4750526658089327,
      "grad_norm": 0.0,
      "learning_rate": 5.252725190515914e-06,
      "loss": 0.0019,
      "step": 33600
    },
    {
      "epoch": 0.47519405053089964,
      "grad_norm": 0.0,
      "learning_rate": 5.251311343296244e-06,
      "loss": 0.0,
      "step": 33610
    },
    {
      "epoch": 0.4753354352528666,
      "grad_norm": 0.0,
      "learning_rate": 5.249897496076575e-06,
      "loss": 0.0086,
      "step": 33620
    },
    {
      "epoch": 0.47547681997483354,
      "grad_norm": 0.0,
      "learning_rate": 5.248483648856905e-06,
      "loss": 0.0027,
      "step": 33630
    },
    {
      "epoch": 0.47561820469680044,
      "grad_norm": 0.036695364862680435,
      "learning_rate": 5.247069801637236e-06,
      "loss": 0.0004,
      "step": 33640
    },
    {
      "epoch": 0.4757595894187674,
      "grad_norm": 0.0,
      "learning_rate": 5.245655954417566e-06,
      "loss": 0.0002,
      "step": 33650
    },
    {
      "epoch": 0.47590097414073435,
      "grad_norm": 0.0,
      "learning_rate": 5.244242107197897e-06,
      "loss": 0.0,
      "step": 33660
    },
    {
      "epoch": 0.4760423588627013,
      "grad_norm": 0.0,
      "learning_rate": 5.2428282599782275e-06,
      "loss": 0.0,
      "step": 33670
    },
    {
      "epoch": 0.47618374358466825,
      "grad_norm": 0.0,
      "learning_rate": 5.2414144127585575e-06,
      "loss": 0.0,
      "step": 33680
    },
    {
      "epoch": 0.4763251283066352,
      "grad_norm": 0.005873082205653191,
      "learning_rate": 5.240000565538888e-06,
      "loss": 0.0,
      "step": 33690
    },
    {
      "epoch": 0.47646651302860216,
      "grad_norm": 4.400554189487593e-06,
      "learning_rate": 5.238586718319218e-06,
      "loss": 0.0021,
      "step": 33700
    },
    {
      "epoch": 0.47660789775056905,
      "grad_norm": 0.0,
      "learning_rate": 5.237172871099549e-06,
      "loss": 0.0,
      "step": 33710
    },
    {
      "epoch": 0.476749282472536,
      "grad_norm": 1.2428874969482422,
      "learning_rate": 5.235759023879879e-06,
      "loss": 0.0174,
      "step": 33720
    },
    {
      "epoch": 0.47689066719450296,
      "grad_norm": 0.0,
      "learning_rate": 5.23434517666021e-06,
      "loss": 0.0005,
      "step": 33730
    },
    {
      "epoch": 0.4770320519164699,
      "grad_norm": 0.0,
      "learning_rate": 5.232931329440542e-06,
      "loss": 0.0,
      "step": 33740
    },
    {
      "epoch": 0.47717343663843687,
      "grad_norm": 0.0,
      "learning_rate": 5.231517482220872e-06,
      "loss": 0.0274,
      "step": 33750
    },
    {
      "epoch": 0.4773148213604038,
      "grad_norm": 0.0,
      "learning_rate": 5.230103635001203e-06,
      "loss": 0.3465,
      "step": 33760
    },
    {
      "epoch": 0.4774562060823707,
      "grad_norm": 0.0,
      "learning_rate": 5.228689787781533e-06,
      "loss": 0.0,
      "step": 33770
    },
    {
      "epoch": 0.47759759080433767,
      "grad_norm": 17.978185653686523,
      "learning_rate": 5.2272759405618635e-06,
      "loss": 0.0023,
      "step": 33780
    },
    {
      "epoch": 0.4777389755263046,
      "grad_norm": 0.0,
      "learning_rate": 5.225862093342194e-06,
      "loss": 0.0,
      "step": 33790
    },
    {
      "epoch": 0.4778803602482716,
      "grad_norm": 0.0,
      "learning_rate": 5.224448246122524e-06,
      "loss": 0.0,
      "step": 33800
    },
    {
      "epoch": 0.4780217449702385,
      "grad_norm": 6.262675015022978e-06,
      "learning_rate": 5.223034398902855e-06,
      "loss": 0.0,
      "step": 33810
    },
    {
      "epoch": 0.4781631296922055,
      "grad_norm": 0.0,
      "learning_rate": 5.221620551683185e-06,
      "loss": 0.0,
      "step": 33820
    },
    {
      "epoch": 0.4783045144141724,
      "grad_norm": 0.0,
      "learning_rate": 5.220206704463516e-06,
      "loss": 0.0001,
      "step": 33830
    },
    {
      "epoch": 0.47844589913613933,
      "grad_norm": 0.0,
      "learning_rate": 5.218792857243846e-06,
      "loss": 0.0008,
      "step": 33840
    },
    {
      "epoch": 0.4785872838581063,
      "grad_norm": 1.0829553604125977,
      "learning_rate": 5.217379010024177e-06,
      "loss": 0.0018,
      "step": 33850
    },
    {
      "epoch": 0.47872866858007324,
      "grad_norm": 0.0,
      "learning_rate": 5.215965162804509e-06,
      "loss": 0.0,
      "step": 33860
    },
    {
      "epoch": 0.4788700533020402,
      "grad_norm": 0.006032263860106468,
      "learning_rate": 5.214551315584838e-06,
      "loss": 0.0,
      "step": 33870
    },
    {
      "epoch": 0.47901143802400714,
      "grad_norm": 0.0,
      "learning_rate": 5.2131374683651695e-06,
      "loss": 0.4789,
      "step": 33880
    },
    {
      "epoch": 0.4791528227459741,
      "grad_norm": 0.0,
      "learning_rate": 5.2117236211454995e-06,
      "loss": 0.0,
      "step": 33890
    },
    {
      "epoch": 0.479294207467941,
      "grad_norm": 0.011198880150914192,
      "learning_rate": 5.21030977392583e-06,
      "loss": 0.0,
      "step": 33900
    },
    {
      "epoch": 0.47943559218990794,
      "grad_norm": 516.0281372070312,
      "learning_rate": 5.20889592670616e-06,
      "loss": 0.2669,
      "step": 33910
    },
    {
      "epoch": 0.4795769769118749,
      "grad_norm": 0.0,
      "learning_rate": 5.207482079486491e-06,
      "loss": 0.0001,
      "step": 33920
    },
    {
      "epoch": 0.47971836163384185,
      "grad_norm": 0.0,
      "learning_rate": 5.206068232266822e-06,
      "loss": 0.0003,
      "step": 33930
    },
    {
      "epoch": 0.4798597463558088,
      "grad_norm": 0.0,
      "learning_rate": 5.204654385047152e-06,
      "loss": 0.0125,
      "step": 33940
    },
    {
      "epoch": 0.48000113107777576,
      "grad_norm": 0.0,
      "learning_rate": 5.203240537827483e-06,
      "loss": 0.0,
      "step": 33950
    },
    {
      "epoch": 0.48014251579974265,
      "grad_norm": 0.0,
      "learning_rate": 5.201826690607813e-06,
      "loss": 0.001,
      "step": 33960
    },
    {
      "epoch": 0.4802839005217096,
      "grad_norm": 0.0,
      "learning_rate": 5.200412843388144e-06,
      "loss": 0.1663,
      "step": 33970
    },
    {
      "epoch": 0.48042528524367656,
      "grad_norm": 0.0,
      "learning_rate": 5.198998996168474e-06,
      "loss": 0.0089,
      "step": 33980
    },
    {
      "epoch": 0.4805666699656435,
      "grad_norm": 7.60455876047672e-09,
      "learning_rate": 5.197585148948805e-06,
      "loss": 0.0,
      "step": 33990
    },
    {
      "epoch": 0.48070805468761046,
      "grad_norm": 0.0,
      "learning_rate": 5.196171301729136e-06,
      "loss": 0.0041,
      "step": 34000
    },
    {
      "epoch": 0.4808494394095774,
      "grad_norm": 0.0,
      "learning_rate": 5.194757454509466e-06,
      "loss": 0.0,
      "step": 34010
    },
    {
      "epoch": 0.48099082413154437,
      "grad_norm": 0.0,
      "learning_rate": 5.193343607289797e-06,
      "loss": 0.0,
      "step": 34020
    },
    {
      "epoch": 0.48113220885351127,
      "grad_norm": 0.0,
      "learning_rate": 5.1919297600701265e-06,
      "loss": 0.0121,
      "step": 34030
    },
    {
      "epoch": 0.4812735935754782,
      "grad_norm": 0.0,
      "learning_rate": 5.190515912850458e-06,
      "loss": 0.0,
      "step": 34040
    },
    {
      "epoch": 0.4814149782974452,
      "grad_norm": 0.0,
      "learning_rate": 5.189102065630789e-06,
      "loss": 0.0075,
      "step": 34050
    },
    {
      "epoch": 0.4815563630194121,
      "grad_norm": 0.0,
      "learning_rate": 5.187688218411119e-06,
      "loss": 0.3881,
      "step": 34060
    },
    {
      "epoch": 0.4816977477413791,
      "grad_norm": 0.008803818374872208,
      "learning_rate": 5.18627437119145e-06,
      "loss": 0.0037,
      "step": 34070
    },
    {
      "epoch": 0.48183913246334603,
      "grad_norm": 0.0,
      "learning_rate": 5.18486052397178e-06,
      "loss": 0.0,
      "step": 34080
    },
    {
      "epoch": 0.48198051718531293,
      "grad_norm": 0.0,
      "learning_rate": 5.183446676752111e-06,
      "loss": 0.0,
      "step": 34090
    },
    {
      "epoch": 0.4821219019072799,
      "grad_norm": 0.0029624970629811287,
      "learning_rate": 5.182032829532441e-06,
      "loss": 0.0,
      "step": 34100
    },
    {
      "epoch": 0.48226328662924683,
      "grad_norm": 0.0,
      "learning_rate": 5.180618982312772e-06,
      "loss": 0.0,
      "step": 34110
    },
    {
      "epoch": 0.4824046713512138,
      "grad_norm": 0.0,
      "learning_rate": 5.1792051350931025e-06,
      "loss": 0.0,
      "step": 34120
    },
    {
      "epoch": 0.48254605607318074,
      "grad_norm": 0.0,
      "learning_rate": 5.1777912878734325e-06,
      "loss": 0.0123,
      "step": 34130
    },
    {
      "epoch": 0.4826874407951477,
      "grad_norm": 0.0,
      "learning_rate": 5.176377440653763e-06,
      "loss": 0.0,
      "step": 34140
    },
    {
      "epoch": 0.48282882551711465,
      "grad_norm": 0.0,
      "learning_rate": 5.174963593434093e-06,
      "loss": 0.0,
      "step": 34150
    },
    {
      "epoch": 0.48297021023908154,
      "grad_norm": 0.005787976086139679,
      "learning_rate": 5.173549746214425e-06,
      "loss": 0.0,
      "step": 34160
    },
    {
      "epoch": 0.4831115949610485,
      "grad_norm": 3.9385787431456265e-07,
      "learning_rate": 5.172135898994754e-06,
      "loss": 0.0,
      "step": 34170
    },
    {
      "epoch": 0.48325297968301545,
      "grad_norm": 0.0,
      "learning_rate": 5.170722051775086e-06,
      "loss": 0.0,
      "step": 34180
    },
    {
      "epoch": 0.4833943644049824,
      "grad_norm": 0.0,
      "learning_rate": 5.169308204555417e-06,
      "loss": 0.0062,
      "step": 34190
    },
    {
      "epoch": 0.48353574912694935,
      "grad_norm": 0.0,
      "learning_rate": 5.167894357335747e-06,
      "loss": 0.0,
      "step": 34200
    },
    {
      "epoch": 0.4836771338489163,
      "grad_norm": 0.0,
      "learning_rate": 5.166480510116078e-06,
      "loss": 0.0,
      "step": 34210
    },
    {
      "epoch": 0.4838185185708832,
      "grad_norm": 5.479714104694722e-08,
      "learning_rate": 5.165066662896408e-06,
      "loss": 0.0002,
      "step": 34220
    },
    {
      "epoch": 0.48395990329285016,
      "grad_norm": 0.0,
      "learning_rate": 5.1636528156767385e-06,
      "loss": 0.0009,
      "step": 34230
    },
    {
      "epoch": 0.4841012880148171,
      "grad_norm": 0.000225571304326877,
      "learning_rate": 5.1622389684570685e-06,
      "loss": 0.0028,
      "step": 34240
    },
    {
      "epoch": 0.48424267273678406,
      "grad_norm": 0.0,
      "learning_rate": 5.160825121237399e-06,
      "loss": 0.0,
      "step": 34250
    },
    {
      "epoch": 0.484384057458751,
      "grad_norm": 0.0,
      "learning_rate": 5.15941127401773e-06,
      "loss": 0.0001,
      "step": 34260
    },
    {
      "epoch": 0.48452544218071797,
      "grad_norm": 0.0,
      "learning_rate": 5.15799742679806e-06,
      "loss": 0.0001,
      "step": 34270
    },
    {
      "epoch": 0.4846668269026849,
      "grad_norm": 0.0,
      "learning_rate": 5.156583579578391e-06,
      "loss": 0.0,
      "step": 34280
    },
    {
      "epoch": 0.4848082116246518,
      "grad_norm": 0.0,
      "learning_rate": 5.155169732358721e-06,
      "loss": 0.0,
      "step": 34290
    },
    {
      "epoch": 0.48494959634661877,
      "grad_norm": 0.0,
      "learning_rate": 5.153755885139052e-06,
      "loss": 0.0,
      "step": 34300
    },
    {
      "epoch": 0.4850909810685857,
      "grad_norm": 0.0004041713254991919,
      "learning_rate": 5.152342037919384e-06,
      "loss": 0.0,
      "step": 34310
    },
    {
      "epoch": 0.4852323657905527,
      "grad_norm": 0.0,
      "learning_rate": 5.150928190699714e-06,
      "loss": 0.0001,
      "step": 34320
    },
    {
      "epoch": 0.48537375051251963,
      "grad_norm": 1.6088556265003717e-07,
      "learning_rate": 5.1495143434800445e-06,
      "loss": 0.0,
      "step": 34330
    },
    {
      "epoch": 0.4855151352344866,
      "grad_norm": 235.15248107910156,
      "learning_rate": 5.1481004962603745e-06,
      "loss": 0.0702,
      "step": 34340
    },
    {
      "epoch": 0.4856565199564535,
      "grad_norm": 0.0,
      "learning_rate": 5.146686649040705e-06,
      "loss": 0.0003,
      "step": 34350
    },
    {
      "epoch": 0.48579790467842043,
      "grad_norm": 0.0,
      "learning_rate": 5.145272801821035e-06,
      "loss": 0.0,
      "step": 34360
    },
    {
      "epoch": 0.4859392894003874,
      "grad_norm": 0.0,
      "learning_rate": 5.143858954601366e-06,
      "loss": 0.008,
      "step": 34370
    },
    {
      "epoch": 0.48608067412235434,
      "grad_norm": 0.0,
      "learning_rate": 5.142445107381697e-06,
      "loss": 0.0,
      "step": 34380
    },
    {
      "epoch": 0.4862220588443213,
      "grad_norm": 0.2788148522377014,
      "learning_rate": 5.141031260162027e-06,
      "loss": 0.0001,
      "step": 34390
    },
    {
      "epoch": 0.48636344356628824,
      "grad_norm": 0.0,
      "learning_rate": 5.139617412942358e-06,
      "loss": 0.0003,
      "step": 34400
    },
    {
      "epoch": 0.4865048282882552,
      "grad_norm": 5.067380698164925e-05,
      "learning_rate": 5.138203565722688e-06,
      "loss": 0.0004,
      "step": 34410
    },
    {
      "epoch": 0.4866462130102221,
      "grad_norm": 0.0,
      "learning_rate": 5.136789718503019e-06,
      "loss": 0.0817,
      "step": 34420
    },
    {
      "epoch": 0.48678759773218905,
      "grad_norm": 0.0,
      "learning_rate": 5.135375871283349e-06,
      "loss": 0.0,
      "step": 34430
    },
    {
      "epoch": 0.486928982454156,
      "grad_norm": 0.0,
      "learning_rate": 5.13396202406368e-06,
      "loss": 0.0,
      "step": 34440
    },
    {
      "epoch": 0.48707036717612295,
      "grad_norm": 3.2733589705458144e-07,
      "learning_rate": 5.132548176844011e-06,
      "loss": 0.0,
      "step": 34450
    },
    {
      "epoch": 0.4872117518980899,
      "grad_norm": 0.0,
      "learning_rate": 5.1311343296243406e-06,
      "loss": 0.0001,
      "step": 34460
    },
    {
      "epoch": 0.48735313662005686,
      "grad_norm": 0.0,
      "learning_rate": 5.129720482404672e-06,
      "loss": 0.0,
      "step": 34470
    },
    {
      "epoch": 0.48749452134202376,
      "grad_norm": 0.0,
      "learning_rate": 5.128306635185002e-06,
      "loss": 0.0001,
      "step": 34480
    },
    {
      "epoch": 0.4876359060639907,
      "grad_norm": 0.0,
      "learning_rate": 5.126892787965333e-06,
      "loss": 0.0,
      "step": 34490
    },
    {
      "epoch": 0.48777729078595766,
      "grad_norm": 0.0,
      "learning_rate": 5.125478940745664e-06,
      "loss": 0.0025,
      "step": 34500
    },
    {
      "epoch": 0.4879186755079246,
      "grad_norm": 0.0,
      "learning_rate": 5.124065093525994e-06,
      "loss": 0.0,
      "step": 34510
    },
    {
      "epoch": 0.48806006022989157,
      "grad_norm": 0.0,
      "learning_rate": 5.122651246306325e-06,
      "loss": 0.0,
      "step": 34520
    },
    {
      "epoch": 0.4882014449518585,
      "grad_norm": 3.7296268939971924,
      "learning_rate": 5.121237399086655e-06,
      "loss": 0.0005,
      "step": 34530
    },
    {
      "epoch": 0.4883428296738255,
      "grad_norm": 0.012826582416892052,
      "learning_rate": 5.119823551866986e-06,
      "loss": 0.0,
      "step": 34540
    },
    {
      "epoch": 0.48848421439579237,
      "grad_norm": 0.0,
      "learning_rate": 5.118409704647316e-06,
      "loss": 0.0,
      "step": 34550
    },
    {
      "epoch": 0.4886255991177593,
      "grad_norm": 0.0,
      "learning_rate": 5.116995857427647e-06,
      "loss": 0.0,
      "step": 34560
    },
    {
      "epoch": 0.4887669838397263,
      "grad_norm": 6.664190976835016e-09,
      "learning_rate": 5.1155820102079775e-06,
      "loss": 0.0,
      "step": 34570
    },
    {
      "epoch": 0.48890836856169323,
      "grad_norm": 0.0,
      "learning_rate": 5.1141681629883075e-06,
      "loss": 0.0,
      "step": 34580
    },
    {
      "epoch": 0.4890497532836602,
      "grad_norm": 0.0,
      "learning_rate": 5.112754315768639e-06,
      "loss": 0.0001,
      "step": 34590
    },
    {
      "epoch": 0.48919113800562714,
      "grad_norm": 0.0,
      "learning_rate": 5.111340468548968e-06,
      "loss": 0.0,
      "step": 34600
    },
    {
      "epoch": 0.48933252272759403,
      "grad_norm": 0.0,
      "learning_rate": 5.1099266213293e-06,
      "loss": 0.0,
      "step": 34610
    },
    {
      "epoch": 0.489473907449561,
      "grad_norm": 0.0,
      "learning_rate": 5.10851277410963e-06,
      "loss": 0.0,
      "step": 34620
    },
    {
      "epoch": 0.48961529217152794,
      "grad_norm": 0.0,
      "learning_rate": 5.107098926889961e-06,
      "loss": 0.0,
      "step": 34630
    },
    {
      "epoch": 0.4897566768934949,
      "grad_norm": 0.0,
      "learning_rate": 5.105685079670292e-06,
      "loss": 0.0,
      "step": 34640
    },
    {
      "epoch": 0.48989806161546184,
      "grad_norm": 9.456500265514478e-06,
      "learning_rate": 5.104271232450622e-06,
      "loss": 0.0015,
      "step": 34650
    },
    {
      "epoch": 0.4900394463374288,
      "grad_norm": 0.0,
      "learning_rate": 5.102857385230953e-06,
      "loss": 0.0,
      "step": 34660
    },
    {
      "epoch": 0.4901808310593957,
      "grad_norm": 0.0,
      "learning_rate": 5.101443538011283e-06,
      "loss": 0.0113,
      "step": 34670
    },
    {
      "epoch": 0.49032221578136265,
      "grad_norm": 0.0,
      "learning_rate": 5.10017107551358e-06,
      "loss": 0.0762,
      "step": 34680
    },
    {
      "epoch": 0.4904636005033296,
      "grad_norm": 0.0,
      "learning_rate": 5.098757228293912e-06,
      "loss": 0.0003,
      "step": 34690
    },
    {
      "epoch": 0.49060498522529655,
      "grad_norm": 0.0,
      "learning_rate": 5.097343381074242e-06,
      "loss": 0.0,
      "step": 34700
    },
    {
      "epoch": 0.4907463699472635,
      "grad_norm": 0.0,
      "learning_rate": 5.0959295338545725e-06,
      "loss": 0.1224,
      "step": 34710
    },
    {
      "epoch": 0.49088775466923046,
      "grad_norm": 0.0,
      "learning_rate": 5.0945156866349025e-06,
      "loss": 0.0,
      "step": 34720
    },
    {
      "epoch": 0.4910291393911974,
      "grad_norm": 0.0,
      "learning_rate": 5.093101839415233e-06,
      "loss": 0.0,
      "step": 34730
    },
    {
      "epoch": 0.4911705241131643,
      "grad_norm": 0.0,
      "learning_rate": 5.091687992195564e-06,
      "loss": 0.0022,
      "step": 34740
    },
    {
      "epoch": 0.49131190883513126,
      "grad_norm": 0.0,
      "learning_rate": 5.090274144975894e-06,
      "loss": 0.0633,
      "step": 34750
    },
    {
      "epoch": 0.4914532935570982,
      "grad_norm": 0.0,
      "learning_rate": 5.088860297756225e-06,
      "loss": 0.0,
      "step": 34760
    },
    {
      "epoch": 0.49159467827906517,
      "grad_norm": 0.0,
      "learning_rate": 5.087446450536555e-06,
      "loss": 0.0,
      "step": 34770
    },
    {
      "epoch": 0.4917360630010321,
      "grad_norm": 0.0,
      "learning_rate": 5.086032603316886e-06,
      "loss": 0.0,
      "step": 34780
    },
    {
      "epoch": 0.4918774477229991,
      "grad_norm": 0.0,
      "learning_rate": 5.084618756097216e-06,
      "loss": 0.0013,
      "step": 34790
    },
    {
      "epoch": 0.49201883244496597,
      "grad_norm": 0.0,
      "learning_rate": 5.083204908877547e-06,
      "loss": 0.0,
      "step": 34800
    },
    {
      "epoch": 0.4921602171669329,
      "grad_norm": 0.0,
      "learning_rate": 5.081791061657878e-06,
      "loss": 0.0094,
      "step": 34810
    },
    {
      "epoch": 0.4923016018888999,
      "grad_norm": 0.0,
      "learning_rate": 5.080377214438208e-06,
      "loss": 0.0,
      "step": 34820
    },
    {
      "epoch": 0.49244298661086683,
      "grad_norm": 4.454298959899461e-06,
      "learning_rate": 5.078963367218539e-06,
      "loss": 0.0,
      "step": 34830
    },
    {
      "epoch": 0.4925843713328338,
      "grad_norm": 0.0,
      "learning_rate": 5.0775495199988685e-06,
      "loss": 0.0,
      "step": 34840
    },
    {
      "epoch": 0.49272575605480073,
      "grad_norm": 0.0,
      "learning_rate": 5.0761356727792e-06,
      "loss": 0.0,
      "step": 34850
    },
    {
      "epoch": 0.4928671407767677,
      "grad_norm": 0.0021643599029630423,
      "learning_rate": 5.074721825559531e-06,
      "loss": 0.0009,
      "step": 34860
    },
    {
      "epoch": 0.4930085254987346,
      "grad_norm": 0.0,
      "learning_rate": 5.073307978339861e-06,
      "loss": 0.0,
      "step": 34870
    },
    {
      "epoch": 0.49314991022070154,
      "grad_norm": 0.09606405347585678,
      "learning_rate": 5.071894131120192e-06,
      "loss": 0.0,
      "step": 34880
    },
    {
      "epoch": 0.4932912949426685,
      "grad_norm": 0.0,
      "learning_rate": 5.070480283900522e-06,
      "loss": 0.0,
      "step": 34890
    },
    {
      "epoch": 0.49343267966463544,
      "grad_norm": 0.0,
      "learning_rate": 5.069066436680853e-06,
      "loss": 0.0001,
      "step": 34900
    },
    {
      "epoch": 0.4935740643866024,
      "grad_norm": 0.0,
      "learning_rate": 5.067652589461183e-06,
      "loss": 0.0,
      "step": 34910
    },
    {
      "epoch": 0.49371544910856935,
      "grad_norm": 8.284153591375798e-05,
      "learning_rate": 5.066238742241514e-06,
      "loss": 0.0,
      "step": 34920
    },
    {
      "epoch": 0.49385683383053625,
      "grad_norm": 0.0,
      "learning_rate": 5.0648248950218445e-06,
      "loss": 0.0,
      "step": 34930
    },
    {
      "epoch": 0.4939982185525032,
      "grad_norm": 0.0,
      "learning_rate": 5.0634110478021745e-06,
      "loss": 0.0,
      "step": 34940
    },
    {
      "epoch": 0.49413960327447015,
      "grad_norm": 0.0,
      "learning_rate": 5.061997200582505e-06,
      "loss": 0.0,
      "step": 34950
    },
    {
      "epoch": 0.4942809879964371,
      "grad_norm": 0.0,
      "learning_rate": 5.060583353362835e-06,
      "loss": 0.0,
      "step": 34960
    },
    {
      "epoch": 0.49442237271840406,
      "grad_norm": 0.0007119185174815357,
      "learning_rate": 5.059169506143167e-06,
      "loss": 0.0001,
      "step": 34970
    },
    {
      "epoch": 0.494563757440371,
      "grad_norm": 0.0,
      "learning_rate": 5.057755658923496e-06,
      "loss": 0.0053,
      "step": 34980
    },
    {
      "epoch": 0.49470514216233796,
      "grad_norm": 0.0,
      "learning_rate": 5.056341811703828e-06,
      "loss": 0.0,
      "step": 34990
    },
    {
      "epoch": 0.49484652688430486,
      "grad_norm": 0.0,
      "learning_rate": 5.054927964484159e-06,
      "loss": 0.0061,
      "step": 35000
    },
    {
      "epoch": 0.4949879116062718,
      "grad_norm": 0.0,
      "learning_rate": 5.053514117264489e-06,
      "loss": 0.0,
      "step": 35010
    },
    {
      "epoch": 0.49512929632823877,
      "grad_norm": 0.0005683198687620461,
      "learning_rate": 5.05210027004482e-06,
      "loss": 0.0,
      "step": 35020
    },
    {
      "epoch": 0.4952706810502057,
      "grad_norm": 0.0,
      "learning_rate": 5.05068642282515e-06,
      "loss": 0.0,
      "step": 35030
    },
    {
      "epoch": 0.49541206577217267,
      "grad_norm": 0.0,
      "learning_rate": 5.0492725756054806e-06,
      "loss": 0.0,
      "step": 35040
    },
    {
      "epoch": 0.4955534504941396,
      "grad_norm": 0.0,
      "learning_rate": 5.047858728385811e-06,
      "loss": 0.0,
      "step": 35050
    },
    {
      "epoch": 0.4956948352161065,
      "grad_norm": 1.7433220148086548,
      "learning_rate": 5.0464448811661414e-06,
      "loss": 0.0008,
      "step": 35060
    },
    {
      "epoch": 0.4958362199380735,
      "grad_norm": 0.0,
      "learning_rate": 5.045031033946472e-06,
      "loss": 0.0,
      "step": 35070
    },
    {
      "epoch": 0.4959776046600404,
      "grad_norm": 0.0,
      "learning_rate": 5.043617186726802e-06,
      "loss": 0.0,
      "step": 35080
    },
    {
      "epoch": 0.4961189893820074,
      "grad_norm": 0.0,
      "learning_rate": 5.042203339507133e-06,
      "loss": 0.0,
      "step": 35090
    },
    {
      "epoch": 0.49626037410397433,
      "grad_norm": 0.0015551913529634476,
      "learning_rate": 5.040789492287463e-06,
      "loss": 1.143,
      "step": 35100
    },
    {
      "epoch": 0.4964017588259413,
      "grad_norm": 4.592556734905884e-08,
      "learning_rate": 5.039375645067794e-06,
      "loss": 0.0091,
      "step": 35110
    },
    {
      "epoch": 0.49654314354790824,
      "grad_norm": 0.0,
      "learning_rate": 5.037961797848126e-06,
      "loss": 0.0,
      "step": 35120
    },
    {
      "epoch": 0.49668452826987514,
      "grad_norm": 0.0,
      "learning_rate": 5.036547950628456e-06,
      "loss": 0.0,
      "step": 35130
    },
    {
      "epoch": 0.4968259129918421,
      "grad_norm": 0.0026837331242859364,
      "learning_rate": 5.035134103408787e-06,
      "loss": 0.0,
      "step": 35140
    },
    {
      "epoch": 0.49696729771380904,
      "grad_norm": 0.0,
      "learning_rate": 5.033720256189117e-06,
      "loss": 0.3205,
      "step": 35150
    },
    {
      "epoch": 0.497108682435776,
      "grad_norm": 9.441027032153215e-06,
      "learning_rate": 5.0323064089694475e-06,
      "loss": 0.0,
      "step": 35160
    },
    {
      "epoch": 0.49725006715774295,
      "grad_norm": 0.0,
      "learning_rate": 5.0308925617497775e-06,
      "loss": 0.0,
      "step": 35170
    },
    {
      "epoch": 0.4973914518797099,
      "grad_norm": 0.0,
      "learning_rate": 5.029478714530108e-06,
      "loss": 0.0025,
      "step": 35180
    },
    {
      "epoch": 0.4975328366016768,
      "grad_norm": 0.0,
      "learning_rate": 5.028064867310439e-06,
      "loss": 0.0,
      "step": 35190
    },
    {
      "epoch": 0.49767422132364375,
      "grad_norm": 0.0,
      "learning_rate": 5.026651020090769e-06,
      "loss": 0.0033,
      "step": 35200
    },
    {
      "epoch": 0.4978156060456107,
      "grad_norm": 0.0,
      "learning_rate": 5.0252371728711e-06,
      "loss": 0.0,
      "step": 35210
    },
    {
      "epoch": 0.49795699076757766,
      "grad_norm": 0.01839914172887802,
      "learning_rate": 5.02382332565143e-06,
      "loss": 0.1191,
      "step": 35220
    },
    {
      "epoch": 0.4980983754895446,
      "grad_norm": 0.002106250962242484,
      "learning_rate": 5.022409478431761e-06,
      "loss": 0.0,
      "step": 35230
    },
    {
      "epoch": 0.49823976021151156,
      "grad_norm": 0.0,
      "learning_rate": 5.020995631212091e-06,
      "loss": 0.0,
      "step": 35240
    },
    {
      "epoch": 0.4983811449334785,
      "grad_norm": 2.3703676106379135e-06,
      "learning_rate": 5.019581783992422e-06,
      "loss": 0.0,
      "step": 35250
    },
    {
      "epoch": 0.4985225296554454,
      "grad_norm": 0.0,
      "learning_rate": 5.0181679367727535e-06,
      "loss": 0.0012,
      "step": 35260
    },
    {
      "epoch": 0.49866391437741237,
      "grad_norm": 152.962890625,
      "learning_rate": 5.0167540895530835e-06,
      "loss": 0.0083,
      "step": 35270
    },
    {
      "epoch": 0.4988052990993793,
      "grad_norm": 0.0,
      "learning_rate": 5.015340242333414e-06,
      "loss": 0.0,
      "step": 35280
    },
    {
      "epoch": 0.49894668382134627,
      "grad_norm": 0.0,
      "learning_rate": 5.013926395113744e-06,
      "loss": 0.0,
      "step": 35290
    },
    {
      "epoch": 0.4990880685433132,
      "grad_norm": 0.0,
      "learning_rate": 5.012512547894075e-06,
      "loss": 0.0,
      "step": 35300
    },
    {
      "epoch": 0.4992294532652802,
      "grad_norm": 0.0,
      "learning_rate": 5.011098700674406e-06,
      "loss": 0.0003,
      "step": 35310
    },
    {
      "epoch": 0.4993708379872471,
      "grad_norm": 0.0,
      "learning_rate": 5.009684853454736e-06,
      "loss": 0.0,
      "step": 35320
    },
    {
      "epoch": 0.499512222709214,
      "grad_norm": 0.0,
      "learning_rate": 5.008271006235067e-06,
      "loss": 0.0,
      "step": 35330
    },
    {
      "epoch": 0.499653607431181,
      "grad_norm": 2.8374706744216383e-06,
      "learning_rate": 5.006857159015397e-06,
      "loss": 0.0,
      "step": 35340
    },
    {
      "epoch": 0.49979499215314793,
      "grad_norm": 0.0,
      "learning_rate": 5.005443311795728e-06,
      "loss": 0.0001,
      "step": 35350
    },
    {
      "epoch": 0.4999363768751149,
      "grad_norm": 0.0,
      "learning_rate": 5.004029464576058e-06,
      "loss": 0.0,
      "step": 35360
    },
    {
      "epoch": 0.5000777615970818,
      "grad_norm": 0.0,
      "learning_rate": 5.002615617356389e-06,
      "loss": 0.0026,
      "step": 35370
    },
    {
      "epoch": 0.5002191463190487,
      "grad_norm": 0.0,
      "learning_rate": 5.0012017701367195e-06,
      "loss": 0.0106,
      "step": 35380
    },
    {
      "epoch": 0.5003605310410157,
      "grad_norm": 0.0,
      "learning_rate": 4.9997879229170495e-06,
      "loss": 0.0017,
      "step": 35390
    },
    {
      "epoch": 0.5005019157629826,
      "grad_norm": 5.228051804806455e-07,
      "learning_rate": 4.99837407569738e-06,
      "loss": 0.0001,
      "step": 35400
    },
    {
      "epoch": 0.5006433004849496,
      "grad_norm": 0.0,
      "learning_rate": 4.996960228477711e-06,
      "loss": 0.0,
      "step": 35410
    },
    {
      "epoch": 0.5007846852069165,
      "grad_norm": 0.0,
      "learning_rate": 4.995546381258042e-06,
      "loss": 0.0,
      "step": 35420
    },
    {
      "epoch": 0.5009260699288834,
      "grad_norm": 0.0,
      "learning_rate": 4.994132534038372e-06,
      "loss": 0.0,
      "step": 35430
    },
    {
      "epoch": 0.5010674546508505,
      "grad_norm": 0.0,
      "learning_rate": 4.992718686818703e-06,
      "loss": 0.0,
      "step": 35440
    },
    {
      "epoch": 0.5012088393728173,
      "grad_norm": 0.0,
      "learning_rate": 4.991304839599033e-06,
      "loss": 1.3883,
      "step": 35450
    },
    {
      "epoch": 0.5013502240947844,
      "grad_norm": 0.0,
      "learning_rate": 4.989890992379364e-06,
      "loss": 0.6984,
      "step": 35460
    },
    {
      "epoch": 0.5014916088167513,
      "grad_norm": 0.0,
      "learning_rate": 4.988477145159695e-06,
      "loss": 0.0,
      "step": 35470
    },
    {
      "epoch": 0.5016329935387182,
      "grad_norm": 0.0,
      "learning_rate": 4.9870632979400255e-06,
      "loss": 0.0566,
      "step": 35480
    },
    {
      "epoch": 0.5017743782606852,
      "grad_norm": 0.0,
      "learning_rate": 4.9856494507203556e-06,
      "loss": 0.0,
      "step": 35490
    },
    {
      "epoch": 0.5019157629826521,
      "grad_norm": 0.0,
      "learning_rate": 4.984235603500686e-06,
      "loss": 0.0,
      "step": 35500
    },
    {
      "epoch": 0.5020571477046191,
      "grad_norm": 6.5523568792968945e-09,
      "learning_rate": 4.982821756281016e-06,
      "loss": 0.0,
      "step": 35510
    },
    {
      "epoch": 0.502198532426586,
      "grad_norm": 0.0,
      "learning_rate": 4.981407909061347e-06,
      "loss": 0.0,
      "step": 35520
    },
    {
      "epoch": 0.502339917148553,
      "grad_norm": 0.0,
      "learning_rate": 4.979994061841677e-06,
      "loss": 0.0001,
      "step": 35530
    },
    {
      "epoch": 0.5024813018705199,
      "grad_norm": 0.0,
      "learning_rate": 4.978580214622009e-06,
      "loss": 0.0175,
      "step": 35540
    },
    {
      "epoch": 0.5026226865924868,
      "grad_norm": 0.0,
      "learning_rate": 4.977166367402339e-06,
      "loss": 0.0,
      "step": 35550
    },
    {
      "epoch": 0.5027640713144538,
      "grad_norm": 0.0,
      "learning_rate": 4.97575252018267e-06,
      "loss": 0.0002,
      "step": 35560
    },
    {
      "epoch": 0.5029054560364207,
      "grad_norm": 0.0,
      "learning_rate": 4.974338672963e-06,
      "loss": 0.0,
      "step": 35570
    },
    {
      "epoch": 0.5030468407583877,
      "grad_norm": 0.00010557363566476852,
      "learning_rate": 4.972924825743331e-06,
      "loss": 0.0059,
      "step": 35580
    },
    {
      "epoch": 0.5031882254803546,
      "grad_norm": 0.0002459108072798699,
      "learning_rate": 4.971510978523661e-06,
      "loss": 0.001,
      "step": 35590
    },
    {
      "epoch": 0.5033296102023216,
      "grad_norm": 0.0,
      "learning_rate": 4.970097131303992e-06,
      "loss": 0.0,
      "step": 35600
    },
    {
      "epoch": 0.5034709949242885,
      "grad_norm": 0.0,
      "learning_rate": 4.9686832840843224e-06,
      "loss": 0.0,
      "step": 35610
    },
    {
      "epoch": 0.5036123796462554,
      "grad_norm": 7.496421261521391e-09,
      "learning_rate": 4.967269436864653e-06,
      "loss": 0.0,
      "step": 35620
    },
    {
      "epoch": 0.5037537643682224,
      "grad_norm": 2.3878311594671686e-07,
      "learning_rate": 4.965855589644983e-06,
      "loss": 0.0018,
      "step": 35630
    },
    {
      "epoch": 0.5038951490901893,
      "grad_norm": 11.518402099609375,
      "learning_rate": 4.964441742425314e-06,
      "loss": 0.0031,
      "step": 35640
    },
    {
      "epoch": 0.5040365338121563,
      "grad_norm": 0.0,
      "learning_rate": 4.963027895205644e-06,
      "loss": 0.0001,
      "step": 35650
    },
    {
      "epoch": 0.5041779185341232,
      "grad_norm": 0.0,
      "learning_rate": 4.961614047985975e-06,
      "loss": 0.0006,
      "step": 35660
    },
    {
      "epoch": 0.5043193032560902,
      "grad_norm": 0.0,
      "learning_rate": 4.960200200766306e-06,
      "loss": 0.0,
      "step": 35670
    },
    {
      "epoch": 0.5044606879780571,
      "grad_norm": 0.0,
      "learning_rate": 4.958786353546636e-06,
      "loss": 0.0,
      "step": 35680
    },
    {
      "epoch": 0.504602072700024,
      "grad_norm": 0.0,
      "learning_rate": 4.957372506326967e-06,
      "loss": 0.0,
      "step": 35690
    },
    {
      "epoch": 0.504743457421991,
      "grad_norm": 4.064681391469094e-08,
      "learning_rate": 4.955958659107298e-06,
      "loss": 0.0547,
      "step": 35700
    },
    {
      "epoch": 0.5048848421439579,
      "grad_norm": 11.486586570739746,
      "learning_rate": 4.954544811887628e-06,
      "loss": 0.0028,
      "step": 35710
    },
    {
      "epoch": 0.5050262268659249,
      "grad_norm": 0.0,
      "learning_rate": 4.9531309646679585e-06,
      "loss": 0.0,
      "step": 35720
    },
    {
      "epoch": 0.5051676115878918,
      "grad_norm": 430.8229675292969,
      "learning_rate": 4.951717117448289e-06,
      "loss": 0.0917,
      "step": 35730
    },
    {
      "epoch": 0.5053089963098587,
      "grad_norm": 0.0,
      "learning_rate": 4.950303270228619e-06,
      "loss": 0.0,
      "step": 35740
    },
    {
      "epoch": 0.5054503810318257,
      "grad_norm": 0.0,
      "learning_rate": 4.94888942300895e-06,
      "loss": 0.0006,
      "step": 35750
    },
    {
      "epoch": 0.5055917657537926,
      "grad_norm": 0.0004626279987860471,
      "learning_rate": 4.94747557578928e-06,
      "loss": 0.0,
      "step": 35760
    },
    {
      "epoch": 0.5057331504757596,
      "grad_norm": 0.0,
      "learning_rate": 4.946061728569611e-06,
      "loss": 0.0129,
      "step": 35770
    },
    {
      "epoch": 0.5058745351977265,
      "grad_norm": 0.0,
      "learning_rate": 4.944647881349942e-06,
      "loss": 0.0089,
      "step": 35780
    },
    {
      "epoch": 0.5060159199196935,
      "grad_norm": 0.0,
      "learning_rate": 4.943234034130272e-06,
      "loss": 0.0,
      "step": 35790
    },
    {
      "epoch": 0.5061573046416604,
      "grad_norm": 0.0,
      "learning_rate": 4.941820186910603e-06,
      "loss": 0.0003,
      "step": 35800
    },
    {
      "epoch": 0.5062986893636273,
      "grad_norm": 0.0,
      "learning_rate": 4.940406339690934e-06,
      "loss": 0.0582,
      "step": 35810
    },
    {
      "epoch": 0.5064400740855943,
      "grad_norm": 0.0,
      "learning_rate": 4.938992492471264e-06,
      "loss": 0.0,
      "step": 35820
    },
    {
      "epoch": 0.5065814588075612,
      "grad_norm": 9.921914897859097e-05,
      "learning_rate": 4.9375786452515945e-06,
      "loss": 0.0,
      "step": 35830
    },
    {
      "epoch": 0.5067228435295282,
      "grad_norm": 3.3745570249266166e-07,
      "learning_rate": 4.936164798031925e-06,
      "loss": 0.0,
      "step": 35840
    },
    {
      "epoch": 0.5068642282514951,
      "grad_norm": 0.0,
      "learning_rate": 4.934750950812255e-06,
      "loss": 0.0,
      "step": 35850
    },
    {
      "epoch": 0.5070056129734621,
      "grad_norm": 0.0,
      "learning_rate": 4.933337103592586e-06,
      "loss": 0.0,
      "step": 35860
    },
    {
      "epoch": 0.507146997695429,
      "grad_norm": 0.0,
      "learning_rate": 4.931923256372917e-06,
      "loss": 0.0,
      "step": 35870
    },
    {
      "epoch": 0.5072883824173959,
      "grad_norm": 0.0,
      "learning_rate": 4.930509409153247e-06,
      "loss": 0.0,
      "step": 35880
    },
    {
      "epoch": 0.5074297671393629,
      "grad_norm": 0.0,
      "learning_rate": 4.929095561933578e-06,
      "loss": 0.0,
      "step": 35890
    },
    {
      "epoch": 0.5075711518613298,
      "grad_norm": 0.0,
      "learning_rate": 4.927681714713908e-06,
      "loss": 0.0177,
      "step": 35900
    },
    {
      "epoch": 0.5077125365832968,
      "grad_norm": 1.374143630528124e-05,
      "learning_rate": 4.926267867494239e-06,
      "loss": 0.0,
      "step": 35910
    },
    {
      "epoch": 0.5078539213052637,
      "grad_norm": 0.0,
      "learning_rate": 4.92485402027457e-06,
      "loss": 0.0,
      "step": 35920
    },
    {
      "epoch": 0.5079953060272306,
      "grad_norm": 1.536607108221233e-08,
      "learning_rate": 4.9234401730549005e-06,
      "loss": 0.0,
      "step": 35930
    },
    {
      "epoch": 0.5081366907491977,
      "grad_norm": 0.00010190613829763606,
      "learning_rate": 4.9220263258352305e-06,
      "loss": 0.005,
      "step": 35940
    },
    {
      "epoch": 0.5082780754711645,
      "grad_norm": 7.208823808468878e-05,
      "learning_rate": 4.920612478615561e-06,
      "loss": 0.0,
      "step": 35950
    },
    {
      "epoch": 0.5084194601931316,
      "grad_norm": 0.061597056686878204,
      "learning_rate": 4.919198631395891e-06,
      "loss": 0.0,
      "step": 35960
    },
    {
      "epoch": 0.5085608449150985,
      "grad_norm": 0.0,
      "learning_rate": 4.917784784176222e-06,
      "loss": 0.0009,
      "step": 35970
    },
    {
      "epoch": 0.5087022296370655,
      "grad_norm": 1.4764705902337028e-08,
      "learning_rate": 4.916370936956552e-06,
      "loss": 0.0,
      "step": 35980
    },
    {
      "epoch": 0.5088436143590324,
      "grad_norm": 0.0,
      "learning_rate": 4.914957089736884e-06,
      "loss": 0.0001,
      "step": 35990
    },
    {
      "epoch": 0.5089849990809993,
      "grad_norm": 0.0,
      "learning_rate": 4.913543242517214e-06,
      "loss": 0.0001,
      "step": 36000
    },
    {
      "epoch": 0.5091263838029663,
      "grad_norm": 0.16010653972625732,
      "learning_rate": 4.912129395297545e-06,
      "loss": 0.3432,
      "step": 36010
    },
    {
      "epoch": 0.5092677685249332,
      "grad_norm": 0.0,
      "learning_rate": 4.910715548077875e-06,
      "loss": 0.0,
      "step": 36020
    },
    {
      "epoch": 0.5094091532469002,
      "grad_norm": 0.0,
      "learning_rate": 4.909301700858206e-06,
      "loss": 0.1023,
      "step": 36030
    },
    {
      "epoch": 0.5095505379688671,
      "grad_norm": 0.0,
      "learning_rate": 4.907887853638536e-06,
      "loss": 0.0086,
      "step": 36040
    },
    {
      "epoch": 0.5096919226908341,
      "grad_norm": 0.0,
      "learning_rate": 4.9064740064188666e-06,
      "loss": 0.2578,
      "step": 36050
    },
    {
      "epoch": 0.509833307412801,
      "grad_norm": 5.257219527265988e-06,
      "learning_rate": 4.9050601591991974e-06,
      "loss": 0.0,
      "step": 36060
    },
    {
      "epoch": 0.5099746921347679,
      "grad_norm": 0.0,
      "learning_rate": 4.903646311979528e-06,
      "loss": 0.3973,
      "step": 36070
    },
    {
      "epoch": 0.5101160768567349,
      "grad_norm": 0.0,
      "learning_rate": 4.902232464759858e-06,
      "loss": 0.0,
      "step": 36080
    },
    {
      "epoch": 0.5102574615787018,
      "grad_norm": 0.0001311636297032237,
      "learning_rate": 4.900818617540189e-06,
      "loss": 0.0,
      "step": 36090
    },
    {
      "epoch": 0.5103988463006688,
      "grad_norm": 0.0006226206896826625,
      "learning_rate": 4.899404770320519e-06,
      "loss": 0.0,
      "step": 36100
    },
    {
      "epoch": 0.5105402310226357,
      "grad_norm": 0.0,
      "learning_rate": 4.89799092310085e-06,
      "loss": 0.0015,
      "step": 36110
    },
    {
      "epoch": 0.5106816157446027,
      "grad_norm": 0.0,
      "learning_rate": 4.896577075881181e-06,
      "loss": 0.0,
      "step": 36120
    },
    {
      "epoch": 0.5108230004665696,
      "grad_norm": 0.0,
      "learning_rate": 4.895163228661512e-06,
      "loss": 0.0,
      "step": 36130
    },
    {
      "epoch": 0.5109643851885365,
      "grad_norm": 0.0,
      "learning_rate": 4.893749381441842e-06,
      "loss": 0.0,
      "step": 36140
    },
    {
      "epoch": 0.5111057699105035,
      "grad_norm": 0.01562289148569107,
      "learning_rate": 4.892335534222173e-06,
      "loss": 0.0007,
      "step": 36150
    },
    {
      "epoch": 0.5112471546324704,
      "grad_norm": 0.0,
      "learning_rate": 4.890921687002503e-06,
      "loss": 0.0093,
      "step": 36160
    },
    {
      "epoch": 0.5113885393544374,
      "grad_norm": 0.0,
      "learning_rate": 4.8895078397828335e-06,
      "loss": 0.01,
      "step": 36170
    },
    {
      "epoch": 0.5115299240764043,
      "grad_norm": 0.0,
      "learning_rate": 4.8880939925631635e-06,
      "loss": 0.0028,
      "step": 36180
    },
    {
      "epoch": 0.5116713087983712,
      "grad_norm": 3.378259094688474e-08,
      "learning_rate": 4.886680145343495e-06,
      "loss": 0.0,
      "step": 36190
    },
    {
      "epoch": 0.5118126935203382,
      "grad_norm": 7.588994321849896e-09,
      "learning_rate": 4.885266298123825e-06,
      "loss": 0.0,
      "step": 36200
    },
    {
      "epoch": 0.5119540782423051,
      "grad_norm": 0.28341996669769287,
      "learning_rate": 4.883852450904156e-06,
      "loss": 0.0,
      "step": 36210
    },
    {
      "epoch": 0.5120954629642721,
      "grad_norm": 7.598547199449968e-06,
      "learning_rate": 4.882438603684486e-06,
      "loss": 0.0001,
      "step": 36220
    },
    {
      "epoch": 0.512236847686239,
      "grad_norm": 6.877885372347237e-09,
      "learning_rate": 4.881024756464817e-06,
      "loss": 0.0007,
      "step": 36230
    },
    {
      "epoch": 0.512378232408206,
      "grad_norm": 0.0,
      "learning_rate": 4.879610909245147e-06,
      "loss": 0.0,
      "step": 36240
    },
    {
      "epoch": 0.5125196171301729,
      "grad_norm": 0.0,
      "learning_rate": 4.878197062025478e-06,
      "loss": 0.0095,
      "step": 36250
    },
    {
      "epoch": 0.5126610018521398,
      "grad_norm": 6.823131843702868e-05,
      "learning_rate": 4.876783214805809e-06,
      "loss": 0.0,
      "step": 36260
    },
    {
      "epoch": 0.5128023865741068,
      "grad_norm": 0.0,
      "learning_rate": 4.8753693675861395e-06,
      "loss": 0.0,
      "step": 36270
    },
    {
      "epoch": 0.5129437712960737,
      "grad_norm": 0.0,
      "learning_rate": 4.8739555203664695e-06,
      "loss": 0.0,
      "step": 36280
    },
    {
      "epoch": 0.5130851560180407,
      "grad_norm": 0.0,
      "learning_rate": 4.8725416731468e-06,
      "loss": 0.0,
      "step": 36290
    },
    {
      "epoch": 0.5132265407400076,
      "grad_norm": 0.0,
      "learning_rate": 4.87112782592713e-06,
      "loss": 0.0024,
      "step": 36300
    },
    {
      "epoch": 0.5133679254619746,
      "grad_norm": 0.0,
      "learning_rate": 4.869713978707461e-06,
      "loss": 0.0,
      "step": 36310
    },
    {
      "epoch": 0.5135093101839415,
      "grad_norm": 0.0,
      "learning_rate": 4.868300131487792e-06,
      "loss": 0.0011,
      "step": 36320
    },
    {
      "epoch": 0.5136506949059084,
      "grad_norm": 0.0,
      "learning_rate": 4.866886284268122e-06,
      "loss": 0.0,
      "step": 36330
    },
    {
      "epoch": 0.5137920796278754,
      "grad_norm": 0.0021062507294118404,
      "learning_rate": 4.865472437048453e-06,
      "loss": 0.0,
      "step": 36340
    },
    {
      "epoch": 0.5139334643498423,
      "grad_norm": 0.0,
      "learning_rate": 4.864058589828784e-06,
      "loss": 0.0,
      "step": 36350
    },
    {
      "epoch": 0.5140748490718093,
      "grad_norm": 0.6672736406326294,
      "learning_rate": 4.862644742609114e-06,
      "loss": 0.0002,
      "step": 36360
    },
    {
      "epoch": 0.5142162337937762,
      "grad_norm": 0.029780764132738113,
      "learning_rate": 4.861230895389445e-06,
      "loss": 0.0758,
      "step": 36370
    },
    {
      "epoch": 0.5143576185157432,
      "grad_norm": 0.0,
      "learning_rate": 4.8598170481697755e-06,
      "loss": 0.0,
      "step": 36380
    },
    {
      "epoch": 0.5144990032377101,
      "grad_norm": 0.0,
      "learning_rate": 4.8584032009501055e-06,
      "loss": 0.3514,
      "step": 36390
    },
    {
      "epoch": 0.514640387959677,
      "grad_norm": 9.470562112312564e-09,
      "learning_rate": 4.856989353730436e-06,
      "loss": 0.0,
      "step": 36400
    },
    {
      "epoch": 0.514781772681644,
      "grad_norm": 0.1940915882587433,
      "learning_rate": 4.855575506510766e-06,
      "loss": 0.0002,
      "step": 36410
    },
    {
      "epoch": 0.5149231574036109,
      "grad_norm": 0.0,
      "learning_rate": 4.854161659291097e-06,
      "loss": 0.0,
      "step": 36420
    },
    {
      "epoch": 0.515064542125578,
      "grad_norm": 0.0,
      "learning_rate": 4.852747812071428e-06,
      "loss": 0.0,
      "step": 36430
    },
    {
      "epoch": 0.5152059268475448,
      "grad_norm": 0.0,
      "learning_rate": 4.851333964851759e-06,
      "loss": 0.0,
      "step": 36440
    },
    {
      "epoch": 0.5153473115695117,
      "grad_norm": 0.0,
      "learning_rate": 4.849920117632089e-06,
      "loss": 0.0,
      "step": 36450
    },
    {
      "epoch": 0.5154886962914788,
      "grad_norm": 0.0,
      "learning_rate": 4.84850627041242e-06,
      "loss": 0.0038,
      "step": 36460
    },
    {
      "epoch": 0.5156300810134457,
      "grad_norm": 8.12293876606418e-09,
      "learning_rate": 4.84709242319275e-06,
      "loss": 0.0,
      "step": 36470
    },
    {
      "epoch": 0.5157714657354127,
      "grad_norm": 0.0,
      "learning_rate": 4.845678575973081e-06,
      "loss": 0.3424,
      "step": 36480
    },
    {
      "epoch": 0.5159128504573796,
      "grad_norm": 310.49957275390625,
      "learning_rate": 4.8442647287534116e-06,
      "loss": 0.0486,
      "step": 36490
    },
    {
      "epoch": 0.5160542351793466,
      "grad_norm": 0.0,
      "learning_rate": 4.8428508815337416e-06,
      "loss": 0.0,
      "step": 36500
    },
    {
      "epoch": 0.5161956199013135,
      "grad_norm": 0.0,
      "learning_rate": 4.841437034314072e-06,
      "loss": 0.0035,
      "step": 36510
    },
    {
      "epoch": 0.5163370046232804,
      "grad_norm": 0.0,
      "learning_rate": 4.840023187094403e-06,
      "loss": 0.0,
      "step": 36520
    },
    {
      "epoch": 0.5164783893452474,
      "grad_norm": 0.0,
      "learning_rate": 4.838609339874733e-06,
      "loss": 0.0,
      "step": 36530
    },
    {
      "epoch": 0.5166197740672143,
      "grad_norm": 0.0,
      "learning_rate": 4.837195492655064e-06,
      "loss": 0.0,
      "step": 36540
    },
    {
      "epoch": 0.5167611587891813,
      "grad_norm": 0.0,
      "learning_rate": 4.835781645435394e-06,
      "loss": 0.0,
      "step": 36550
    },
    {
      "epoch": 0.5169025435111482,
      "grad_norm": 0.0,
      "learning_rate": 4.834367798215725e-06,
      "loss": 0.0,
      "step": 36560
    },
    {
      "epoch": 0.5170439282331152,
      "grad_norm": 3.572593598732965e-08,
      "learning_rate": 4.832953950996056e-06,
      "loss": 0.0001,
      "step": 36570
    },
    {
      "epoch": 0.5171853129550821,
      "grad_norm": 0.0,
      "learning_rate": 4.831540103776387e-06,
      "loss": 0.0,
      "step": 36580
    },
    {
      "epoch": 0.517326697677049,
      "grad_norm": 0.0,
      "learning_rate": 4.830126256556717e-06,
      "loss": 0.0,
      "step": 36590
    },
    {
      "epoch": 0.517468082399016,
      "grad_norm": 4.791266974280006e-07,
      "learning_rate": 4.828712409337048e-06,
      "loss": 0.0002,
      "step": 36600
    },
    {
      "epoch": 0.5176094671209829,
      "grad_norm": 8.247950518125435e-07,
      "learning_rate": 4.827298562117378e-06,
      "loss": 0.0,
      "step": 36610
    },
    {
      "epoch": 0.5177508518429499,
      "grad_norm": 0.0,
      "learning_rate": 4.8258847148977085e-06,
      "loss": 0.0,
      "step": 36620
    },
    {
      "epoch": 0.5178922365649168,
      "grad_norm": 0.0,
      "learning_rate": 4.8244708676780385e-06,
      "loss": 0.0,
      "step": 36630
    },
    {
      "epoch": 0.5180336212868838,
      "grad_norm": 0.0,
      "learning_rate": 4.82305702045837e-06,
      "loss": 0.0,
      "step": 36640
    },
    {
      "epoch": 0.5181750060088507,
      "grad_norm": 0.0,
      "learning_rate": 4.8216431732387e-06,
      "loss": 0.0001,
      "step": 36650
    },
    {
      "epoch": 0.5183163907308176,
      "grad_norm": 0.0,
      "learning_rate": 4.820229326019031e-06,
      "loss": 0.0003,
      "step": 36660
    },
    {
      "epoch": 0.5184577754527846,
      "grad_norm": 0.0,
      "learning_rate": 4.818815478799361e-06,
      "loss": 0.0474,
      "step": 36670
    },
    {
      "epoch": 0.5185991601747515,
      "grad_norm": 6.073997127487019e-08,
      "learning_rate": 4.817401631579692e-06,
      "loss": 0.0026,
      "step": 36680
    },
    {
      "epoch": 0.5187405448967185,
      "grad_norm": 0.014402773231267929,
      "learning_rate": 4.815987784360022e-06,
      "loss": 0.4219,
      "step": 36690
    },
    {
      "epoch": 0.5188819296186854,
      "grad_norm": 0.0,
      "learning_rate": 4.814573937140354e-06,
      "loss": 0.0,
      "step": 36700
    },
    {
      "epoch": 0.5190233143406523,
      "grad_norm": 0.006350591778755188,
      "learning_rate": 4.813160089920684e-06,
      "loss": 0.0001,
      "step": 36710
    },
    {
      "epoch": 0.5191646990626193,
      "grad_norm": 0.0,
      "learning_rate": 4.8117462427010145e-06,
      "loss": 0.0079,
      "step": 36720
    },
    {
      "epoch": 0.5193060837845862,
      "grad_norm": 1.353811285298434e-06,
      "learning_rate": 4.8103323954813445e-06,
      "loss": 0.0,
      "step": 36730
    },
    {
      "epoch": 0.5194474685065532,
      "grad_norm": 0.0,
      "learning_rate": 4.808918548261675e-06,
      "loss": 0.0,
      "step": 36740
    },
    {
      "epoch": 0.5195888532285201,
      "grad_norm": 0.005487664602696896,
      "learning_rate": 4.807504701042005e-06,
      "loss": 0.0001,
      "step": 36750
    },
    {
      "epoch": 0.5197302379504871,
      "grad_norm": 0.1493157297372818,
      "learning_rate": 4.806090853822336e-06,
      "loss": 0.0,
      "step": 36760
    },
    {
      "epoch": 0.519871622672454,
      "grad_norm": 0.0,
      "learning_rate": 4.804677006602667e-06,
      "loss": 0.0,
      "step": 36770
    },
    {
      "epoch": 0.5200130073944209,
      "grad_norm": 0.0,
      "learning_rate": 4.803263159382998e-06,
      "loss": 0.0012,
      "step": 36780
    },
    {
      "epoch": 0.5201543921163879,
      "grad_norm": 0.0,
      "learning_rate": 4.801849312163328e-06,
      "loss": 0.0039,
      "step": 36790
    },
    {
      "epoch": 0.5202957768383548,
      "grad_norm": 0.0,
      "learning_rate": 4.800435464943659e-06,
      "loss": 0.0418,
      "step": 36800
    },
    {
      "epoch": 0.5204371615603218,
      "grad_norm": 0.00024700837093405426,
      "learning_rate": 4.799021617723989e-06,
      "loss": 0.0,
      "step": 36810
    },
    {
      "epoch": 0.5205785462822887,
      "grad_norm": 0.0,
      "learning_rate": 4.79760777050432e-06,
      "loss": 0.0,
      "step": 36820
    },
    {
      "epoch": 0.5207199310042557,
      "grad_norm": 0.0,
      "learning_rate": 4.7961939232846505e-06,
      "loss": 0.0001,
      "step": 36830
    },
    {
      "epoch": 0.5208613157262226,
      "grad_norm": 0.0,
      "learning_rate": 4.794780076064981e-06,
      "loss": 0.5055,
      "step": 36840
    },
    {
      "epoch": 0.5210027004481895,
      "grad_norm": 0.0,
      "learning_rate": 4.793366228845311e-06,
      "loss": 0.0002,
      "step": 36850
    },
    {
      "epoch": 0.5211440851701565,
      "grad_norm": 0.0,
      "learning_rate": 4.791952381625642e-06,
      "loss": 0.0,
      "step": 36860
    },
    {
      "epoch": 0.5212854698921234,
      "grad_norm": 0.0,
      "learning_rate": 4.790538534405972e-06,
      "loss": 0.0,
      "step": 36870
    },
    {
      "epoch": 0.5214268546140904,
      "grad_norm": 0.0,
      "learning_rate": 4.789124687186303e-06,
      "loss": 0.018,
      "step": 36880
    },
    {
      "epoch": 0.5215682393360573,
      "grad_norm": 0.0,
      "learning_rate": 4.787710839966633e-06,
      "loss": 0.0,
      "step": 36890
    },
    {
      "epoch": 0.5217096240580242,
      "grad_norm": 3.692917016451247e-05,
      "learning_rate": 4.786296992746964e-06,
      "loss": 0.0,
      "step": 36900
    },
    {
      "epoch": 0.5218510087799912,
      "grad_norm": 0.0,
      "learning_rate": 4.784883145527295e-06,
      "loss": 1.0461,
      "step": 36910
    },
    {
      "epoch": 0.5219923935019581,
      "grad_norm": 0.0,
      "learning_rate": 4.783469298307626e-06,
      "loss": 0.0,
      "step": 36920
    },
    {
      "epoch": 0.5221337782239251,
      "grad_norm": 0.0,
      "learning_rate": 4.782055451087956e-06,
      "loss": 0.0002,
      "step": 36930
    },
    {
      "epoch": 0.522275162945892,
      "grad_norm": 0.0,
      "learning_rate": 4.7806416038682865e-06,
      "loss": 0.0,
      "step": 36940
    },
    {
      "epoch": 0.522416547667859,
      "grad_norm": 7.125428965082392e-05,
      "learning_rate": 4.7792277566486166e-06,
      "loss": 0.0001,
      "step": 36950
    },
    {
      "epoch": 0.522557932389826,
      "grad_norm": 0.00044480100041255355,
      "learning_rate": 4.777813909428947e-06,
      "loss": 0.0029,
      "step": 36960
    },
    {
      "epoch": 0.5226993171117928,
      "grad_norm": 0.982993483543396,
      "learning_rate": 4.776400062209278e-06,
      "loss": 0.3744,
      "step": 36970
    },
    {
      "epoch": 0.5228407018337599,
      "grad_norm": 0.0,
      "learning_rate": 4.774986214989608e-06,
      "loss": 0.0011,
      "step": 36980
    },
    {
      "epoch": 0.5229820865557268,
      "grad_norm": 0.0,
      "learning_rate": 4.773572367769939e-06,
      "loss": 0.0,
      "step": 36990
    },
    {
      "epoch": 0.5231234712776938,
      "grad_norm": 4.345201531918974e-09,
      "learning_rate": 4.77215852055027e-06,
      "loss": 0.6535,
      "step": 37000
    },
    {
      "epoch": 0.5232648559996607,
      "grad_norm": 0.0,
      "learning_rate": 4.7707446733306e-06,
      "loss": 0.0,
      "step": 37010
    },
    {
      "epoch": 0.5234062407216277,
      "grad_norm": 9.335384021369464e-09,
      "learning_rate": 4.769330826110931e-06,
      "loss": 0.0,
      "step": 37020
    },
    {
      "epoch": 0.5235476254435946,
      "grad_norm": 0.0,
      "learning_rate": 4.767916978891262e-06,
      "loss": 0.0001,
      "step": 37030
    },
    {
      "epoch": 0.5236890101655615,
      "grad_norm": 0.0,
      "learning_rate": 4.766503131671592e-06,
      "loss": 0.0,
      "step": 37040
    },
    {
      "epoch": 0.5238303948875285,
      "grad_norm": 0.0,
      "learning_rate": 4.7650892844519226e-06,
      "loss": 0.0,
      "step": 37050
    },
    {
      "epoch": 0.5239717796094954,
      "grad_norm": 6.552958075189963e-05,
      "learning_rate": 4.763675437232253e-06,
      "loss": 0.0,
      "step": 37060
    },
    {
      "epoch": 0.5241131643314624,
      "grad_norm": 12.271563529968262,
      "learning_rate": 4.7622615900125834e-06,
      "loss": 0.0027,
      "step": 37070
    },
    {
      "epoch": 0.5242545490534293,
      "grad_norm": 0.0,
      "learning_rate": 4.760847742792914e-06,
      "loss": 0.0,
      "step": 37080
    },
    {
      "epoch": 0.5243959337753963,
      "grad_norm": 0.0,
      "learning_rate": 4.759433895573245e-06,
      "loss": 0.0,
      "step": 37090
    },
    {
      "epoch": 0.5245373184973632,
      "grad_norm": 0.0,
      "learning_rate": 4.758020048353575e-06,
      "loss": 0.0,
      "step": 37100
    },
    {
      "epoch": 0.5246787032193301,
      "grad_norm": 0.0,
      "learning_rate": 4.756606201133906e-06,
      "loss": 0.0,
      "step": 37110
    },
    {
      "epoch": 0.5248200879412971,
      "grad_norm": 8.851928214426152e-06,
      "learning_rate": 4.755192353914236e-06,
      "loss": 0.0,
      "step": 37120
    },
    {
      "epoch": 0.524961472663264,
      "grad_norm": 0.0,
      "learning_rate": 4.753778506694567e-06,
      "loss": 0.0005,
      "step": 37130
    },
    {
      "epoch": 0.525102857385231,
      "grad_norm": 0.0,
      "learning_rate": 4.752364659474898e-06,
      "loss": 0.0,
      "step": 37140
    },
    {
      "epoch": 0.5252442421071979,
      "grad_norm": 0.0,
      "learning_rate": 4.750950812255229e-06,
      "loss": 0.0,
      "step": 37150
    },
    {
      "epoch": 0.5253856268291648,
      "grad_norm": 0.0,
      "learning_rate": 4.749536965035559e-06,
      "loss": 0.0001,
      "step": 37160
    },
    {
      "epoch": 0.5255270115511318,
      "grad_norm": 5.169102951185778e-05,
      "learning_rate": 4.7481231178158895e-06,
      "loss": 0.0,
      "step": 37170
    },
    {
      "epoch": 0.5256683962730987,
      "grad_norm": 0.0,
      "learning_rate": 4.7467092705962195e-06,
      "loss": 0.0001,
      "step": 37180
    },
    {
      "epoch": 0.5258097809950657,
      "grad_norm": 0.0,
      "learning_rate": 4.74529542337655e-06,
      "loss": 0.0,
      "step": 37190
    },
    {
      "epoch": 0.5259511657170326,
      "grad_norm": 0.0,
      "learning_rate": 4.74388157615688e-06,
      "loss": 0.1824,
      "step": 37200
    },
    {
      "epoch": 0.5260925504389996,
      "grad_norm": 0.0,
      "learning_rate": 4.742467728937211e-06,
      "loss": 0.0,
      "step": 37210
    },
    {
      "epoch": 0.5262339351609665,
      "grad_norm": 0.0,
      "learning_rate": 4.741053881717542e-06,
      "loss": 0.0,
      "step": 37220
    },
    {
      "epoch": 0.5263753198829334,
      "grad_norm": 0.0,
      "learning_rate": 4.739640034497873e-06,
      "loss": 0.0,
      "step": 37230
    },
    {
      "epoch": 0.5265167046049004,
      "grad_norm": 0.0,
      "learning_rate": 4.738226187278203e-06,
      "loss": 0.0,
      "step": 37240
    },
    {
      "epoch": 0.5266580893268673,
      "grad_norm": 0.0,
      "learning_rate": 4.736812340058534e-06,
      "loss": 0.0,
      "step": 37250
    },
    {
      "epoch": 0.5267994740488343,
      "grad_norm": 0.9178866147994995,
      "learning_rate": 4.735398492838864e-06,
      "loss": 0.0002,
      "step": 37260
    },
    {
      "epoch": 0.5269408587708012,
      "grad_norm": 0.0,
      "learning_rate": 4.733984645619195e-06,
      "loss": 0.0,
      "step": 37270
    },
    {
      "epoch": 0.5270822434927682,
      "grad_norm": 0.00021435471717268229,
      "learning_rate": 4.7325707983995255e-06,
      "loss": 0.0,
      "step": 37280
    },
    {
      "epoch": 0.5272236282147351,
      "grad_norm": 0.0,
      "learning_rate": 4.731156951179856e-06,
      "loss": 0.2566,
      "step": 37290
    },
    {
      "epoch": 0.527365012936702,
      "grad_norm": 0.0,
      "learning_rate": 4.729743103960186e-06,
      "loss": 0.0001,
      "step": 37300
    },
    {
      "epoch": 0.527506397658669,
      "grad_norm": 0.0,
      "learning_rate": 4.728329256740517e-06,
      "loss": 0.0,
      "step": 37310
    },
    {
      "epoch": 0.5276477823806359,
      "grad_norm": 4.187134639010992e-09,
      "learning_rate": 4.726915409520847e-06,
      "loss": 0.0,
      "step": 37320
    },
    {
      "epoch": 0.5277891671026029,
      "grad_norm": 0.0,
      "learning_rate": 4.725501562301178e-06,
      "loss": 0.0002,
      "step": 37330
    },
    {
      "epoch": 0.5279305518245698,
      "grad_norm": 0.0,
      "learning_rate": 4.724087715081508e-06,
      "loss": 0.0,
      "step": 37340
    },
    {
      "epoch": 0.5280719365465368,
      "grad_norm": 0.0,
      "learning_rate": 4.72267386786184e-06,
      "loss": 0.0,
      "step": 37350
    },
    {
      "epoch": 0.5282133212685037,
      "grad_norm": 0.0,
      "learning_rate": 4.72126002064217e-06,
      "loss": 0.0,
      "step": 37360
    },
    {
      "epoch": 0.5283547059904706,
      "grad_norm": 31.154726028442383,
      "learning_rate": 4.719846173422501e-06,
      "loss": 0.0049,
      "step": 37370
    },
    {
      "epoch": 0.5284960907124376,
      "grad_norm": 0.0,
      "learning_rate": 4.718432326202831e-06,
      "loss": 0.0078,
      "step": 37380
    },
    {
      "epoch": 0.5286374754344045,
      "grad_norm": 0.0,
      "learning_rate": 4.7170184789831615e-06,
      "loss": 0.0012,
      "step": 37390
    },
    {
      "epoch": 0.5287788601563715,
      "grad_norm": 0.00032964171259664,
      "learning_rate": 4.7156046317634915e-06,
      "loss": 0.0,
      "step": 37400
    },
    {
      "epoch": 0.5289202448783384,
      "grad_norm": 0.0,
      "learning_rate": 4.714190784543823e-06,
      "loss": 0.0011,
      "step": 37410
    },
    {
      "epoch": 0.5290616296003053,
      "grad_norm": 0.0,
      "learning_rate": 4.712776937324153e-06,
      "loss": 0.0,
      "step": 37420
    },
    {
      "epoch": 0.5292030143222723,
      "grad_norm": 116.68013000488281,
      "learning_rate": 4.711363090104484e-06,
      "loss": 0.7111,
      "step": 37430
    },
    {
      "epoch": 0.5293443990442392,
      "grad_norm": 2.0497784802842034e-08,
      "learning_rate": 4.709949242884814e-06,
      "loss": 0.1406,
      "step": 37440
    },
    {
      "epoch": 0.5294857837662063,
      "grad_norm": 1145.9996337890625,
      "learning_rate": 4.708535395665145e-06,
      "loss": 0.2469,
      "step": 37450
    },
    {
      "epoch": 0.5296271684881732,
      "grad_norm": 0.0,
      "learning_rate": 4.707121548445475e-06,
      "loss": 0.0,
      "step": 37460
    },
    {
      "epoch": 0.5297685532101402,
      "grad_norm": 0.0,
      "learning_rate": 4.705707701225806e-06,
      "loss": 0.0057,
      "step": 37470
    },
    {
      "epoch": 0.5299099379321071,
      "grad_norm": 0.00011446577991591766,
      "learning_rate": 4.704293854006137e-06,
      "loss": 0.0,
      "step": 37480
    },
    {
      "epoch": 0.530051322654074,
      "grad_norm": 0.0,
      "learning_rate": 4.7028800067864676e-06,
      "loss": 0.0,
      "step": 37490
    },
    {
      "epoch": 0.530192707376041,
      "grad_norm": 0.0,
      "learning_rate": 4.7014661595667976e-06,
      "loss": 0.0,
      "step": 37500
    },
    {
      "epoch": 0.5303340920980079,
      "grad_norm": 0.0,
      "learning_rate": 4.700052312347128e-06,
      "loss": 0.0,
      "step": 37510
    },
    {
      "epoch": 0.5304754768199749,
      "grad_norm": 0.0,
      "learning_rate": 4.6986384651274584e-06,
      "loss": 0.0,
      "step": 37520
    },
    {
      "epoch": 0.5306168615419418,
      "grad_norm": 0.0,
      "learning_rate": 4.697224617907789e-06,
      "loss": 0.2365,
      "step": 37530
    },
    {
      "epoch": 0.5307582462639088,
      "grad_norm": 0.0,
      "learning_rate": 4.69581077068812e-06,
      "loss": 0.0,
      "step": 37540
    },
    {
      "epoch": 0.5308996309858757,
      "grad_norm": 0.07230549305677414,
      "learning_rate": 4.69439692346845e-06,
      "loss": 0.0,
      "step": 37550
    },
    {
      "epoch": 0.5310410157078426,
      "grad_norm": 0.0,
      "learning_rate": 4.692983076248781e-06,
      "loss": 0.0,
      "step": 37560
    },
    {
      "epoch": 0.5311824004298096,
      "grad_norm": 0.09767601639032364,
      "learning_rate": 4.691569229029112e-06,
      "loss": 0.0183,
      "step": 37570
    },
    {
      "epoch": 0.5313237851517765,
      "grad_norm": 0.0,
      "learning_rate": 4.690155381809442e-06,
      "loss": 0.0,
      "step": 37580
    },
    {
      "epoch": 0.5314651698737435,
      "grad_norm": 0.0009413033840246499,
      "learning_rate": 4.688741534589773e-06,
      "loss": 0.0,
      "step": 37590
    },
    {
      "epoch": 0.5316065545957104,
      "grad_norm": 0.0,
      "learning_rate": 4.687327687370103e-06,
      "loss": 0.0001,
      "step": 37600
    },
    {
      "epoch": 0.5317479393176774,
      "grad_norm": 0.0,
      "learning_rate": 4.685913840150434e-06,
      "loss": 0.0,
      "step": 37610
    },
    {
      "epoch": 0.5318893240396443,
      "grad_norm": 0.003000852884724736,
      "learning_rate": 4.6844999929307645e-06,
      "loss": 0.0,
      "step": 37620
    },
    {
      "epoch": 0.5320307087616112,
      "grad_norm": 0.22290639579296112,
      "learning_rate": 4.6830861457110945e-06,
      "loss": 0.0001,
      "step": 37630
    },
    {
      "epoch": 0.5321720934835782,
      "grad_norm": 0.0009086300851777196,
      "learning_rate": 4.681672298491425e-06,
      "loss": 0.0015,
      "step": 37640
    },
    {
      "epoch": 0.5323134782055451,
      "grad_norm": 0.0,
      "learning_rate": 4.680258451271756e-06,
      "loss": 0.0,
      "step": 37650
    },
    {
      "epoch": 0.5324548629275121,
      "grad_norm": 8.86887718820617e-09,
      "learning_rate": 4.678844604052086e-06,
      "loss": 0.0,
      "step": 37660
    },
    {
      "epoch": 0.532596247649479,
      "grad_norm": 0.0,
      "learning_rate": 4.677430756832417e-06,
      "loss": 0.0012,
      "step": 37670
    },
    {
      "epoch": 0.5327376323714459,
      "grad_norm": 665.8759765625,
      "learning_rate": 4.676016909612748e-06,
      "loss": 0.3787,
      "step": 37680
    },
    {
      "epoch": 0.5328790170934129,
      "grad_norm": 0.0,
      "learning_rate": 4.674603062393078e-06,
      "loss": 0.0,
      "step": 37690
    },
    {
      "epoch": 0.5330204018153798,
      "grad_norm": 0.0,
      "learning_rate": 4.673189215173409e-06,
      "loss": 0.0,
      "step": 37700
    },
    {
      "epoch": 0.5331617865373468,
      "grad_norm": 0.0,
      "learning_rate": 4.67177536795374e-06,
      "loss": 0.0,
      "step": 37710
    },
    {
      "epoch": 0.5333031712593137,
      "grad_norm": 26.676815032958984,
      "learning_rate": 4.67036152073407e-06,
      "loss": 0.0108,
      "step": 37720
    },
    {
      "epoch": 0.5334445559812807,
      "grad_norm": 1.538719658356058e-07,
      "learning_rate": 4.6689476735144005e-06,
      "loss": 0.0003,
      "step": 37730
    },
    {
      "epoch": 0.5335859407032476,
      "grad_norm": 0.0,
      "learning_rate": 4.667533826294731e-06,
      "loss": 0.0622,
      "step": 37740
    },
    {
      "epoch": 0.5337273254252145,
      "grad_norm": 0.0010573618346825242,
      "learning_rate": 4.666119979075061e-06,
      "loss": 0.0,
      "step": 37750
    },
    {
      "epoch": 0.5338687101471815,
      "grad_norm": 1.4077442756388336e-05,
      "learning_rate": 4.664706131855392e-06,
      "loss": 0.0052,
      "step": 37760
    },
    {
      "epoch": 0.5340100948691484,
      "grad_norm": 8.04377214080887e-06,
      "learning_rate": 4.663292284635722e-06,
      "loss": 0.0037,
      "step": 37770
    },
    {
      "epoch": 0.5341514795911154,
      "grad_norm": 0.0,
      "learning_rate": 4.661878437416053e-06,
      "loss": 0.0,
      "step": 37780
    },
    {
      "epoch": 0.5342928643130823,
      "grad_norm": 0.0,
      "learning_rate": 4.660464590196384e-06,
      "loss": 0.0,
      "step": 37790
    },
    {
      "epoch": 0.5344342490350493,
      "grad_norm": 0.0,
      "learning_rate": 4.659050742976715e-06,
      "loss": 0.0,
      "step": 37800
    },
    {
      "epoch": 0.5345756337570162,
      "grad_norm": 0.0,
      "learning_rate": 4.657636895757045e-06,
      "loss": 0.0,
      "step": 37810
    },
    {
      "epoch": 0.5347170184789831,
      "grad_norm": 0.0,
      "learning_rate": 4.656223048537376e-06,
      "loss": 0.0,
      "step": 37820
    },
    {
      "epoch": 0.5348584032009501,
      "grad_norm": 0.0,
      "learning_rate": 4.654809201317706e-06,
      "loss": 0.0,
      "step": 37830
    },
    {
      "epoch": 0.534999787922917,
      "grad_norm": 0.0,
      "learning_rate": 4.6533953540980365e-06,
      "loss": 0.0,
      "step": 37840
    },
    {
      "epoch": 0.535141172644884,
      "grad_norm": 0.0,
      "learning_rate": 4.6519815068783665e-06,
      "loss": 0.0002,
      "step": 37850
    },
    {
      "epoch": 0.5352825573668509,
      "grad_norm": 0.0,
      "learning_rate": 4.650567659658697e-06,
      "loss": 0.0,
      "step": 37860
    },
    {
      "epoch": 0.5354239420888178,
      "grad_norm": 0.0,
      "learning_rate": 4.649153812439028e-06,
      "loss": 0.0,
      "step": 37870
    },
    {
      "epoch": 0.5355653268107848,
      "grad_norm": 0.002233438193798065,
      "learning_rate": 4.647739965219359e-06,
      "loss": 0.0,
      "step": 37880
    },
    {
      "epoch": 0.5357067115327517,
      "grad_norm": 3.6709217887676004e-09,
      "learning_rate": 4.646326117999689e-06,
      "loss": 0.0,
      "step": 37890
    },
    {
      "epoch": 0.5358480962547187,
      "grad_norm": 0.0,
      "learning_rate": 4.64491227078002e-06,
      "loss": 0.0,
      "step": 37900
    },
    {
      "epoch": 0.5359894809766856,
      "grad_norm": 25.50032615661621,
      "learning_rate": 4.64349842356035e-06,
      "loss": 0.0064,
      "step": 37910
    },
    {
      "epoch": 0.5361308656986526,
      "grad_norm": 0.09119933098554611,
      "learning_rate": 4.642084576340681e-06,
      "loss": 0.0,
      "step": 37920
    },
    {
      "epoch": 0.5362722504206195,
      "grad_norm": 2.877994802474859e-07,
      "learning_rate": 4.640670729121012e-06,
      "loss": 0.017,
      "step": 37930
    },
    {
      "epoch": 0.5364136351425864,
      "grad_norm": 0.0,
      "learning_rate": 4.6392568819013425e-06,
      "loss": 0.0,
      "step": 37940
    },
    {
      "epoch": 0.5365550198645535,
      "grad_norm": 7.403077262324587e-09,
      "learning_rate": 4.6378430346816726e-06,
      "loss": 0.0,
      "step": 37950
    },
    {
      "epoch": 0.5366964045865203,
      "grad_norm": 0.0,
      "learning_rate": 4.636429187462003e-06,
      "loss": 0.0001,
      "step": 37960
    },
    {
      "epoch": 0.5368377893084874,
      "grad_norm": 0.0,
      "learning_rate": 4.635015340242333e-06,
      "loss": 0.0095,
      "step": 37970
    },
    {
      "epoch": 0.5369791740304543,
      "grad_norm": 2.1141463548701722e-06,
      "learning_rate": 4.633601493022664e-06,
      "loss": 0.0,
      "step": 37980
    },
    {
      "epoch": 0.5371205587524213,
      "grad_norm": 0.1370207518339157,
      "learning_rate": 4.632187645802995e-06,
      "loss": 0.0,
      "step": 37990
    },
    {
      "epoch": 0.5372619434743882,
      "grad_norm": 0.0,
      "learning_rate": 4.630773798583326e-06,
      "loss": 0.0004,
      "step": 38000
    },
    {
      "epoch": 0.5374033281963551,
      "grad_norm": 0.0,
      "learning_rate": 4.629359951363656e-06,
      "loss": 0.0,
      "step": 38010
    },
    {
      "epoch": 0.5375447129183221,
      "grad_norm": 0.0,
      "learning_rate": 4.627946104143987e-06,
      "loss": 0.0,
      "step": 38020
    },
    {
      "epoch": 0.537686097640289,
      "grad_norm": 0.0,
      "learning_rate": 4.626532256924317e-06,
      "loss": 0.0029,
      "step": 38030
    },
    {
      "epoch": 0.537827482362256,
      "grad_norm": 0.0,
      "learning_rate": 4.625118409704648e-06,
      "loss": 0.2862,
      "step": 38040
    },
    {
      "epoch": 0.5379688670842229,
      "grad_norm": 0.0,
      "learning_rate": 4.623704562484978e-06,
      "loss": 0.01,
      "step": 38050
    },
    {
      "epoch": 0.5381102518061899,
      "grad_norm": 0.0,
      "learning_rate": 4.6222907152653094e-06,
      "loss": 0.0,
      "step": 38060
    },
    {
      "epoch": 0.5382516365281568,
      "grad_norm": 0.0,
      "learning_rate": 4.6208768680456394e-06,
      "loss": 0.0,
      "step": 38070
    },
    {
      "epoch": 0.5383930212501237,
      "grad_norm": 0.0,
      "learning_rate": 4.61946302082597e-06,
      "loss": 0.0,
      "step": 38080
    },
    {
      "epoch": 0.5385344059720907,
      "grad_norm": 0.0,
      "learning_rate": 4.6180491736063e-06,
      "loss": 0.0001,
      "step": 38090
    },
    {
      "epoch": 0.5386757906940576,
      "grad_norm": 0.0,
      "learning_rate": 4.616635326386631e-06,
      "loss": 0.0,
      "step": 38100
    },
    {
      "epoch": 0.5388171754160246,
      "grad_norm": 0.0,
      "learning_rate": 4.615221479166961e-06,
      "loss": 0.0037,
      "step": 38110
    },
    {
      "epoch": 0.5389585601379915,
      "grad_norm": 0.0,
      "learning_rate": 4.613807631947292e-06,
      "loss": 0.0137,
      "step": 38120
    },
    {
      "epoch": 0.5390999448599584,
      "grad_norm": 0.0,
      "learning_rate": 4.612393784727623e-06,
      "loss": 0.0,
      "step": 38130
    },
    {
      "epoch": 0.5392413295819254,
      "grad_norm": 0.0,
      "learning_rate": 4.610979937507954e-06,
      "loss": 0.0,
      "step": 38140
    },
    {
      "epoch": 0.5393827143038923,
      "grad_norm": 7.44171835265206e-09,
      "learning_rate": 4.609566090288284e-06,
      "loss": 0.0,
      "step": 38150
    },
    {
      "epoch": 0.5395240990258593,
      "grad_norm": 0.0,
      "learning_rate": 4.608152243068615e-06,
      "loss": 0.0,
      "step": 38160
    },
    {
      "epoch": 0.5396654837478262,
      "grad_norm": 0.0,
      "learning_rate": 4.606738395848945e-06,
      "loss": 0.0007,
      "step": 38170
    },
    {
      "epoch": 0.5398068684697932,
      "grad_norm": 1.3598235845565796,
      "learning_rate": 4.6053245486292755e-06,
      "loss": 0.0004,
      "step": 38180
    },
    {
      "epoch": 0.5399482531917601,
      "grad_norm": 0.0,
      "learning_rate": 4.603910701409606e-06,
      "loss": 0.0034,
      "step": 38190
    },
    {
      "epoch": 0.540089637913727,
      "grad_norm": 0.0,
      "learning_rate": 4.602496854189936e-06,
      "loss": 0.0,
      "step": 38200
    },
    {
      "epoch": 0.540231022635694,
      "grad_norm": 0.0,
      "learning_rate": 4.601083006970267e-06,
      "loss": 0.0,
      "step": 38210
    },
    {
      "epoch": 0.5403724073576609,
      "grad_norm": 0.0,
      "learning_rate": 4.599669159750598e-06,
      "loss": 0.0006,
      "step": 38220
    },
    {
      "epoch": 0.5405137920796279,
      "grad_norm": 4.061580405334553e-09,
      "learning_rate": 4.598255312530928e-06,
      "loss": 0.0,
      "step": 38230
    },
    {
      "epoch": 0.5406551768015948,
      "grad_norm": 0.0,
      "learning_rate": 4.596841465311259e-06,
      "loss": 0.0,
      "step": 38240
    },
    {
      "epoch": 0.5407965615235618,
      "grad_norm": 0.00013333620154298842,
      "learning_rate": 4.59542761809159e-06,
      "loss": 0.0,
      "step": 38250
    },
    {
      "epoch": 0.5409379462455287,
      "grad_norm": 0.0,
      "learning_rate": 4.59401377087192e-06,
      "loss": 0.4148,
      "step": 38260
    },
    {
      "epoch": 0.5410793309674956,
      "grad_norm": 1.138042193815636e-06,
      "learning_rate": 4.592599923652251e-06,
      "loss": 0.0,
      "step": 38270
    },
    {
      "epoch": 0.5412207156894626,
      "grad_norm": 0.0,
      "learning_rate": 4.591186076432581e-06,
      "loss": 0.0004,
      "step": 38280
    },
    {
      "epoch": 0.5413621004114295,
      "grad_norm": 3.553984218029882e-09,
      "learning_rate": 4.5897722292129115e-06,
      "loss": 0.0,
      "step": 38290
    },
    {
      "epoch": 0.5415034851333965,
      "grad_norm": 0.0,
      "learning_rate": 4.588358381993242e-06,
      "loss": 0.0003,
      "step": 38300
    },
    {
      "epoch": 0.5416448698553634,
      "grad_norm": 0.0,
      "learning_rate": 4.586944534773572e-06,
      "loss": 0.001,
      "step": 38310
    },
    {
      "epoch": 0.5417862545773304,
      "grad_norm": 2.755059469450316e-08,
      "learning_rate": 4.585530687553903e-06,
      "loss": 0.0,
      "step": 38320
    },
    {
      "epoch": 0.5419276392992973,
      "grad_norm": 0.0,
      "learning_rate": 4.584116840334234e-06,
      "loss": 0.0,
      "step": 38330
    },
    {
      "epoch": 0.5420690240212642,
      "grad_norm": 0.0,
      "learning_rate": 4.582702993114564e-06,
      "loss": 0.0,
      "step": 38340
    },
    {
      "epoch": 0.5422104087432312,
      "grad_norm": 0.0,
      "learning_rate": 4.581289145894895e-06,
      "loss": 0.0,
      "step": 38350
    },
    {
      "epoch": 0.5423517934651981,
      "grad_norm": 0.29293301701545715,
      "learning_rate": 4.579875298675226e-06,
      "loss": 0.0,
      "step": 38360
    },
    {
      "epoch": 0.5424931781871651,
      "grad_norm": 0.0,
      "learning_rate": 4.578461451455556e-06,
      "loss": 0.0013,
      "step": 38370
    },
    {
      "epoch": 0.542634562909132,
      "grad_norm": 0.0,
      "learning_rate": 4.577047604235887e-06,
      "loss": 0.0,
      "step": 38380
    },
    {
      "epoch": 0.5427759476310989,
      "grad_norm": 0.0,
      "learning_rate": 4.5756337570162175e-06,
      "loss": 0.0,
      "step": 38390
    },
    {
      "epoch": 0.5429173323530659,
      "grad_norm": 0.11080855876207352,
      "learning_rate": 4.5742199097965475e-06,
      "loss": 0.0,
      "step": 38400
    },
    {
      "epoch": 0.5430587170750328,
      "grad_norm": 0.0,
      "learning_rate": 4.572806062576878e-06,
      "loss": 0.0,
      "step": 38410
    },
    {
      "epoch": 0.5432001017969998,
      "grad_norm": 0.0,
      "learning_rate": 4.571392215357208e-06,
      "loss": 0.0,
      "step": 38420
    },
    {
      "epoch": 0.5433414865189667,
      "grad_norm": 0.0,
      "learning_rate": 4.569978368137539e-06,
      "loss": 0.0,
      "step": 38430
    },
    {
      "epoch": 0.5434828712409338,
      "grad_norm": 0.0,
      "learning_rate": 4.56856452091787e-06,
      "loss": 0.4266,
      "step": 38440
    },
    {
      "epoch": 0.5436242559629006,
      "grad_norm": 0.0,
      "learning_rate": 4.567150673698201e-06,
      "loss": 0.0112,
      "step": 38450
    },
    {
      "epoch": 0.5437656406848675,
      "grad_norm": 0.0,
      "learning_rate": 4.565736826478531e-06,
      "loss": 0.0,
      "step": 38460
    },
    {
      "epoch": 0.5439070254068346,
      "grad_norm": 0.00015319841622840613,
      "learning_rate": 4.564322979258862e-06,
      "loss": 0.0,
      "step": 38470
    },
    {
      "epoch": 0.5440484101288015,
      "grad_norm": 0.0,
      "learning_rate": 4.562909132039192e-06,
      "loss": 0.0,
      "step": 38480
    },
    {
      "epoch": 0.5441897948507685,
      "grad_norm": 3.892547617567743e-09,
      "learning_rate": 4.561495284819523e-06,
      "loss": 0.0,
      "step": 38490
    },
    {
      "epoch": 0.5443311795727354,
      "grad_norm": 2.2543656541529344e-06,
      "learning_rate": 4.560081437599853e-06,
      "loss": 0.0,
      "step": 38500
    },
    {
      "epoch": 0.5444725642947024,
      "grad_norm": 0.0,
      "learning_rate": 4.558667590380184e-06,
      "loss": 0.0,
      "step": 38510
    },
    {
      "epoch": 0.5446139490166693,
      "grad_norm": 0.0,
      "learning_rate": 4.5572537431605144e-06,
      "loss": 0.0,
      "step": 38520
    },
    {
      "epoch": 0.5447553337386362,
      "grad_norm": 0.003527140012010932,
      "learning_rate": 4.555839895940845e-06,
      "loss": 0.0,
      "step": 38530
    },
    {
      "epoch": 0.5448967184606032,
      "grad_norm": 443.0439147949219,
      "learning_rate": 4.554426048721175e-06,
      "loss": 0.1511,
      "step": 38540
    },
    {
      "epoch": 0.5450381031825701,
      "grad_norm": 1.1749306167985196e-06,
      "learning_rate": 4.553012201501506e-06,
      "loss": 0.0,
      "step": 38550
    },
    {
      "epoch": 0.5451794879045371,
      "grad_norm": 0.0,
      "learning_rate": 4.551598354281836e-06,
      "loss": 0.0,
      "step": 38560
    },
    {
      "epoch": 0.545320872626504,
      "grad_norm": 3.6739018582920835e-07,
      "learning_rate": 4.550184507062167e-06,
      "loss": 0.1336,
      "step": 38570
    },
    {
      "epoch": 0.5454622573484709,
      "grad_norm": 0.0,
      "learning_rate": 4.548770659842498e-06,
      "loss": 0.0,
      "step": 38580
    },
    {
      "epoch": 0.5456036420704379,
      "grad_norm": 2.3306771254283376e-05,
      "learning_rate": 4.547356812622829e-06,
      "loss": 0.0,
      "step": 38590
    },
    {
      "epoch": 0.5457450267924048,
      "grad_norm": 0.14973807334899902,
      "learning_rate": 4.545942965403159e-06,
      "loss": 0.0,
      "step": 38600
    },
    {
      "epoch": 0.5458864115143718,
      "grad_norm": 0.0,
      "learning_rate": 4.54452911818349e-06,
      "loss": 0.0003,
      "step": 38610
    },
    {
      "epoch": 0.5460277962363387,
      "grad_norm": 0.0,
      "learning_rate": 4.54311527096382e-06,
      "loss": 0.0,
      "step": 38620
    },
    {
      "epoch": 0.5461691809583057,
      "grad_norm": 0.0,
      "learning_rate": 4.5417014237441505e-06,
      "loss": 0.0,
      "step": 38630
    },
    {
      "epoch": 0.5463105656802726,
      "grad_norm": 4.216592408567976e-09,
      "learning_rate": 4.540287576524481e-06,
      "loss": 0.0,
      "step": 38640
    },
    {
      "epoch": 0.5464519504022395,
      "grad_norm": 6.956308789085597e-05,
      "learning_rate": 4.538873729304812e-06,
      "loss": 0.0,
      "step": 38650
    },
    {
      "epoch": 0.5465933351242065,
      "grad_norm": 666.5427856445312,
      "learning_rate": 4.537459882085142e-06,
      "loss": 0.5033,
      "step": 38660
    },
    {
      "epoch": 0.5467347198461734,
      "grad_norm": 0.0012233025627210736,
      "learning_rate": 4.536046034865473e-06,
      "loss": 0.0,
      "step": 38670
    },
    {
      "epoch": 0.5468761045681404,
      "grad_norm": 0.0,
      "learning_rate": 4.534632187645803e-06,
      "loss": 0.0057,
      "step": 38680
    },
    {
      "epoch": 0.5470174892901073,
      "grad_norm": 0.0,
      "learning_rate": 4.533218340426134e-06,
      "loss": 0.0003,
      "step": 38690
    },
    {
      "epoch": 0.5471588740120743,
      "grad_norm": 0.0,
      "learning_rate": 4.531804493206464e-06,
      "loss": 0.0,
      "step": 38700
    },
    {
      "epoch": 0.5473002587340412,
      "grad_norm": 0.0,
      "learning_rate": 4.530390645986796e-06,
      "loss": 0.0,
      "step": 38710
    },
    {
      "epoch": 0.5474416434560081,
      "grad_norm": 8.559860731338631e-08,
      "learning_rate": 4.528976798767126e-06,
      "loss": 0.0,
      "step": 38720
    },
    {
      "epoch": 0.5475830281779751,
      "grad_norm": 0.16088107228279114,
      "learning_rate": 4.5275629515474565e-06,
      "loss": 0.4993,
      "step": 38730
    },
    {
      "epoch": 0.547724412899942,
      "grad_norm": 0.005933794192969799,
      "learning_rate": 4.5261491043277865e-06,
      "loss": 0.0,
      "step": 38740
    },
    {
      "epoch": 0.547865797621909,
      "grad_norm": 0.0,
      "learning_rate": 4.524735257108117e-06,
      "loss": 0.1478,
      "step": 38750
    },
    {
      "epoch": 0.5480071823438759,
      "grad_norm": 1.058197085512802e-08,
      "learning_rate": 4.523321409888447e-06,
      "loss": 0.2428,
      "step": 38760
    },
    {
      "epoch": 0.5481485670658429,
      "grad_norm": 0.0,
      "learning_rate": 4.521907562668778e-06,
      "loss": 0.0,
      "step": 38770
    },
    {
      "epoch": 0.5482899517878098,
      "grad_norm": 7.406245276797563e-05,
      "learning_rate": 4.520493715449109e-06,
      "loss": 0.0,
      "step": 38780
    },
    {
      "epoch": 0.5484313365097767,
      "grad_norm": 0.004862528759986162,
      "learning_rate": 4.51907986822944e-06,
      "loss": 0.0002,
      "step": 38790
    },
    {
      "epoch": 0.5485727212317437,
      "grad_norm": 0.0,
      "learning_rate": 4.51766602100977e-06,
      "loss": 0.0074,
      "step": 38800
    },
    {
      "epoch": 0.5487141059537106,
      "grad_norm": 0.0,
      "learning_rate": 4.516252173790101e-06,
      "loss": 0.0,
      "step": 38810
    },
    {
      "epoch": 0.5488554906756776,
      "grad_norm": 1.7428828380516848e-09,
      "learning_rate": 4.514838326570431e-06,
      "loss": 0.0,
      "step": 38820
    },
    {
      "epoch": 0.5489968753976445,
      "grad_norm": 0.0,
      "learning_rate": 4.513424479350762e-06,
      "loss": 0.7719,
      "step": 38830
    },
    {
      "epoch": 0.5491382601196114,
      "grad_norm": 0.0,
      "learning_rate": 4.5120106321310925e-06,
      "loss": 0.0,
      "step": 38840
    },
    {
      "epoch": 0.5492796448415784,
      "grad_norm": 1.994693077378429e-09,
      "learning_rate": 4.5105967849114225e-06,
      "loss": 0.0,
      "step": 38850
    },
    {
      "epoch": 0.5494210295635453,
      "grad_norm": 0.0,
      "learning_rate": 4.509182937691753e-06,
      "loss": 0.0,
      "step": 38860
    },
    {
      "epoch": 0.5495624142855123,
      "grad_norm": 0.0006632870645262301,
      "learning_rate": 4.507769090472084e-06,
      "loss": 0.0,
      "step": 38870
    },
    {
      "epoch": 0.5497037990074792,
      "grad_norm": 0.0,
      "learning_rate": 4.506355243252414e-06,
      "loss": 0.0002,
      "step": 38880
    },
    {
      "epoch": 0.5498451837294462,
      "grad_norm": 0.0,
      "learning_rate": 4.504941396032745e-06,
      "loss": 0.0,
      "step": 38890
    },
    {
      "epoch": 0.5499865684514131,
      "grad_norm": 0.0,
      "learning_rate": 4.503527548813076e-06,
      "loss": 0.0,
      "step": 38900
    },
    {
      "epoch": 0.55012795317338,
      "grad_norm": 0.0,
      "learning_rate": 4.502113701593406e-06,
      "loss": 0.0,
      "step": 38910
    },
    {
      "epoch": 0.550269337895347,
      "grad_norm": 1.392085414408939e-05,
      "learning_rate": 4.500699854373737e-06,
      "loss": 0.0,
      "step": 38920
    },
    {
      "epoch": 0.5504107226173139,
      "grad_norm": 0.0,
      "learning_rate": 4.499286007154068e-06,
      "loss": 0.0,
      "step": 38930
    },
    {
      "epoch": 0.550552107339281,
      "grad_norm": 0.0,
      "learning_rate": 4.497872159934398e-06,
      "loss": 0.0372,
      "step": 38940
    },
    {
      "epoch": 0.5506934920612478,
      "grad_norm": 0.0,
      "learning_rate": 4.4964583127147286e-06,
      "loss": 0.0001,
      "step": 38950
    },
    {
      "epoch": 0.5508348767832149,
      "grad_norm": 0.0,
      "learning_rate": 4.495044465495059e-06,
      "loss": 0.0,
      "step": 38960
    },
    {
      "epoch": 0.5509762615051818,
      "grad_norm": 0.0,
      "learning_rate": 4.493630618275389e-06,
      "loss": 0.0,
      "step": 38970
    },
    {
      "epoch": 0.5511176462271486,
      "grad_norm": 0.0,
      "learning_rate": 4.49221677105572e-06,
      "loss": 0.0,
      "step": 38980
    },
    {
      "epoch": 0.5512590309491157,
      "grad_norm": 0.0,
      "learning_rate": 4.49080292383605e-06,
      "loss": 0.0,
      "step": 38990
    },
    {
      "epoch": 0.5514004156710826,
      "grad_norm": 0.0,
      "learning_rate": 4.489389076616381e-06,
      "loss": 0.0297,
      "step": 39000
    },
    {
      "epoch": 0.5515418003930496,
      "grad_norm": 0.0,
      "learning_rate": 4.487975229396712e-06,
      "loss": 0.0,
      "step": 39010
    },
    {
      "epoch": 0.5516831851150165,
      "grad_norm": 0.0,
      "learning_rate": 4.486561382177042e-06,
      "loss": 0.0,
      "step": 39020
    },
    {
      "epoch": 0.5518245698369835,
      "grad_norm": 0.0,
      "learning_rate": 4.485147534957373e-06,
      "loss": 0.0,
      "step": 39030
    },
    {
      "epoch": 0.5519659545589504,
      "grad_norm": 0.0,
      "learning_rate": 4.483733687737704e-06,
      "loss": 0.0,
      "step": 39040
    },
    {
      "epoch": 0.5521073392809173,
      "grad_norm": 0.0,
      "learning_rate": 4.482319840518034e-06,
      "loss": 0.0,
      "step": 39050
    },
    {
      "epoch": 0.5522487240028843,
      "grad_norm": 0.0,
      "learning_rate": 4.480905993298365e-06,
      "loss": 0.0,
      "step": 39060
    },
    {
      "epoch": 0.5523901087248512,
      "grad_norm": 9.247423804481514e-08,
      "learning_rate": 4.479492146078695e-06,
      "loss": 0.0,
      "step": 39070
    },
    {
      "epoch": 0.5525314934468182,
      "grad_norm": 0.001767209148965776,
      "learning_rate": 4.4780782988590255e-06,
      "loss": 0.0,
      "step": 39080
    },
    {
      "epoch": 0.5526728781687851,
      "grad_norm": 0.0,
      "learning_rate": 4.476664451639356e-06,
      "loss": 0.0,
      "step": 39090
    },
    {
      "epoch": 0.552814262890752,
      "grad_norm": 0.06833095848560333,
      "learning_rate": 4.475250604419687e-06,
      "loss": 0.0,
      "step": 39100
    },
    {
      "epoch": 0.552955647612719,
      "grad_norm": 0.0,
      "learning_rate": 4.473836757200017e-06,
      "loss": 0.0055,
      "step": 39110
    },
    {
      "epoch": 0.5530970323346859,
      "grad_norm": 0.0,
      "learning_rate": 4.472422909980348e-06,
      "loss": 0.0002,
      "step": 39120
    },
    {
      "epoch": 0.5532384170566529,
      "grad_norm": 0.0,
      "learning_rate": 4.471009062760678e-06,
      "loss": 0.0,
      "step": 39130
    },
    {
      "epoch": 0.5533798017786198,
      "grad_norm": 0.0,
      "learning_rate": 4.469595215541009e-06,
      "loss": 0.0248,
      "step": 39140
    },
    {
      "epoch": 0.5535211865005868,
      "grad_norm": 2.7930227020078746e-07,
      "learning_rate": 4.468181368321339e-06,
      "loss": 0.0,
      "step": 39150
    },
    {
      "epoch": 0.5536625712225537,
      "grad_norm": 2.1912715965299867e-05,
      "learning_rate": 4.466767521101671e-06,
      "loss": 0.0,
      "step": 39160
    },
    {
      "epoch": 0.5538039559445206,
      "grad_norm": 0.0005949631449766457,
      "learning_rate": 4.465353673882001e-06,
      "loss": 0.0,
      "step": 39170
    },
    {
      "epoch": 0.5539453406664876,
      "grad_norm": 1.8866144202434043e-09,
      "learning_rate": 4.4639398266623315e-06,
      "loss": 0.0001,
      "step": 39180
    },
    {
      "epoch": 0.5540867253884545,
      "grad_norm": 0.0,
      "learning_rate": 4.4625259794426615e-06,
      "loss": 0.0,
      "step": 39190
    },
    {
      "epoch": 0.5542281101104215,
      "grad_norm": 0.0,
      "learning_rate": 4.461112132222992e-06,
      "loss": 0.0,
      "step": 39200
    },
    {
      "epoch": 0.5543694948323884,
      "grad_norm": 1.792037851444661e-09,
      "learning_rate": 4.459698285003322e-06,
      "loss": 0.0,
      "step": 39210
    },
    {
      "epoch": 0.5545108795543554,
      "grad_norm": 0.0,
      "learning_rate": 4.458284437783654e-06,
      "loss": 0.0001,
      "step": 39220
    },
    {
      "epoch": 0.5546522642763223,
      "grad_norm": 0.0,
      "learning_rate": 4.456870590563984e-06,
      "loss": 0.0,
      "step": 39230
    },
    {
      "epoch": 0.5547936489982892,
      "grad_norm": 0.0,
      "learning_rate": 4.455456743344315e-06,
      "loss": 0.0,
      "step": 39240
    },
    {
      "epoch": 0.5549350337202562,
      "grad_norm": 0.0,
      "learning_rate": 4.454042896124645e-06,
      "loss": 0.0,
      "step": 39250
    },
    {
      "epoch": 0.5550764184422231,
      "grad_norm": 0.0001754987461026758,
      "learning_rate": 4.452629048904976e-06,
      "loss": 0.0002,
      "step": 39260
    },
    {
      "epoch": 0.5552178031641901,
      "grad_norm": 0.0,
      "learning_rate": 4.451215201685306e-06,
      "loss": 0.0,
      "step": 39270
    },
    {
      "epoch": 0.555359187886157,
      "grad_norm": 0.0,
      "learning_rate": 4.449801354465637e-06,
      "loss": 0.0,
      "step": 39280
    },
    {
      "epoch": 0.555500572608124,
      "grad_norm": 0.0,
      "learning_rate": 4.4483875072459675e-06,
      "loss": 0.0,
      "step": 39290
    },
    {
      "epoch": 0.5556419573300909,
      "grad_norm": 0.0,
      "learning_rate": 4.446973660026298e-06,
      "loss": 0.0,
      "step": 39300
    },
    {
      "epoch": 0.5557833420520578,
      "grad_norm": 0.0,
      "learning_rate": 4.445559812806628e-06,
      "loss": 0.0024,
      "step": 39310
    },
    {
      "epoch": 0.5559247267740248,
      "grad_norm": 7.294019042092259e-07,
      "learning_rate": 4.444145965586959e-06,
      "loss": 0.0003,
      "step": 39320
    },
    {
      "epoch": 0.5560661114959917,
      "grad_norm": 9.038098505698144e-05,
      "learning_rate": 4.442732118367289e-06,
      "loss": 0.0,
      "step": 39330
    },
    {
      "epoch": 0.5562074962179587,
      "grad_norm": 0.0,
      "learning_rate": 4.44131827114762e-06,
      "loss": 0.0,
      "step": 39340
    },
    {
      "epoch": 0.5563488809399256,
      "grad_norm": 1.077061824616976e-05,
      "learning_rate": 4.439904423927951e-06,
      "loss": 0.0001,
      "step": 39350
    },
    {
      "epoch": 0.5564902656618925,
      "grad_norm": 0.0,
      "learning_rate": 4.438490576708282e-06,
      "loss": 0.0,
      "step": 39360
    },
    {
      "epoch": 0.5566316503838595,
      "grad_norm": 0.0010667084716260433,
      "learning_rate": 4.437076729488612e-06,
      "loss": 0.002,
      "step": 39370
    },
    {
      "epoch": 0.5567730351058264,
      "grad_norm": 22.979541778564453,
      "learning_rate": 4.435662882268943e-06,
      "loss": 0.0034,
      "step": 39380
    },
    {
      "epoch": 0.5569144198277934,
      "grad_norm": 0.0,
      "learning_rate": 4.434249035049273e-06,
      "loss": 0.0,
      "step": 39390
    },
    {
      "epoch": 0.5570558045497603,
      "grad_norm": 0.0,
      "learning_rate": 4.4328351878296035e-06,
      "loss": 1.2203,
      "step": 39400
    },
    {
      "epoch": 0.5571971892717273,
      "grad_norm": 0.0,
      "learning_rate": 4.4314213406099336e-06,
      "loss": 0.0022,
      "step": 39410
    },
    {
      "epoch": 0.5573385739936942,
      "grad_norm": 6.322505669231759e-06,
      "learning_rate": 4.430007493390264e-06,
      "loss": 0.0,
      "step": 39420
    },
    {
      "epoch": 0.5574799587156611,
      "grad_norm": 0.0,
      "learning_rate": 4.428593646170595e-06,
      "loss": 0.0,
      "step": 39430
    },
    {
      "epoch": 0.5576213434376281,
      "grad_norm": 0.0,
      "learning_rate": 4.427179798950926e-06,
      "loss": 0.0011,
      "step": 39440
    },
    {
      "epoch": 0.557762728159595,
      "grad_norm": 0.0,
      "learning_rate": 4.425765951731256e-06,
      "loss": 0.0,
      "step": 39450
    },
    {
      "epoch": 0.557904112881562,
      "grad_norm": 0.0,
      "learning_rate": 4.424352104511587e-06,
      "loss": 0.0,
      "step": 39460
    },
    {
      "epoch": 0.558045497603529,
      "grad_norm": 0.0,
      "learning_rate": 4.422938257291917e-06,
      "loss": 0.0,
      "step": 39470
    },
    {
      "epoch": 0.558186882325496,
      "grad_norm": 0.0,
      "learning_rate": 4.421524410072248e-06,
      "loss": 0.0,
      "step": 39480
    },
    {
      "epoch": 0.5583282670474629,
      "grad_norm": 0.0,
      "learning_rate": 4.420110562852579e-06,
      "loss": 0.0,
      "step": 39490
    },
    {
      "epoch": 0.5584696517694298,
      "grad_norm": 0.0,
      "learning_rate": 4.418696715632909e-06,
      "loss": 0.0,
      "step": 39500
    },
    {
      "epoch": 0.5586110364913968,
      "grad_norm": 0.0,
      "learning_rate": 4.4172828684132396e-06,
      "loss": 0.0,
      "step": 39510
    },
    {
      "epoch": 0.5587524212133637,
      "grad_norm": 0.0,
      "learning_rate": 4.4158690211935704e-06,
      "loss": 0.0429,
      "step": 39520
    },
    {
      "epoch": 0.5588938059353307,
      "grad_norm": 8.607823787087909e-08,
      "learning_rate": 4.4144551739739004e-06,
      "loss": 0.0,
      "step": 39530
    },
    {
      "epoch": 0.5590351906572976,
      "grad_norm": 0.0,
      "learning_rate": 4.413041326754231e-06,
      "loss": 0.0,
      "step": 39540
    },
    {
      "epoch": 0.5591765753792645,
      "grad_norm": 0.0,
      "learning_rate": 4.411627479534562e-06,
      "loss": 0.0,
      "step": 39550
    },
    {
      "epoch": 0.5593179601012315,
      "grad_norm": 0.0,
      "learning_rate": 4.410213632314892e-06,
      "loss": 0.0678,
      "step": 39560
    },
    {
      "epoch": 0.5594593448231984,
      "grad_norm": 3.2088105399452616e-06,
      "learning_rate": 4.408799785095223e-06,
      "loss": 0.4822,
      "step": 39570
    },
    {
      "epoch": 0.5596007295451654,
      "grad_norm": 0.0,
      "learning_rate": 4.407385937875554e-06,
      "loss": 0.0,
      "step": 39580
    },
    {
      "epoch": 0.5597421142671323,
      "grad_norm": 0.0,
      "learning_rate": 4.405972090655884e-06,
      "loss": 0.0004,
      "step": 39590
    },
    {
      "epoch": 0.5598834989890993,
      "grad_norm": 6.142014399301843e-07,
      "learning_rate": 4.404558243436215e-06,
      "loss": 0.6789,
      "step": 39600
    },
    {
      "epoch": 0.5600248837110662,
      "grad_norm": 0.020278289914131165,
      "learning_rate": 4.403144396216546e-06,
      "loss": 0.0003,
      "step": 39610
    },
    {
      "epoch": 0.5601662684330331,
      "grad_norm": 1.9980084253745645e-09,
      "learning_rate": 4.401730548996876e-06,
      "loss": 0.0,
      "step": 39620
    },
    {
      "epoch": 0.5603076531550001,
      "grad_norm": 0.0,
      "learning_rate": 4.4003167017772065e-06,
      "loss": 0.0,
      "step": 39630
    },
    {
      "epoch": 0.560449037876967,
      "grad_norm": 0.0,
      "learning_rate": 4.3989028545575365e-06,
      "loss": 0.0,
      "step": 39640
    },
    {
      "epoch": 0.560590422598934,
      "grad_norm": 0.2776833176612854,
      "learning_rate": 4.397489007337867e-06,
      "loss": 0.0001,
      "step": 39650
    },
    {
      "epoch": 0.5607318073209009,
      "grad_norm": 0.0,
      "learning_rate": 4.396075160118198e-06,
      "loss": 0.0,
      "step": 39660
    },
    {
      "epoch": 0.5608731920428679,
      "grad_norm": 0.0,
      "learning_rate": 4.394661312898529e-06,
      "loss": 0.0,
      "step": 39670
    },
    {
      "epoch": 0.5610145767648348,
      "grad_norm": 0.0,
      "learning_rate": 4.393247465678859e-06,
      "loss": 0.0028,
      "step": 39680
    },
    {
      "epoch": 0.5611559614868017,
      "grad_norm": 0.0,
      "learning_rate": 4.39183361845919e-06,
      "loss": 0.0,
      "step": 39690
    },
    {
      "epoch": 0.5612973462087687,
      "grad_norm": 0.0,
      "learning_rate": 4.39041977123952e-06,
      "loss": 0.0,
      "step": 39700
    },
    {
      "epoch": 0.5614387309307356,
      "grad_norm": 0.08888682723045349,
      "learning_rate": 4.389005924019851e-06,
      "loss": 0.0,
      "step": 39710
    },
    {
      "epoch": 0.5615801156527026,
      "grad_norm": 0.0,
      "learning_rate": 4.387592076800181e-06,
      "loss": 0.0,
      "step": 39720
    },
    {
      "epoch": 0.5617215003746695,
      "grad_norm": 0.0027801182586699724,
      "learning_rate": 4.386178229580512e-06,
      "loss": 0.139,
      "step": 39730
    },
    {
      "epoch": 0.5618628850966365,
      "grad_norm": 0.0,
      "learning_rate": 4.3847643823608425e-06,
      "loss": 0.0,
      "step": 39740
    },
    {
      "epoch": 0.5620042698186034,
      "grad_norm": 1.1542295652589019e-07,
      "learning_rate": 4.383350535141173e-06,
      "loss": 0.0,
      "step": 39750
    },
    {
      "epoch": 0.5621456545405703,
      "grad_norm": 0.0,
      "learning_rate": 4.381936687921503e-06,
      "loss": 0.0,
      "step": 39760
    },
    {
      "epoch": 0.5622870392625373,
      "grad_norm": 1.694827611586902e-09,
      "learning_rate": 4.380522840701834e-06,
      "loss": 0.0,
      "step": 39770
    },
    {
      "epoch": 0.5624284239845042,
      "grad_norm": 0.0,
      "learning_rate": 4.379108993482164e-06,
      "loss": 0.0,
      "step": 39780
    },
    {
      "epoch": 0.5625698087064712,
      "grad_norm": 0.0,
      "learning_rate": 4.377695146262495e-06,
      "loss": 0.0,
      "step": 39790
    },
    {
      "epoch": 0.5627111934284381,
      "grad_norm": 0.0,
      "learning_rate": 4.376281299042826e-06,
      "loss": 0.0,
      "step": 39800
    },
    {
      "epoch": 0.562852578150405,
      "grad_norm": 0.0,
      "learning_rate": 4.374867451823157e-06,
      "loss": 0.0,
      "step": 39810
    },
    {
      "epoch": 0.562993962872372,
      "grad_norm": 0.0,
      "learning_rate": 4.373453604603487e-06,
      "loss": 0.0,
      "step": 39820
    },
    {
      "epoch": 0.5631353475943389,
      "grad_norm": 0.0,
      "learning_rate": 4.372039757383818e-06,
      "loss": 0.0,
      "step": 39830
    },
    {
      "epoch": 0.5632767323163059,
      "grad_norm": 0.0,
      "learning_rate": 4.370625910164148e-06,
      "loss": 0.0,
      "step": 39840
    },
    {
      "epoch": 0.5634181170382728,
      "grad_norm": 4.022752353449732e-09,
      "learning_rate": 4.3692120629444785e-06,
      "loss": 0.0,
      "step": 39850
    },
    {
      "epoch": 0.5635595017602398,
      "grad_norm": 0.0,
      "learning_rate": 4.3677982157248085e-06,
      "loss": 0.0,
      "step": 39860
    },
    {
      "epoch": 0.5637008864822067,
      "grad_norm": 0.10267210751771927,
      "learning_rate": 4.36638436850514e-06,
      "loss": 0.0002,
      "step": 39870
    },
    {
      "epoch": 0.5638422712041736,
      "grad_norm": 0.0,
      "learning_rate": 4.36497052128547e-06,
      "loss": 0.0,
      "step": 39880
    },
    {
      "epoch": 0.5639836559261406,
      "grad_norm": 6.115643918747082e-05,
      "learning_rate": 4.363556674065801e-06,
      "loss": 0.0002,
      "step": 39890
    },
    {
      "epoch": 0.5641250406481075,
      "grad_norm": 0.0,
      "learning_rate": 4.362142826846131e-06,
      "loss": 0.0,
      "step": 39900
    },
    {
      "epoch": 0.5642664253700745,
      "grad_norm": 0.0,
      "learning_rate": 4.360728979626462e-06,
      "loss": 0.0,
      "step": 39910
    },
    {
      "epoch": 0.5644078100920414,
      "grad_norm": 1.9040267140724154e-09,
      "learning_rate": 4.359315132406792e-06,
      "loss": 0.0,
      "step": 39920
    },
    {
      "epoch": 0.5645491948140084,
      "grad_norm": 5.434164449980017e-06,
      "learning_rate": 4.357901285187124e-06,
      "loss": 0.0001,
      "step": 39930
    },
    {
      "epoch": 0.5646905795359753,
      "grad_norm": 0.0,
      "learning_rate": 4.356487437967454e-06,
      "loss": 0.0,
      "step": 39940
    },
    {
      "epoch": 0.5648319642579422,
      "grad_norm": 0.0,
      "learning_rate": 4.3550735907477846e-06,
      "loss": 0.0,
      "step": 39950
    },
    {
      "epoch": 0.5649733489799093,
      "grad_norm": 2.242390655737836e-05,
      "learning_rate": 4.3536597435281146e-06,
      "loss": 0.5113,
      "step": 39960
    },
    {
      "epoch": 0.5651147337018761,
      "grad_norm": 1.2082047078365576e-06,
      "learning_rate": 4.352245896308445e-06,
      "loss": 0.0,
      "step": 39970
    },
    {
      "epoch": 0.5652561184238432,
      "grad_norm": 1.6470427510739682e-07,
      "learning_rate": 4.3508320490887754e-06,
      "loss": 0.0004,
      "step": 39980
    },
    {
      "epoch": 0.56539750314581,
      "grad_norm": 0.0,
      "learning_rate": 4.349418201869106e-06,
      "loss": 0.0,
      "step": 39990
    },
    {
      "epoch": 0.5655388878677771,
      "grad_norm": 0.0,
      "learning_rate": 4.348004354649437e-06,
      "loss": 0.0097,
      "step": 40000
    },
    {
      "epoch": 0.565680272589744,
      "grad_norm": 0.0,
      "learning_rate": 4.346590507429768e-06,
      "loss": 0.0,
      "step": 40010
    },
    {
      "epoch": 0.5658216573117109,
      "grad_norm": 2.0043204074227106e-07,
      "learning_rate": 4.345176660210098e-06,
      "loss": 0.0,
      "step": 40020
    },
    {
      "epoch": 0.5659630420336779,
      "grad_norm": 4.888324383500731e-06,
      "learning_rate": 4.343762812990429e-06,
      "loss": 0.0,
      "step": 40030
    },
    {
      "epoch": 0.5661044267556448,
      "grad_norm": 0.0,
      "learning_rate": 4.342348965770759e-06,
      "loss": 0.2135,
      "step": 40040
    },
    {
      "epoch": 0.5662458114776118,
      "grad_norm": 1.8306659521627466e-09,
      "learning_rate": 4.34093511855109e-06,
      "loss": 0.0003,
      "step": 40050
    },
    {
      "epoch": 0.5663871961995787,
      "grad_norm": 0.0,
      "learning_rate": 4.339521271331421e-06,
      "loss": 0.0,
      "step": 40060
    },
    {
      "epoch": 0.5665285809215456,
      "grad_norm": 0.0,
      "learning_rate": 4.338107424111751e-06,
      "loss": 0.0,
      "step": 40070
    },
    {
      "epoch": 0.5666699656435126,
      "grad_norm": 0.0,
      "learning_rate": 4.3366935768920815e-06,
      "loss": 0.0,
      "step": 40080
    },
    {
      "epoch": 0.5668113503654795,
      "grad_norm": 1.8508741206346713e-09,
      "learning_rate": 4.335279729672412e-06,
      "loss": 0.1564,
      "step": 40090
    },
    {
      "epoch": 0.5669527350874465,
      "grad_norm": 0.0,
      "learning_rate": 4.333865882452742e-06,
      "loss": 0.0,
      "step": 40100
    },
    {
      "epoch": 0.5670941198094134,
      "grad_norm": 0.0,
      "learning_rate": 4.332452035233073e-06,
      "loss": 0.0,
      "step": 40110
    },
    {
      "epoch": 0.5672355045313804,
      "grad_norm": 1.8035241078351305e-09,
      "learning_rate": 4.331038188013403e-06,
      "loss": 0.0,
      "step": 40120
    },
    {
      "epoch": 0.5673768892533473,
      "grad_norm": 2.0850034161412623e-06,
      "learning_rate": 4.329624340793734e-06,
      "loss": 0.0001,
      "step": 40130
    },
    {
      "epoch": 0.5675182739753142,
      "grad_norm": 1.2131470938925304e-08,
      "learning_rate": 4.328210493574065e-06,
      "loss": 0.0147,
      "step": 40140
    },
    {
      "epoch": 0.5676596586972812,
      "grad_norm": 0.0,
      "learning_rate": 4.326796646354396e-06,
      "loss": 0.0057,
      "step": 40150
    },
    {
      "epoch": 0.5678010434192481,
      "grad_norm": 8.357678638049038e-08,
      "learning_rate": 4.325382799134726e-06,
      "loss": 0.0,
      "step": 40160
    },
    {
      "epoch": 0.5679424281412151,
      "grad_norm": 0.0,
      "learning_rate": 4.323968951915057e-06,
      "loss": 0.0,
      "step": 40170
    },
    {
      "epoch": 0.568083812863182,
      "grad_norm": 0.0,
      "learning_rate": 4.322555104695387e-06,
      "loss": 0.0,
      "step": 40180
    },
    {
      "epoch": 0.568225197585149,
      "grad_norm": 0.0,
      "learning_rate": 4.3211412574757175e-06,
      "loss": 0.0,
      "step": 40190
    },
    {
      "epoch": 0.5683665823071159,
      "grad_norm": 0.0,
      "learning_rate": 4.319727410256048e-06,
      "loss": 0.0002,
      "step": 40200
    },
    {
      "epoch": 0.5685079670290828,
      "grad_norm": 0.0,
      "learning_rate": 4.318313563036378e-06,
      "loss": 0.0,
      "step": 40210
    },
    {
      "epoch": 0.5686493517510498,
      "grad_norm": 0.0,
      "learning_rate": 4.316899715816709e-06,
      "loss": 0.0,
      "step": 40220
    },
    {
      "epoch": 0.5687907364730167,
      "grad_norm": 0.0,
      "learning_rate": 4.31548586859704e-06,
      "loss": 0.0,
      "step": 40230
    },
    {
      "epoch": 0.5689321211949837,
      "grad_norm": 1.0988087524310686e-05,
      "learning_rate": 4.31407202137737e-06,
      "loss": 0.0,
      "step": 40240
    },
    {
      "epoch": 0.5690735059169506,
      "grad_norm": 0.0,
      "learning_rate": 4.312658174157701e-06,
      "loss": 0.0,
      "step": 40250
    },
    {
      "epoch": 0.5692148906389175,
      "grad_norm": 0.0009252988966181874,
      "learning_rate": 4.311244326938032e-06,
      "loss": 0.0,
      "step": 40260
    },
    {
      "epoch": 0.5693562753608845,
      "grad_norm": 2.659546589711681e-05,
      "learning_rate": 4.309830479718362e-06,
      "loss": 0.0001,
      "step": 40270
    },
    {
      "epoch": 0.5694976600828514,
      "grad_norm": 1.770320667837666e-09,
      "learning_rate": 4.308416632498693e-06,
      "loss": 0.0,
      "step": 40280
    },
    {
      "epoch": 0.5696390448048184,
      "grad_norm": 0.03546358272433281,
      "learning_rate": 4.307002785279023e-06,
      "loss": 0.0,
      "step": 40290
    },
    {
      "epoch": 0.5697804295267853,
      "grad_norm": 0.0,
      "learning_rate": 4.3055889380593535e-06,
      "loss": 0.0,
      "step": 40300
    },
    {
      "epoch": 0.5699218142487523,
      "grad_norm": 0.08369642496109009,
      "learning_rate": 4.304175090839684e-06,
      "loss": 0.0,
      "step": 40310
    },
    {
      "epoch": 0.5700631989707192,
      "grad_norm": 0.0,
      "learning_rate": 4.302761243620015e-06,
      "loss": 0.0,
      "step": 40320
    },
    {
      "epoch": 0.5702045836926861,
      "grad_norm": 0.0,
      "learning_rate": 4.301347396400345e-06,
      "loss": 0.0,
      "step": 40330
    },
    {
      "epoch": 0.5703459684146531,
      "grad_norm": 0.0,
      "learning_rate": 4.299933549180676e-06,
      "loss": 0.0,
      "step": 40340
    },
    {
      "epoch": 0.57048735313662,
      "grad_norm": 0.0,
      "learning_rate": 4.298519701961006e-06,
      "loss": 0.0001,
      "step": 40350
    },
    {
      "epoch": 0.570628737858587,
      "grad_norm": 0.0,
      "learning_rate": 4.297105854741337e-06,
      "loss": 0.0,
      "step": 40360
    },
    {
      "epoch": 0.5707701225805539,
      "grad_norm": 0.0,
      "learning_rate": 4.295692007521667e-06,
      "loss": 0.0006,
      "step": 40370
    },
    {
      "epoch": 0.5709115073025209,
      "grad_norm": 865.190673828125,
      "learning_rate": 4.294278160301998e-06,
      "loss": 0.3665,
      "step": 40380
    },
    {
      "epoch": 0.5710528920244878,
      "grad_norm": 3.455061914792168e-06,
      "learning_rate": 4.292864313082329e-06,
      "loss": 0.0,
      "step": 40390
    },
    {
      "epoch": 0.5711942767464547,
      "grad_norm": 0.0,
      "learning_rate": 4.2914504658626595e-06,
      "loss": 0.0004,
      "step": 40400
    },
    {
      "epoch": 0.5713356614684217,
      "grad_norm": 0.004169742576777935,
      "learning_rate": 4.2900366186429896e-06,
      "loss": 0.0,
      "step": 40410
    },
    {
      "epoch": 0.5714770461903886,
      "grad_norm": 0.0028192216996103525,
      "learning_rate": 4.28862277142332e-06,
      "loss": 0.0,
      "step": 40420
    },
    {
      "epoch": 0.5716184309123556,
      "grad_norm": 4.8962894652504474e-05,
      "learning_rate": 4.28720892420365e-06,
      "loss": 0.0,
      "step": 40430
    },
    {
      "epoch": 0.5717598156343225,
      "grad_norm": 0.0,
      "learning_rate": 4.285795076983981e-06,
      "loss": 0.0,
      "step": 40440
    },
    {
      "epoch": 0.5719012003562896,
      "grad_norm": 0.0,
      "learning_rate": 4.284381229764312e-06,
      "loss": 0.0,
      "step": 40450
    },
    {
      "epoch": 0.5720425850782564,
      "grad_norm": 0.0,
      "learning_rate": 4.282967382544643e-06,
      "loss": 0.0,
      "step": 40460
    },
    {
      "epoch": 0.5721839698002233,
      "grad_norm": 0.0,
      "learning_rate": 4.281553535324973e-06,
      "loss": 0.0,
      "step": 40470
    },
    {
      "epoch": 0.5723253545221904,
      "grad_norm": 0.0,
      "learning_rate": 4.280139688105304e-06,
      "loss": 0.0001,
      "step": 40480
    },
    {
      "epoch": 0.5724667392441573,
      "grad_norm": 0.0,
      "learning_rate": 4.278725840885634e-06,
      "loss": 0.0,
      "step": 40490
    },
    {
      "epoch": 0.5726081239661243,
      "grad_norm": 0.0,
      "learning_rate": 4.277311993665965e-06,
      "loss": 0.0,
      "step": 40500
    },
    {
      "epoch": 0.5727495086880912,
      "grad_norm": 0.0,
      "learning_rate": 4.2758981464462956e-06,
      "loss": 0.0,
      "step": 40510
    },
    {
      "epoch": 0.572890893410058,
      "grad_norm": 0.00014094682410359383,
      "learning_rate": 4.2744842992266264e-06,
      "loss": 0.045,
      "step": 40520
    },
    {
      "epoch": 0.5730322781320251,
      "grad_norm": 0.0,
      "learning_rate": 4.2730704520069564e-06,
      "loss": 0.003,
      "step": 40530
    },
    {
      "epoch": 0.573173662853992,
      "grad_norm": 1.4433207695674355e-07,
      "learning_rate": 4.271656604787287e-06,
      "loss": 0.1564,
      "step": 40540
    },
    {
      "epoch": 0.573315047575959,
      "grad_norm": 5.698295353795402e-05,
      "learning_rate": 4.270242757567617e-06,
      "loss": 0.0,
      "step": 40550
    },
    {
      "epoch": 0.5734564322979259,
      "grad_norm": 0.00011665008059935644,
      "learning_rate": 4.268828910347948e-06,
      "loss": 0.0001,
      "step": 40560
    },
    {
      "epoch": 0.5735978170198929,
      "grad_norm": 0.0,
      "learning_rate": 4.267415063128278e-06,
      "loss": 0.0,
      "step": 40570
    },
    {
      "epoch": 0.5737392017418598,
      "grad_norm": 0.0,
      "learning_rate": 4.26600121590861e-06,
      "loss": 0.0,
      "step": 40580
    },
    {
      "epoch": 0.5738805864638267,
      "grad_norm": 0.0,
      "learning_rate": 4.26458736868894e-06,
      "loss": 0.0,
      "step": 40590
    },
    {
      "epoch": 0.5740219711857937,
      "grad_norm": 7.902583121222051e-08,
      "learning_rate": 4.263173521469271e-06,
      "loss": 0.0,
      "step": 40600
    },
    {
      "epoch": 0.5741633559077606,
      "grad_norm": 0.0,
      "learning_rate": 4.261759674249601e-06,
      "loss": 0.0,
      "step": 40610
    },
    {
      "epoch": 0.5743047406297276,
      "grad_norm": 0.0,
      "learning_rate": 4.260345827029932e-06,
      "loss": 0.0,
      "step": 40620
    },
    {
      "epoch": 0.5744461253516945,
      "grad_norm": 0.00014350650599226356,
      "learning_rate": 4.258931979810262e-06,
      "loss": 0.0104,
      "step": 40630
    },
    {
      "epoch": 0.5745875100736615,
      "grad_norm": 0.0,
      "learning_rate": 4.2575181325905925e-06,
      "loss": 0.5613,
      "step": 40640
    },
    {
      "epoch": 0.5747288947956284,
      "grad_norm": 0.03940139338374138,
      "learning_rate": 4.256104285370923e-06,
      "loss": 0.0,
      "step": 40650
    },
    {
      "epoch": 0.5748702795175953,
      "grad_norm": 6.076989222947304e-08,
      "learning_rate": 4.254690438151254e-06,
      "loss": 0.0,
      "step": 40660
    },
    {
      "epoch": 0.5750116642395623,
      "grad_norm": 0.0,
      "learning_rate": 4.253276590931584e-06,
      "loss": 0.0,
      "step": 40670
    },
    {
      "epoch": 0.5751530489615292,
      "grad_norm": 3.6932731291017262e-06,
      "learning_rate": 4.251862743711915e-06,
      "loss": 0.0,
      "step": 40680
    },
    {
      "epoch": 0.5752944336834962,
      "grad_norm": 0.00010727710468927398,
      "learning_rate": 4.250448896492245e-06,
      "loss": 0.0,
      "step": 40690
    },
    {
      "epoch": 0.5754358184054631,
      "grad_norm": 2.8000574111938477,
      "learning_rate": 4.249035049272576e-06,
      "loss": 0.0003,
      "step": 40700
    },
    {
      "epoch": 0.5755772031274301,
      "grad_norm": 0.0,
      "learning_rate": 4.247621202052907e-06,
      "loss": 0.0,
      "step": 40710
    },
    {
      "epoch": 0.575718587849397,
      "grad_norm": 8.135552343446761e-05,
      "learning_rate": 4.246207354833237e-06,
      "loss": 0.0025,
      "step": 40720
    },
    {
      "epoch": 0.5758599725713639,
      "grad_norm": 5.99299312398216e-07,
      "learning_rate": 4.244793507613568e-06,
      "loss": 0.0,
      "step": 40730
    },
    {
      "epoch": 0.5760013572933309,
      "grad_norm": 0.0,
      "learning_rate": 4.2433796603938985e-06,
      "loss": 0.0006,
      "step": 40740
    },
    {
      "epoch": 0.5761427420152978,
      "grad_norm": 0.0,
      "learning_rate": 4.2419658131742285e-06,
      "loss": 0.0,
      "step": 40750
    },
    {
      "epoch": 0.5762841267372648,
      "grad_norm": 0.0,
      "learning_rate": 4.240551965954559e-06,
      "loss": 0.0,
      "step": 40760
    },
    {
      "epoch": 0.5764255114592317,
      "grad_norm": 0.0,
      "learning_rate": 4.23913811873489e-06,
      "loss": 0.0,
      "step": 40770
    },
    {
      "epoch": 0.5765668961811986,
      "grad_norm": 0.000443299300968647,
      "learning_rate": 4.23772427151522e-06,
      "loss": 0.0,
      "step": 40780
    },
    {
      "epoch": 0.5767082809031656,
      "grad_norm": 0.0,
      "learning_rate": 4.236310424295551e-06,
      "loss": 0.0,
      "step": 40790
    },
    {
      "epoch": 0.5768496656251325,
      "grad_norm": 0.0,
      "learning_rate": 4.234896577075882e-06,
      "loss": 0.0006,
      "step": 40800
    },
    {
      "epoch": 0.5769910503470995,
      "grad_norm": 0.0,
      "learning_rate": 4.233482729856212e-06,
      "loss": 0.0,
      "step": 40810
    },
    {
      "epoch": 0.5771324350690664,
      "grad_norm": 0.0,
      "learning_rate": 4.232068882636543e-06,
      "loss": 0.0701,
      "step": 40820
    },
    {
      "epoch": 0.5772738197910334,
      "grad_norm": 4.246792316436768,
      "learning_rate": 4.230655035416873e-06,
      "loss": 0.2693,
      "step": 40830
    },
    {
      "epoch": 0.5774152045130003,
      "grad_norm": 0.0,
      "learning_rate": 4.229241188197204e-06,
      "loss": 0.0003,
      "step": 40840
    },
    {
      "epoch": 0.5775565892349672,
      "grad_norm": 0.0,
      "learning_rate": 4.2278273409775345e-06,
      "loss": 0.0,
      "step": 40850
    },
    {
      "epoch": 0.5776979739569342,
      "grad_norm": 0.0,
      "learning_rate": 4.2264134937578645e-06,
      "loss": 0.0,
      "step": 40860
    },
    {
      "epoch": 0.5778393586789011,
      "grad_norm": 0.012757350690662861,
      "learning_rate": 4.224999646538195e-06,
      "loss": 0.0,
      "step": 40870
    },
    {
      "epoch": 0.5779807434008681,
      "grad_norm": 0.0,
      "learning_rate": 4.223585799318526e-06,
      "loss": 0.0256,
      "step": 40880
    },
    {
      "epoch": 0.578122128122835,
      "grad_norm": 1.2530027504453756e-07,
      "learning_rate": 4.222171952098856e-06,
      "loss": 0.0,
      "step": 40890
    },
    {
      "epoch": 0.578263512844802,
      "grad_norm": 1057.1048583984375,
      "learning_rate": 4.220758104879187e-06,
      "loss": 0.3834,
      "step": 40900
    },
    {
      "epoch": 0.5784048975667689,
      "grad_norm": 0.005243917927145958,
      "learning_rate": 4.219344257659518e-06,
      "loss": 0.0001,
      "step": 40910
    },
    {
      "epoch": 0.5785462822887358,
      "grad_norm": 0.0,
      "learning_rate": 4.217930410439848e-06,
      "loss": 0.0003,
      "step": 40920
    },
    {
      "epoch": 0.5786876670107028,
      "grad_norm": 0.0,
      "learning_rate": 4.216516563220179e-06,
      "loss": 0.0,
      "step": 40930
    },
    {
      "epoch": 0.5788290517326697,
      "grad_norm": 1.797559576743879e-07,
      "learning_rate": 4.215102716000509e-06,
      "loss": 0.0001,
      "step": 40940
    },
    {
      "epoch": 0.5789704364546367,
      "grad_norm": 8.224440506410247e-08,
      "learning_rate": 4.21368886878084e-06,
      "loss": 0.0,
      "step": 40950
    },
    {
      "epoch": 0.5791118211766036,
      "grad_norm": 6.8493168328132015e-06,
      "learning_rate": 4.2122750215611706e-06,
      "loss": 0.0044,
      "step": 40960
    },
    {
      "epoch": 0.5792532058985707,
      "grad_norm": 0.0,
      "learning_rate": 4.210861174341501e-06,
      "loss": 0.0004,
      "step": 40970
    },
    {
      "epoch": 0.5793945906205376,
      "grad_norm": 0.0,
      "learning_rate": 4.2094473271218314e-06,
      "loss": 0.0,
      "step": 40980
    },
    {
      "epoch": 0.5795359753425045,
      "grad_norm": 0.0,
      "learning_rate": 4.208033479902162e-06,
      "loss": 0.0001,
      "step": 40990
    },
    {
      "epoch": 0.5796773600644715,
      "grad_norm": 0.0,
      "learning_rate": 4.206619632682492e-06,
      "loss": 0.0079,
      "step": 41000
    },
    {
      "epoch": 0.5798187447864384,
      "grad_norm": 1.0430795116533886e-09,
      "learning_rate": 4.205205785462823e-06,
      "loss": 0.0,
      "step": 41010
    },
    {
      "epoch": 0.5799601295084054,
      "grad_norm": 0.0,
      "learning_rate": 4.203791938243153e-06,
      "loss": 0.0,
      "step": 41020
    },
    {
      "epoch": 0.5801015142303723,
      "grad_norm": 0.0,
      "learning_rate": 4.202378091023485e-06,
      "loss": 0.0,
      "step": 41030
    },
    {
      "epoch": 0.5802428989523392,
      "grad_norm": 1.62256219482515e-05,
      "learning_rate": 4.200964243803815e-06,
      "loss": 0.0,
      "step": 41040
    },
    {
      "epoch": 0.5803842836743062,
      "grad_norm": 3.826584361377172e-05,
      "learning_rate": 4.199550396584146e-06,
      "loss": 0.0,
      "step": 41050
    },
    {
      "epoch": 0.5805256683962731,
      "grad_norm": 0.0,
      "learning_rate": 4.198136549364476e-06,
      "loss": 0.0,
      "step": 41060
    },
    {
      "epoch": 0.5806670531182401,
      "grad_norm": 0.0,
      "learning_rate": 4.196722702144807e-06,
      "loss": 0.0,
      "step": 41070
    },
    {
      "epoch": 0.580808437840207,
      "grad_norm": 0.0,
      "learning_rate": 4.195308854925137e-06,
      "loss": 0.0,
      "step": 41080
    },
    {
      "epoch": 0.580949822562174,
      "grad_norm": 0.012164698913693428,
      "learning_rate": 4.1938950077054675e-06,
      "loss": 0.0001,
      "step": 41090
    },
    {
      "epoch": 0.5810912072841409,
      "grad_norm": 0.0003468912618700415,
      "learning_rate": 4.192481160485798e-06,
      "loss": 0.0001,
      "step": 41100
    },
    {
      "epoch": 0.5812325920061078,
      "grad_norm": 0.0,
      "learning_rate": 4.191067313266129e-06,
      "loss": 0.0001,
      "step": 41110
    },
    {
      "epoch": 0.5813739767280748,
      "grad_norm": 0.0,
      "learning_rate": 4.189653466046459e-06,
      "loss": 0.0007,
      "step": 41120
    },
    {
      "epoch": 0.5815153614500417,
      "grad_norm": 0.0017206569900736213,
      "learning_rate": 4.18823961882679e-06,
      "loss": 0.0331,
      "step": 41130
    },
    {
      "epoch": 0.5816567461720087,
      "grad_norm": 0.0,
      "learning_rate": 4.18682577160712e-06,
      "loss": 0.0143,
      "step": 41140
    },
    {
      "epoch": 0.5817981308939756,
      "grad_norm": 0.0,
      "learning_rate": 4.185411924387451e-06,
      "loss": 0.001,
      "step": 41150
    },
    {
      "epoch": 0.5819395156159426,
      "grad_norm": 3.5403554647928104e-06,
      "learning_rate": 4.183998077167782e-06,
      "loss": 0.0,
      "step": 41160
    },
    {
      "epoch": 0.5820809003379095,
      "grad_norm": 3.238674821659515e-07,
      "learning_rate": 4.182584229948113e-06,
      "loss": 0.0,
      "step": 41170
    },
    {
      "epoch": 0.5822222850598764,
      "grad_norm": 0.017445651814341545,
      "learning_rate": 4.181170382728443e-06,
      "loss": 0.0,
      "step": 41180
    },
    {
      "epoch": 0.5823636697818434,
      "grad_norm": 3.042958496735082e-06,
      "learning_rate": 4.1797565355087735e-06,
      "loss": 0.0,
      "step": 41190
    },
    {
      "epoch": 0.5825050545038103,
      "grad_norm": 0.0,
      "learning_rate": 4.1783426882891035e-06,
      "loss": 0.0002,
      "step": 41200
    },
    {
      "epoch": 0.5826464392257773,
      "grad_norm": 2.037662261145101e-09,
      "learning_rate": 4.176928841069434e-06,
      "loss": 0.0,
      "step": 41210
    },
    {
      "epoch": 0.5827878239477442,
      "grad_norm": 0.0,
      "learning_rate": 4.175514993849764e-06,
      "loss": 0.001,
      "step": 41220
    },
    {
      "epoch": 0.5829292086697111,
      "grad_norm": 0.0,
      "learning_rate": 4.174101146630096e-06,
      "loss": 0.0,
      "step": 41230
    },
    {
      "epoch": 0.5830705933916781,
      "grad_norm": 0.0,
      "learning_rate": 4.172687299410426e-06,
      "loss": 0.0,
      "step": 41240
    },
    {
      "epoch": 0.583211978113645,
      "grad_norm": 0.0,
      "learning_rate": 4.171273452190757e-06,
      "loss": 0.0,
      "step": 41250
    },
    {
      "epoch": 0.583353362835612,
      "grad_norm": 0.0,
      "learning_rate": 4.169859604971087e-06,
      "loss": 0.0,
      "step": 41260
    },
    {
      "epoch": 0.5834947475575789,
      "grad_norm": 0.0,
      "learning_rate": 4.168445757751418e-06,
      "loss": 0.0,
      "step": 41270
    },
    {
      "epoch": 0.5836361322795459,
      "grad_norm": 1.0057268351459925e-09,
      "learning_rate": 4.167031910531748e-06,
      "loss": 0.0,
      "step": 41280
    },
    {
      "epoch": 0.5837775170015128,
      "grad_norm": 0.0,
      "learning_rate": 4.165618063312079e-06,
      "loss": 0.1014,
      "step": 41290
    },
    {
      "epoch": 0.5839189017234797,
      "grad_norm": 0.00028206463321112096,
      "learning_rate": 4.1642042160924095e-06,
      "loss": 0.0,
      "step": 41300
    },
    {
      "epoch": 0.5840602864454467,
      "grad_norm": 0.0,
      "learning_rate": 4.16279036887274e-06,
      "loss": 0.0,
      "step": 41310
    },
    {
      "epoch": 0.5842016711674136,
      "grad_norm": 0.0,
      "learning_rate": 4.16137652165307e-06,
      "loss": 0.0006,
      "step": 41320
    },
    {
      "epoch": 0.5843430558893806,
      "grad_norm": 0.0,
      "learning_rate": 4.159962674433401e-06,
      "loss": 0.0,
      "step": 41330
    },
    {
      "epoch": 0.5844844406113475,
      "grad_norm": 0.0,
      "learning_rate": 4.158548827213731e-06,
      "loss": 0.0016,
      "step": 41340
    },
    {
      "epoch": 0.5846258253333145,
      "grad_norm": 0.0,
      "learning_rate": 4.157134979994062e-06,
      "loss": 0.0,
      "step": 41350
    },
    {
      "epoch": 0.5847672100552814,
      "grad_norm": 0.0001466879330109805,
      "learning_rate": 4.155721132774393e-06,
      "loss": 0.0,
      "step": 41360
    },
    {
      "epoch": 0.5849085947772483,
      "grad_norm": 0.0,
      "learning_rate": 4.154307285554724e-06,
      "loss": 0.0,
      "step": 41370
    },
    {
      "epoch": 0.5850499794992153,
      "grad_norm": 4.88086516270414e-05,
      "learning_rate": 4.152893438335054e-06,
      "loss": 0.0,
      "step": 41380
    },
    {
      "epoch": 0.5851913642211822,
      "grad_norm": 0.0,
      "learning_rate": 4.151479591115385e-06,
      "loss": 0.0,
      "step": 41390
    },
    {
      "epoch": 0.5853327489431492,
      "grad_norm": 0.0,
      "learning_rate": 4.150065743895715e-06,
      "loss": 0.0,
      "step": 41400
    },
    {
      "epoch": 0.5854741336651161,
      "grad_norm": 0.0,
      "learning_rate": 4.1486518966760456e-06,
      "loss": 0.0001,
      "step": 41410
    },
    {
      "epoch": 0.5856155183870831,
      "grad_norm": 0.00023894122568890452,
      "learning_rate": 4.147238049456376e-06,
      "loss": 0.0,
      "step": 41420
    },
    {
      "epoch": 0.58575690310905,
      "grad_norm": 0.0,
      "learning_rate": 4.145824202236706e-06,
      "loss": 0.0,
      "step": 41430
    },
    {
      "epoch": 0.5858982878310169,
      "grad_norm": 0.0,
      "learning_rate": 4.144410355017037e-06,
      "loss": 0.0001,
      "step": 41440
    },
    {
      "epoch": 0.586039672552984,
      "grad_norm": 0.0,
      "learning_rate": 4.142996507797368e-06,
      "loss": 0.0,
      "step": 41450
    },
    {
      "epoch": 0.5861810572749508,
      "grad_norm": 0.0,
      "learning_rate": 4.141582660577698e-06,
      "loss": 0.0,
      "step": 41460
    },
    {
      "epoch": 0.5863224419969179,
      "grad_norm": 0.30817005038261414,
      "learning_rate": 4.140168813358029e-06,
      "loss": 0.0,
      "step": 41470
    },
    {
      "epoch": 0.5864638267188848,
      "grad_norm": 0.0,
      "learning_rate": 4.13875496613836e-06,
      "loss": 0.0002,
      "step": 41480
    },
    {
      "epoch": 0.5866052114408516,
      "grad_norm": 0.0002878283557947725,
      "learning_rate": 4.13734111891869e-06,
      "loss": 0.0,
      "step": 41490
    },
    {
      "epoch": 0.5867465961628187,
      "grad_norm": 0.0,
      "learning_rate": 4.135927271699021e-06,
      "loss": 0.1713,
      "step": 41500
    },
    {
      "epoch": 0.5868879808847856,
      "grad_norm": 0.0008205598569475114,
      "learning_rate": 4.134513424479351e-06,
      "loss": 0.0,
      "step": 41510
    },
    {
      "epoch": 0.5870293656067526,
      "grad_norm": 0.0,
      "learning_rate": 4.133099577259682e-06,
      "loss": 0.0,
      "step": 41520
    },
    {
      "epoch": 0.5871707503287195,
      "grad_norm": 0.0,
      "learning_rate": 4.1316857300400124e-06,
      "loss": 0.0001,
      "step": 41530
    },
    {
      "epoch": 0.5873121350506865,
      "grad_norm": 0.0,
      "learning_rate": 4.1302718828203425e-06,
      "loss": 0.0,
      "step": 41540
    },
    {
      "epoch": 0.5874535197726534,
      "grad_norm": 9.318184002182761e-10,
      "learning_rate": 4.128858035600673e-06,
      "loss": 0.0,
      "step": 41550
    },
    {
      "epoch": 0.5875949044946203,
      "grad_norm": 0.0,
      "learning_rate": 4.127444188381004e-06,
      "loss": 0.0,
      "step": 41560
    },
    {
      "epoch": 0.5877362892165873,
      "grad_norm": 9.31601334741572e-06,
      "learning_rate": 4.126030341161334e-06,
      "loss": 0.0,
      "step": 41570
    },
    {
      "epoch": 0.5878776739385542,
      "grad_norm": 9.70380109599489e-10,
      "learning_rate": 4.124616493941665e-06,
      "loss": 0.0,
      "step": 41580
    },
    {
      "epoch": 0.5880190586605212,
      "grad_norm": 0.0006712886970490217,
      "learning_rate": 4.123202646721995e-06,
      "loss": 0.0,
      "step": 41590
    },
    {
      "epoch": 0.5881604433824881,
      "grad_norm": 3.3650641853455454e-05,
      "learning_rate": 4.121788799502326e-06,
      "loss": 0.0,
      "step": 41600
    },
    {
      "epoch": 0.5883018281044551,
      "grad_norm": 0.027795586735010147,
      "learning_rate": 4.120374952282657e-06,
      "loss": 0.0,
      "step": 41610
    },
    {
      "epoch": 0.588443212826422,
      "grad_norm": 0.0,
      "learning_rate": 4.118961105062988e-06,
      "loss": 0.0,
      "step": 41620
    },
    {
      "epoch": 0.5885845975483889,
      "grad_norm": 0.0007984726107679307,
      "learning_rate": 4.117547257843318e-06,
      "loss": 0.0039,
      "step": 41630
    },
    {
      "epoch": 0.5887259822703559,
      "grad_norm": 0.0,
      "learning_rate": 4.1161334106236485e-06,
      "loss": 0.0,
      "step": 41640
    },
    {
      "epoch": 0.5888673669923228,
      "grad_norm": 0.0,
      "learning_rate": 4.1147195634039785e-06,
      "loss": 0.0003,
      "step": 41650
    },
    {
      "epoch": 0.5890087517142898,
      "grad_norm": 0.0,
      "learning_rate": 4.113305716184309e-06,
      "loss": 0.0,
      "step": 41660
    },
    {
      "epoch": 0.5891501364362567,
      "grad_norm": 0.0,
      "learning_rate": 4.111891868964639e-06,
      "loss": 0.0,
      "step": 41670
    },
    {
      "epoch": 0.5892915211582237,
      "grad_norm": 1.2712723219010513e-05,
      "learning_rate": 4.110478021744971e-06,
      "loss": 0.2113,
      "step": 41680
    },
    {
      "epoch": 0.5894329058801906,
      "grad_norm": 0.0,
      "learning_rate": 4.109064174525301e-06,
      "loss": 0.0,
      "step": 41690
    },
    {
      "epoch": 0.5895742906021575,
      "grad_norm": 0.0,
      "learning_rate": 4.107650327305632e-06,
      "loss": 0.0,
      "step": 41700
    },
    {
      "epoch": 0.5897156753241245,
      "grad_norm": 0.0,
      "learning_rate": 4.106236480085962e-06,
      "loss": 0.0,
      "step": 41710
    },
    {
      "epoch": 0.5898570600460914,
      "grad_norm": 57.2813606262207,
      "learning_rate": 4.104822632866293e-06,
      "loss": 0.0178,
      "step": 41720
    },
    {
      "epoch": 0.5899984447680584,
      "grad_norm": 101.524658203125,
      "learning_rate": 4.103408785646623e-06,
      "loss": 0.0125,
      "step": 41730
    },
    {
      "epoch": 0.5901398294900253,
      "grad_norm": 0.0,
      "learning_rate": 4.1019949384269545e-06,
      "loss": 0.0001,
      "step": 41740
    },
    {
      "epoch": 0.5902812142119922,
      "grad_norm": 0.0,
      "learning_rate": 4.1005810912072845e-06,
      "loss": 0.0,
      "step": 41750
    },
    {
      "epoch": 0.5904225989339592,
      "grad_norm": 39.112220764160156,
      "learning_rate": 4.099167243987615e-06,
      "loss": 0.0099,
      "step": 41760
    },
    {
      "epoch": 0.5905639836559261,
      "grad_norm": 0.00041281036101281643,
      "learning_rate": 4.097753396767945e-06,
      "loss": 0.0,
      "step": 41770
    },
    {
      "epoch": 0.5907053683778931,
      "grad_norm": 0.0,
      "learning_rate": 4.096339549548276e-06,
      "loss": 0.0,
      "step": 41780
    },
    {
      "epoch": 0.59084675309986,
      "grad_norm": 0.0,
      "learning_rate": 4.094925702328606e-06,
      "loss": 0.0,
      "step": 41790
    },
    {
      "epoch": 0.590988137821827,
      "grad_norm": 0.0,
      "learning_rate": 4.093511855108937e-06,
      "loss": 0.0,
      "step": 41800
    },
    {
      "epoch": 0.5911295225437939,
      "grad_norm": 0.0,
      "learning_rate": 4.092098007889268e-06,
      "loss": 0.3303,
      "step": 41810
    },
    {
      "epoch": 0.5912709072657608,
      "grad_norm": 0.0,
      "learning_rate": 4.090684160669599e-06,
      "loss": 0.0,
      "step": 41820
    },
    {
      "epoch": 0.5914122919877278,
      "grad_norm": 0.0,
      "learning_rate": 4.089270313449929e-06,
      "loss": 0.0,
      "step": 41830
    },
    {
      "epoch": 0.5915536767096947,
      "grad_norm": 0.0,
      "learning_rate": 4.08785646623026e-06,
      "loss": 0.0,
      "step": 41840
    },
    {
      "epoch": 0.5916950614316617,
      "grad_norm": 0.0,
      "learning_rate": 4.08644261901059e-06,
      "loss": 0.0,
      "step": 41850
    },
    {
      "epoch": 0.5918364461536286,
      "grad_norm": 0.007186184637248516,
      "learning_rate": 4.0850287717909205e-06,
      "loss": 0.0,
      "step": 41860
    },
    {
      "epoch": 0.5919778308755956,
      "grad_norm": 1.0271991035537553e-09,
      "learning_rate": 4.083614924571251e-06,
      "loss": 0.0,
      "step": 41870
    },
    {
      "epoch": 0.5921192155975625,
      "grad_norm": 0.0,
      "learning_rate": 4.082201077351582e-06,
      "loss": 0.0,
      "step": 41880
    },
    {
      "epoch": 0.5922606003195294,
      "grad_norm": 0.0,
      "learning_rate": 4.080787230131912e-06,
      "loss": 0.0,
      "step": 41890
    },
    {
      "epoch": 0.5924019850414964,
      "grad_norm": 0.0,
      "learning_rate": 4.079373382912243e-06,
      "loss": 0.0,
      "step": 41900
    },
    {
      "epoch": 0.5925433697634633,
      "grad_norm": 0.0,
      "learning_rate": 4.077959535692573e-06,
      "loss": 0.0,
      "step": 41910
    },
    {
      "epoch": 0.5926847544854303,
      "grad_norm": 4.369437363038742e-07,
      "learning_rate": 4.076545688472904e-06,
      "loss": 0.0,
      "step": 41920
    },
    {
      "epoch": 0.5928261392073972,
      "grad_norm": 0.0,
      "learning_rate": 4.075131841253234e-06,
      "loss": 0.0529,
      "step": 41930
    },
    {
      "epoch": 0.5929675239293641,
      "grad_norm": 0.0,
      "learning_rate": 4.073717994033565e-06,
      "loss": 0.0051,
      "step": 41940
    },
    {
      "epoch": 0.5931089086513311,
      "grad_norm": 0.0,
      "learning_rate": 4.072304146813896e-06,
      "loss": 0.3363,
      "step": 41950
    },
    {
      "epoch": 0.593250293373298,
      "grad_norm": 3.399794445968496e-09,
      "learning_rate": 4.0708902995942266e-06,
      "loss": 0.0,
      "step": 41960
    },
    {
      "epoch": 0.593391678095265,
      "grad_norm": 0.0,
      "learning_rate": 4.0694764523745566e-06,
      "loss": 0.0,
      "step": 41970
    },
    {
      "epoch": 0.593533062817232,
      "grad_norm": 0.0,
      "learning_rate": 4.0680626051548874e-06,
      "loss": 0.0,
      "step": 41980
    },
    {
      "epoch": 0.593674447539199,
      "grad_norm": 9.848821640014648,
      "learning_rate": 4.0666487579352174e-06,
      "loss": 0.0017,
      "step": 41990
    },
    {
      "epoch": 0.5938158322611659,
      "grad_norm": 0.0,
      "learning_rate": 4.065234910715548e-06,
      "loss": 0.0,
      "step": 42000
    },
    {
      "epoch": 0.5939572169831328,
      "grad_norm": 0.0,
      "learning_rate": 4.063821063495879e-06,
      "loss": 0.0,
      "step": 42010
    },
    {
      "epoch": 0.5940986017050998,
      "grad_norm": 0.0,
      "learning_rate": 4.06240721627621e-06,
      "loss": 0.0,
      "step": 42020
    },
    {
      "epoch": 0.5942399864270667,
      "grad_norm": 0.0006871916120871902,
      "learning_rate": 4.06099336905654e-06,
      "loss": 0.0001,
      "step": 42030
    },
    {
      "epoch": 0.5943813711490337,
      "grad_norm": 0.0,
      "learning_rate": 4.059579521836871e-06,
      "loss": 0.0,
      "step": 42040
    },
    {
      "epoch": 0.5945227558710006,
      "grad_norm": 0.0,
      "learning_rate": 4.058165674617201e-06,
      "loss": 0.0,
      "step": 42050
    },
    {
      "epoch": 0.5946641405929676,
      "grad_norm": 0.0,
      "learning_rate": 4.056751827397532e-06,
      "loss": 0.0,
      "step": 42060
    },
    {
      "epoch": 0.5948055253149345,
      "grad_norm": 2.286377593918587e-07,
      "learning_rate": 4.055337980177863e-06,
      "loss": 0.0,
      "step": 42070
    },
    {
      "epoch": 0.5949469100369014,
      "grad_norm": 4.7000661140828015e-09,
      "learning_rate": 4.053924132958193e-06,
      "loss": 0.0,
      "step": 42080
    },
    {
      "epoch": 0.5950882947588684,
      "grad_norm": 2.2629181728461845e-07,
      "learning_rate": 4.0525102857385235e-06,
      "loss": 0.0,
      "step": 42090
    },
    {
      "epoch": 0.5952296794808353,
      "grad_norm": 0.0,
      "learning_rate": 4.051096438518854e-06,
      "loss": 0.0089,
      "step": 42100
    },
    {
      "epoch": 0.5953710642028023,
      "grad_norm": 0.0,
      "learning_rate": 4.049682591299184e-06,
      "loss": 0.0,
      "step": 42110
    },
    {
      "epoch": 0.5955124489247692,
      "grad_norm": 0.0,
      "learning_rate": 4.048268744079515e-06,
      "loss": 0.0,
      "step": 42120
    },
    {
      "epoch": 0.5956538336467362,
      "grad_norm": 0.0,
      "learning_rate": 4.046854896859846e-06,
      "loss": 0.0,
      "step": 42130
    },
    {
      "epoch": 0.5957952183687031,
      "grad_norm": 7.528867244720459,
      "learning_rate": 4.045441049640176e-06,
      "loss": 0.0009,
      "step": 42140
    },
    {
      "epoch": 0.59593660309067,
      "grad_norm": 1.5060994940085948e-07,
      "learning_rate": 4.044027202420507e-06,
      "loss": 0.0062,
      "step": 42150
    },
    {
      "epoch": 0.596077987812637,
      "grad_norm": 7.274927611433668e-07,
      "learning_rate": 4.042613355200837e-06,
      "loss": 0.0001,
      "step": 42160
    },
    {
      "epoch": 0.5962193725346039,
      "grad_norm": 0.0,
      "learning_rate": 4.041199507981168e-06,
      "loss": 0.0225,
      "step": 42170
    },
    {
      "epoch": 0.5963607572565709,
      "grad_norm": 4.5416314264912216e-07,
      "learning_rate": 4.039785660761499e-06,
      "loss": 0.0,
      "step": 42180
    },
    {
      "epoch": 0.5965021419785378,
      "grad_norm": 0.07198867201805115,
      "learning_rate": 4.0383718135418295e-06,
      "loss": 0.0,
      "step": 42190
    },
    {
      "epoch": 0.5966435267005047,
      "grad_norm": 0.0,
      "learning_rate": 4.0369579663221595e-06,
      "loss": 0.2565,
      "step": 42200
    },
    {
      "epoch": 0.5967849114224717,
      "grad_norm": 0.0,
      "learning_rate": 4.03554411910249e-06,
      "loss": 0.0,
      "step": 42210
    },
    {
      "epoch": 0.5969262961444386,
      "grad_norm": 0.0,
      "learning_rate": 4.03413027188282e-06,
      "loss": 0.0,
      "step": 42220
    },
    {
      "epoch": 0.5970676808664056,
      "grad_norm": 0.0,
      "learning_rate": 4.032716424663151e-06,
      "loss": 0.0024,
      "step": 42230
    },
    {
      "epoch": 0.5972090655883725,
      "grad_norm": 0.0,
      "learning_rate": 4.031302577443481e-06,
      "loss": 0.0898,
      "step": 42240
    },
    {
      "epoch": 0.5973504503103395,
      "grad_norm": 0.0,
      "learning_rate": 4.029888730223812e-06,
      "loss": 0.0,
      "step": 42250
    },
    {
      "epoch": 0.5974918350323064,
      "grad_norm": 0.0,
      "learning_rate": 4.028474883004143e-06,
      "loss": 0.0,
      "step": 42260
    },
    {
      "epoch": 0.5976332197542733,
      "grad_norm": 0.0021977487485855818,
      "learning_rate": 4.027061035784474e-06,
      "loss": 0.0004,
      "step": 42270
    },
    {
      "epoch": 0.5977746044762403,
      "grad_norm": 0.0,
      "learning_rate": 4.025647188564804e-06,
      "loss": 0.0,
      "step": 42280
    },
    {
      "epoch": 0.5979159891982072,
      "grad_norm": 0.0,
      "learning_rate": 4.024233341345135e-06,
      "loss": 0.0,
      "step": 42290
    },
    {
      "epoch": 0.5980573739201742,
      "grad_norm": 0.0,
      "learning_rate": 4.022819494125465e-06,
      "loss": 0.0001,
      "step": 42300
    },
    {
      "epoch": 0.5981987586421411,
      "grad_norm": 51.64496994018555,
      "learning_rate": 4.0214056469057955e-06,
      "loss": 0.0105,
      "step": 42310
    },
    {
      "epoch": 0.5983401433641081,
      "grad_norm": 1.848496689049739e-09,
      "learning_rate": 4.019991799686126e-06,
      "loss": 0.0,
      "step": 42320
    },
    {
      "epoch": 0.598481528086075,
      "grad_norm": 6.51242544336128e-07,
      "learning_rate": 4.018577952466457e-06,
      "loss": 0.0,
      "step": 42330
    },
    {
      "epoch": 0.5986229128080419,
      "grad_norm": 6.150582976260921e-07,
      "learning_rate": 4.017164105246787e-06,
      "loss": 0.0004,
      "step": 42340
    },
    {
      "epoch": 0.5987642975300089,
      "grad_norm": 0.0,
      "learning_rate": 4.015750258027118e-06,
      "loss": 0.001,
      "step": 42350
    },
    {
      "epoch": 0.5989056822519758,
      "grad_norm": 0.0,
      "learning_rate": 4.014336410807448e-06,
      "loss": 0.4168,
      "step": 42360
    },
    {
      "epoch": 0.5990470669739428,
      "grad_norm": 0.00040261526009999216,
      "learning_rate": 4.012922563587779e-06,
      "loss": 0.0,
      "step": 42370
    },
    {
      "epoch": 0.5991884516959097,
      "grad_norm": 0.0,
      "learning_rate": 4.011508716368109e-06,
      "loss": 0.0002,
      "step": 42380
    },
    {
      "epoch": 0.5993298364178767,
      "grad_norm": 0.0,
      "learning_rate": 4.010094869148441e-06,
      "loss": 0.0,
      "step": 42390
    },
    {
      "epoch": 0.5994712211398436,
      "grad_norm": 0.0,
      "learning_rate": 4.008681021928771e-06,
      "loss": 0.0,
      "step": 42400
    },
    {
      "epoch": 0.5996126058618105,
      "grad_norm": 0.0,
      "learning_rate": 4.0072671747091016e-06,
      "loss": 0.0,
      "step": 42410
    },
    {
      "epoch": 0.5997539905837775,
      "grad_norm": 2.1864287191419862e-05,
      "learning_rate": 4.0058533274894316e-06,
      "loss": 0.0,
      "step": 42420
    },
    {
      "epoch": 0.5998953753057444,
      "grad_norm": 0.0,
      "learning_rate": 4.004439480269762e-06,
      "loss": 0.0016,
      "step": 42430
    },
    {
      "epoch": 0.6000367600277114,
      "grad_norm": 0.00011295573494862765,
      "learning_rate": 4.0030256330500924e-06,
      "loss": 0.0014,
      "step": 42440
    },
    {
      "epoch": 0.6001781447496783,
      "grad_norm": 6.283508469095977e-07,
      "learning_rate": 4.001611785830424e-06,
      "loss": 0.0,
      "step": 42450
    },
    {
      "epoch": 0.6003195294716452,
      "grad_norm": 6.162401255949135e-09,
      "learning_rate": 4.000197938610754e-06,
      "loss": 0.0,
      "step": 42460
    },
    {
      "epoch": 0.6004609141936122,
      "grad_norm": 0.0,
      "learning_rate": 3.998784091391085e-06,
      "loss": 0.0001,
      "step": 42470
    },
    {
      "epoch": 0.6006022989155791,
      "grad_norm": 1.715717790062854e-09,
      "learning_rate": 3.997370244171415e-06,
      "loss": 0.0,
      "step": 42480
    },
    {
      "epoch": 0.6007436836375462,
      "grad_norm": 0.0,
      "learning_rate": 3.995956396951746e-06,
      "loss": 0.0,
      "step": 42490
    },
    {
      "epoch": 0.600885068359513,
      "grad_norm": 0.0,
      "learning_rate": 3.994542549732076e-06,
      "loss": 0.0,
      "step": 42500
    },
    {
      "epoch": 0.6010264530814801,
      "grad_norm": 1.2562878737298888e-08,
      "learning_rate": 3.993128702512407e-06,
      "loss": 0.0,
      "step": 42510
    },
    {
      "epoch": 0.601167837803447,
      "grad_norm": 0.001316602691076696,
      "learning_rate": 3.991714855292738e-06,
      "loss": 0.0,
      "step": 42520
    },
    {
      "epoch": 0.6013092225254139,
      "grad_norm": 0.0,
      "learning_rate": 3.9903010080730684e-06,
      "loss": 0.0,
      "step": 42530
    },
    {
      "epoch": 0.6014506072473809,
      "grad_norm": 0.0,
      "learning_rate": 3.9888871608533985e-06,
      "loss": 0.006,
      "step": 42540
    },
    {
      "epoch": 0.6015919919693478,
      "grad_norm": 0.0,
      "learning_rate": 3.987473313633729e-06,
      "loss": 0.0,
      "step": 42550
    },
    {
      "epoch": 0.6017333766913148,
      "grad_norm": 0.0,
      "learning_rate": 3.986059466414059e-06,
      "loss": 0.0,
      "step": 42560
    },
    {
      "epoch": 0.6018747614132817,
      "grad_norm": 0.0,
      "learning_rate": 3.98464561919439e-06,
      "loss": 0.0,
      "step": 42570
    },
    {
      "epoch": 0.6020161461352487,
      "grad_norm": 0.0,
      "learning_rate": 3.983231771974721e-06,
      "loss": 0.0,
      "step": 42580
    },
    {
      "epoch": 0.6021575308572156,
      "grad_norm": 0.0,
      "learning_rate": 3.981817924755052e-06,
      "loss": 0.0006,
      "step": 42590
    },
    {
      "epoch": 0.6022989155791825,
      "grad_norm": 3.4557503525434186e-09,
      "learning_rate": 3.980404077535382e-06,
      "loss": 0.0,
      "step": 42600
    },
    {
      "epoch": 0.6024403003011495,
      "grad_norm": 0.008141821250319481,
      "learning_rate": 3.978990230315713e-06,
      "loss": 0.0,
      "step": 42610
    },
    {
      "epoch": 0.6025816850231164,
      "grad_norm": 0.0,
      "learning_rate": 3.977576383096043e-06,
      "loss": 0.5055,
      "step": 42620
    },
    {
      "epoch": 0.6027230697450834,
      "grad_norm": 0.0,
      "learning_rate": 3.976162535876374e-06,
      "loss": 0.0,
      "step": 42630
    },
    {
      "epoch": 0.6028644544670503,
      "grad_norm": 0.0,
      "learning_rate": 3.974748688656704e-06,
      "loss": 0.0,
      "step": 42640
    },
    {
      "epoch": 0.6030058391890173,
      "grad_norm": 0.0,
      "learning_rate": 3.9733348414370345e-06,
      "loss": 0.0,
      "step": 42650
    },
    {
      "epoch": 0.6031472239109842,
      "grad_norm": 3.5739255963562755e-06,
      "learning_rate": 3.971920994217365e-06,
      "loss": 0.0,
      "step": 42660
    },
    {
      "epoch": 0.6032886086329511,
      "grad_norm": 0.0,
      "learning_rate": 3.970507146997696e-06,
      "loss": 0.0,
      "step": 42670
    },
    {
      "epoch": 0.6034299933549181,
      "grad_norm": 0.0,
      "learning_rate": 3.969093299778026e-06,
      "loss": 0.0,
      "step": 42680
    },
    {
      "epoch": 0.603571378076885,
      "grad_norm": 365.924072265625,
      "learning_rate": 3.967679452558357e-06,
      "loss": 0.065,
      "step": 42690
    },
    {
      "epoch": 0.603712762798852,
      "grad_norm": 0.0010465322993695736,
      "learning_rate": 3.966265605338687e-06,
      "loss": 0.0,
      "step": 42700
    },
    {
      "epoch": 0.6038541475208189,
      "grad_norm": 0.0,
      "learning_rate": 3.964851758119018e-06,
      "loss": 0.0,
      "step": 42710
    },
    {
      "epoch": 0.6039955322427858,
      "grad_norm": 0.0,
      "learning_rate": 3.963437910899349e-06,
      "loss": 0.0019,
      "step": 42720
    },
    {
      "epoch": 0.6041369169647528,
      "grad_norm": 8.465478629204881e-08,
      "learning_rate": 3.962024063679679e-06,
      "loss": 0.0,
      "step": 42730
    },
    {
      "epoch": 0.6042783016867197,
      "grad_norm": 626.8239135742188,
      "learning_rate": 3.96061021646001e-06,
      "loss": 0.2574,
      "step": 42740
    },
    {
      "epoch": 0.6044196864086867,
      "grad_norm": 0.35746997594833374,
      "learning_rate": 3.9591963692403405e-06,
      "loss": 0.0002,
      "step": 42750
    },
    {
      "epoch": 0.6045610711306536,
      "grad_norm": 0.0,
      "learning_rate": 3.9577825220206705e-06,
      "loss": 0.0,
      "step": 42760
    },
    {
      "epoch": 0.6047024558526206,
      "grad_norm": 3.002479687097548e-08,
      "learning_rate": 3.956368674801001e-06,
      "loss": 0.0,
      "step": 42770
    },
    {
      "epoch": 0.6048438405745875,
      "grad_norm": 0.0,
      "learning_rate": 3.954954827581332e-06,
      "loss": 0.0,
      "step": 42780
    },
    {
      "epoch": 0.6049852252965544,
      "grad_norm": 5.9272606449667364e-06,
      "learning_rate": 3.953540980361662e-06,
      "loss": 0.0,
      "step": 42790
    },
    {
      "epoch": 0.6051266100185214,
      "grad_norm": 0.0,
      "learning_rate": 3.952127133141993e-06,
      "loss": 0.0,
      "step": 42800
    },
    {
      "epoch": 0.6052679947404883,
      "grad_norm": 0.0,
      "learning_rate": 3.950713285922323e-06,
      "loss": 0.0001,
      "step": 42810
    },
    {
      "epoch": 0.6054093794624553,
      "grad_norm": 0.0,
      "learning_rate": 3.949299438702654e-06,
      "loss": 0.0,
      "step": 42820
    },
    {
      "epoch": 0.6055507641844222,
      "grad_norm": 0.0,
      "learning_rate": 3.947885591482985e-06,
      "loss": 0.0,
      "step": 42830
    },
    {
      "epoch": 0.6056921489063892,
      "grad_norm": 0.0,
      "learning_rate": 3.946471744263316e-06,
      "loss": 0.0889,
      "step": 42840
    },
    {
      "epoch": 0.6058335336283561,
      "grad_norm": 8.984670230027803e-10,
      "learning_rate": 3.945057897043646e-06,
      "loss": 0.0,
      "step": 42850
    },
    {
      "epoch": 0.605974918350323,
      "grad_norm": 0.0,
      "learning_rate": 3.9436440498239765e-06,
      "loss": 0.0,
      "step": 42860
    },
    {
      "epoch": 0.60611630307229,
      "grad_norm": 0.0,
      "learning_rate": 3.9422302026043066e-06,
      "loss": 0.0,
      "step": 42870
    },
    {
      "epoch": 0.6062576877942569,
      "grad_norm": 0.0,
      "learning_rate": 3.940816355384637e-06,
      "loss": 0.0,
      "step": 42880
    },
    {
      "epoch": 0.6063990725162239,
      "grad_norm": 0.009771745651960373,
      "learning_rate": 3.939402508164967e-06,
      "loss": 0.0,
      "step": 42890
    },
    {
      "epoch": 0.6065404572381908,
      "grad_norm": 0.0,
      "learning_rate": 3.937988660945298e-06,
      "loss": 0.0,
      "step": 42900
    },
    {
      "epoch": 0.6066818419601577,
      "grad_norm": 0.0,
      "learning_rate": 3.936574813725629e-06,
      "loss": 0.0,
      "step": 42910
    },
    {
      "epoch": 0.6068232266821247,
      "grad_norm": 0.0,
      "learning_rate": 3.93516096650596e-06,
      "loss": 0.0008,
      "step": 42920
    },
    {
      "epoch": 0.6069646114040916,
      "grad_norm": 0.0,
      "learning_rate": 3.93374711928629e-06,
      "loss": 0.0,
      "step": 42930
    },
    {
      "epoch": 0.6071059961260586,
      "grad_norm": 0.0,
      "learning_rate": 3.932333272066621e-06,
      "loss": 0.0,
      "step": 42940
    },
    {
      "epoch": 0.6072473808480255,
      "grad_norm": 4.779865392379179e-10,
      "learning_rate": 3.930919424846951e-06,
      "loss": 0.0001,
      "step": 42950
    },
    {
      "epoch": 0.6073887655699926,
      "grad_norm": 0.0,
      "learning_rate": 3.929505577627282e-06,
      "loss": 0.0,
      "step": 42960
    },
    {
      "epoch": 0.6075301502919594,
      "grad_norm": 0.0,
      "learning_rate": 3.9280917304076126e-06,
      "loss": 0.0002,
      "step": 42970
    },
    {
      "epoch": 0.6076715350139263,
      "grad_norm": 0.0,
      "learning_rate": 3.9266778831879434e-06,
      "loss": 0.0,
      "step": 42980
    },
    {
      "epoch": 0.6078129197358934,
      "grad_norm": 0.0,
      "learning_rate": 3.9252640359682734e-06,
      "loss": 0.0,
      "step": 42990
    },
    {
      "epoch": 0.6079543044578603,
      "grad_norm": 2.40903176518259e-07,
      "learning_rate": 3.923850188748604e-06,
      "loss": 0.0,
      "step": 43000
    },
    {
      "epoch": 0.6080956891798273,
      "grad_norm": 8.803048845429373e-10,
      "learning_rate": 3.922436341528934e-06,
      "loss": 0.0,
      "step": 43010
    },
    {
      "epoch": 0.6082370739017942,
      "grad_norm": 0.0,
      "learning_rate": 3.921022494309265e-06,
      "loss": 0.0,
      "step": 43020
    },
    {
      "epoch": 0.6083784586237612,
      "grad_norm": 0.0,
      "learning_rate": 3.919608647089595e-06,
      "loss": 0.0,
      "step": 43030
    },
    {
      "epoch": 0.6085198433457281,
      "grad_norm": 0.0,
      "learning_rate": 3.918194799869927e-06,
      "loss": 0.0,
      "step": 43040
    },
    {
      "epoch": 0.608661228067695,
      "grad_norm": 0.007588997017592192,
      "learning_rate": 3.916780952650257e-06,
      "loss": 0.0,
      "step": 43050
    },
    {
      "epoch": 0.608802612789662,
      "grad_norm": 4.603355030496914e-09,
      "learning_rate": 3.915367105430588e-06,
      "loss": 0.0,
      "step": 43060
    },
    {
      "epoch": 0.6089439975116289,
      "grad_norm": 0.0,
      "learning_rate": 3.913953258210918e-06,
      "loss": 0.0,
      "step": 43070
    },
    {
      "epoch": 0.6090853822335959,
      "grad_norm": 0.0,
      "learning_rate": 3.912539410991249e-06,
      "loss": 0.0,
      "step": 43080
    },
    {
      "epoch": 0.6092267669555628,
      "grad_norm": 0.0,
      "learning_rate": 3.911125563771579e-06,
      "loss": 0.0,
      "step": 43090
    },
    {
      "epoch": 0.6093681516775298,
      "grad_norm": 0.0,
      "learning_rate": 3.90971171655191e-06,
      "loss": 0.0199,
      "step": 43100
    },
    {
      "epoch": 0.6095095363994967,
      "grad_norm": 0.0,
      "learning_rate": 3.90829786933224e-06,
      "loss": 0.0,
      "step": 43110
    },
    {
      "epoch": 0.6096509211214636,
      "grad_norm": 0.0,
      "learning_rate": 3.906884022112571e-06,
      "loss": 0.0,
      "step": 43120
    },
    {
      "epoch": 0.6097923058434306,
      "grad_norm": 0.0,
      "learning_rate": 3.905470174892901e-06,
      "loss": 0.0,
      "step": 43130
    },
    {
      "epoch": 0.6099336905653975,
      "grad_norm": 0.0,
      "learning_rate": 3.904056327673232e-06,
      "loss": 0.0835,
      "step": 43140
    },
    {
      "epoch": 0.6100750752873645,
      "grad_norm": 0.0,
      "learning_rate": 3.902642480453562e-06,
      "loss": 0.0,
      "step": 43150
    },
    {
      "epoch": 0.6102164600093314,
      "grad_norm": 0.0,
      "learning_rate": 3.901228633233893e-06,
      "loss": 0.0,
      "step": 43160
    },
    {
      "epoch": 0.6103578447312983,
      "grad_norm": 0.0,
      "learning_rate": 3.899814786014224e-06,
      "loss": 0.0017,
      "step": 43170
    },
    {
      "epoch": 0.6104992294532653,
      "grad_norm": 0.0,
      "learning_rate": 3.898400938794555e-06,
      "loss": 0.0,
      "step": 43180
    },
    {
      "epoch": 0.6106406141752322,
      "grad_norm": 0.0,
      "learning_rate": 3.896987091574885e-06,
      "loss": 0.0,
      "step": 43190
    },
    {
      "epoch": 0.6107819988971992,
      "grad_norm": 0.0,
      "learning_rate": 3.8955732443552155e-06,
      "loss": 0.0,
      "step": 43200
    },
    {
      "epoch": 0.6109233836191661,
      "grad_norm": 2.485364358761899e-09,
      "learning_rate": 3.8941593971355455e-06,
      "loss": 0.0,
      "step": 43210
    },
    {
      "epoch": 0.6110647683411331,
      "grad_norm": 0.0,
      "learning_rate": 3.892745549915876e-06,
      "loss": 0.0,
      "step": 43220
    },
    {
      "epoch": 0.6112061530631,
      "grad_norm": 0.0,
      "learning_rate": 3.891331702696207e-06,
      "loss": 0.0559,
      "step": 43230
    },
    {
      "epoch": 0.6113475377850669,
      "grad_norm": 0.0,
      "learning_rate": 3.889917855476538e-06,
      "loss": 0.0,
      "step": 43240
    },
    {
      "epoch": 0.6114889225070339,
      "grad_norm": 0.0,
      "learning_rate": 3.888504008256868e-06,
      "loss": 0.0001,
      "step": 43250
    },
    {
      "epoch": 0.6116303072290008,
      "grad_norm": 0.0,
      "learning_rate": 3.887090161037199e-06,
      "loss": 0.0984,
      "step": 43260
    },
    {
      "epoch": 0.6117716919509678,
      "grad_norm": 0.0,
      "learning_rate": 3.885676313817529e-06,
      "loss": 0.0004,
      "step": 43270
    },
    {
      "epoch": 0.6119130766729347,
      "grad_norm": 0.0,
      "learning_rate": 3.88426246659786e-06,
      "loss": 0.0,
      "step": 43280
    },
    {
      "epoch": 0.6120544613949017,
      "grad_norm": 4.0480131247022655e-07,
      "learning_rate": 3.882848619378191e-06,
      "loss": 0.0,
      "step": 43290
    },
    {
      "epoch": 0.6121958461168686,
      "grad_norm": 0.0,
      "learning_rate": 3.881434772158521e-06,
      "loss": 0.0,
      "step": 43300
    },
    {
      "epoch": 0.6123372308388355,
      "grad_norm": 0.0,
      "learning_rate": 3.8800209249388515e-06,
      "loss": 0.0,
      "step": 43310
    },
    {
      "epoch": 0.6124786155608025,
      "grad_norm": 0.0,
      "learning_rate": 3.878607077719182e-06,
      "loss": 0.0,
      "step": 43320
    },
    {
      "epoch": 0.6126200002827694,
      "grad_norm": 0.0,
      "learning_rate": 3.877193230499512e-06,
      "loss": 0.0002,
      "step": 43330
    },
    {
      "epoch": 0.6127613850047364,
      "grad_norm": 0.0,
      "learning_rate": 3.875779383279843e-06,
      "loss": 0.0,
      "step": 43340
    },
    {
      "epoch": 0.6129027697267033,
      "grad_norm": 0.0,
      "learning_rate": 3.874365536060173e-06,
      "loss": 0.0,
      "step": 43350
    },
    {
      "epoch": 0.6130441544486703,
      "grad_norm": 0.0,
      "learning_rate": 3.872951688840504e-06,
      "loss": 0.0,
      "step": 43360
    },
    {
      "epoch": 0.6131855391706372,
      "grad_norm": 0.0020187669433653355,
      "learning_rate": 3.871537841620835e-06,
      "loss": 0.0003,
      "step": 43370
    },
    {
      "epoch": 0.6133269238926041,
      "grad_norm": 0.0,
      "learning_rate": 3.870123994401165e-06,
      "loss": 0.0,
      "step": 43380
    },
    {
      "epoch": 0.6134683086145711,
      "grad_norm": 0.0,
      "learning_rate": 3.868710147181496e-06,
      "loss": 0.0,
      "step": 43390
    },
    {
      "epoch": 0.613609693336538,
      "grad_norm": 0.0,
      "learning_rate": 3.867296299961827e-06,
      "loss": 0.0002,
      "step": 43400
    },
    {
      "epoch": 0.613751078058505,
      "grad_norm": 0.0,
      "learning_rate": 3.865882452742157e-06,
      "loss": 0.0,
      "step": 43410
    },
    {
      "epoch": 0.6138924627804719,
      "grad_norm": 0.0,
      "learning_rate": 3.8644686055224876e-06,
      "loss": 0.0,
      "step": 43420
    },
    {
      "epoch": 0.6140338475024388,
      "grad_norm": 0.0,
      "learning_rate": 3.863054758302818e-06,
      "loss": 0.0001,
      "step": 43430
    },
    {
      "epoch": 0.6141752322244058,
      "grad_norm": 0.0,
      "learning_rate": 3.8616409110831484e-06,
      "loss": 0.0,
      "step": 43440
    },
    {
      "epoch": 0.6143166169463727,
      "grad_norm": 0.0,
      "learning_rate": 3.860227063863479e-06,
      "loss": 0.0,
      "step": 43450
    },
    {
      "epoch": 0.6144580016683397,
      "grad_norm": 0.0,
      "learning_rate": 3.858813216643809e-06,
      "loss": 0.0,
      "step": 43460
    },
    {
      "epoch": 0.6145993863903066,
      "grad_norm": 1.8024481960310368e-06,
      "learning_rate": 3.85739936942414e-06,
      "loss": 0.0,
      "step": 43470
    },
    {
      "epoch": 0.6147407711122737,
      "grad_norm": 50.45518493652344,
      "learning_rate": 3.855985522204471e-06,
      "loss": 0.0131,
      "step": 43480
    },
    {
      "epoch": 0.6148821558342406,
      "grad_norm": 0.00028774779639206827,
      "learning_rate": 3.854571674984802e-06,
      "loss": 0.0,
      "step": 43490
    },
    {
      "epoch": 0.6150235405562074,
      "grad_norm": 8.115366654237732e-05,
      "learning_rate": 3.853157827765132e-06,
      "loss": 0.0,
      "step": 43500
    },
    {
      "epoch": 0.6151649252781745,
      "grad_norm": 0.0,
      "learning_rate": 3.851743980545463e-06,
      "loss": 0.0035,
      "step": 43510
    },
    {
      "epoch": 0.6153063100001414,
      "grad_norm": 5.674495241692057e-06,
      "learning_rate": 3.850330133325793e-06,
      "loss": 0.0127,
      "step": 43520
    },
    {
      "epoch": 0.6154476947221084,
      "grad_norm": 0.0,
      "learning_rate": 3.848916286106124e-06,
      "loss": 0.0042,
      "step": 43530
    },
    {
      "epoch": 0.6155890794440753,
      "grad_norm": 0.0,
      "learning_rate": 3.847502438886454e-06,
      "loss": 0.0,
      "step": 43540
    },
    {
      "epoch": 0.6157304641660423,
      "grad_norm": 0.0,
      "learning_rate": 3.846088591666785e-06,
      "loss": 0.0,
      "step": 43550
    },
    {
      "epoch": 0.6158718488880092,
      "grad_norm": 0.0,
      "learning_rate": 3.844674744447115e-06,
      "loss": 0.0,
      "step": 43560
    },
    {
      "epoch": 0.6160132336099761,
      "grad_norm": 1.078733213866201e-09,
      "learning_rate": 3.843260897227446e-06,
      "loss": 0.0,
      "step": 43570
    },
    {
      "epoch": 0.6161546183319431,
      "grad_norm": 8.520102710463107e-05,
      "learning_rate": 3.841847050007776e-06,
      "loss": 0.001,
      "step": 43580
    },
    {
      "epoch": 0.61629600305391,
      "grad_norm": 0.0,
      "learning_rate": 3.840433202788107e-06,
      "loss": 0.0,
      "step": 43590
    },
    {
      "epoch": 0.616437387775877,
      "grad_norm": 0.0,
      "learning_rate": 3.839019355568437e-06,
      "loss": 0.0005,
      "step": 43600
    },
    {
      "epoch": 0.6165787724978439,
      "grad_norm": 0.0,
      "learning_rate": 3.837605508348768e-06,
      "loss": 0.0,
      "step": 43610
    },
    {
      "epoch": 0.6167201572198108,
      "grad_norm": 0.0,
      "learning_rate": 3.836191661129099e-06,
      "loss": 0.0,
      "step": 43620
    },
    {
      "epoch": 0.6168615419417778,
      "grad_norm": 0.0,
      "learning_rate": 3.83477781390943e-06,
      "loss": 0.0,
      "step": 43630
    },
    {
      "epoch": 0.6170029266637447,
      "grad_norm": 0.8621752262115479,
      "learning_rate": 3.83336396668976e-06,
      "loss": 0.0007,
      "step": 43640
    },
    {
      "epoch": 0.6171443113857117,
      "grad_norm": 0.0008992551593109965,
      "learning_rate": 3.8319501194700905e-06,
      "loss": 0.0001,
      "step": 43650
    },
    {
      "epoch": 0.6172856961076786,
      "grad_norm": 0.0,
      "learning_rate": 3.8305362722504205e-06,
      "loss": 0.0,
      "step": 43660
    },
    {
      "epoch": 0.6174270808296456,
      "grad_norm": 0.0,
      "learning_rate": 3.829122425030751e-06,
      "loss": 0.0001,
      "step": 43670
    },
    {
      "epoch": 0.6175684655516125,
      "grad_norm": 0.0,
      "learning_rate": 3.827708577811082e-06,
      "loss": 0.4141,
      "step": 43680
    },
    {
      "epoch": 0.6177098502735794,
      "grad_norm": 3.029706974189139e-08,
      "learning_rate": 3.826294730591413e-06,
      "loss": 0.0,
      "step": 43690
    },
    {
      "epoch": 0.6178512349955464,
      "grad_norm": 0.0,
      "learning_rate": 3.824880883371743e-06,
      "loss": 0.0,
      "step": 43700
    },
    {
      "epoch": 0.6179926197175133,
      "grad_norm": 2.8919771022373197e-09,
      "learning_rate": 3.823467036152074e-06,
      "loss": 0.0,
      "step": 43710
    },
    {
      "epoch": 0.6181340044394803,
      "grad_norm": 0.0,
      "learning_rate": 3.822053188932404e-06,
      "loss": 0.0,
      "step": 43720
    },
    {
      "epoch": 0.6182753891614472,
      "grad_norm": 0.0,
      "learning_rate": 3.820639341712735e-06,
      "loss": 0.001,
      "step": 43730
    },
    {
      "epoch": 0.6184167738834142,
      "grad_norm": 0.0,
      "learning_rate": 3.819225494493065e-06,
      "loss": 0.0002,
      "step": 43740
    },
    {
      "epoch": 0.6185581586053811,
      "grad_norm": 0.2572053074836731,
      "learning_rate": 3.8178116472733965e-06,
      "loss": 0.0,
      "step": 43750
    },
    {
      "epoch": 0.618699543327348,
      "grad_norm": 0.0,
      "learning_rate": 3.8163978000537265e-06,
      "loss": 0.0691,
      "step": 43760
    },
    {
      "epoch": 0.618840928049315,
      "grad_norm": 0.0,
      "learning_rate": 3.814983952834057e-06,
      "loss": 0.0,
      "step": 43770
    },
    {
      "epoch": 0.6189823127712819,
      "grad_norm": 0.0,
      "learning_rate": 3.8135701056143874e-06,
      "loss": 0.0024,
      "step": 43780
    },
    {
      "epoch": 0.6191236974932489,
      "grad_norm": 0.0,
      "learning_rate": 3.8121562583947182e-06,
      "loss": 0.0,
      "step": 43790
    },
    {
      "epoch": 0.6192650822152158,
      "grad_norm": 0.021678932011127472,
      "learning_rate": 3.8107424111750487e-06,
      "loss": 0.0001,
      "step": 43800
    },
    {
      "epoch": 0.6194064669371828,
      "grad_norm": 0.0,
      "learning_rate": 3.8093285639553795e-06,
      "loss": 0.002,
      "step": 43810
    },
    {
      "epoch": 0.6195478516591497,
      "grad_norm": 0.0,
      "learning_rate": 3.80791471673571e-06,
      "loss": 0.477,
      "step": 43820
    },
    {
      "epoch": 0.6196892363811166,
      "grad_norm": 0.0,
      "learning_rate": 3.8065008695160404e-06,
      "loss": 0.0,
      "step": 43830
    },
    {
      "epoch": 0.6198306211030836,
      "grad_norm": 0.0,
      "learning_rate": 3.805087022296371e-06,
      "loss": 0.0,
      "step": 43840
    },
    {
      "epoch": 0.6199720058250505,
      "grad_norm": 0.0,
      "learning_rate": 3.8036731750767013e-06,
      "loss": 0.6814,
      "step": 43850
    },
    {
      "epoch": 0.6201133905470175,
      "grad_norm": 0.0,
      "learning_rate": 3.8022593278570317e-06,
      "loss": 0.0092,
      "step": 43860
    },
    {
      "epoch": 0.6202547752689844,
      "grad_norm": 6.392829418182373,
      "learning_rate": 3.8008454806373626e-06,
      "loss": 0.038,
      "step": 43870
    },
    {
      "epoch": 0.6203961599909513,
      "grad_norm": 0.0,
      "learning_rate": 3.7994316334176934e-06,
      "loss": 0.0001,
      "step": 43880
    },
    {
      "epoch": 0.6205375447129183,
      "grad_norm": 0.0,
      "learning_rate": 3.798017786198024e-06,
      "loss": 0.0,
      "step": 43890
    },
    {
      "epoch": 0.6206789294348852,
      "grad_norm": 0.0,
      "learning_rate": 3.7966039389783543e-06,
      "loss": 0.0,
      "step": 43900
    },
    {
      "epoch": 0.6208203141568522,
      "grad_norm": 0.0017950076144188643,
      "learning_rate": 3.7951900917586847e-06,
      "loss": 0.0,
      "step": 43910
    },
    {
      "epoch": 0.6209616988788191,
      "grad_norm": 0.0,
      "learning_rate": 3.793776244539015e-06,
      "loss": 0.0,
      "step": 43920
    },
    {
      "epoch": 0.6211030836007861,
      "grad_norm": 0.0,
      "learning_rate": 3.7923623973193456e-06,
      "loss": 0.0,
      "step": 43930
    },
    {
      "epoch": 0.621244468322753,
      "grad_norm": 0.16484929621219635,
      "learning_rate": 3.790948550099677e-06,
      "loss": 0.0274,
      "step": 43940
    },
    {
      "epoch": 0.6213858530447199,
      "grad_norm": 0.0,
      "learning_rate": 3.7895347028800073e-06,
      "loss": 0.0,
      "step": 43950
    },
    {
      "epoch": 0.621527237766687,
      "grad_norm": 0.0,
      "learning_rate": 3.7881208556603377e-06,
      "loss": 0.0,
      "step": 43960
    },
    {
      "epoch": 0.6216686224886538,
      "grad_norm": 0.0,
      "learning_rate": 3.786707008440668e-06,
      "loss": 0.0,
      "step": 43970
    },
    {
      "epoch": 0.6218100072106209,
      "grad_norm": 0.0,
      "learning_rate": 3.7852931612209986e-06,
      "loss": 0.0,
      "step": 43980
    },
    {
      "epoch": 0.6219513919325877,
      "grad_norm": 0.0,
      "learning_rate": 3.783879314001329e-06,
      "loss": 0.0014,
      "step": 43990
    },
    {
      "epoch": 0.6220927766545548,
      "grad_norm": 0.0,
      "learning_rate": 3.7824654667816603e-06,
      "loss": 0.0,
      "step": 44000
    },
    {
      "epoch": 0.6222341613765217,
      "grad_norm": 0.0,
      "learning_rate": 3.7810516195619907e-06,
      "loss": 0.3873,
      "step": 44010
    },
    {
      "epoch": 0.6223755460984886,
      "grad_norm": 0.00024318833311554044,
      "learning_rate": 3.779637772342321e-06,
      "loss": 0.0,
      "step": 44020
    },
    {
      "epoch": 0.6225169308204556,
      "grad_norm": 0.0,
      "learning_rate": 3.7782239251226516e-06,
      "loss": 0.0,
      "step": 44030
    },
    {
      "epoch": 0.6226583155424225,
      "grad_norm": 0.0,
      "learning_rate": 3.776810077902982e-06,
      "loss": 0.0,
      "step": 44040
    },
    {
      "epoch": 0.6227997002643895,
      "grad_norm": 0.0,
      "learning_rate": 3.7753962306833125e-06,
      "loss": 0.0011,
      "step": 44050
    },
    {
      "epoch": 0.6229410849863564,
      "grad_norm": 0.0,
      "learning_rate": 3.773982383463643e-06,
      "loss": 0.0,
      "step": 44060
    },
    {
      "epoch": 0.6230824697083234,
      "grad_norm": 3.2633017781336093e-06,
      "learning_rate": 3.772568536243974e-06,
      "loss": 0.0,
      "step": 44070
    },
    {
      "epoch": 0.6232238544302903,
      "grad_norm": 0.0,
      "learning_rate": 3.7711546890243046e-06,
      "loss": 0.0001,
      "step": 44080
    },
    {
      "epoch": 0.6233652391522572,
      "grad_norm": 7.769900278731257e-09,
      "learning_rate": 3.769740841804635e-06,
      "loss": 0.0004,
      "step": 44090
    },
    {
      "epoch": 0.6235066238742242,
      "grad_norm": 3.0048149710637517e-06,
      "learning_rate": 3.7683269945849655e-06,
      "loss": 0.1085,
      "step": 44100
    },
    {
      "epoch": 0.6236480085961911,
      "grad_norm": 0.0,
      "learning_rate": 3.766913147365296e-06,
      "loss": 0.7,
      "step": 44110
    },
    {
      "epoch": 0.6237893933181581,
      "grad_norm": 0.0,
      "learning_rate": 3.7654993001456263e-06,
      "loss": 0.0,
      "step": 44120
    },
    {
      "epoch": 0.623930778040125,
      "grad_norm": 0.0,
      "learning_rate": 3.764085452925957e-06,
      "loss": 0.0,
      "step": 44130
    },
    {
      "epoch": 0.6240721627620919,
      "grad_norm": 0.0,
      "learning_rate": 3.762671605706288e-06,
      "loss": 0.0009,
      "step": 44140
    },
    {
      "epoch": 0.6242135474840589,
      "grad_norm": 0.0,
      "learning_rate": 3.7612577584866185e-06,
      "loss": 0.0,
      "step": 44150
    },
    {
      "epoch": 0.6243549322060258,
      "grad_norm": 9.263211309118446e-10,
      "learning_rate": 3.759843911266949e-06,
      "loss": 0.0,
      "step": 44160
    },
    {
      "epoch": 0.6244963169279928,
      "grad_norm": 11.290754318237305,
      "learning_rate": 3.7584300640472794e-06,
      "loss": 0.0018,
      "step": 44170
    },
    {
      "epoch": 0.6246377016499597,
      "grad_norm": 0.0,
      "learning_rate": 3.7570162168276098e-06,
      "loss": 0.0892,
      "step": 44180
    },
    {
      "epoch": 0.6247790863719267,
      "grad_norm": 0.0,
      "learning_rate": 3.7556023696079402e-06,
      "loss": 0.0,
      "step": 44190
    },
    {
      "epoch": 0.6249204710938936,
      "grad_norm": 0.0,
      "learning_rate": 3.754188522388271e-06,
      "loss": 0.0001,
      "step": 44200
    },
    {
      "epoch": 0.6250618558158605,
      "grad_norm": 0.0,
      "learning_rate": 3.752774675168602e-06,
      "loss": 0.0,
      "step": 44210
    },
    {
      "epoch": 0.6252032405378275,
      "grad_norm": 0.0,
      "learning_rate": 3.7513608279489324e-06,
      "loss": 0.0,
      "step": 44220
    },
    {
      "epoch": 0.6253446252597944,
      "grad_norm": 0.1788487583398819,
      "learning_rate": 3.749946980729263e-06,
      "loss": 0.0,
      "step": 44230
    },
    {
      "epoch": 0.6254860099817614,
      "grad_norm": 489.7900695800781,
      "learning_rate": 3.7485331335095932e-06,
      "loss": 0.1303,
      "step": 44240
    },
    {
      "epoch": 0.6256273947037283,
      "grad_norm": 0.0,
      "learning_rate": 3.7471192862899237e-06,
      "loss": 0.0002,
      "step": 44250
    },
    {
      "epoch": 0.6257687794256953,
      "grad_norm": 0.0,
      "learning_rate": 3.7457054390702545e-06,
      "loss": 0.0,
      "step": 44260
    },
    {
      "epoch": 0.6259101641476622,
      "grad_norm": 0.0,
      "learning_rate": 3.744291591850585e-06,
      "loss": 0.0,
      "step": 44270
    },
    {
      "epoch": 0.6260515488696291,
      "grad_norm": 8.159182773859186e-10,
      "learning_rate": 3.7428777446309154e-06,
      "loss": 0.0123,
      "step": 44280
    },
    {
      "epoch": 0.6261929335915961,
      "grad_norm": 2.552613409534388e-07,
      "learning_rate": 3.7414638974112462e-06,
      "loss": 0.0,
      "step": 44290
    },
    {
      "epoch": 0.626334318313563,
      "grad_norm": 0.0,
      "learning_rate": 3.7400500501915767e-06,
      "loss": 0.0004,
      "step": 44300
    },
    {
      "epoch": 0.62647570303553,
      "grad_norm": 0.0,
      "learning_rate": 3.738636202971907e-06,
      "loss": 0.0,
      "step": 44310
    },
    {
      "epoch": 0.6266170877574969,
      "grad_norm": 0.0,
      "learning_rate": 3.7372223557522375e-06,
      "loss": 0.0,
      "step": 44320
    },
    {
      "epoch": 0.6267584724794639,
      "grad_norm": 0.0,
      "learning_rate": 3.7358085085325684e-06,
      "loss": 0.0,
      "step": 44330
    },
    {
      "epoch": 0.6268998572014308,
      "grad_norm": 1.2333962331467774e-05,
      "learning_rate": 3.734394661312899e-06,
      "loss": 0.0,
      "step": 44340
    },
    {
      "epoch": 0.6270412419233977,
      "grad_norm": 8.295782727429923e-09,
      "learning_rate": 3.7329808140932293e-06,
      "loss": 0.0125,
      "step": 44350
    },
    {
      "epoch": 0.6271826266453647,
      "grad_norm": 0.0041649495251476765,
      "learning_rate": 3.7315669668735597e-06,
      "loss": 0.0,
      "step": 44360
    },
    {
      "epoch": 0.6273240113673316,
      "grad_norm": 0.0002810999285429716,
      "learning_rate": 3.7301531196538906e-06,
      "loss": 0.0,
      "step": 44370
    },
    {
      "epoch": 0.6274653960892986,
      "grad_norm": 0.0,
      "learning_rate": 3.728739272434221e-06,
      "loss": 0.0,
      "step": 44380
    },
    {
      "epoch": 0.6276067808112655,
      "grad_norm": 5.8584948419593275e-05,
      "learning_rate": 3.727325425214552e-06,
      "loss": 0.0,
      "step": 44390
    },
    {
      "epoch": 0.6277481655332324,
      "grad_norm": 0.0,
      "learning_rate": 3.7259115779948823e-06,
      "loss": 0.0,
      "step": 44400
    },
    {
      "epoch": 0.6278895502551994,
      "grad_norm": 0.0,
      "learning_rate": 3.7244977307752127e-06,
      "loss": 0.0,
      "step": 44410
    },
    {
      "epoch": 0.6280309349771663,
      "grad_norm": 0.0,
      "learning_rate": 3.723083883555543e-06,
      "loss": 0.0011,
      "step": 44420
    },
    {
      "epoch": 0.6281723196991333,
      "grad_norm": 9.238535767508438e-07,
      "learning_rate": 3.7216700363358736e-06,
      "loss": 0.0,
      "step": 44430
    },
    {
      "epoch": 0.6283137044211002,
      "grad_norm": 0.0,
      "learning_rate": 3.7202561891162044e-06,
      "loss": 0.0,
      "step": 44440
    },
    {
      "epoch": 0.6284550891430672,
      "grad_norm": 2.418993858555041e-07,
      "learning_rate": 3.718842341896535e-06,
      "loss": 0.0,
      "step": 44450
    },
    {
      "epoch": 0.6285964738650341,
      "grad_norm": 0.0,
      "learning_rate": 3.7174284946768657e-06,
      "loss": 0.0,
      "step": 44460
    },
    {
      "epoch": 0.628737858587001,
      "grad_norm": 0.0,
      "learning_rate": 3.716014647457196e-06,
      "loss": 0.0,
      "step": 44470
    },
    {
      "epoch": 0.628879243308968,
      "grad_norm": 0.0,
      "learning_rate": 3.7146008002375266e-06,
      "loss": 0.0,
      "step": 44480
    },
    {
      "epoch": 0.629020628030935,
      "grad_norm": 0.0,
      "learning_rate": 3.713186953017857e-06,
      "loss": 0.0,
      "step": 44490
    },
    {
      "epoch": 0.629162012752902,
      "grad_norm": 9.760721695784014e-06,
      "learning_rate": 3.7117731057981875e-06,
      "loss": 0.0,
      "step": 44500
    },
    {
      "epoch": 0.6293033974748689,
      "grad_norm": 0.0008754776208661497,
      "learning_rate": 3.710359258578518e-06,
      "loss": 0.0,
      "step": 44510
    },
    {
      "epoch": 0.6294447821968359,
      "grad_norm": 0.0,
      "learning_rate": 3.708945411358849e-06,
      "loss": 0.0001,
      "step": 44520
    },
    {
      "epoch": 0.6295861669188028,
      "grad_norm": 0.0,
      "learning_rate": 3.7075315641391796e-06,
      "loss": 0.0,
      "step": 44530
    },
    {
      "epoch": 0.6297275516407697,
      "grad_norm": 0.0,
      "learning_rate": 3.70611771691951e-06,
      "loss": 0.0001,
      "step": 44540
    },
    {
      "epoch": 0.6298689363627367,
      "grad_norm": 132.12220764160156,
      "learning_rate": 3.7047038696998405e-06,
      "loss": 0.0219,
      "step": 44550
    },
    {
      "epoch": 0.6300103210847036,
      "grad_norm": 0.0,
      "learning_rate": 3.703290022480171e-06,
      "loss": 0.0001,
      "step": 44560
    },
    {
      "epoch": 0.6301517058066706,
      "grad_norm": 4.37344660486616e-10,
      "learning_rate": 3.7018761752605013e-06,
      "loss": 0.0,
      "step": 44570
    },
    {
      "epoch": 0.6302930905286375,
      "grad_norm": 0.0,
      "learning_rate": 3.7004623280408318e-06,
      "loss": 0.0,
      "step": 44580
    },
    {
      "epoch": 0.6304344752506044,
      "grad_norm": 0.0,
      "learning_rate": 3.699048480821163e-06,
      "loss": 0.0001,
      "step": 44590
    },
    {
      "epoch": 0.6305758599725714,
      "grad_norm": 0.0,
      "learning_rate": 3.6976346336014935e-06,
      "loss": 0.0002,
      "step": 44600
    },
    {
      "epoch": 0.6307172446945383,
      "grad_norm": 0.0,
      "learning_rate": 3.696220786381824e-06,
      "loss": 0.0,
      "step": 44610
    },
    {
      "epoch": 0.6308586294165053,
      "grad_norm": 3.049599172300077e-06,
      "learning_rate": 3.6948069391621543e-06,
      "loss": 0.0,
      "step": 44620
    },
    {
      "epoch": 0.6310000141384722,
      "grad_norm": 0.0,
      "learning_rate": 3.6933930919424848e-06,
      "loss": 0.0,
      "step": 44630
    },
    {
      "epoch": 0.6311413988604392,
      "grad_norm": 0.0,
      "learning_rate": 3.691979244722815e-06,
      "loss": 0.0,
      "step": 44640
    },
    {
      "epoch": 0.6312827835824061,
      "grad_norm": 0.0,
      "learning_rate": 3.6905653975031465e-06,
      "loss": 0.0,
      "step": 44650
    },
    {
      "epoch": 0.631424168304373,
      "grad_norm": 0.0,
      "learning_rate": 3.689151550283477e-06,
      "loss": 0.0005,
      "step": 44660
    },
    {
      "epoch": 0.63156555302634,
      "grad_norm": 0.517427384853363,
      "learning_rate": 3.6877377030638074e-06,
      "loss": 0.0005,
      "step": 44670
    },
    {
      "epoch": 0.6317069377483069,
      "grad_norm": 0.0,
      "learning_rate": 3.6863238558441378e-06,
      "loss": 0.0027,
      "step": 44680
    },
    {
      "epoch": 0.6318483224702739,
      "grad_norm": 0.0009999987669289112,
      "learning_rate": 3.6849100086244682e-06,
      "loss": 0.0001,
      "step": 44690
    },
    {
      "epoch": 0.6319897071922408,
      "grad_norm": 6.7092305471305735e-06,
      "learning_rate": 3.6834961614047987e-06,
      "loss": 0.0,
      "step": 44700
    },
    {
      "epoch": 0.6321310919142078,
      "grad_norm": 0.0,
      "learning_rate": 3.6822236989070963e-06,
      "loss": 0.7359,
      "step": 44710
    },
    {
      "epoch": 0.6322724766361747,
      "grad_norm": 0.0,
      "learning_rate": 3.6808098516874268e-06,
      "loss": 0.0,
      "step": 44720
    },
    {
      "epoch": 0.6324138613581416,
      "grad_norm": 0.0,
      "learning_rate": 3.679396004467757e-06,
      "loss": 0.0,
      "step": 44730
    },
    {
      "epoch": 0.6325552460801086,
      "grad_norm": 0.0,
      "learning_rate": 3.6779821572480876e-06,
      "loss": 0.0001,
      "step": 44740
    },
    {
      "epoch": 0.6326966308020755,
      "grad_norm": 0.0,
      "learning_rate": 3.676568310028419e-06,
      "loss": 0.0,
      "step": 44750
    },
    {
      "epoch": 0.6328380155240425,
      "grad_norm": 0.0,
      "learning_rate": 3.6751544628087494e-06,
      "loss": 0.0,
      "step": 44760
    },
    {
      "epoch": 0.6329794002460094,
      "grad_norm": 0.0,
      "learning_rate": 3.6737406155890798e-06,
      "loss": 0.0,
      "step": 44770
    },
    {
      "epoch": 0.6331207849679764,
      "grad_norm": 8.030809840420261e-05,
      "learning_rate": 3.6723267683694102e-06,
      "loss": 0.0,
      "step": 44780
    },
    {
      "epoch": 0.6332621696899433,
      "grad_norm": 5.62361856282223e-05,
      "learning_rate": 3.6709129211497406e-06,
      "loss": 0.0149,
      "step": 44790
    },
    {
      "epoch": 0.6334035544119102,
      "grad_norm": 6.729673032168648e-07,
      "learning_rate": 3.669499073930071e-06,
      "loss": 0.0005,
      "step": 44800
    },
    {
      "epoch": 0.6335449391338772,
      "grad_norm": 0.0,
      "learning_rate": 3.6680852267104024e-06,
      "loss": 0.0011,
      "step": 44810
    },
    {
      "epoch": 0.6336863238558441,
      "grad_norm": 0.0,
      "learning_rate": 3.666671379490733e-06,
      "loss": 0.0004,
      "step": 44820
    },
    {
      "epoch": 0.6338277085778111,
      "grad_norm": 5.225671007025312e-10,
      "learning_rate": 3.6652575322710632e-06,
      "loss": 0.0,
      "step": 44830
    },
    {
      "epoch": 0.633969093299778,
      "grad_norm": 0.004376371391117573,
      "learning_rate": 3.6638436850513937e-06,
      "loss": 0.0,
      "step": 44840
    },
    {
      "epoch": 0.6341104780217449,
      "grad_norm": 0.0002228384546469897,
      "learning_rate": 3.662429837831724e-06,
      "loss": 0.0007,
      "step": 44850
    },
    {
      "epoch": 0.6342518627437119,
      "grad_norm": 0.014830597676336765,
      "learning_rate": 3.6610159906120545e-06,
      "loss": 0.001,
      "step": 44860
    },
    {
      "epoch": 0.6343932474656788,
      "grad_norm": 0.0,
      "learning_rate": 3.659602143392385e-06,
      "loss": 0.0,
      "step": 44870
    },
    {
      "epoch": 0.6345346321876458,
      "grad_norm": 0.0,
      "learning_rate": 3.6581882961727162e-06,
      "loss": 0.0,
      "step": 44880
    },
    {
      "epoch": 0.6346760169096127,
      "grad_norm": 0.0,
      "learning_rate": 3.6567744489530467e-06,
      "loss": 0.0,
      "step": 44890
    },
    {
      "epoch": 0.6348174016315797,
      "grad_norm": 0.0,
      "learning_rate": 3.655360601733377e-06,
      "loss": 0.0,
      "step": 44900
    },
    {
      "epoch": 0.6349587863535466,
      "grad_norm": 0.0,
      "learning_rate": 3.6539467545137075e-06,
      "loss": 0.0,
      "step": 44910
    },
    {
      "epoch": 0.6351001710755135,
      "grad_norm": 0.0,
      "learning_rate": 3.652532907294038e-06,
      "loss": 0.0,
      "step": 44920
    },
    {
      "epoch": 0.6352415557974805,
      "grad_norm": 0.0,
      "learning_rate": 3.6511190600743684e-06,
      "loss": 0.0,
      "step": 44930
    },
    {
      "epoch": 0.6353829405194474,
      "grad_norm": 0.0,
      "learning_rate": 3.6497052128546997e-06,
      "loss": 0.0004,
      "step": 44940
    },
    {
      "epoch": 0.6355243252414144,
      "grad_norm": 0.0,
      "learning_rate": 3.64829136563503e-06,
      "loss": 0.0,
      "step": 44950
    },
    {
      "epoch": 0.6356657099633813,
      "grad_norm": 5.216185261502915e-10,
      "learning_rate": 3.6468775184153606e-06,
      "loss": 0.0923,
      "step": 44960
    },
    {
      "epoch": 0.6358070946853484,
      "grad_norm": 0.0,
      "learning_rate": 3.645463671195691e-06,
      "loss": 0.0,
      "step": 44970
    },
    {
      "epoch": 0.6359484794073152,
      "grad_norm": 6.720245437463745e-05,
      "learning_rate": 3.6440498239760214e-06,
      "loss": 0.0,
      "step": 44980
    },
    {
      "epoch": 0.6360898641292821,
      "grad_norm": 0.0,
      "learning_rate": 3.642635976756352e-06,
      "loss": 0.0,
      "step": 44990
    },
    {
      "epoch": 0.6362312488512492,
      "grad_norm": 0.0,
      "learning_rate": 3.6412221295366823e-06,
      "loss": 0.0001,
      "step": 45000
    },
    {
      "epoch": 0.636372633573216,
      "grad_norm": 0.0,
      "learning_rate": 3.639808282317013e-06,
      "loss": 0.0,
      "step": 45010
    },
    {
      "epoch": 0.6365140182951831,
      "grad_norm": 24.773508071899414,
      "learning_rate": 3.638394435097344e-06,
      "loss": 0.0038,
      "step": 45020
    },
    {
      "epoch": 0.63665540301715,
      "grad_norm": 0.0,
      "learning_rate": 3.6369805878776744e-06,
      "loss": 0.0,
      "step": 45030
    },
    {
      "epoch": 0.636796787739117,
      "grad_norm": 0.0,
      "learning_rate": 3.635566740658005e-06,
      "loss": 0.0094,
      "step": 45040
    },
    {
      "epoch": 0.6369381724610839,
      "grad_norm": 0.0,
      "learning_rate": 3.6341528934383353e-06,
      "loss": 0.0012,
      "step": 45050
    },
    {
      "epoch": 0.6370795571830508,
      "grad_norm": 0.0012263822136446834,
      "learning_rate": 3.6327390462186657e-06,
      "loss": 0.0,
      "step": 45060
    },
    {
      "epoch": 0.6372209419050178,
      "grad_norm": 6.905621376063209e-06,
      "learning_rate": 3.6313251989989966e-06,
      "loss": 0.0,
      "step": 45070
    },
    {
      "epoch": 0.6373623266269847,
      "grad_norm": 0.010561809875071049,
      "learning_rate": 3.629911351779327e-06,
      "loss": 0.0,
      "step": 45080
    },
    {
      "epoch": 0.6375037113489517,
      "grad_norm": 0.0039110369980335236,
      "learning_rate": 3.6284975045596575e-06,
      "loss": 0.0,
      "step": 45090
    },
    {
      "epoch": 0.6376450960709186,
      "grad_norm": 2.7649064577417448e-05,
      "learning_rate": 3.6270836573399883e-06,
      "loss": 0.0,
      "step": 45100
    },
    {
      "epoch": 0.6377864807928855,
      "grad_norm": 0.0,
      "learning_rate": 3.6256698101203187e-06,
      "loss": 0.0,
      "step": 45110
    },
    {
      "epoch": 0.6379278655148525,
      "grad_norm": 0.00015850916679482907,
      "learning_rate": 3.624255962900649e-06,
      "loss": 0.0,
      "step": 45120
    },
    {
      "epoch": 0.6380692502368194,
      "grad_norm": 0.0,
      "learning_rate": 3.6228421156809796e-06,
      "loss": 0.0,
      "step": 45130
    },
    {
      "epoch": 0.6382106349587864,
      "grad_norm": 0.0,
      "learning_rate": 3.6214282684613105e-06,
      "loss": 0.0,
      "step": 45140
    },
    {
      "epoch": 0.6383520196807533,
      "grad_norm": 0.002966279862448573,
      "learning_rate": 3.620014421241641e-06,
      "loss": 0.0002,
      "step": 45150
    },
    {
      "epoch": 0.6384934044027203,
      "grad_norm": 0.0,
      "learning_rate": 3.6186005740219713e-06,
      "loss": 0.0,
      "step": 45160
    },
    {
      "epoch": 0.6386347891246872,
      "grad_norm": 0.0,
      "learning_rate": 3.617186726802302e-06,
      "loss": 0.3449,
      "step": 45170
    },
    {
      "epoch": 0.6387761738466541,
      "grad_norm": 0.0,
      "learning_rate": 3.6157728795826326e-06,
      "loss": 0.0001,
      "step": 45180
    },
    {
      "epoch": 0.6389175585686211,
      "grad_norm": 7.934576728985121e-07,
      "learning_rate": 3.614359032362963e-06,
      "loss": 0.0,
      "step": 45190
    },
    {
      "epoch": 0.639058943290588,
      "grad_norm": 0.0,
      "learning_rate": 3.612945185143294e-06,
      "loss": 0.0,
      "step": 45200
    },
    {
      "epoch": 0.639200328012555,
      "grad_norm": 0.0,
      "learning_rate": 3.6115313379236243e-06,
      "loss": 0.0,
      "step": 45210
    },
    {
      "epoch": 0.6393417127345219,
      "grad_norm": 0.0,
      "learning_rate": 3.6101174907039548e-06,
      "loss": 0.0001,
      "step": 45220
    },
    {
      "epoch": 0.6394830974564889,
      "grad_norm": 0.0,
      "learning_rate": 3.608703643484285e-06,
      "loss": 0.0092,
      "step": 45230
    },
    {
      "epoch": 0.6396244821784558,
      "grad_norm": 0.0,
      "learning_rate": 3.6072897962646156e-06,
      "loss": 0.0,
      "step": 45240
    },
    {
      "epoch": 0.6397658669004227,
      "grad_norm": 1.0220277317785076e-06,
      "learning_rate": 3.6058759490449465e-06,
      "loss": 0.0,
      "step": 45250
    },
    {
      "epoch": 0.6399072516223897,
      "grad_norm": 4.862804603433801e-10,
      "learning_rate": 3.604462101825277e-06,
      "loss": 0.0071,
      "step": 45260
    },
    {
      "epoch": 0.6400486363443566,
      "grad_norm": 0.0,
      "learning_rate": 3.6030482546056078e-06,
      "loss": 0.0036,
      "step": 45270
    },
    {
      "epoch": 0.6401900210663236,
      "grad_norm": 1.1731185913085938,
      "learning_rate": 3.6016344073859382e-06,
      "loss": 0.0002,
      "step": 45280
    },
    {
      "epoch": 0.6403314057882905,
      "grad_norm": 0.0001220154736074619,
      "learning_rate": 3.6002205601662687e-06,
      "loss": 0.0016,
      "step": 45290
    },
    {
      "epoch": 0.6404727905102575,
      "grad_norm": 0.0,
      "learning_rate": 3.598806712946599e-06,
      "loss": 0.0012,
      "step": 45300
    },
    {
      "epoch": 0.6406141752322244,
      "grad_norm": 0.0,
      "learning_rate": 3.5973928657269295e-06,
      "loss": 0.0011,
      "step": 45310
    },
    {
      "epoch": 0.6407555599541913,
      "grad_norm": 5.321909611666342e-06,
      "learning_rate": 3.5959790185072604e-06,
      "loss": 0.0309,
      "step": 45320
    },
    {
      "epoch": 0.6408969446761583,
      "grad_norm": 2.022819217017968e-06,
      "learning_rate": 3.5945651712875912e-06,
      "loss": 0.0,
      "step": 45330
    },
    {
      "epoch": 0.6410383293981252,
      "grad_norm": 0.8735533356666565,
      "learning_rate": 3.5931513240679217e-06,
      "loss": 0.0008,
      "step": 45340
    },
    {
      "epoch": 0.6411797141200922,
      "grad_norm": 0.0,
      "learning_rate": 3.591737476848252e-06,
      "loss": 0.0002,
      "step": 45350
    },
    {
      "epoch": 0.6413210988420591,
      "grad_norm": 0.0,
      "learning_rate": 3.5903236296285825e-06,
      "loss": 0.0,
      "step": 45360
    },
    {
      "epoch": 0.641462483564026,
      "grad_norm": 0.03121117129921913,
      "learning_rate": 3.588909782408913e-06,
      "loss": 0.0,
      "step": 45370
    },
    {
      "epoch": 0.641603868285993,
      "grad_norm": 1.235353153106189e-07,
      "learning_rate": 3.5874959351892434e-06,
      "loss": 0.0039,
      "step": 45380
    },
    {
      "epoch": 0.6417452530079599,
      "grad_norm": 0.0,
      "learning_rate": 3.5860820879695747e-06,
      "loss": 0.0,
      "step": 45390
    },
    {
      "epoch": 0.6418866377299269,
      "grad_norm": 0.0,
      "learning_rate": 3.584668240749905e-06,
      "loss": 0.0,
      "step": 45400
    },
    {
      "epoch": 0.6420280224518938,
      "grad_norm": 0.0,
      "learning_rate": 3.5832543935302355e-06,
      "loss": 0.0655,
      "step": 45410
    },
    {
      "epoch": 0.6421694071738608,
      "grad_norm": 5.535103397846797e-08,
      "learning_rate": 3.581840546310566e-06,
      "loss": 0.0,
      "step": 45420
    },
    {
      "epoch": 0.6423107918958277,
      "grad_norm": 0.0,
      "learning_rate": 3.5804266990908964e-06,
      "loss": 0.0002,
      "step": 45430
    },
    {
      "epoch": 0.6424521766177946,
      "grad_norm": 0.0,
      "learning_rate": 3.579012851871227e-06,
      "loss": 0.0001,
      "step": 45440
    },
    {
      "epoch": 0.6425935613397616,
      "grad_norm": 0.0,
      "learning_rate": 3.5775990046515573e-06,
      "loss": 0.0002,
      "step": 45450
    },
    {
      "epoch": 0.6427349460617285,
      "grad_norm": 0.0,
      "learning_rate": 3.5761851574318886e-06,
      "loss": 0.0,
      "step": 45460
    },
    {
      "epoch": 0.6428763307836955,
      "grad_norm": 0.0,
      "learning_rate": 3.574771310212219e-06,
      "loss": 0.0,
      "step": 45470
    },
    {
      "epoch": 0.6430177155056624,
      "grad_norm": 0.0,
      "learning_rate": 3.5733574629925494e-06,
      "loss": 0.8578,
      "step": 45480
    },
    {
      "epoch": 0.6431591002276295,
      "grad_norm": 0.0,
      "learning_rate": 3.57194361577288e-06,
      "loss": 0.0,
      "step": 45490
    },
    {
      "epoch": 0.6433004849495964,
      "grad_norm": 0.0,
      "learning_rate": 3.5705297685532103e-06,
      "loss": 0.0,
      "step": 45500
    },
    {
      "epoch": 0.6434418696715632,
      "grad_norm": 0.19594496488571167,
      "learning_rate": 3.5691159213335407e-06,
      "loss": 0.0,
      "step": 45510
    },
    {
      "epoch": 0.6435832543935303,
      "grad_norm": 0.0,
      "learning_rate": 3.567702074113872e-06,
      "loss": 0.0,
      "step": 45520
    },
    {
      "epoch": 0.6437246391154972,
      "grad_norm": 4.521912289057184e-10,
      "learning_rate": 3.5662882268942024e-06,
      "loss": 0.0,
      "step": 45530
    },
    {
      "epoch": 0.6438660238374642,
      "grad_norm": 0.0,
      "learning_rate": 3.564874379674533e-06,
      "loss": 0.0,
      "step": 45540
    },
    {
      "epoch": 0.6440074085594311,
      "grad_norm": 0.0,
      "learning_rate": 3.5634605324548633e-06,
      "loss": 0.0,
      "step": 45550
    },
    {
      "epoch": 0.644148793281398,
      "grad_norm": 0.0,
      "learning_rate": 3.5620466852351937e-06,
      "loss": 0.0,
      "step": 45560
    },
    {
      "epoch": 0.644290178003365,
      "grad_norm": 0.00012085979687981308,
      "learning_rate": 3.560632838015524e-06,
      "loss": 0.0,
      "step": 45570
    },
    {
      "epoch": 0.6444315627253319,
      "grad_norm": 9.171171044819459e-10,
      "learning_rate": 3.5592189907958546e-06,
      "loss": 0.0,
      "step": 45580
    },
    {
      "epoch": 0.6445729474472989,
      "grad_norm": 4.5910777957125504e-10,
      "learning_rate": 3.557805143576186e-06,
      "loss": 0.0229,
      "step": 45590
    },
    {
      "epoch": 0.6447143321692658,
      "grad_norm": 0.0814618170261383,
      "learning_rate": 3.5563912963565163e-06,
      "loss": 0.0,
      "step": 45600
    },
    {
      "epoch": 0.6448557168912328,
      "grad_norm": 0.0,
      "learning_rate": 3.5549774491368467e-06,
      "loss": 0.0,
      "step": 45610
    },
    {
      "epoch": 0.6449971016131997,
      "grad_norm": 0.0,
      "learning_rate": 3.553563601917177e-06,
      "loss": 0.0,
      "step": 45620
    },
    {
      "epoch": 0.6451384863351666,
      "grad_norm": 0.0,
      "learning_rate": 3.5521497546975076e-06,
      "loss": 0.359,
      "step": 45630
    },
    {
      "epoch": 0.6452798710571336,
      "grad_norm": 0.0,
      "learning_rate": 3.550735907477838e-06,
      "loss": 0.0002,
      "step": 45640
    },
    {
      "epoch": 0.6454212557791005,
      "grad_norm": 0.40963828563690186,
      "learning_rate": 3.549322060258169e-06,
      "loss": 0.0001,
      "step": 45650
    },
    {
      "epoch": 0.6455626405010675,
      "grad_norm": 6.712524314167467e-09,
      "learning_rate": 3.5479082130384993e-06,
      "loss": 0.0,
      "step": 45660
    },
    {
      "epoch": 0.6457040252230344,
      "grad_norm": 0.0,
      "learning_rate": 3.54649436581883e-06,
      "loss": 0.0002,
      "step": 45670
    },
    {
      "epoch": 0.6458454099450014,
      "grad_norm": 0.0,
      "learning_rate": 3.5450805185991606e-06,
      "loss": 0.0,
      "step": 45680
    },
    {
      "epoch": 0.6459867946669683,
      "grad_norm": 0.0002088045293930918,
      "learning_rate": 3.543666671379491e-06,
      "loss": 0.9531,
      "step": 45690
    },
    {
      "epoch": 0.6461281793889352,
      "grad_norm": 0.0,
      "learning_rate": 3.5422528241598215e-06,
      "loss": 0.0005,
      "step": 45700
    },
    {
      "epoch": 0.6462695641109022,
      "grad_norm": 0.0,
      "learning_rate": 3.540838976940152e-06,
      "loss": 0.0,
      "step": 45710
    },
    {
      "epoch": 0.6464109488328691,
      "grad_norm": 8.348417281922593e-07,
      "learning_rate": 3.5394251297204828e-06,
      "loss": 0.0,
      "step": 45720
    },
    {
      "epoch": 0.6465523335548361,
      "grad_norm": 0.0,
      "learning_rate": 3.538011282500813e-06,
      "loss": 0.0,
      "step": 45730
    },
    {
      "epoch": 0.646693718276803,
      "grad_norm": 0.0,
      "learning_rate": 3.536597435281144e-06,
      "loss": 0.0,
      "step": 45740
    },
    {
      "epoch": 0.64683510299877,
      "grad_norm": 0.0,
      "learning_rate": 3.5351835880614745e-06,
      "loss": 0.0,
      "step": 45750
    },
    {
      "epoch": 0.6469764877207369,
      "grad_norm": 1.5507982969284058,
      "learning_rate": 3.533769740841805e-06,
      "loss": 0.0002,
      "step": 45760
    },
    {
      "epoch": 0.6471178724427038,
      "grad_norm": 0.0,
      "learning_rate": 3.5323558936221354e-06,
      "loss": 0.0,
      "step": 45770
    },
    {
      "epoch": 0.6472592571646708,
      "grad_norm": 0.0,
      "learning_rate": 3.5309420464024662e-06,
      "loss": 0.0,
      "step": 45780
    },
    {
      "epoch": 0.6474006418866377,
      "grad_norm": 0.0,
      "learning_rate": 3.5295281991827967e-06,
      "loss": 0.0,
      "step": 45790
    },
    {
      "epoch": 0.6475420266086047,
      "grad_norm": 0.0,
      "learning_rate": 3.528114351963127e-06,
      "loss": 0.0,
      "step": 45800
    },
    {
      "epoch": 0.6476834113305716,
      "grad_norm": 0.0,
      "learning_rate": 3.5267005047434575e-06,
      "loss": 0.2646,
      "step": 45810
    },
    {
      "epoch": 0.6478247960525385,
      "grad_norm": 409.03759765625,
      "learning_rate": 3.5252866575237884e-06,
      "loss": 0.1818,
      "step": 45820
    },
    {
      "epoch": 0.6479661807745055,
      "grad_norm": 2.54584762160448e-07,
      "learning_rate": 3.523872810304119e-06,
      "loss": 0.0,
      "step": 45830
    },
    {
      "epoch": 0.6481075654964724,
      "grad_norm": 0.0,
      "learning_rate": 3.5224589630844492e-06,
      "loss": 0.3996,
      "step": 45840
    },
    {
      "epoch": 0.6482489502184394,
      "grad_norm": 0.0,
      "learning_rate": 3.52104511586478e-06,
      "loss": 0.0,
      "step": 45850
    },
    {
      "epoch": 0.6483903349404063,
      "grad_norm": 0.0,
      "learning_rate": 3.5196312686451105e-06,
      "loss": 0.0,
      "step": 45860
    },
    {
      "epoch": 0.6485317196623733,
      "grad_norm": 7.571069886580517e-07,
      "learning_rate": 3.518217421425441e-06,
      "loss": 0.0,
      "step": 45870
    },
    {
      "epoch": 0.6486731043843402,
      "grad_norm": 0.0,
      "learning_rate": 3.5168035742057714e-06,
      "loss": 0.0,
      "step": 45880
    },
    {
      "epoch": 0.6488144891063071,
      "grad_norm": 1.5780329704284668e-05,
      "learning_rate": 3.515389726986102e-06,
      "loss": 0.0,
      "step": 45890
    },
    {
      "epoch": 0.6489558738282741,
      "grad_norm": 0.0,
      "learning_rate": 3.5139758797664327e-06,
      "loss": 0.0,
      "step": 45900
    },
    {
      "epoch": 0.649097258550241,
      "grad_norm": 0.0,
      "learning_rate": 3.5125620325467635e-06,
      "loss": 0.0,
      "step": 45910
    },
    {
      "epoch": 0.649238643272208,
      "grad_norm": 0.0,
      "learning_rate": 3.511148185327094e-06,
      "loss": 0.0,
      "step": 45920
    },
    {
      "epoch": 0.6493800279941749,
      "grad_norm": 0.0,
      "learning_rate": 3.5097343381074244e-06,
      "loss": 0.0,
      "step": 45930
    },
    {
      "epoch": 0.6495214127161419,
      "grad_norm": 0.0,
      "learning_rate": 3.508320490887755e-06,
      "loss": 0.0,
      "step": 45940
    },
    {
      "epoch": 0.6496627974381088,
      "grad_norm": 0.0,
      "learning_rate": 3.5069066436680853e-06,
      "loss": 0.0,
      "step": 45950
    },
    {
      "epoch": 0.6498041821600757,
      "grad_norm": 0.0,
      "learning_rate": 3.5054927964484157e-06,
      "loss": 0.0027,
      "step": 45960
    },
    {
      "epoch": 0.6499455668820427,
      "grad_norm": 53.4996337890625,
      "learning_rate": 3.5040789492287466e-06,
      "loss": 0.0115,
      "step": 45970
    },
    {
      "epoch": 0.6500869516040096,
      "grad_norm": 0.0,
      "learning_rate": 3.5026651020090774e-06,
      "loss": 0.0,
      "step": 45980
    },
    {
      "epoch": 0.6502283363259767,
      "grad_norm": 0.0,
      "learning_rate": 3.501251254789408e-06,
      "loss": 0.0002,
      "step": 45990
    },
    {
      "epoch": 0.6503697210479435,
      "grad_norm": 4.1344443161506206e-05,
      "learning_rate": 3.4998374075697383e-06,
      "loss": 0.0,
      "step": 46000
    },
    {
      "epoch": 0.6505111057699106,
      "grad_norm": 0.0,
      "learning_rate": 3.4984235603500687e-06,
      "loss": 0.0001,
      "step": 46010
    },
    {
      "epoch": 0.6506524904918775,
      "grad_norm": 0.0,
      "learning_rate": 3.497009713130399e-06,
      "loss": 0.0,
      "step": 46020
    },
    {
      "epoch": 0.6507938752138444,
      "grad_norm": 0.0,
      "learning_rate": 3.4955958659107296e-06,
      "loss": 0.0044,
      "step": 46030
    },
    {
      "epoch": 0.6509352599358114,
      "grad_norm": 0.06635527312755585,
      "learning_rate": 3.494182018691061e-06,
      "loss": 0.0,
      "step": 46040
    },
    {
      "epoch": 0.6510766446577783,
      "grad_norm": 0.00013548217248171568,
      "learning_rate": 3.4927681714713913e-06,
      "loss": 0.0001,
      "step": 46050
    },
    {
      "epoch": 0.6512180293797453,
      "grad_norm": 0.0,
      "learning_rate": 3.4913543242517217e-06,
      "loss": 0.0,
      "step": 46060
    },
    {
      "epoch": 0.6513594141017122,
      "grad_norm": 0.0,
      "learning_rate": 3.489940477032052e-06,
      "loss": 0.0,
      "step": 46070
    },
    {
      "epoch": 0.6515007988236791,
      "grad_norm": 0.0,
      "learning_rate": 3.4885266298123826e-06,
      "loss": 0.0,
      "step": 46080
    },
    {
      "epoch": 0.6516421835456461,
      "grad_norm": 0.0,
      "learning_rate": 3.487112782592713e-06,
      "loss": 0.1006,
      "step": 46090
    },
    {
      "epoch": 0.651783568267613,
      "grad_norm": 0.0,
      "learning_rate": 3.4856989353730435e-06,
      "loss": 0.0,
      "step": 46100
    },
    {
      "epoch": 0.65192495298958,
      "grad_norm": 1.0320836305618286,
      "learning_rate": 3.4842850881533747e-06,
      "loss": 0.0002,
      "step": 46110
    },
    {
      "epoch": 0.6520663377115469,
      "grad_norm": 0.0,
      "learning_rate": 3.482871240933705e-06,
      "loss": 0.0,
      "step": 46120
    },
    {
      "epoch": 0.6522077224335139,
      "grad_norm": 9.425487223779783e-05,
      "learning_rate": 3.4814573937140356e-06,
      "loss": 0.0,
      "step": 46130
    },
    {
      "epoch": 0.6523491071554808,
      "grad_norm": 0.0,
      "learning_rate": 3.480043546494366e-06,
      "loss": 0.0,
      "step": 46140
    },
    {
      "epoch": 0.6524904918774477,
      "grad_norm": 0.0,
      "learning_rate": 3.4786296992746965e-06,
      "loss": 0.0,
      "step": 46150
    },
    {
      "epoch": 0.6526318765994147,
      "grad_norm": 0.0,
      "learning_rate": 3.477215852055027e-06,
      "loss": 0.0,
      "step": 46160
    },
    {
      "epoch": 0.6527732613213816,
      "grad_norm": 0.0,
      "learning_rate": 3.475802004835358e-06,
      "loss": 0.0,
      "step": 46170
    },
    {
      "epoch": 0.6529146460433486,
      "grad_norm": 0.0,
      "learning_rate": 3.4743881576156886e-06,
      "loss": 0.0,
      "step": 46180
    },
    {
      "epoch": 0.6530560307653155,
      "grad_norm": 2.0853516957686224e-07,
      "learning_rate": 3.472974310396019e-06,
      "loss": 0.0,
      "step": 46190
    },
    {
      "epoch": 0.6531974154872825,
      "grad_norm": 0.0,
      "learning_rate": 3.4715604631763495e-06,
      "loss": 0.0001,
      "step": 46200
    },
    {
      "epoch": 0.6533388002092494,
      "grad_norm": 0.0,
      "learning_rate": 3.47014661595668e-06,
      "loss": 0.0,
      "step": 46210
    },
    {
      "epoch": 0.6534801849312163,
      "grad_norm": 0.0,
      "learning_rate": 3.4687327687370104e-06,
      "loss": 0.0,
      "step": 46220
    },
    {
      "epoch": 0.6536215696531833,
      "grad_norm": 0.7588031888008118,
      "learning_rate": 3.467318921517341e-06,
      "loss": 0.0001,
      "step": 46230
    },
    {
      "epoch": 0.6537629543751502,
      "grad_norm": 0.0,
      "learning_rate": 3.465905074297672e-06,
      "loss": 0.0,
      "step": 46240
    },
    {
      "epoch": 0.6539043390971172,
      "grad_norm": 0.0,
      "learning_rate": 3.4644912270780025e-06,
      "loss": 0.0,
      "step": 46250
    },
    {
      "epoch": 0.6540457238190841,
      "grad_norm": 0.0,
      "learning_rate": 3.463077379858333e-06,
      "loss": 0.0,
      "step": 46260
    },
    {
      "epoch": 0.654187108541051,
      "grad_norm": 0.0001060556314769201,
      "learning_rate": 3.4616635326386634e-06,
      "loss": 0.0,
      "step": 46270
    },
    {
      "epoch": 0.654328493263018,
      "grad_norm": 0.0,
      "learning_rate": 3.460249685418994e-06,
      "loss": 0.0,
      "step": 46280
    },
    {
      "epoch": 0.6544698779849849,
      "grad_norm": 0.0,
      "learning_rate": 3.4588358381993242e-06,
      "loss": 0.0,
      "step": 46290
    },
    {
      "epoch": 0.6546112627069519,
      "grad_norm": 0.0,
      "learning_rate": 3.457421990979655e-06,
      "loss": 0.0,
      "step": 46300
    },
    {
      "epoch": 0.6547526474289188,
      "grad_norm": 0.0,
      "learning_rate": 3.4560081437599855e-06,
      "loss": 0.001,
      "step": 46310
    },
    {
      "epoch": 0.6548940321508858,
      "grad_norm": 0.0,
      "learning_rate": 3.4545942965403164e-06,
      "loss": 0.0,
      "step": 46320
    },
    {
      "epoch": 0.6550354168728527,
      "grad_norm": 0.0,
      "learning_rate": 3.453180449320647e-06,
      "loss": 0.0002,
      "step": 46330
    },
    {
      "epoch": 0.6551768015948196,
      "grad_norm": 6.381125672305643e-07,
      "learning_rate": 3.4517666021009772e-06,
      "loss": 0.6145,
      "step": 46340
    },
    {
      "epoch": 0.6553181863167866,
      "grad_norm": 8.746406376936022e-10,
      "learning_rate": 3.4503527548813077e-06,
      "loss": 0.0,
      "step": 46350
    },
    {
      "epoch": 0.6554595710387535,
      "grad_norm": 0.0,
      "learning_rate": 3.4489389076616385e-06,
      "loss": 0.0,
      "step": 46360
    },
    {
      "epoch": 0.6556009557607205,
      "grad_norm": 0.0,
      "learning_rate": 3.447525060441969e-06,
      "loss": 0.0,
      "step": 46370
    },
    {
      "epoch": 0.6557423404826874,
      "grad_norm": 0.0,
      "learning_rate": 3.4461112132222994e-06,
      "loss": 0.0025,
      "step": 46380
    },
    {
      "epoch": 0.6558837252046544,
      "grad_norm": 0.0,
      "learning_rate": 3.4446973660026303e-06,
      "loss": 0.0,
      "step": 46390
    },
    {
      "epoch": 0.6560251099266213,
      "grad_norm": 0.0,
      "learning_rate": 3.4432835187829607e-06,
      "loss": 0.0,
      "step": 46400
    },
    {
      "epoch": 0.6561664946485882,
      "grad_norm": 0.0,
      "learning_rate": 3.441869671563291e-06,
      "loss": 0.0,
      "step": 46410
    },
    {
      "epoch": 0.6563078793705552,
      "grad_norm": 0.05408221110701561,
      "learning_rate": 3.4404558243436216e-06,
      "loss": 0.0004,
      "step": 46420
    },
    {
      "epoch": 0.6564492640925221,
      "grad_norm": 0.0,
      "learning_rate": 3.4390419771239524e-06,
      "loss": 0.0004,
      "step": 46430
    },
    {
      "epoch": 0.6565906488144891,
      "grad_norm": 0.0,
      "learning_rate": 3.437628129904283e-06,
      "loss": 0.0003,
      "step": 46440
    },
    {
      "epoch": 0.656732033536456,
      "grad_norm": 0.0,
      "learning_rate": 3.4362142826846133e-06,
      "loss": 0.0044,
      "step": 46450
    },
    {
      "epoch": 0.656873418258423,
      "grad_norm": 0.0,
      "learning_rate": 3.4348004354649437e-06,
      "loss": 0.0,
      "step": 46460
    },
    {
      "epoch": 0.65701480298039,
      "grad_norm": 0.0,
      "learning_rate": 3.4333865882452746e-06,
      "loss": 0.0,
      "step": 46470
    },
    {
      "epoch": 0.6571561877023568,
      "grad_norm": 0.0,
      "learning_rate": 3.431972741025605e-06,
      "loss": 0.0,
      "step": 46480
    },
    {
      "epoch": 0.6572975724243239,
      "grad_norm": 22.985477447509766,
      "learning_rate": 3.430558893805936e-06,
      "loss": 0.0036,
      "step": 46490
    },
    {
      "epoch": 0.6574389571462907,
      "grad_norm": 0.0,
      "learning_rate": 3.4291450465862663e-06,
      "loss": 0.0,
      "step": 46500
    },
    {
      "epoch": 0.6575803418682578,
      "grad_norm": 0.032615385949611664,
      "learning_rate": 3.4277311993665967e-06,
      "loss": 0.0272,
      "step": 46510
    },
    {
      "epoch": 0.6577217265902247,
      "grad_norm": 0.10531293600797653,
      "learning_rate": 3.426317352146927e-06,
      "loss": 0.0,
      "step": 46520
    },
    {
      "epoch": 0.6578631113121916,
      "grad_norm": 0.0,
      "learning_rate": 3.4249035049272576e-06,
      "loss": 0.0,
      "step": 46530
    },
    {
      "epoch": 0.6580044960341586,
      "grad_norm": 8.027704461710528e-05,
      "learning_rate": 3.4234896577075884e-06,
      "loss": 0.0,
      "step": 46540
    },
    {
      "epoch": 0.6581458807561255,
      "grad_norm": 0.0,
      "learning_rate": 3.422075810487919e-06,
      "loss": 0.0,
      "step": 46550
    },
    {
      "epoch": 0.6582872654780925,
      "grad_norm": 0.0,
      "learning_rate": 3.4206619632682497e-06,
      "loss": 0.0011,
      "step": 46560
    },
    {
      "epoch": 0.6584286502000594,
      "grad_norm": 0.0,
      "learning_rate": 3.41924811604858e-06,
      "loss": 0.0,
      "step": 46570
    },
    {
      "epoch": 0.6585700349220264,
      "grad_norm": 1.6608382793492638e-05,
      "learning_rate": 3.4178342688289106e-06,
      "loss": 0.001,
      "step": 46580
    },
    {
      "epoch": 0.6587114196439933,
      "grad_norm": 0.0,
      "learning_rate": 3.416420421609241e-06,
      "loss": 0.0,
      "step": 46590
    },
    {
      "epoch": 0.6588528043659602,
      "grad_norm": 1.0158664736081846e-06,
      "learning_rate": 3.4150065743895715e-06,
      "loss": 0.0,
      "step": 46600
    },
    {
      "epoch": 0.6589941890879272,
      "grad_norm": 0.0,
      "learning_rate": 3.413592727169902e-06,
      "loss": 0.5481,
      "step": 46610
    },
    {
      "epoch": 0.6591355738098941,
      "grad_norm": 7.358831766168805e-09,
      "learning_rate": 3.412178879950233e-06,
      "loss": 0.0,
      "step": 46620
    },
    {
      "epoch": 0.6592769585318611,
      "grad_norm": 5.351907020667568e-06,
      "learning_rate": 3.4107650327305636e-06,
      "loss": 0.0103,
      "step": 46630
    },
    {
      "epoch": 0.659418343253828,
      "grad_norm": 0.0,
      "learning_rate": 3.409351185510894e-06,
      "loss": 0.0,
      "step": 46640
    },
    {
      "epoch": 0.659559727975795,
      "grad_norm": 0.0,
      "learning_rate": 3.4079373382912245e-06,
      "loss": 0.0008,
      "step": 46650
    },
    {
      "epoch": 0.6597011126977619,
      "grad_norm": 0.0,
      "learning_rate": 3.406523491071555e-06,
      "loss": 0.1137,
      "step": 46660
    },
    {
      "epoch": 0.6598424974197288,
      "grad_norm": 0.0,
      "learning_rate": 3.4051096438518853e-06,
      "loss": 0.0038,
      "step": 46670
    },
    {
      "epoch": 0.6599838821416958,
      "grad_norm": 0.0,
      "learning_rate": 3.4036957966322158e-06,
      "loss": 0.0486,
      "step": 46680
    },
    {
      "epoch": 0.6601252668636627,
      "grad_norm": 0.0,
      "learning_rate": 3.402281949412547e-06,
      "loss": 0.0,
      "step": 46690
    },
    {
      "epoch": 0.6602666515856297,
      "grad_norm": 0.0,
      "learning_rate": 3.4008681021928775e-06,
      "loss": 0.0,
      "step": 46700
    },
    {
      "epoch": 0.6604080363075966,
      "grad_norm": 1.3578875268649426e-06,
      "learning_rate": 3.399454254973208e-06,
      "loss": 0.0005,
      "step": 46710
    },
    {
      "epoch": 0.6605494210295636,
      "grad_norm": 4.2944407141476404e-07,
      "learning_rate": 3.3980404077535384e-06,
      "loss": 0.0,
      "step": 46720
    },
    {
      "epoch": 0.6606908057515305,
      "grad_norm": 134.25872802734375,
      "learning_rate": 3.3966265605338688e-06,
      "loss": 0.0232,
      "step": 46730
    },
    {
      "epoch": 0.6608321904734974,
      "grad_norm": 0.0,
      "learning_rate": 3.3952127133141992e-06,
      "loss": 0.0,
      "step": 46740
    },
    {
      "epoch": 0.6609735751954644,
      "grad_norm": 0.0,
      "learning_rate": 3.3937988660945305e-06,
      "loss": 0.0,
      "step": 46750
    },
    {
      "epoch": 0.6611149599174313,
      "grad_norm": 0.00026442919624969363,
      "learning_rate": 3.392385018874861e-06,
      "loss": 0.0,
      "step": 46760
    },
    {
      "epoch": 0.6612563446393983,
      "grad_norm": 0.030715547502040863,
      "learning_rate": 3.3911125563771578e-06,
      "loss": 0.7805,
      "step": 46770
    },
    {
      "epoch": 0.6613977293613652,
      "grad_norm": 0.0,
      "learning_rate": 3.3896987091574886e-06,
      "loss": 0.0022,
      "step": 46780
    },
    {
      "epoch": 0.6615391140833321,
      "grad_norm": 8.056298611336388e-06,
      "learning_rate": 3.3882848619378195e-06,
      "loss": 0.0,
      "step": 46790
    },
    {
      "epoch": 0.6616804988052991,
      "grad_norm": 0.0,
      "learning_rate": 3.38687101471815e-06,
      "loss": 0.0,
      "step": 46800
    },
    {
      "epoch": 0.661821883527266,
      "grad_norm": 9.206931883554148e-10,
      "learning_rate": 3.3854571674984803e-06,
      "loss": 0.0,
      "step": 46810
    },
    {
      "epoch": 0.661963268249233,
      "grad_norm": 0.0,
      "learning_rate": 3.3840433202788108e-06,
      "loss": 0.0569,
      "step": 46820
    },
    {
      "epoch": 0.6621046529711999,
      "grad_norm": 0.0,
      "learning_rate": 3.3826294730591412e-06,
      "loss": 0.0005,
      "step": 46830
    },
    {
      "epoch": 0.6622460376931669,
      "grad_norm": 0.0,
      "learning_rate": 3.3812156258394716e-06,
      "loss": 0.0,
      "step": 46840
    },
    {
      "epoch": 0.6623874224151338,
      "grad_norm": 5.916868772715134e-09,
      "learning_rate": 3.379801778619803e-06,
      "loss": 0.0,
      "step": 46850
    },
    {
      "epoch": 0.6625288071371007,
      "grad_norm": 0.0,
      "learning_rate": 3.3783879314001334e-06,
      "loss": 0.0,
      "step": 46860
    },
    {
      "epoch": 0.6626701918590677,
      "grad_norm": 6.250405931496061e-06,
      "learning_rate": 3.376974084180464e-06,
      "loss": 0.0002,
      "step": 46870
    },
    {
      "epoch": 0.6628115765810346,
      "grad_norm": 0.0,
      "learning_rate": 3.3755602369607942e-06,
      "loss": 0.0,
      "step": 46880
    },
    {
      "epoch": 0.6629529613030016,
      "grad_norm": 0.0,
      "learning_rate": 3.3741463897411247e-06,
      "loss": 0.0025,
      "step": 46890
    },
    {
      "epoch": 0.6630943460249685,
      "grad_norm": 5.250404000456399e-10,
      "learning_rate": 3.372732542521455e-06,
      "loss": 0.1054,
      "step": 46900
    },
    {
      "epoch": 0.6632357307469355,
      "grad_norm": 0.0,
      "learning_rate": 3.3713186953017864e-06,
      "loss": 0.0008,
      "step": 46910
    },
    {
      "epoch": 0.6633771154689024,
      "grad_norm": 0.0,
      "learning_rate": 3.369904848082117e-06,
      "loss": 0.0,
      "step": 46920
    },
    {
      "epoch": 0.6635185001908693,
      "grad_norm": 0.0,
      "learning_rate": 3.3684910008624472e-06,
      "loss": 0.0,
      "step": 46930
    },
    {
      "epoch": 0.6636598849128363,
      "grad_norm": 1.2033259721988543e-08,
      "learning_rate": 3.3670771536427777e-06,
      "loss": 0.0,
      "step": 46940
    },
    {
      "epoch": 0.6638012696348032,
      "grad_norm": 0.0,
      "learning_rate": 3.365663306423108e-06,
      "loss": 0.0007,
      "step": 46950
    },
    {
      "epoch": 0.6639426543567702,
      "grad_norm": 0.0,
      "learning_rate": 3.3642494592034385e-06,
      "loss": 0.0002,
      "step": 46960
    },
    {
      "epoch": 0.6640840390787371,
      "grad_norm": 0.0,
      "learning_rate": 3.362835611983769e-06,
      "loss": 0.0,
      "step": 46970
    },
    {
      "epoch": 0.6642254238007042,
      "grad_norm": 0.0,
      "learning_rate": 3.3614217647641002e-06,
      "loss": 0.0,
      "step": 46980
    },
    {
      "epoch": 0.664366808522671,
      "grad_norm": 0.0,
      "learning_rate": 3.3600079175444307e-06,
      "loss": 0.0,
      "step": 46990
    },
    {
      "epoch": 0.664508193244638,
      "grad_norm": 3.684453758978634e-06,
      "learning_rate": 3.358594070324761e-06,
      "loss": 0.0,
      "step": 47000
    },
    {
      "epoch": 0.664649577966605,
      "grad_norm": 0.020937256515026093,
      "learning_rate": 3.3571802231050915e-06,
      "loss": 0.0,
      "step": 47010
    },
    {
      "epoch": 0.6647909626885719,
      "grad_norm": 2.8062189016964112e-08,
      "learning_rate": 3.355766375885422e-06,
      "loss": 0.0,
      "step": 47020
    },
    {
      "epoch": 0.6649323474105389,
      "grad_norm": 0.005717391148209572,
      "learning_rate": 3.3543525286657524e-06,
      "loss": 0.0043,
      "step": 47030
    },
    {
      "epoch": 0.6650737321325058,
      "grad_norm": 0.0,
      "learning_rate": 3.3529386814460833e-06,
      "loss": 0.0,
      "step": 47040
    },
    {
      "epoch": 0.6652151168544727,
      "grad_norm": 0.0,
      "learning_rate": 3.351524834226414e-06,
      "loss": 0.0421,
      "step": 47050
    },
    {
      "epoch": 0.6653565015764397,
      "grad_norm": 0.0,
      "learning_rate": 3.3501109870067446e-06,
      "loss": 0.0242,
      "step": 47060
    },
    {
      "epoch": 0.6654978862984066,
      "grad_norm": 2.2869066924613435e-06,
      "learning_rate": 3.348697139787075e-06,
      "loss": 0.0,
      "step": 47070
    },
    {
      "epoch": 0.6656392710203736,
      "grad_norm": 9.167731036541227e-07,
      "learning_rate": 3.3472832925674054e-06,
      "loss": 0.003,
      "step": 47080
    },
    {
      "epoch": 0.6657806557423405,
      "grad_norm": 1.3204293281887658e-05,
      "learning_rate": 3.345869445347736e-06,
      "loss": 0.0,
      "step": 47090
    },
    {
      "epoch": 0.6659220404643075,
      "grad_norm": 3.4138611937351016e-08,
      "learning_rate": 3.3444555981280663e-06,
      "loss": 0.0017,
      "step": 47100
    },
    {
      "epoch": 0.6660634251862744,
      "grad_norm": 0.0,
      "learning_rate": 3.343183135630364e-06,
      "loss": 0.0499,
      "step": 47110
    },
    {
      "epoch": 0.6662048099082413,
      "grad_norm": 9.354006458295316e-10,
      "learning_rate": 3.3417692884106944e-06,
      "loss": 0.0005,
      "step": 47120
    },
    {
      "epoch": 0.6663461946302083,
      "grad_norm": 0.007982775568962097,
      "learning_rate": 3.340355441191025e-06,
      "loss": 0.0,
      "step": 47130
    },
    {
      "epoch": 0.6664875793521752,
      "grad_norm": 0.0,
      "learning_rate": 3.338941593971356e-06,
      "loss": 0.0,
      "step": 47140
    },
    {
      "epoch": 0.6666289640741422,
      "grad_norm": 0.0,
      "learning_rate": 3.3375277467516866e-06,
      "loss": 0.0,
      "step": 47150
    },
    {
      "epoch": 0.6667703487961091,
      "grad_norm": 0.0,
      "learning_rate": 3.336113899532017e-06,
      "loss": 0.0,
      "step": 47160
    },
    {
      "epoch": 0.6669117335180761,
      "grad_norm": 1.8239958876620221e-07,
      "learning_rate": 3.3347000523123474e-06,
      "loss": 0.0,
      "step": 47170
    },
    {
      "epoch": 0.667053118240043,
      "grad_norm": 0.0,
      "learning_rate": 3.333286205092678e-06,
      "loss": 0.0,
      "step": 47180
    },
    {
      "epoch": 0.6671945029620099,
      "grad_norm": 0.0,
      "learning_rate": 3.3318723578730083e-06,
      "loss": 0.1029,
      "step": 47190
    },
    {
      "epoch": 0.6673358876839769,
      "grad_norm": 0.0,
      "learning_rate": 3.3304585106533387e-06,
      "loss": 0.0,
      "step": 47200
    },
    {
      "epoch": 0.6674772724059438,
      "grad_norm": 0.0,
      "learning_rate": 3.32904466343367e-06,
      "loss": 0.0001,
      "step": 47210
    },
    {
      "epoch": 0.6676186571279108,
      "grad_norm": 0.0,
      "learning_rate": 3.3276308162140004e-06,
      "loss": 0.0733,
      "step": 47220
    },
    {
      "epoch": 0.6677600418498777,
      "grad_norm": 0.0006639301427640021,
      "learning_rate": 3.326216968994331e-06,
      "loss": 0.0001,
      "step": 47230
    },
    {
      "epoch": 0.6679014265718446,
      "grad_norm": 0.0,
      "learning_rate": 3.3248031217746613e-06,
      "loss": 0.0012,
      "step": 47240
    },
    {
      "epoch": 0.6680428112938116,
      "grad_norm": 0.0,
      "learning_rate": 3.3233892745549917e-06,
      "loss": 0.0,
      "step": 47250
    },
    {
      "epoch": 0.6681841960157785,
      "grad_norm": 1.606752562111069e-06,
      "learning_rate": 3.321975427335322e-06,
      "loss": 0.0043,
      "step": 47260
    },
    {
      "epoch": 0.6683255807377455,
      "grad_norm": 0.0064505962654948235,
      "learning_rate": 3.320561580115653e-06,
      "loss": 0.0,
      "step": 47270
    },
    {
      "epoch": 0.6684669654597124,
      "grad_norm": 0.0,
      "learning_rate": 3.319147732895984e-06,
      "loss": 0.0,
      "step": 47280
    },
    {
      "epoch": 0.6686083501816794,
      "grad_norm": 0.0,
      "learning_rate": 3.3177338856763143e-06,
      "loss": 0.0001,
      "step": 47290
    },
    {
      "epoch": 0.6687497349036463,
      "grad_norm": 0.0,
      "learning_rate": 3.3163200384566447e-06,
      "loss": 0.0,
      "step": 47300
    },
    {
      "epoch": 0.6688911196256132,
      "grad_norm": 3.6558279991149902,
      "learning_rate": 3.314906191236975e-06,
      "loss": 0.0006,
      "step": 47310
    },
    {
      "epoch": 0.6690325043475802,
      "grad_norm": 9.586063054456417e-10,
      "learning_rate": 3.3134923440173056e-06,
      "loss": 0.0005,
      "step": 47320
    },
    {
      "epoch": 0.6691738890695471,
      "grad_norm": 1.1199768223946194e-08,
      "learning_rate": 3.312078496797636e-06,
      "loss": 0.0,
      "step": 47330
    },
    {
      "epoch": 0.6693152737915141,
      "grad_norm": 0.0,
      "learning_rate": 3.310664649577967e-06,
      "loss": 0.0051,
      "step": 47340
    },
    {
      "epoch": 0.669456658513481,
      "grad_norm": 3.381529495527502e-06,
      "learning_rate": 3.3092508023582973e-06,
      "loss": 0.0,
      "step": 47350
    },
    {
      "epoch": 0.669598043235448,
      "grad_norm": 0.0,
      "learning_rate": 3.307836955138628e-06,
      "loss": 0.0,
      "step": 47360
    },
    {
      "epoch": 0.6697394279574149,
      "grad_norm": 95.7496109008789,
      "learning_rate": 3.3064231079189586e-06,
      "loss": 0.0156,
      "step": 47370
    },
    {
      "epoch": 0.6698808126793818,
      "grad_norm": 0.00022967967379372567,
      "learning_rate": 3.305009260699289e-06,
      "loss": 0.0,
      "step": 47380
    },
    {
      "epoch": 0.6700221974013488,
      "grad_norm": 1.0242026291962247e-05,
      "learning_rate": 3.3035954134796195e-06,
      "loss": 0.0,
      "step": 47390
    },
    {
      "epoch": 0.6701635821233157,
      "grad_norm": 0.0,
      "learning_rate": 3.3021815662599503e-06,
      "loss": 0.0003,
      "step": 47400
    },
    {
      "epoch": 0.6703049668452827,
      "grad_norm": 0.0,
      "learning_rate": 3.3007677190402808e-06,
      "loss": 0.0,
      "step": 47410
    },
    {
      "epoch": 0.6704463515672496,
      "grad_norm": 0.0,
      "learning_rate": 3.299353871820611e-06,
      "loss": 0.0059,
      "step": 47420
    },
    {
      "epoch": 0.6705877362892166,
      "grad_norm": 0.0,
      "learning_rate": 3.297940024600942e-06,
      "loss": 0.0,
      "step": 47430
    },
    {
      "epoch": 0.6707291210111835,
      "grad_norm": 0.0,
      "learning_rate": 3.2965261773812725e-06,
      "loss": 0.0,
      "step": 47440
    },
    {
      "epoch": 0.6708705057331504,
      "grad_norm": 0.0,
      "learning_rate": 3.295112330161603e-06,
      "loss": 0.0044,
      "step": 47450
    },
    {
      "epoch": 0.6710118904551174,
      "grad_norm": 0.0,
      "learning_rate": 3.293698482941934e-06,
      "loss": 0.0,
      "step": 47460
    },
    {
      "epoch": 0.6711532751770843,
      "grad_norm": 0.0,
      "learning_rate": 3.2922846357222642e-06,
      "loss": 0.0,
      "step": 47470
    },
    {
      "epoch": 0.6712946598990513,
      "grad_norm": 0.00010686707537388429,
      "learning_rate": 3.2908707885025947e-06,
      "loss": 0.0,
      "step": 47480
    },
    {
      "epoch": 0.6714360446210182,
      "grad_norm": 0.0,
      "learning_rate": 3.289456941282925e-06,
      "loss": 0.0,
      "step": 47490
    },
    {
      "epoch": 0.6715774293429851,
      "grad_norm": 1.3392373432452587e-07,
      "learning_rate": 3.2880430940632555e-06,
      "loss": 0.0,
      "step": 47500
    },
    {
      "epoch": 0.6717188140649522,
      "grad_norm": 0.08309878408908844,
      "learning_rate": 3.2866292468435864e-06,
      "loss": 0.0,
      "step": 47510
    },
    {
      "epoch": 0.671860198786919,
      "grad_norm": 0.0,
      "learning_rate": 3.285215399623917e-06,
      "loss": 0.0,
      "step": 47520
    },
    {
      "epoch": 0.6720015835088861,
      "grad_norm": 0.0,
      "learning_rate": 3.2838015524042477e-06,
      "loss": 0.0,
      "step": 47530
    },
    {
      "epoch": 0.672142968230853,
      "grad_norm": 0.0,
      "learning_rate": 3.282387705184578e-06,
      "loss": 0.0007,
      "step": 47540
    },
    {
      "epoch": 0.67228435295282,
      "grad_norm": 0.0,
      "learning_rate": 3.2809738579649085e-06,
      "loss": 0.0,
      "step": 47550
    },
    {
      "epoch": 0.6724257376747869,
      "grad_norm": 0.0,
      "learning_rate": 3.279560010745239e-06,
      "loss": 0.0,
      "step": 47560
    },
    {
      "epoch": 0.6725671223967538,
      "grad_norm": 0.0,
      "learning_rate": 3.2781461635255694e-06,
      "loss": 0.0,
      "step": 47570
    },
    {
      "epoch": 0.6727085071187208,
      "grad_norm": 0.00017090828623622656,
      "learning_rate": 3.2767323163059003e-06,
      "loss": 0.0,
      "step": 47580
    },
    {
      "epoch": 0.6728498918406877,
      "grad_norm": 0.0,
      "learning_rate": 3.275318469086231e-06,
      "loss": 0.0297,
      "step": 47590
    },
    {
      "epoch": 0.6729912765626547,
      "grad_norm": 1.0648951054292866e-08,
      "learning_rate": 3.2739046218665615e-06,
      "loss": 0.0,
      "step": 47600
    },
    {
      "epoch": 0.6731326612846216,
      "grad_norm": 0.0,
      "learning_rate": 3.272490774646892e-06,
      "loss": 0.0,
      "step": 47610
    },
    {
      "epoch": 0.6732740460065886,
      "grad_norm": 0.0,
      "learning_rate": 3.2710769274272224e-06,
      "loss": 0.009,
      "step": 47620
    },
    {
      "epoch": 0.6734154307285555,
      "grad_norm": 0.0,
      "learning_rate": 3.269663080207553e-06,
      "loss": 0.0,
      "step": 47630
    },
    {
      "epoch": 0.6735568154505224,
      "grad_norm": 0.0,
      "learning_rate": 3.2682492329878833e-06,
      "loss": 0.0,
      "step": 47640
    },
    {
      "epoch": 0.6736982001724894,
      "grad_norm": 0.0,
      "learning_rate": 3.2668353857682137e-06,
      "loss": 0.2797,
      "step": 47650
    },
    {
      "epoch": 0.6738395848944563,
      "grad_norm": 0.0,
      "learning_rate": 3.265421538548545e-06,
      "loss": 0.0029,
      "step": 47660
    },
    {
      "epoch": 0.6739809696164233,
      "grad_norm": 0.0,
      "learning_rate": 3.2640076913288754e-06,
      "loss": 0.0,
      "step": 47670
    },
    {
      "epoch": 0.6741223543383902,
      "grad_norm": 8.971348108843813e-10,
      "learning_rate": 3.262593844109206e-06,
      "loss": 0.0,
      "step": 47680
    },
    {
      "epoch": 0.6742637390603572,
      "grad_norm": 0.03584689646959305,
      "learning_rate": 3.2611799968895363e-06,
      "loss": 0.0001,
      "step": 47690
    },
    {
      "epoch": 0.6744051237823241,
      "grad_norm": 0.017039185389876366,
      "learning_rate": 3.2597661496698667e-06,
      "loss": 0.0,
      "step": 47700
    },
    {
      "epoch": 0.674546508504291,
      "grad_norm": 9.038225168289671e-10,
      "learning_rate": 3.258352302450197e-06,
      "loss": 0.0,
      "step": 47710
    },
    {
      "epoch": 0.674687893226258,
      "grad_norm": 0.0,
      "learning_rate": 3.2569384552305284e-06,
      "loss": 0.0001,
      "step": 47720
    },
    {
      "epoch": 0.6748292779482249,
      "grad_norm": 0.0003568472748156637,
      "learning_rate": 3.255524608010859e-06,
      "loss": 0.0,
      "step": 47730
    },
    {
      "epoch": 0.6749706626701919,
      "grad_norm": 0.0,
      "learning_rate": 3.2541107607911893e-06,
      "loss": 0.0,
      "step": 47740
    },
    {
      "epoch": 0.6751120473921588,
      "grad_norm": 0.0,
      "learning_rate": 3.2526969135715197e-06,
      "loss": 0.0,
      "step": 47750
    },
    {
      "epoch": 0.6752534321141257,
      "grad_norm": 2.685431581994635e-06,
      "learning_rate": 3.25128306635185e-06,
      "loss": 0.0,
      "step": 47760
    },
    {
      "epoch": 0.6753948168360927,
      "grad_norm": 0.0,
      "learning_rate": 3.2498692191321806e-06,
      "loss": 0.0,
      "step": 47770
    },
    {
      "epoch": 0.6755362015580596,
      "grad_norm": 0.0,
      "learning_rate": 3.248455371912511e-06,
      "loss": 0.0,
      "step": 47780
    },
    {
      "epoch": 0.6756775862800266,
      "grad_norm": 2.540350806157221e-06,
      "learning_rate": 3.2470415246928423e-06,
      "loss": 0.0,
      "step": 47790
    },
    {
      "epoch": 0.6758189710019935,
      "grad_norm": 0.0,
      "learning_rate": 3.2456276774731727e-06,
      "loss": 0.0,
      "step": 47800
    },
    {
      "epoch": 0.6759603557239605,
      "grad_norm": 0.0,
      "learning_rate": 3.244213830253503e-06,
      "loss": 0.0008,
      "step": 47810
    },
    {
      "epoch": 0.6761017404459274,
      "grad_norm": 0.0,
      "learning_rate": 3.2427999830338336e-06,
      "loss": 0.0,
      "step": 47820
    },
    {
      "epoch": 0.6762431251678943,
      "grad_norm": 0.0,
      "learning_rate": 3.241386135814164e-06,
      "loss": 0.0004,
      "step": 47830
    },
    {
      "epoch": 0.6763845098898613,
      "grad_norm": 5.078301182948053e-05,
      "learning_rate": 3.2399722885944945e-06,
      "loss": 0.0001,
      "step": 47840
    },
    {
      "epoch": 0.6765258946118282,
      "grad_norm": 0.0,
      "learning_rate": 3.2385584413748258e-06,
      "loss": 0.165,
      "step": 47850
    },
    {
      "epoch": 0.6766672793337952,
      "grad_norm": 1.1265199439947082e-08,
      "learning_rate": 3.237144594155156e-06,
      "loss": 0.0,
      "step": 47860
    },
    {
      "epoch": 0.6768086640557621,
      "grad_norm": 0.0,
      "learning_rate": 3.2357307469354866e-06,
      "loss": 0.2279,
      "step": 47870
    },
    {
      "epoch": 0.6769500487777291,
      "grad_norm": 0.0,
      "learning_rate": 3.234316899715817e-06,
      "loss": 0.0,
      "step": 47880
    },
    {
      "epoch": 0.677091433499696,
      "grad_norm": 0.028772270306944847,
      "learning_rate": 3.2329030524961475e-06,
      "loss": 0.0003,
      "step": 47890
    },
    {
      "epoch": 0.6772328182216629,
      "grad_norm": 0.0,
      "learning_rate": 3.231489205276478e-06,
      "loss": 0.0768,
      "step": 47900
    },
    {
      "epoch": 0.6773742029436299,
      "grad_norm": 0.0,
      "learning_rate": 3.2300753580568084e-06,
      "loss": 0.004,
      "step": 47910
    },
    {
      "epoch": 0.6775155876655968,
      "grad_norm": 0.0,
      "learning_rate": 3.228661510837139e-06,
      "loss": 0.0,
      "step": 47920
    },
    {
      "epoch": 0.6776569723875638,
      "grad_norm": 0.0,
      "learning_rate": 3.22724766361747e-06,
      "loss": 0.0006,
      "step": 47930
    },
    {
      "epoch": 0.6777983571095307,
      "grad_norm": 0.0,
      "learning_rate": 3.2258338163978005e-06,
      "loss": 0.0,
      "step": 47940
    },
    {
      "epoch": 0.6779397418314976,
      "grad_norm": 0.0,
      "learning_rate": 3.224419969178131e-06,
      "loss": 0.0046,
      "step": 47950
    },
    {
      "epoch": 0.6780811265534646,
      "grad_norm": 0.0,
      "learning_rate": 3.2230061219584614e-06,
      "loss": 0.0,
      "step": 47960
    },
    {
      "epoch": 0.6782225112754315,
      "grad_norm": 0.0,
      "learning_rate": 3.221592274738792e-06,
      "loss": 0.0005,
      "step": 47970
    },
    {
      "epoch": 0.6783638959973985,
      "grad_norm": 0.0,
      "learning_rate": 3.2201784275191227e-06,
      "loss": 0.016,
      "step": 47980
    },
    {
      "epoch": 0.6785052807193654,
      "grad_norm": 0.00022648254525847733,
      "learning_rate": 3.218764580299453e-06,
      "loss": 0.0025,
      "step": 47990
    },
    {
      "epoch": 0.6786466654413325,
      "grad_norm": 1.608692946319934e-05,
      "learning_rate": 3.217350733079784e-06,
      "loss": 0.0,
      "step": 48000
    },
    {
      "epoch": 0.6787880501632994,
      "grad_norm": 0.0,
      "learning_rate": 3.2159368858601144e-06,
      "loss": 0.0,
      "step": 48010
    },
    {
      "epoch": 0.6789294348852662,
      "grad_norm": 0.0,
      "learning_rate": 3.214523038640445e-06,
      "loss": 0.0021,
      "step": 48020
    },
    {
      "epoch": 0.6790708196072333,
      "grad_norm": 0.0,
      "learning_rate": 3.2131091914207752e-06,
      "loss": 0.0006,
      "step": 48030
    },
    {
      "epoch": 0.6792122043292002,
      "grad_norm": 0.0,
      "learning_rate": 3.2116953442011057e-06,
      "loss": 0.0134,
      "step": 48040
    },
    {
      "epoch": 0.6793535890511672,
      "grad_norm": 0.0,
      "learning_rate": 3.2102814969814365e-06,
      "loss": 0.0,
      "step": 48050
    },
    {
      "epoch": 0.6794949737731341,
      "grad_norm": 0.0,
      "learning_rate": 3.208867649761767e-06,
      "loss": 0.0007,
      "step": 48060
    },
    {
      "epoch": 0.6796363584951011,
      "grad_norm": 0.0,
      "learning_rate": 3.2074538025420974e-06,
      "loss": 0.0106,
      "step": 48070
    },
    {
      "epoch": 0.679777743217068,
      "grad_norm": 0.0,
      "learning_rate": 3.2060399553224283e-06,
      "loss": 0.006,
      "step": 48080
    },
    {
      "epoch": 0.6799191279390349,
      "grad_norm": 0.0,
      "learning_rate": 3.2046261081027587e-06,
      "loss": 0.0,
      "step": 48090
    },
    {
      "epoch": 0.6800605126610019,
      "grad_norm": 0.0,
      "learning_rate": 3.203212260883089e-06,
      "loss": 0.019,
      "step": 48100
    },
    {
      "epoch": 0.6802018973829688,
      "grad_norm": 0.0,
      "learning_rate": 3.20179841366342e-06,
      "loss": 0.0,
      "step": 48110
    },
    {
      "epoch": 0.6803432821049358,
      "grad_norm": 0.0,
      "learning_rate": 3.2003845664437504e-06,
      "loss": 0.0,
      "step": 48120
    },
    {
      "epoch": 0.6804846668269027,
      "grad_norm": 8.51517979754135e-05,
      "learning_rate": 3.198970719224081e-06,
      "loss": 0.0,
      "step": 48130
    },
    {
      "epoch": 0.6806260515488697,
      "grad_norm": 0.0,
      "learning_rate": 3.1975568720044113e-06,
      "loss": 0.0,
      "step": 48140
    },
    {
      "epoch": 0.6807674362708366,
      "grad_norm": 0.0,
      "learning_rate": 3.1961430247847417e-06,
      "loss": 0.0,
      "step": 48150
    },
    {
      "epoch": 0.6809088209928035,
      "grad_norm": 0.0,
      "learning_rate": 3.1947291775650726e-06,
      "loss": 0.0955,
      "step": 48160
    },
    {
      "epoch": 0.6810502057147705,
      "grad_norm": 0.0,
      "learning_rate": 3.193315330345403e-06,
      "loss": 0.0001,
      "step": 48170
    },
    {
      "epoch": 0.6811915904367374,
      "grad_norm": 0.0,
      "learning_rate": 3.191901483125734e-06,
      "loss": 0.0,
      "step": 48180
    },
    {
      "epoch": 0.6813329751587044,
      "grad_norm": 0.0,
      "learning_rate": 3.1904876359060643e-06,
      "loss": 0.0,
      "step": 48190
    },
    {
      "epoch": 0.6814743598806713,
      "grad_norm": 3.184997154903613e-08,
      "learning_rate": 3.1890737886863947e-06,
      "loss": 0.0007,
      "step": 48200
    },
    {
      "epoch": 0.6816157446026382,
      "grad_norm": 0.0,
      "learning_rate": 3.187659941466725e-06,
      "loss": 0.0,
      "step": 48210
    },
    {
      "epoch": 0.6817571293246052,
      "grad_norm": 0.0,
      "learning_rate": 3.1862460942470556e-06,
      "loss": 0.0,
      "step": 48220
    },
    {
      "epoch": 0.6818985140465721,
      "grad_norm": 6.161593773867935e-06,
      "learning_rate": 3.1848322470273864e-06,
      "loss": 0.0,
      "step": 48230
    },
    {
      "epoch": 0.6820398987685391,
      "grad_norm": 0.0,
      "learning_rate": 3.1834183998077173e-06,
      "loss": 0.0,
      "step": 48240
    },
    {
      "epoch": 0.682181283490506,
      "grad_norm": 0.0,
      "learning_rate": 3.1820045525880477e-06,
      "loss": 0.0,
      "step": 48250
    },
    {
      "epoch": 0.682322668212473,
      "grad_norm": 9.383630094816908e-05,
      "learning_rate": 3.180590705368378e-06,
      "loss": 0.0,
      "step": 48260
    },
    {
      "epoch": 0.6824640529344399,
      "grad_norm": 0.0,
      "learning_rate": 3.1791768581487086e-06,
      "loss": 0.0,
      "step": 48270
    },
    {
      "epoch": 0.6826054376564068,
      "grad_norm": 0.0,
      "learning_rate": 3.177763010929039e-06,
      "loss": 0.0054,
      "step": 48280
    },
    {
      "epoch": 0.6827468223783738,
      "grad_norm": 0.00198855297639966,
      "learning_rate": 3.1763491637093695e-06,
      "loss": 0.0,
      "step": 48290
    },
    {
      "epoch": 0.6828882071003407,
      "grad_norm": 0.0,
      "learning_rate": 3.1749353164897007e-06,
      "loss": 0.0,
      "step": 48300
    },
    {
      "epoch": 0.6830295918223077,
      "grad_norm": 0.0,
      "learning_rate": 3.173521469270031e-06,
      "loss": 0.0061,
      "step": 48310
    },
    {
      "epoch": 0.6831709765442746,
      "grad_norm": 0.0,
      "learning_rate": 3.1721076220503616e-06,
      "loss": 0.0,
      "step": 48320
    },
    {
      "epoch": 0.6833123612662416,
      "grad_norm": 9.337759454552952e-10,
      "learning_rate": 3.170693774830692e-06,
      "loss": 0.0002,
      "step": 48330
    },
    {
      "epoch": 0.6834537459882085,
      "grad_norm": 0.0,
      "learning_rate": 3.1692799276110225e-06,
      "loss": 0.0,
      "step": 48340
    },
    {
      "epoch": 0.6835951307101754,
      "grad_norm": 0.0,
      "learning_rate": 3.167866080391353e-06,
      "loss": 0.0,
      "step": 48350
    },
    {
      "epoch": 0.6837365154321424,
      "grad_norm": 0.0,
      "learning_rate": 3.1664522331716833e-06,
      "loss": 0.0,
      "step": 48360
    },
    {
      "epoch": 0.6838779001541093,
      "grad_norm": 0.0,
      "learning_rate": 3.1650383859520146e-06,
      "loss": 0.0,
      "step": 48370
    },
    {
      "epoch": 0.6840192848760763,
      "grad_norm": 8.18564060978133e-08,
      "learning_rate": 3.163624538732345e-06,
      "loss": 0.0,
      "step": 48380
    },
    {
      "epoch": 0.6841606695980432,
      "grad_norm": 1.0075698053668702e-09,
      "learning_rate": 3.1622106915126755e-06,
      "loss": 0.0,
      "step": 48390
    },
    {
      "epoch": 0.6843020543200102,
      "grad_norm": 0.0,
      "learning_rate": 3.160796844293006e-06,
      "loss": 0.0445,
      "step": 48400
    },
    {
      "epoch": 0.6844434390419771,
      "grad_norm": 0.0,
      "learning_rate": 3.1593829970733364e-06,
      "loss": 0.0,
      "step": 48410
    },
    {
      "epoch": 0.684584823763944,
      "grad_norm": 0.0,
      "learning_rate": 3.1579691498536668e-06,
      "loss": 0.0,
      "step": 48420
    },
    {
      "epoch": 0.684726208485911,
      "grad_norm": 0.0,
      "learning_rate": 3.156555302633998e-06,
      "loss": 0.0,
      "step": 48430
    },
    {
      "epoch": 0.6848675932078779,
      "grad_norm": 0.0,
      "learning_rate": 3.1551414554143285e-06,
      "loss": 0.0011,
      "step": 48440
    },
    {
      "epoch": 0.6850089779298449,
      "grad_norm": 4.687758803356701e-08,
      "learning_rate": 3.153727608194659e-06,
      "loss": 0.0,
      "step": 48450
    },
    {
      "epoch": 0.6851503626518118,
      "grad_norm": 0.0,
      "learning_rate": 3.1523137609749894e-06,
      "loss": 0.0,
      "step": 48460
    },
    {
      "epoch": 0.6852917473737787,
      "grad_norm": 0.003800105070695281,
      "learning_rate": 3.15089991375532e-06,
      "loss": 0.0007,
      "step": 48470
    },
    {
      "epoch": 0.6854331320957457,
      "grad_norm": 0.0,
      "learning_rate": 3.1494860665356502e-06,
      "loss": 0.0,
      "step": 48480
    },
    {
      "epoch": 0.6855745168177126,
      "grad_norm": 3.0861872346577e-06,
      "learning_rate": 3.1480722193159807e-06,
      "loss": 0.0511,
      "step": 48490
    },
    {
      "epoch": 0.6857159015396797,
      "grad_norm": 6.078627734495967e-07,
      "learning_rate": 3.146658372096312e-06,
      "loss": 0.0,
      "step": 48500
    },
    {
      "epoch": 0.6858572862616465,
      "grad_norm": 0.0,
      "learning_rate": 3.1452445248766424e-06,
      "loss": 0.0,
      "step": 48510
    },
    {
      "epoch": 0.6859986709836136,
      "grad_norm": 0.0,
      "learning_rate": 3.143830677656973e-06,
      "loss": 0.0,
      "step": 48520
    },
    {
      "epoch": 0.6861400557055805,
      "grad_norm": 0.0,
      "learning_rate": 3.1424168304373032e-06,
      "loss": 0.0,
      "step": 48530
    },
    {
      "epoch": 0.6862814404275474,
      "grad_norm": 7.789146184222773e-05,
      "learning_rate": 3.1410029832176337e-06,
      "loss": 0.0,
      "step": 48540
    },
    {
      "epoch": 0.6864228251495144,
      "grad_norm": 0.0,
      "learning_rate": 3.139589135997964e-06,
      "loss": 0.0,
      "step": 48550
    },
    {
      "epoch": 0.6865642098714813,
      "grad_norm": 0.0,
      "learning_rate": 3.138175288778295e-06,
      "loss": 0.0,
      "step": 48560
    },
    {
      "epoch": 0.6867055945934483,
      "grad_norm": 3.699597073136829e-06,
      "learning_rate": 3.1367614415586254e-06,
      "loss": 0.0,
      "step": 48570
    },
    {
      "epoch": 0.6868469793154152,
      "grad_norm": 0.0,
      "learning_rate": 3.1353475943389563e-06,
      "loss": 0.0,
      "step": 48580
    },
    {
      "epoch": 0.6869883640373822,
      "grad_norm": 5.904820682189893e-06,
      "learning_rate": 3.1339337471192867e-06,
      "loss": 0.0,
      "step": 48590
    },
    {
      "epoch": 0.6871297487593491,
      "grad_norm": 0.001016785274259746,
      "learning_rate": 3.132519899899617e-06,
      "loss": 0.0,
      "step": 48600
    },
    {
      "epoch": 0.687271133481316,
      "grad_norm": 0.0,
      "learning_rate": 3.1311060526799476e-06,
      "loss": 0.0,
      "step": 48610
    },
    {
      "epoch": 0.687412518203283,
      "grad_norm": 0.0,
      "learning_rate": 3.129692205460278e-06,
      "loss": 0.0,
      "step": 48620
    },
    {
      "epoch": 0.6875539029252499,
      "grad_norm": 9.134796252752153e-10,
      "learning_rate": 3.128278358240609e-06,
      "loss": 0.0002,
      "step": 48630
    },
    {
      "epoch": 0.6876952876472169,
      "grad_norm": 0.007604690734297037,
      "learning_rate": 3.1268645110209393e-06,
      "loss": 0.0,
      "step": 48640
    },
    {
      "epoch": 0.6878366723691838,
      "grad_norm": 0.0,
      "learning_rate": 3.12545066380127e-06,
      "loss": 0.5738,
      "step": 48650
    },
    {
      "epoch": 0.6879780570911508,
      "grad_norm": 0.0,
      "learning_rate": 3.1240368165816006e-06,
      "loss": 0.0,
      "step": 48660
    },
    {
      "epoch": 0.6881194418131177,
      "grad_norm": 0.0,
      "learning_rate": 3.122622969361931e-06,
      "loss": 0.0,
      "step": 48670
    },
    {
      "epoch": 0.6882608265350846,
      "grad_norm": 0.0,
      "learning_rate": 3.1212091221422614e-06,
      "loss": 0.0007,
      "step": 48680
    },
    {
      "epoch": 0.6884022112570516,
      "grad_norm": 3.3447020541643724e-05,
      "learning_rate": 3.1197952749225923e-06,
      "loss": 0.0,
      "step": 48690
    },
    {
      "epoch": 0.6885435959790185,
      "grad_norm": 0.0,
      "learning_rate": 3.1183814277029227e-06,
      "loss": 0.0,
      "step": 48700
    },
    {
      "epoch": 0.6886849807009855,
      "grad_norm": 0.0,
      "learning_rate": 3.116967580483253e-06,
      "loss": 0.0,
      "step": 48710
    },
    {
      "epoch": 0.6888263654229524,
      "grad_norm": 6.191709189806716e-07,
      "learning_rate": 3.1155537332635836e-06,
      "loss": 0.0005,
      "step": 48720
    },
    {
      "epoch": 0.6889677501449193,
      "grad_norm": 0.008764658123254776,
      "learning_rate": 3.1141398860439144e-06,
      "loss": 0.0015,
      "step": 48730
    },
    {
      "epoch": 0.6891091348668863,
      "grad_norm": 9.570743086939615e-10,
      "learning_rate": 3.112726038824245e-06,
      "loss": 0.0004,
      "step": 48740
    },
    {
      "epoch": 0.6892505195888532,
      "grad_norm": 0.00035290728555992246,
      "learning_rate": 3.1113121916045753e-06,
      "loss": 0.0,
      "step": 48750
    },
    {
      "epoch": 0.6893919043108202,
      "grad_norm": 0.0,
      "learning_rate": 3.109898344384906e-06,
      "loss": 0.0,
      "step": 48760
    },
    {
      "epoch": 0.6895332890327871,
      "grad_norm": 8.733920253689575e-07,
      "learning_rate": 3.1084844971652366e-06,
      "loss": 0.0,
      "step": 48770
    },
    {
      "epoch": 0.6896746737547541,
      "grad_norm": 0.0,
      "learning_rate": 3.107070649945567e-06,
      "loss": 0.0,
      "step": 48780
    },
    {
      "epoch": 0.689816058476721,
      "grad_norm": 8.813709229116284e-08,
      "learning_rate": 3.1056568027258975e-06,
      "loss": 0.0,
      "step": 48790
    },
    {
      "epoch": 0.6899574431986879,
      "grad_norm": 0.017586516216397285,
      "learning_rate": 3.1042429555062283e-06,
      "loss": 0.0,
      "step": 48800
    },
    {
      "epoch": 0.6900988279206549,
      "grad_norm": 9.595546579532765e-10,
      "learning_rate": 3.1028291082865588e-06,
      "loss": 0.0,
      "step": 48810
    },
    {
      "epoch": 0.6902402126426218,
      "grad_norm": 0.0,
      "learning_rate": 3.1014152610668896e-06,
      "loss": 0.0,
      "step": 48820
    },
    {
      "epoch": 0.6903815973645888,
      "grad_norm": 0.0,
      "learning_rate": 3.10000141384722e-06,
      "loss": 0.0,
      "step": 48830
    },
    {
      "epoch": 0.6905229820865557,
      "grad_norm": 0.0,
      "learning_rate": 3.0985875666275505e-06,
      "loss": 0.0,
      "step": 48840
    },
    {
      "epoch": 0.6906643668085227,
      "grad_norm": 0.0,
      "learning_rate": 3.097173719407881e-06,
      "loss": 0.0002,
      "step": 48850
    },
    {
      "epoch": 0.6908057515304896,
      "grad_norm": 0.0,
      "learning_rate": 3.0957598721882113e-06,
      "loss": 0.0012,
      "step": 48860
    },
    {
      "epoch": 0.6909471362524565,
      "grad_norm": 0.0,
      "learning_rate": 3.0943460249685418e-06,
      "loss": 0.0091,
      "step": 48870
    },
    {
      "epoch": 0.6910885209744235,
      "grad_norm": 5.132471869728761e-08,
      "learning_rate": 3.0929321777488726e-06,
      "loss": 0.0,
      "step": 48880
    },
    {
      "epoch": 0.6912299056963904,
      "grad_norm": 0.0,
      "learning_rate": 3.0915183305292035e-06,
      "loss": 0.001,
      "step": 48890
    },
    {
      "epoch": 0.6913712904183574,
      "grad_norm": 0.00017325289081782103,
      "learning_rate": 3.090104483309534e-06,
      "loss": 0.0,
      "step": 48900
    },
    {
      "epoch": 0.6915126751403243,
      "grad_norm": 0.0,
      "learning_rate": 3.0886906360898644e-06,
      "loss": 0.0,
      "step": 48910
    },
    {
      "epoch": 0.6916540598622912,
      "grad_norm": 0.0,
      "learning_rate": 3.087276788870195e-06,
      "loss": 0.1935,
      "step": 48920
    },
    {
      "epoch": 0.6917954445842582,
      "grad_norm": 0.028297271579504013,
      "learning_rate": 3.0858629416505252e-06,
      "loss": 0.0,
      "step": 48930
    },
    {
      "epoch": 0.6919368293062251,
      "grad_norm": 0.00012077394057996571,
      "learning_rate": 3.0844490944308557e-06,
      "loss": 0.0,
      "step": 48940
    },
    {
      "epoch": 0.6920782140281921,
      "grad_norm": 0.06410489231348038,
      "learning_rate": 3.083035247211187e-06,
      "loss": 0.0,
      "step": 48950
    },
    {
      "epoch": 0.692219598750159,
      "grad_norm": 0.0,
      "learning_rate": 3.0816213999915174e-06,
      "loss": 0.3693,
      "step": 48960
    },
    {
      "epoch": 0.692360983472126,
      "grad_norm": 0.0,
      "learning_rate": 3.080207552771848e-06,
      "loss": 0.0,
      "step": 48970
    },
    {
      "epoch": 0.6925023681940929,
      "grad_norm": 0.03956642374396324,
      "learning_rate": 3.0787937055521782e-06,
      "loss": 0.0,
      "step": 48980
    },
    {
      "epoch": 0.6926437529160598,
      "grad_norm": 0.0,
      "learning_rate": 3.0773798583325087e-06,
      "loss": 0.0,
      "step": 48990
    },
    {
      "epoch": 0.6927851376380268,
      "grad_norm": 0.0,
      "learning_rate": 3.075966011112839e-06,
      "loss": 0.0,
      "step": 49000
    },
    {
      "epoch": 0.6929265223599937,
      "grad_norm": 1.8060861825942993,
      "learning_rate": 3.0745521638931695e-06,
      "loss": 0.0444,
      "step": 49010
    },
    {
      "epoch": 0.6930679070819608,
      "grad_norm": 0.0,
      "learning_rate": 3.073138316673501e-06,
      "loss": 0.0,
      "step": 49020
    },
    {
      "epoch": 0.6932092918039277,
      "grad_norm": 2.6631412765709683e-05,
      "learning_rate": 3.0717244694538312e-06,
      "loss": 0.0,
      "step": 49030
    },
    {
      "epoch": 0.6933506765258947,
      "grad_norm": 0.011137612164020538,
      "learning_rate": 3.0703106222341617e-06,
      "loss": 0.0,
      "step": 49040
    },
    {
      "epoch": 0.6934920612478616,
      "grad_norm": 0.0,
      "learning_rate": 3.068896775014492e-06,
      "loss": 0.002,
      "step": 49050
    },
    {
      "epoch": 0.6936334459698285,
      "grad_norm": 0.9982007145881653,
      "learning_rate": 3.0674829277948225e-06,
      "loss": 0.4824,
      "step": 49060
    },
    {
      "epoch": 0.6937748306917955,
      "grad_norm": 0.0,
      "learning_rate": 3.066069080575153e-06,
      "loss": 0.0,
      "step": 49070
    },
    {
      "epoch": 0.6939162154137624,
      "grad_norm": 0.008142488077282906,
      "learning_rate": 3.0646552333554843e-06,
      "loss": 0.0324,
      "step": 49080
    },
    {
      "epoch": 0.6940576001357294,
      "grad_norm": 0.0003238036297261715,
      "learning_rate": 3.0632413861358147e-06,
      "loss": 0.0407,
      "step": 49090
    },
    {
      "epoch": 0.6941989848576963,
      "grad_norm": 0.0,
      "learning_rate": 3.061827538916145e-06,
      "loss": 0.0,
      "step": 49100
    },
    {
      "epoch": 0.6943403695796633,
      "grad_norm": 0.0411418154835701,
      "learning_rate": 3.0604136916964756e-06,
      "loss": 0.0,
      "step": 49110
    },
    {
      "epoch": 0.6944817543016302,
      "grad_norm": 0.3113756477832794,
      "learning_rate": 3.058999844476806e-06,
      "loss": 0.0,
      "step": 49120
    },
    {
      "epoch": 0.6946231390235971,
      "grad_norm": 1.4557803851289464e-08,
      "learning_rate": 3.0575859972571364e-06,
      "loss": 0.0,
      "step": 49130
    },
    {
      "epoch": 0.6947645237455641,
      "grad_norm": 8.746182720642537e-05,
      "learning_rate": 3.0561721500374673e-06,
      "loss": 0.0006,
      "step": 49140
    },
    {
      "epoch": 0.694905908467531,
      "grad_norm": 6.511602812508954e-09,
      "learning_rate": 3.054758302817798e-06,
      "loss": 0.0,
      "step": 49150
    },
    {
      "epoch": 0.695047293189498,
      "grad_norm": 0.0,
      "learning_rate": 3.0533444555981286e-06,
      "loss": 0.0,
      "step": 49160
    },
    {
      "epoch": 0.6951886779114649,
      "grad_norm": 0.0,
      "learning_rate": 3.051930608378459e-06,
      "loss": 0.0,
      "step": 49170
    },
    {
      "epoch": 0.6953300626334318,
      "grad_norm": 5.832412739437132e-07,
      "learning_rate": 3.0505167611587894e-06,
      "loss": 0.0196,
      "step": 49180
    },
    {
      "epoch": 0.6954714473553988,
      "grad_norm": 0.0,
      "learning_rate": 3.04910291393912e-06,
      "loss": 0.0,
      "step": 49190
    },
    {
      "epoch": 0.6956128320773657,
      "grad_norm": 0.0,
      "learning_rate": 3.0476890667194503e-06,
      "loss": 0.0,
      "step": 49200
    },
    {
      "epoch": 0.6957542167993327,
      "grad_norm": 11.114969253540039,
      "learning_rate": 3.046275219499781e-06,
      "loss": 0.0011,
      "step": 49210
    },
    {
      "epoch": 0.6958956015212996,
      "grad_norm": 0.04375641420483589,
      "learning_rate": 3.044861372280112e-06,
      "loss": 0.0,
      "step": 49220
    },
    {
      "epoch": 0.6960369862432666,
      "grad_norm": 0.0,
      "learning_rate": 3.0434475250604424e-06,
      "loss": 0.0,
      "step": 49230
    },
    {
      "epoch": 0.6961783709652335,
      "grad_norm": 0.0,
      "learning_rate": 3.042033677840773e-06,
      "loss": 0.0002,
      "step": 49240
    },
    {
      "epoch": 0.6963197556872004,
      "grad_norm": 1.0418810703072268e-08,
      "learning_rate": 3.0406198306211033e-06,
      "loss": 0.0,
      "step": 49250
    },
    {
      "epoch": 0.6964611404091674,
      "grad_norm": 0.0,
      "learning_rate": 3.0392059834014337e-06,
      "loss": 0.0,
      "step": 49260
    },
    {
      "epoch": 0.6966025251311343,
      "grad_norm": 0.00014200927398633212,
      "learning_rate": 3.0377921361817646e-06,
      "loss": 0.0,
      "step": 49270
    },
    {
      "epoch": 0.6967439098531013,
      "grad_norm": 0.0,
      "learning_rate": 3.036378288962095e-06,
      "loss": 0.0,
      "step": 49280
    },
    {
      "epoch": 0.6968852945750682,
      "grad_norm": 0.010792872868478298,
      "learning_rate": 3.0349644417424255e-06,
      "loss": 0.0002,
      "step": 49290
    },
    {
      "epoch": 0.6970266792970352,
      "grad_norm": 0.0,
      "learning_rate": 3.0335505945227563e-06,
      "loss": 0.0,
      "step": 49300
    },
    {
      "epoch": 0.6971680640190021,
      "grad_norm": 0.0,
      "learning_rate": 3.0321367473030868e-06,
      "loss": 0.0019,
      "step": 49310
    },
    {
      "epoch": 0.697309448740969,
      "grad_norm": 33.06269073486328,
      "learning_rate": 3.030722900083417e-06,
      "loss": 0.0056,
      "step": 49320
    },
    {
      "epoch": 0.697450833462936,
      "grad_norm": 8.022779184102546e-06,
      "learning_rate": 3.0293090528637476e-06,
      "loss": 0.1116,
      "step": 49330
    },
    {
      "epoch": 0.6975922181849029,
      "grad_norm": 3.5777344703674316,
      "learning_rate": 3.0278952056440785e-06,
      "loss": 0.0005,
      "step": 49340
    },
    {
      "epoch": 0.6977336029068699,
      "grad_norm": 2.236273758171592e-06,
      "learning_rate": 3.026481358424409e-06,
      "loss": 0.0,
      "step": 49350
    },
    {
      "epoch": 0.6978749876288368,
      "grad_norm": 4.158158262157485e-08,
      "learning_rate": 3.0250675112047393e-06,
      "loss": 0.0001,
      "step": 49360
    },
    {
      "epoch": 0.6980163723508038,
      "grad_norm": 0.0,
      "learning_rate": 3.0236536639850698e-06,
      "loss": 0.0,
      "step": 49370
    },
    {
      "epoch": 0.6981577570727707,
      "grad_norm": 0.0005777091719210148,
      "learning_rate": 3.0222398167654006e-06,
      "loss": 0.0,
      "step": 49380
    },
    {
      "epoch": 0.6982991417947376,
      "grad_norm": 0.0,
      "learning_rate": 3.020825969545731e-06,
      "loss": 0.0,
      "step": 49390
    },
    {
      "epoch": 0.6984405265167046,
      "grad_norm": 6.108679144745111e-07,
      "learning_rate": 3.019412122326062e-06,
      "loss": 0.0,
      "step": 49400
    },
    {
      "epoch": 0.6985819112386715,
      "grad_norm": 6.032098099240102e-05,
      "learning_rate": 3.0179982751063924e-06,
      "loss": 0.0,
      "step": 49410
    },
    {
      "epoch": 0.6987232959606385,
      "grad_norm": 0.0,
      "learning_rate": 3.016584427886723e-06,
      "loss": 0.0,
      "step": 49420
    },
    {
      "epoch": 0.6988646806826054,
      "grad_norm": 1.9211647510528564,
      "learning_rate": 3.0151705806670532e-06,
      "loss": 0.0003,
      "step": 49430
    },
    {
      "epoch": 0.6990060654045723,
      "grad_norm": 3.411774862627226e-09,
      "learning_rate": 3.0137567334473837e-06,
      "loss": 0.0,
      "step": 49440
    },
    {
      "epoch": 0.6991474501265393,
      "grad_norm": 0.0,
      "learning_rate": 3.0123428862277145e-06,
      "loss": 0.0,
      "step": 49450
    },
    {
      "epoch": 0.6992888348485062,
      "grad_norm": 0.0006610859418287873,
      "learning_rate": 3.010929039008045e-06,
      "loss": 0.0005,
      "step": 49460
    },
    {
      "epoch": 0.6994302195704732,
      "grad_norm": 0.00721152825281024,
      "learning_rate": 3.009515191788376e-06,
      "loss": 0.0003,
      "step": 49470
    },
    {
      "epoch": 0.6995716042924401,
      "grad_norm": 0.0,
      "learning_rate": 3.0081013445687062e-06,
      "loss": 0.0007,
      "step": 49480
    },
    {
      "epoch": 0.6997129890144071,
      "grad_norm": 0.0,
      "learning_rate": 3.0066874973490367e-06,
      "loss": 0.0006,
      "step": 49490
    },
    {
      "epoch": 0.699854373736374,
      "grad_norm": 0.0,
      "learning_rate": 3.005273650129367e-06,
      "loss": 0.0,
      "step": 49500
    },
    {
      "epoch": 0.6999957584583409,
      "grad_norm": 0.0,
      "learning_rate": 3.0038598029096975e-06,
      "loss": 0.0,
      "step": 49510
    },
    {
      "epoch": 0.700137143180308,
      "grad_norm": 0.0,
      "learning_rate": 3.002445955690028e-06,
      "loss": 0.0,
      "step": 49520
    },
    {
      "epoch": 0.7002785279022748,
      "grad_norm": 0.0,
      "learning_rate": 3.0010321084703592e-06,
      "loss": 0.0,
      "step": 49530
    },
    {
      "epoch": 0.7004199126242419,
      "grad_norm": 0.0,
      "learning_rate": 2.9996182612506897e-06,
      "loss": 0.0,
      "step": 49540
    },
    {
      "epoch": 0.7005612973462088,
      "grad_norm": 2.927705367028466e-08,
      "learning_rate": 2.99820441403102e-06,
      "loss": 0.0,
      "step": 49550
    },
    {
      "epoch": 0.7007026820681758,
      "grad_norm": 0.0,
      "learning_rate": 2.9967905668113505e-06,
      "loss": 0.0,
      "step": 49560
    },
    {
      "epoch": 0.7008440667901427,
      "grad_norm": 0.0,
      "learning_rate": 2.995376719591681e-06,
      "loss": 0.0894,
      "step": 49570
    },
    {
      "epoch": 0.7009854515121096,
      "grad_norm": 7.530388757004403e-08,
      "learning_rate": 2.9939628723720114e-06,
      "loss": 0.9657,
      "step": 49580
    },
    {
      "epoch": 0.7011268362340766,
      "grad_norm": 0.0,
      "learning_rate": 2.992549025152342e-06,
      "loss": 0.0,
      "step": 49590
    },
    {
      "epoch": 0.7012682209560435,
      "grad_norm": 0.0,
      "learning_rate": 2.991135177932673e-06,
      "loss": 0.0007,
      "step": 49600
    },
    {
      "epoch": 0.7014096056780105,
      "grad_norm": 0.2220059037208557,
      "learning_rate": 2.9897213307130036e-06,
      "loss": 0.0001,
      "step": 49610
    },
    {
      "epoch": 0.7015509903999774,
      "grad_norm": 0.0,
      "learning_rate": 2.988307483493334e-06,
      "loss": 0.0,
      "step": 49620
    },
    {
      "epoch": 0.7016923751219443,
      "grad_norm": 0.0,
      "learning_rate": 2.9868936362736644e-06,
      "loss": 0.0001,
      "step": 49630
    },
    {
      "epoch": 0.7018337598439113,
      "grad_norm": 0.0,
      "learning_rate": 2.985479789053995e-06,
      "loss": 0.0,
      "step": 49640
    },
    {
      "epoch": 0.7019751445658782,
      "grad_norm": 0.0,
      "learning_rate": 2.9840659418343253e-06,
      "loss": 0.0,
      "step": 49650
    },
    {
      "epoch": 0.7021165292878452,
      "grad_norm": 0.0,
      "learning_rate": 2.9826520946146566e-06,
      "loss": 0.0,
      "step": 49660
    },
    {
      "epoch": 0.7022579140098121,
      "grad_norm": 0.0,
      "learning_rate": 2.981238247394987e-06,
      "loss": 0.0,
      "step": 49670
    },
    {
      "epoch": 0.7023992987317791,
      "grad_norm": 0.0,
      "learning_rate": 2.9798244001753174e-06,
      "loss": 0.0,
      "step": 49680
    },
    {
      "epoch": 0.702540683453746,
      "grad_norm": 0.0,
      "learning_rate": 2.978410552955648e-06,
      "loss": 0.0116,
      "step": 49690
    },
    {
      "epoch": 0.7026820681757129,
      "grad_norm": 0.0013630858156830072,
      "learning_rate": 2.9769967057359783e-06,
      "loss": 0.0004,
      "step": 49700
    },
    {
      "epoch": 0.7028234528976799,
      "grad_norm": 0.0,
      "learning_rate": 2.9755828585163087e-06,
      "loss": 0.5801,
      "step": 49710
    },
    {
      "epoch": 0.7029648376196468,
      "grad_norm": 0.0,
      "learning_rate": 2.974169011296639e-06,
      "loss": 0.0,
      "step": 49720
    },
    {
      "epoch": 0.7031062223416138,
      "grad_norm": 0.09077633172273636,
      "learning_rate": 2.9727551640769704e-06,
      "loss": 0.0,
      "step": 49730
    },
    {
      "epoch": 0.7032476070635807,
      "grad_norm": 0.0,
      "learning_rate": 2.971341316857301e-06,
      "loss": 0.0,
      "step": 49740
    },
    {
      "epoch": 0.7033889917855477,
      "grad_norm": 0.0,
      "learning_rate": 2.9699274696376313e-06,
      "loss": 0.0,
      "step": 49750
    },
    {
      "epoch": 0.7035303765075146,
      "grad_norm": 2.196843817969807e-09,
      "learning_rate": 2.9685136224179617e-06,
      "loss": 0.0,
      "step": 49760
    },
    {
      "epoch": 0.7036717612294815,
      "grad_norm": 4.0203601558985724e-10,
      "learning_rate": 2.967099775198292e-06,
      "loss": 0.0,
      "step": 49770
    },
    {
      "epoch": 0.7038131459514485,
      "grad_norm": 0.0,
      "learning_rate": 2.9656859279786226e-06,
      "loss": 0.0,
      "step": 49780
    },
    {
      "epoch": 0.7039545306734154,
      "grad_norm": 0.0004453324363566935,
      "learning_rate": 2.9642720807589535e-06,
      "loss": 0.0,
      "step": 49790
    },
    {
      "epoch": 0.7040959153953824,
      "grad_norm": 0.0,
      "learning_rate": 2.9628582335392843e-06,
      "loss": 0.0,
      "step": 49800
    },
    {
      "epoch": 0.7042373001173493,
      "grad_norm": 0.0,
      "learning_rate": 2.9614443863196148e-06,
      "loss": 0.0,
      "step": 49810
    },
    {
      "epoch": 0.7043786848393163,
      "grad_norm": 0.008914920501410961,
      "learning_rate": 2.960030539099945e-06,
      "loss": 0.0,
      "step": 49820
    },
    {
      "epoch": 0.7045200695612832,
      "grad_norm": 4.135687049711123e-06,
      "learning_rate": 2.9586166918802756e-06,
      "loss": 0.0,
      "step": 49830
    },
    {
      "epoch": 0.7046614542832501,
      "grad_norm": 0.0,
      "learning_rate": 2.957202844660606e-06,
      "loss": 0.0001,
      "step": 49840
    },
    {
      "epoch": 0.7048028390052171,
      "grad_norm": 0.0,
      "learning_rate": 2.9557889974409365e-06,
      "loss": 0.0,
      "step": 49850
    },
    {
      "epoch": 0.704944223727184,
      "grad_norm": 0.0,
      "learning_rate": 2.9543751502212673e-06,
      "loss": 0.0019,
      "step": 49860
    },
    {
      "epoch": 0.705085608449151,
      "grad_norm": 0.0,
      "learning_rate": 2.952961303001598e-06,
      "loss": 0.0,
      "step": 49870
    },
    {
      "epoch": 0.7052269931711179,
      "grad_norm": 0.0,
      "learning_rate": 2.9515474557819286e-06,
      "loss": 0.0,
      "step": 49880
    },
    {
      "epoch": 0.7053683778930848,
      "grad_norm": 0.0,
      "learning_rate": 2.950133608562259e-06,
      "loss": 0.0,
      "step": 49890
    },
    {
      "epoch": 0.7055097626150518,
      "grad_norm": 0.0,
      "learning_rate": 2.9487197613425895e-06,
      "loss": 0.0,
      "step": 49900
    },
    {
      "epoch": 0.7056511473370187,
      "grad_norm": 0.0,
      "learning_rate": 2.94730591412292e-06,
      "loss": 0.0202,
      "step": 49910
    },
    {
      "epoch": 0.7057925320589857,
      "grad_norm": 0.0,
      "learning_rate": 2.945892066903251e-06,
      "loss": 0.0,
      "step": 49920
    },
    {
      "epoch": 0.7059339167809526,
      "grad_norm": 0.0,
      "learning_rate": 2.9444782196835812e-06,
      "loss": 0.0,
      "step": 49930
    },
    {
      "epoch": 0.7060753015029196,
      "grad_norm": 0.0,
      "learning_rate": 2.9430643724639117e-06,
      "loss": 0.0,
      "step": 49940
    },
    {
      "epoch": 0.7062166862248865,
      "grad_norm": 0.0,
      "learning_rate": 2.9416505252442425e-06,
      "loss": 0.0,
      "step": 49950
    },
    {
      "epoch": 0.7063580709468534,
      "grad_norm": 1.0487155321925457e-07,
      "learning_rate": 2.940236678024573e-06,
      "loss": 0.0025,
      "step": 49960
    },
    {
      "epoch": 0.7064994556688204,
      "grad_norm": 0.0,
      "learning_rate": 2.9388228308049034e-06,
      "loss": 0.0,
      "step": 49970
    },
    {
      "epoch": 0.7066408403907873,
      "grad_norm": 0.0,
      "learning_rate": 2.9374089835852342e-06,
      "loss": 0.0016,
      "step": 49980
    },
    {
      "epoch": 0.7067822251127543,
      "grad_norm": 0.0,
      "learning_rate": 2.9359951363655647e-06,
      "loss": 0.0,
      "step": 49990
    },
    {
      "epoch": 0.7069236098347212,
      "grad_norm": 0.05767301842570305,
      "learning_rate": 2.934581289145895e-06,
      "loss": 0.0001,
      "step": 50000
    },
    {
      "epoch": 0.7070649945566883,
      "grad_norm": 4.731272595925873e-10,
      "learning_rate": 2.9331674419262255e-06,
      "loss": 0.0,
      "step": 50010
    },
    {
      "epoch": 0.7072063792786552,
      "grad_norm": 0.0,
      "learning_rate": 2.9317535947065564e-06,
      "loss": 0.0,
      "step": 50020
    },
    {
      "epoch": 0.707347764000622,
      "grad_norm": 1.3483730554580688,
      "learning_rate": 2.930339747486887e-06,
      "loss": 0.0003,
      "step": 50030
    },
    {
      "epoch": 0.7074891487225891,
      "grad_norm": 0.0,
      "learning_rate": 2.9289259002672173e-06,
      "loss": 0.0,
      "step": 50040
    },
    {
      "epoch": 0.707630533444556,
      "grad_norm": 0.000706653343513608,
      "learning_rate": 2.927512053047548e-06,
      "loss": 0.0,
      "step": 50050
    },
    {
      "epoch": 0.707771918166523,
      "grad_norm": 0.0,
      "learning_rate": 2.9260982058278785e-06,
      "loss": 0.0,
      "step": 50060
    },
    {
      "epoch": 0.7079133028884899,
      "grad_norm": 7.687055040150881e-05,
      "learning_rate": 2.924684358608209e-06,
      "loss": 0.0,
      "step": 50070
    },
    {
      "epoch": 0.7080546876104569,
      "grad_norm": 4.972570133432441e-10,
      "learning_rate": 2.9232705113885394e-06,
      "loss": 0.0,
      "step": 50080
    },
    {
      "epoch": 0.7081960723324238,
      "grad_norm": 0.0,
      "learning_rate": 2.92185666416887e-06,
      "loss": 0.0,
      "step": 50090
    },
    {
      "epoch": 0.7083374570543907,
      "grad_norm": 0.0,
      "learning_rate": 2.9204428169492007e-06,
      "loss": 0.0,
      "step": 50100
    },
    {
      "epoch": 0.7084788417763577,
      "grad_norm": 2.90957232209621e-05,
      "learning_rate": 2.9190289697295316e-06,
      "loss": 0.0,
      "step": 50110
    },
    {
      "epoch": 0.7086202264983246,
      "grad_norm": 0.0,
      "learning_rate": 2.917615122509862e-06,
      "loss": 0.0,
      "step": 50120
    },
    {
      "epoch": 0.7087616112202916,
      "grad_norm": 0.24551048874855042,
      "learning_rate": 2.9162012752901924e-06,
      "loss": 0.0001,
      "step": 50130
    },
    {
      "epoch": 0.7089029959422585,
      "grad_norm": 0.0,
      "learning_rate": 2.914787428070523e-06,
      "loss": 0.0,
      "step": 50140
    },
    {
      "epoch": 0.7090443806642254,
      "grad_norm": 0.0,
      "learning_rate": 2.9133735808508533e-06,
      "loss": 0.0153,
      "step": 50150
    },
    {
      "epoch": 0.7091857653861924,
      "grad_norm": 0.0,
      "learning_rate": 2.9119597336311837e-06,
      "loss": 0.0,
      "step": 50160
    },
    {
      "epoch": 0.7093271501081593,
      "grad_norm": 0.0,
      "learning_rate": 2.910545886411514e-06,
      "loss": 0.0002,
      "step": 50170
    },
    {
      "epoch": 0.7094685348301263,
      "grad_norm": 0.0,
      "learning_rate": 2.9091320391918454e-06,
      "loss": 0.0,
      "step": 50180
    },
    {
      "epoch": 0.7096099195520932,
      "grad_norm": 1.1809282085550876e-07,
      "learning_rate": 2.907718191972176e-06,
      "loss": 0.0,
      "step": 50190
    },
    {
      "epoch": 0.7097513042740602,
      "grad_norm": 0.12116429954767227,
      "learning_rate": 2.9063043447525063e-06,
      "loss": 0.0,
      "step": 50200
    },
    {
      "epoch": 0.7098926889960271,
      "grad_norm": 0.0,
      "learning_rate": 2.9048904975328367e-06,
      "loss": 0.0,
      "step": 50210
    },
    {
      "epoch": 0.710034073717994,
      "grad_norm": 0.0,
      "learning_rate": 2.903476650313167e-06,
      "loss": 0.0003,
      "step": 50220
    },
    {
      "epoch": 0.710175458439961,
      "grad_norm": 0.0,
      "learning_rate": 2.9020628030934976e-06,
      "loss": 0.0,
      "step": 50230
    },
    {
      "epoch": 0.7103168431619279,
      "grad_norm": 0.0,
      "learning_rate": 2.900648955873829e-06,
      "loss": 0.0,
      "step": 50240
    },
    {
      "epoch": 0.7104582278838949,
      "grad_norm": 0.0,
      "learning_rate": 2.8992351086541593e-06,
      "loss": 0.4213,
      "step": 50250
    },
    {
      "epoch": 0.7105996126058618,
      "grad_norm": 40.28670120239258,
      "learning_rate": 2.8978212614344897e-06,
      "loss": 0.0072,
      "step": 50260
    },
    {
      "epoch": 0.7107409973278288,
      "grad_norm": 0.0,
      "learning_rate": 2.89640741421482e-06,
      "loss": 0.0,
      "step": 50270
    },
    {
      "epoch": 0.7108823820497957,
      "grad_norm": 0.0,
      "learning_rate": 2.8949935669951506e-06,
      "loss": 0.0,
      "step": 50280
    },
    {
      "epoch": 0.7110237667717626,
      "grad_norm": 0.0,
      "learning_rate": 2.893579719775481e-06,
      "loss": 0.0,
      "step": 50290
    },
    {
      "epoch": 0.7111651514937296,
      "grad_norm": 0.0,
      "learning_rate": 2.8921658725558115e-06,
      "loss": 0.0,
      "step": 50300
    },
    {
      "epoch": 0.7113065362156965,
      "grad_norm": 0.0,
      "learning_rate": 2.8907520253361428e-06,
      "loss": 0.0002,
      "step": 50310
    },
    {
      "epoch": 0.7114479209376635,
      "grad_norm": 0.0,
      "learning_rate": 2.889338178116473e-06,
      "loss": 0.0,
      "step": 50320
    },
    {
      "epoch": 0.7115893056596304,
      "grad_norm": 0.0,
      "learning_rate": 2.8879243308968036e-06,
      "loss": 0.0,
      "step": 50330
    },
    {
      "epoch": 0.7117306903815974,
      "grad_norm": 0.0,
      "learning_rate": 2.886510483677134e-06,
      "loss": 0.0,
      "step": 50340
    },
    {
      "epoch": 0.7118720751035643,
      "grad_norm": 0.0,
      "learning_rate": 2.8850966364574645e-06,
      "loss": 0.0,
      "step": 50350
    },
    {
      "epoch": 0.7120134598255312,
      "grad_norm": 0.0,
      "learning_rate": 2.883682789237795e-06,
      "loss": 0.0,
      "step": 50360
    },
    {
      "epoch": 0.7121548445474982,
      "grad_norm": 0.0,
      "learning_rate": 2.882268942018126e-06,
      "loss": 0.0,
      "step": 50370
    },
    {
      "epoch": 0.7122962292694651,
      "grad_norm": 0.0,
      "learning_rate": 2.8808550947984566e-06,
      "loss": 0.0,
      "step": 50380
    },
    {
      "epoch": 0.7124376139914321,
      "grad_norm": 0.0,
      "learning_rate": 2.879441247578787e-06,
      "loss": 0.0,
      "step": 50390
    },
    {
      "epoch": 0.712578998713399,
      "grad_norm": 0.0,
      "learning_rate": 2.8780274003591175e-06,
      "loss": 0.0001,
      "step": 50400
    },
    {
      "epoch": 0.7127203834353659,
      "grad_norm": 0.0,
      "learning_rate": 2.876613553139448e-06,
      "loss": 0.001,
      "step": 50410
    },
    {
      "epoch": 0.7128617681573329,
      "grad_norm": 0.0,
      "learning_rate": 2.8751997059197784e-06,
      "loss": 0.0,
      "step": 50420
    },
    {
      "epoch": 0.7130031528792998,
      "grad_norm": 0.0,
      "learning_rate": 2.873785858700109e-06,
      "loss": 0.0011,
      "step": 50430
    },
    {
      "epoch": 0.7131445376012668,
      "grad_norm": 146.60003662109375,
      "learning_rate": 2.87237201148044e-06,
      "loss": 0.0277,
      "step": 50440
    },
    {
      "epoch": 0.7132859223232337,
      "grad_norm": 0.0,
      "learning_rate": 2.8709581642607705e-06,
      "loss": 0.0,
      "step": 50450
    },
    {
      "epoch": 0.7134273070452007,
      "grad_norm": 0.0,
      "learning_rate": 2.869544317041101e-06,
      "loss": 0.0,
      "step": 50460
    },
    {
      "epoch": 0.7135686917671676,
      "grad_norm": 0.0,
      "learning_rate": 2.8681304698214314e-06,
      "loss": 1.1053,
      "step": 50470
    },
    {
      "epoch": 0.7137100764891345,
      "grad_norm": 0.0,
      "learning_rate": 2.866716622601762e-06,
      "loss": 0.0003,
      "step": 50480
    },
    {
      "epoch": 0.7138514612111015,
      "grad_norm": 0.0,
      "learning_rate": 2.8653027753820922e-06,
      "loss": 0.0259,
      "step": 50490
    },
    {
      "epoch": 0.7139928459330684,
      "grad_norm": 0.0,
      "learning_rate": 2.863888928162423e-06,
      "loss": 0.0,
      "step": 50500
    },
    {
      "epoch": 0.7141342306550355,
      "grad_norm": 7.785960769979283e-05,
      "learning_rate": 2.8624750809427535e-06,
      "loss": 0.0,
      "step": 50510
    },
    {
      "epoch": 0.7142756153770023,
      "grad_norm": 0.0,
      "learning_rate": 2.8610612337230844e-06,
      "loss": 0.0,
      "step": 50520
    },
    {
      "epoch": 0.7144170000989694,
      "grad_norm": 0.0,
      "learning_rate": 2.859647386503415e-06,
      "loss": 0.0,
      "step": 50530
    },
    {
      "epoch": 0.7145583848209363,
      "grad_norm": 0.0,
      "learning_rate": 2.8582335392837453e-06,
      "loss": 0.0001,
      "step": 50540
    },
    {
      "epoch": 0.7146997695429032,
      "grad_norm": 0.0,
      "learning_rate": 2.8568196920640757e-06,
      "loss": 0.6578,
      "step": 50550
    },
    {
      "epoch": 0.7148411542648702,
      "grad_norm": 0.0,
      "learning_rate": 2.855405844844406e-06,
      "loss": 0.6258,
      "step": 50560
    },
    {
      "epoch": 0.7149825389868371,
      "grad_norm": 0.0,
      "learning_rate": 2.853991997624737e-06,
      "loss": 0.0,
      "step": 50570
    },
    {
      "epoch": 0.7151239237088041,
      "grad_norm": 0.0,
      "learning_rate": 2.8525781504050674e-06,
      "loss": 0.0,
      "step": 50580
    },
    {
      "epoch": 0.715265308430771,
      "grad_norm": 9.820651030167937e-06,
      "learning_rate": 2.851164303185398e-06,
      "loss": 0.0,
      "step": 50590
    },
    {
      "epoch": 0.7154066931527379,
      "grad_norm": 0.0,
      "learning_rate": 2.8497504559657287e-06,
      "loss": 0.0,
      "step": 50600
    },
    {
      "epoch": 0.7155480778747049,
      "grad_norm": 0.0,
      "learning_rate": 2.848336608746059e-06,
      "loss": 0.0,
      "step": 50610
    },
    {
      "epoch": 0.7156894625966718,
      "grad_norm": 1.0928515621344559e-06,
      "learning_rate": 2.8469227615263896e-06,
      "loss": 0.0231,
      "step": 50620
    },
    {
      "epoch": 0.7158308473186388,
      "grad_norm": 0.0,
      "learning_rate": 2.8455089143067204e-06,
      "loss": 0.0,
      "step": 50630
    },
    {
      "epoch": 0.7159722320406057,
      "grad_norm": 0.0,
      "learning_rate": 2.844095067087051e-06,
      "loss": 0.0,
      "step": 50640
    },
    {
      "epoch": 0.7161136167625727,
      "grad_norm": 0.0,
      "learning_rate": 2.8426812198673813e-06,
      "loss": 0.0003,
      "step": 50650
    },
    {
      "epoch": 0.7162550014845396,
      "grad_norm": 5.28167176661043e-10,
      "learning_rate": 2.8412673726477117e-06,
      "loss": 0.0,
      "step": 50660
    },
    {
      "epoch": 0.7163963862065065,
      "grad_norm": 0.08912783861160278,
      "learning_rate": 2.8398535254280426e-06,
      "loss": 0.0,
      "step": 50670
    },
    {
      "epoch": 0.7165377709284735,
      "grad_norm": 1.0384513249306337e-09,
      "learning_rate": 2.838439678208373e-06,
      "loss": 0.0018,
      "step": 50680
    },
    {
      "epoch": 0.7166791556504404,
      "grad_norm": 0.0,
      "learning_rate": 2.8370258309887034e-06,
      "loss": 0.0,
      "step": 50690
    },
    {
      "epoch": 0.7168205403724074,
      "grad_norm": 0.0,
      "learning_rate": 2.8356119837690343e-06,
      "loss": 0.0001,
      "step": 50700
    },
    {
      "epoch": 0.7169619250943743,
      "grad_norm": 0.0,
      "learning_rate": 2.8341981365493647e-06,
      "loss": 0.0,
      "step": 50710
    },
    {
      "epoch": 0.7171033098163413,
      "grad_norm": 0.0,
      "learning_rate": 2.832784289329695e-06,
      "loss": 0.0,
      "step": 50720
    },
    {
      "epoch": 0.7172446945383082,
      "grad_norm": 0.0,
      "learning_rate": 2.8313704421100256e-06,
      "loss": 0.0,
      "step": 50730
    },
    {
      "epoch": 0.7173860792602751,
      "grad_norm": 0.0,
      "learning_rate": 2.829956594890356e-06,
      "loss": 0.0,
      "step": 50740
    },
    {
      "epoch": 0.7175274639822421,
      "grad_norm": 0.0,
      "learning_rate": 2.828542747670687e-06,
      "loss": 0.0,
      "step": 50750
    },
    {
      "epoch": 0.717668848704209,
      "grad_norm": 1.9175189663656056e-05,
      "learning_rate": 2.8271289004510177e-06,
      "loss": 0.0,
      "step": 50760
    },
    {
      "epoch": 0.717810233426176,
      "grad_norm": 0.0,
      "learning_rate": 2.825715053231348e-06,
      "loss": 0.0,
      "step": 50770
    },
    {
      "epoch": 0.7179516181481429,
      "grad_norm": 0.0,
      "learning_rate": 2.8243012060116786e-06,
      "loss": 0.0,
      "step": 50780
    },
    {
      "epoch": 0.7180930028701099,
      "grad_norm": 0.0,
      "learning_rate": 2.822887358792009e-06,
      "loss": 0.0,
      "step": 50790
    },
    {
      "epoch": 0.7182343875920768,
      "grad_norm": 0.0,
      "learning_rate": 2.8214735115723395e-06,
      "loss": 0.0017,
      "step": 50800
    },
    {
      "epoch": 0.7183757723140437,
      "grad_norm": 4.107832580757531e-07,
      "learning_rate": 2.82005966435267e-06,
      "loss": 0.0,
      "step": 50810
    },
    {
      "epoch": 0.7185171570360107,
      "grad_norm": 0.0,
      "learning_rate": 2.8186458171330008e-06,
      "loss": 0.0,
      "step": 50820
    },
    {
      "epoch": 0.7186585417579776,
      "grad_norm": 0.0,
      "learning_rate": 2.8172319699133316e-06,
      "loss": 0.0009,
      "step": 50830
    },
    {
      "epoch": 0.7187999264799446,
      "grad_norm": 0.0,
      "learning_rate": 2.815818122693662e-06,
      "loss": 0.0,
      "step": 50840
    },
    {
      "epoch": 0.7189413112019115,
      "grad_norm": 0.0,
      "learning_rate": 2.8144042754739925e-06,
      "loss": 0.0,
      "step": 50850
    },
    {
      "epoch": 0.7190826959238784,
      "grad_norm": 0.0,
      "learning_rate": 2.812990428254323e-06,
      "loss": 0.0,
      "step": 50860
    },
    {
      "epoch": 0.7192240806458454,
      "grad_norm": 0.0,
      "learning_rate": 2.8115765810346534e-06,
      "loss": 0.0,
      "step": 50870
    },
    {
      "epoch": 0.7193654653678123,
      "grad_norm": 0.0,
      "learning_rate": 2.8101627338149838e-06,
      "loss": 0.0,
      "step": 50880
    },
    {
      "epoch": 0.7195068500897793,
      "grad_norm": 0.0,
      "learning_rate": 2.808748886595315e-06,
      "loss": 0.0001,
      "step": 50890
    },
    {
      "epoch": 0.7196482348117462,
      "grad_norm": 1.095351649382792e-06,
      "learning_rate": 2.8073350393756455e-06,
      "loss": 0.0,
      "step": 50900
    },
    {
      "epoch": 0.7197896195337132,
      "grad_norm": 0.0,
      "learning_rate": 2.805921192155976e-06,
      "loss": 0.0,
      "step": 50910
    },
    {
      "epoch": 0.7199310042556801,
      "grad_norm": 0.0,
      "learning_rate": 2.8045073449363064e-06,
      "loss": 0.0,
      "step": 50920
    },
    {
      "epoch": 0.720072388977647,
      "grad_norm": 0.0,
      "learning_rate": 2.803093497716637e-06,
      "loss": 0.0067,
      "step": 50930
    },
    {
      "epoch": 0.720213773699614,
      "grad_norm": 0.0,
      "learning_rate": 2.8016796504969672e-06,
      "loss": 0.0016,
      "step": 50940
    },
    {
      "epoch": 0.7203551584215809,
      "grad_norm": 0.0,
      "learning_rate": 2.8002658032772985e-06,
      "loss": 0.0,
      "step": 50950
    },
    {
      "epoch": 0.7204965431435479,
      "grad_norm": 0.0,
      "learning_rate": 2.798851956057629e-06,
      "loss": 0.0,
      "step": 50960
    },
    {
      "epoch": 0.7206379278655148,
      "grad_norm": 0.0,
      "learning_rate": 2.7974381088379594e-06,
      "loss": 0.0,
      "step": 50970
    },
    {
      "epoch": 0.7207793125874818,
      "grad_norm": 0.0,
      "learning_rate": 2.79602426161829e-06,
      "loss": 0.0,
      "step": 50980
    },
    {
      "epoch": 0.7209206973094487,
      "grad_norm": 9.319094385062954e-10,
      "learning_rate": 2.7946104143986202e-06,
      "loss": 0.0024,
      "step": 50990
    },
    {
      "epoch": 0.7210620820314156,
      "grad_norm": 0.0,
      "learning_rate": 2.7931965671789507e-06,
      "loss": 0.0139,
      "step": 51000
    },
    {
      "epoch": 0.7212034667533826,
      "grad_norm": 0.0,
      "learning_rate": 2.791782719959281e-06,
      "loss": 0.0,
      "step": 51010
    },
    {
      "epoch": 0.7213448514753495,
      "grad_norm": 0.0,
      "learning_rate": 2.7903688727396124e-06,
      "loss": 0.0,
      "step": 51020
    },
    {
      "epoch": 0.7214862361973166,
      "grad_norm": 0.0,
      "learning_rate": 2.788955025519943e-06,
      "loss": 0.0,
      "step": 51030
    },
    {
      "epoch": 0.7216276209192835,
      "grad_norm": 0.0,
      "learning_rate": 2.7875411783002733e-06,
      "loss": 0.0,
      "step": 51040
    },
    {
      "epoch": 0.7217690056412505,
      "grad_norm": 0.0,
      "learning_rate": 2.7861273310806037e-06,
      "loss": 0.2988,
      "step": 51050
    },
    {
      "epoch": 0.7219103903632174,
      "grad_norm": 0.30592647194862366,
      "learning_rate": 2.784713483860934e-06,
      "loss": 0.0487,
      "step": 51060
    },
    {
      "epoch": 0.7220517750851843,
      "grad_norm": 0.0,
      "learning_rate": 2.7832996366412646e-06,
      "loss": 0.0,
      "step": 51070
    },
    {
      "epoch": 0.7221931598071513,
      "grad_norm": 0.0,
      "learning_rate": 2.7818857894215954e-06,
      "loss": 0.0126,
      "step": 51080
    },
    {
      "epoch": 0.7223345445291182,
      "grad_norm": 0.0,
      "learning_rate": 2.7804719422019263e-06,
      "loss": 0.0,
      "step": 51090
    },
    {
      "epoch": 0.7224759292510852,
      "grad_norm": 0.007275630719959736,
      "learning_rate": 2.7790580949822567e-06,
      "loss": 0.0,
      "step": 51100
    },
    {
      "epoch": 0.7226173139730521,
      "grad_norm": 0.0,
      "learning_rate": 2.777644247762587e-06,
      "loss": 0.0007,
      "step": 51110
    },
    {
      "epoch": 0.722758698695019,
      "grad_norm": 0.0,
      "learning_rate": 2.7762304005429176e-06,
      "loss": 0.0,
      "step": 51120
    },
    {
      "epoch": 0.722900083416986,
      "grad_norm": 0.0,
      "learning_rate": 2.774816553323248e-06,
      "loss": 0.0,
      "step": 51130
    },
    {
      "epoch": 0.7230414681389529,
      "grad_norm": 2.1356962032204052e-10,
      "learning_rate": 2.7734027061035784e-06,
      "loss": 0.0,
      "step": 51140
    },
    {
      "epoch": 0.7231828528609199,
      "grad_norm": 7.047134999993432e-07,
      "learning_rate": 2.7719888588839093e-06,
      "loss": 0.0,
      "step": 51150
    },
    {
      "epoch": 0.7233242375828868,
      "grad_norm": 7.694927626289427e-05,
      "learning_rate": 2.7705750116642397e-06,
      "loss": 0.0007,
      "step": 51160
    },
    {
      "epoch": 0.7234656223048538,
      "grad_norm": 0.0,
      "learning_rate": 2.7691611644445706e-06,
      "loss": 0.0,
      "step": 51170
    },
    {
      "epoch": 0.7236070070268207,
      "grad_norm": 3.879696578223957e-06,
      "learning_rate": 2.767747317224901e-06,
      "loss": 0.0315,
      "step": 51180
    },
    {
      "epoch": 0.7237483917487876,
      "grad_norm": 1.7011786894727265e-06,
      "learning_rate": 2.7663334700052314e-06,
      "loss": 0.0483,
      "step": 51190
    },
    {
      "epoch": 0.7238897764707546,
      "grad_norm": 0.0,
      "learning_rate": 2.764919622785562e-06,
      "loss": 0.0,
      "step": 51200
    },
    {
      "epoch": 0.7240311611927215,
      "grad_norm": 0.0,
      "learning_rate": 2.7635057755658927e-06,
      "loss": 0.0,
      "step": 51210
    },
    {
      "epoch": 0.7241725459146885,
      "grad_norm": 0.0,
      "learning_rate": 2.762091928346223e-06,
      "loss": 0.0,
      "step": 51220
    },
    {
      "epoch": 0.7243139306366554,
      "grad_norm": 0.0,
      "learning_rate": 2.7606780811265536e-06,
      "loss": 0.0,
      "step": 51230
    },
    {
      "epoch": 0.7244553153586224,
      "grad_norm": 2.2407685429381985e-10,
      "learning_rate": 2.7592642339068845e-06,
      "loss": 0.0,
      "step": 51240
    },
    {
      "epoch": 0.7245967000805893,
      "grad_norm": 0.007402906194329262,
      "learning_rate": 2.757850386687215e-06,
      "loss": 0.0029,
      "step": 51250
    },
    {
      "epoch": 0.7247380848025562,
      "grad_norm": 0.0,
      "learning_rate": 2.7564365394675453e-06,
      "loss": 0.0,
      "step": 51260
    },
    {
      "epoch": 0.7248794695245232,
      "grad_norm": 2.2189713490661234e-05,
      "learning_rate": 2.7550226922478758e-06,
      "loss": 0.0,
      "step": 51270
    },
    {
      "epoch": 0.7250208542464901,
      "grad_norm": 0.0,
      "learning_rate": 2.7536088450282066e-06,
      "loss": 0.0,
      "step": 51280
    },
    {
      "epoch": 0.7251622389684571,
      "grad_norm": 0.0002566586481407285,
      "learning_rate": 2.752194997808537e-06,
      "loss": 0.0,
      "step": 51290
    },
    {
      "epoch": 0.725303623690424,
      "grad_norm": 2.4616502614094316e-08,
      "learning_rate": 2.7507811505888675e-06,
      "loss": 0.0003,
      "step": 51300
    },
    {
      "epoch": 0.725445008412391,
      "grad_norm": 0.0005819625221192837,
      "learning_rate": 2.749367303369198e-06,
      "loss": 0.0232,
      "step": 51310
    },
    {
      "epoch": 0.7255863931343579,
      "grad_norm": 2.4812960019993113e-10,
      "learning_rate": 2.7479534561495288e-06,
      "loss": 0.0,
      "step": 51320
    },
    {
      "epoch": 0.7257277778563248,
      "grad_norm": 0.0,
      "learning_rate": 2.746539608929859e-06,
      "loss": 0.0,
      "step": 51330
    },
    {
      "epoch": 0.7258691625782918,
      "grad_norm": 0.0,
      "learning_rate": 2.74512576171019e-06,
      "loss": 0.0001,
      "step": 51340
    },
    {
      "epoch": 0.7260105473002587,
      "grad_norm": 0.0,
      "learning_rate": 2.7437119144905205e-06,
      "loss": 0.0121,
      "step": 51350
    },
    {
      "epoch": 0.7261519320222257,
      "grad_norm": 0.0,
      "learning_rate": 2.742298067270851e-06,
      "loss": 0.0,
      "step": 51360
    },
    {
      "epoch": 0.7262933167441926,
      "grad_norm": 0.0,
      "learning_rate": 2.7408842200511814e-06,
      "loss": 0.0,
      "step": 51370
    },
    {
      "epoch": 0.7264347014661595,
      "grad_norm": 0.0,
      "learning_rate": 2.739470372831512e-06,
      "loss": 0.0,
      "step": 51380
    },
    {
      "epoch": 0.7265760861881265,
      "grad_norm": 0.00043455694685690105,
      "learning_rate": 2.7380565256118422e-06,
      "loss": 0.0,
      "step": 51390
    },
    {
      "epoch": 0.7267174709100934,
      "grad_norm": 0.0,
      "learning_rate": 2.736642678392173e-06,
      "loss": 0.0,
      "step": 51400
    },
    {
      "epoch": 0.7268588556320604,
      "grad_norm": 1.2477555344503344e-07,
      "learning_rate": 2.735228831172504e-06,
      "loss": 0.0,
      "step": 51410
    },
    {
      "epoch": 0.7270002403540273,
      "grad_norm": 0.0,
      "learning_rate": 2.7338149839528344e-06,
      "loss": 0.0,
      "step": 51420
    },
    {
      "epoch": 0.7271416250759943,
      "grad_norm": 0.0,
      "learning_rate": 2.732401136733165e-06,
      "loss": 0.0,
      "step": 51430
    },
    {
      "epoch": 0.7272830097979612,
      "grad_norm": 0.0,
      "learning_rate": 2.7309872895134952e-06,
      "loss": 0.0,
      "step": 51440
    },
    {
      "epoch": 0.7274243945199281,
      "grad_norm": 0.5919610857963562,
      "learning_rate": 2.7295734422938257e-06,
      "loss": 0.0003,
      "step": 51450
    },
    {
      "epoch": 0.7275657792418951,
      "grad_norm": 0.0,
      "learning_rate": 2.728159595074156e-06,
      "loss": 0.0,
      "step": 51460
    },
    {
      "epoch": 0.727707163963862,
      "grad_norm": 0.0,
      "learning_rate": 2.7267457478544874e-06,
      "loss": 0.0035,
      "step": 51470
    },
    {
      "epoch": 0.727848548685829,
      "grad_norm": 0.0,
      "learning_rate": 2.725331900634818e-06,
      "loss": 0.0,
      "step": 51480
    },
    {
      "epoch": 0.7279899334077959,
      "grad_norm": 0.0,
      "learning_rate": 2.7239180534151482e-06,
      "loss": 0.0,
      "step": 51490
    },
    {
      "epoch": 0.728131318129763,
      "grad_norm": 2.3918410079204477e-06,
      "learning_rate": 2.7225042061954787e-06,
      "loss": 0.0,
      "step": 51500
    },
    {
      "epoch": 0.7282727028517298,
      "grad_norm": 0.0,
      "learning_rate": 2.721090358975809e-06,
      "loss": 0.0,
      "step": 51510
    },
    {
      "epoch": 0.7284140875736967,
      "grad_norm": 0.0,
      "learning_rate": 2.7196765117561395e-06,
      "loss": 0.0,
      "step": 51520
    },
    {
      "epoch": 0.7285554722956638,
      "grad_norm": 0.0,
      "learning_rate": 2.71826266453647e-06,
      "loss": 0.0,
      "step": 51530
    },
    {
      "epoch": 0.7286968570176307,
      "grad_norm": 6.204494184203213e-06,
      "learning_rate": 2.7168488173168013e-06,
      "loss": 0.0,
      "step": 51540
    },
    {
      "epoch": 0.7288382417395977,
      "grad_norm": 0.0,
      "learning_rate": 2.7155763548190985e-06,
      "loss": 0.1236,
      "step": 51550
    },
    {
      "epoch": 0.7289796264615646,
      "grad_norm": 0.0,
      "learning_rate": 2.714162507599429e-06,
      "loss": 0.0228,
      "step": 51560
    },
    {
      "epoch": 0.7291210111835315,
      "grad_norm": 0.0,
      "learning_rate": 2.71274866037976e-06,
      "loss": 0.0,
      "step": 51570
    },
    {
      "epoch": 0.7292623959054985,
      "grad_norm": 4.4925982378707374e-10,
      "learning_rate": 2.7113348131600902e-06,
      "loss": 0.0,
      "step": 51580
    },
    {
      "epoch": 0.7294037806274654,
      "grad_norm": 0.0,
      "learning_rate": 2.7099209659404207e-06,
      "loss": 0.0,
      "step": 51590
    },
    {
      "epoch": 0.7295451653494324,
      "grad_norm": 0.0,
      "learning_rate": 2.708507118720751e-06,
      "loss": 0.0,
      "step": 51600
    },
    {
      "epoch": 0.7296865500713993,
      "grad_norm": 0.0,
      "learning_rate": 2.7070932715010815e-06,
      "loss": 0.0,
      "step": 51610
    },
    {
      "epoch": 0.7298279347933663,
      "grad_norm": 4.2362826206954196e-05,
      "learning_rate": 2.705679424281412e-06,
      "loss": 0.0,
      "step": 51620
    },
    {
      "epoch": 0.7299693195153332,
      "grad_norm": 0.0,
      "learning_rate": 2.7042655770617433e-06,
      "loss": 0.0,
      "step": 51630
    },
    {
      "epoch": 0.7301107042373001,
      "grad_norm": 0.0,
      "learning_rate": 2.7028517298420737e-06,
      "loss": 0.0,
      "step": 51640
    },
    {
      "epoch": 0.7302520889592671,
      "grad_norm": 0.0,
      "learning_rate": 2.701437882622404e-06,
      "loss": 0.0,
      "step": 51650
    },
    {
      "epoch": 0.730393473681234,
      "grad_norm": 3.905052947672516e-10,
      "learning_rate": 2.7000240354027346e-06,
      "loss": 0.0003,
      "step": 51660
    },
    {
      "epoch": 0.730534858403201,
      "grad_norm": 0.0,
      "learning_rate": 2.698610188183065e-06,
      "loss": 0.0,
      "step": 51670
    },
    {
      "epoch": 0.7306762431251679,
      "grad_norm": 1.1689147427773605e-09,
      "learning_rate": 2.6971963409633954e-06,
      "loss": 0.0,
      "step": 51680
    },
    {
      "epoch": 0.7308176278471349,
      "grad_norm": 0.0,
      "learning_rate": 2.695782493743726e-06,
      "loss": 0.0001,
      "step": 51690
    },
    {
      "epoch": 0.7309590125691018,
      "grad_norm": 9.565438813297078e-05,
      "learning_rate": 2.694368646524057e-06,
      "loss": 0.0,
      "step": 51700
    },
    {
      "epoch": 0.7311003972910687,
      "grad_norm": 4.637669563293457,
      "learning_rate": 2.6929547993043876e-06,
      "loss": 0.0006,
      "step": 51710
    },
    {
      "epoch": 0.7312417820130357,
      "grad_norm": 0.0,
      "learning_rate": 2.691540952084718e-06,
      "loss": 0.0002,
      "step": 51720
    },
    {
      "epoch": 0.7313831667350026,
      "grad_norm": 1.2991756648261799e-06,
      "learning_rate": 2.6901271048650484e-06,
      "loss": 0.0,
      "step": 51730
    },
    {
      "epoch": 0.7315245514569696,
      "grad_norm": 9.174393085231713e-07,
      "learning_rate": 2.688713257645379e-06,
      "loss": 0.0,
      "step": 51740
    },
    {
      "epoch": 0.7316659361789365,
      "grad_norm": 0.0,
      "learning_rate": 2.6872994104257093e-06,
      "loss": 0.0,
      "step": 51750
    },
    {
      "epoch": 0.7318073209009035,
      "grad_norm": 0.0,
      "learning_rate": 2.6858855632060406e-06,
      "loss": 0.0,
      "step": 51760
    },
    {
      "epoch": 0.7319487056228704,
      "grad_norm": 0.0,
      "learning_rate": 2.684471715986371e-06,
      "loss": 0.0,
      "step": 51770
    },
    {
      "epoch": 0.7320900903448373,
      "grad_norm": 43.8358268737793,
      "learning_rate": 2.6830578687667014e-06,
      "loss": 0.005,
      "step": 51780
    },
    {
      "epoch": 0.7322314750668043,
      "grad_norm": 0.0,
      "learning_rate": 2.681644021547032e-06,
      "loss": 0.0,
      "step": 51790
    },
    {
      "epoch": 0.7323728597887712,
      "grad_norm": 0.0,
      "learning_rate": 2.6802301743273623e-06,
      "loss": 0.0,
      "step": 51800
    },
    {
      "epoch": 0.7325142445107382,
      "grad_norm": 0.0,
      "learning_rate": 2.6788163271076927e-06,
      "loss": 0.0002,
      "step": 51810
    },
    {
      "epoch": 0.7326556292327051,
      "grad_norm": 1.2793105241826197e-08,
      "learning_rate": 2.677402479888023e-06,
      "loss": 0.0,
      "step": 51820
    },
    {
      "epoch": 0.732797013954672,
      "grad_norm": 1.794216615280675e-07,
      "learning_rate": 2.6759886326683545e-06,
      "loss": 0.0,
      "step": 51830
    },
    {
      "epoch": 0.732938398676639,
      "grad_norm": 6.046614362276159e-05,
      "learning_rate": 2.674574785448685e-06,
      "loss": 0.0001,
      "step": 51840
    },
    {
      "epoch": 0.7330797833986059,
      "grad_norm": 2.7171536203240976e-05,
      "learning_rate": 2.6731609382290153e-06,
      "loss": 0.0,
      "step": 51850
    },
    {
      "epoch": 0.7332211681205729,
      "grad_norm": 0.0,
      "learning_rate": 2.6717470910093458e-06,
      "loss": 0.0001,
      "step": 51860
    },
    {
      "epoch": 0.7333625528425398,
      "grad_norm": 0.0,
      "learning_rate": 2.670333243789676e-06,
      "loss": 0.0,
      "step": 51870
    },
    {
      "epoch": 0.7335039375645068,
      "grad_norm": 0.0,
      "learning_rate": 2.6689193965700066e-06,
      "loss": 0.0004,
      "step": 51880
    },
    {
      "epoch": 0.7336453222864737,
      "grad_norm": 0.0,
      "learning_rate": 2.6675055493503375e-06,
      "loss": 0.0,
      "step": 51890
    },
    {
      "epoch": 0.7337867070084406,
      "grad_norm": 2.1629593671690373e-09,
      "learning_rate": 2.6660917021306683e-06,
      "loss": 0.0,
      "step": 51900
    },
    {
      "epoch": 0.7339280917304076,
      "grad_norm": 4.615833548715642e-10,
      "learning_rate": 2.6646778549109988e-06,
      "loss": 0.0,
      "step": 51910
    },
    {
      "epoch": 0.7340694764523745,
      "grad_norm": 0.0,
      "learning_rate": 2.663264007691329e-06,
      "loss": 0.0,
      "step": 51920
    },
    {
      "epoch": 0.7342108611743415,
      "grad_norm": 0.0,
      "learning_rate": 2.6618501604716596e-06,
      "loss": 0.0,
      "step": 51930
    },
    {
      "epoch": 0.7343522458963084,
      "grad_norm": 0.0,
      "learning_rate": 2.66043631325199e-06,
      "loss": 0.0,
      "step": 51940
    },
    {
      "epoch": 0.7344936306182754,
      "grad_norm": 0.0,
      "learning_rate": 2.6590224660323205e-06,
      "loss": 0.0,
      "step": 51950
    },
    {
      "epoch": 0.7346350153402423,
      "grad_norm": 0.0,
      "learning_rate": 2.6576086188126514e-06,
      "loss": 0.0,
      "step": 51960
    },
    {
      "epoch": 0.7347764000622092,
      "grad_norm": 0.5237599611282349,
      "learning_rate": 2.656194771592982e-06,
      "loss": 0.0001,
      "step": 51970
    },
    {
      "epoch": 0.7349177847841762,
      "grad_norm": 0.0,
      "learning_rate": 2.6547809243733126e-06,
      "loss": 0.0,
      "step": 51980
    },
    {
      "epoch": 0.7350591695061431,
      "grad_norm": 0.0,
      "learning_rate": 2.653367077153643e-06,
      "loss": 0.0002,
      "step": 51990
    },
    {
      "epoch": 0.7352005542281101,
      "grad_norm": 0.0,
      "learning_rate": 2.6519532299339735e-06,
      "loss": 0.0048,
      "step": 52000
    },
    {
      "epoch": 0.735341938950077,
      "grad_norm": 0.0,
      "learning_rate": 2.650539382714304e-06,
      "loss": 0.0,
      "step": 52010
    },
    {
      "epoch": 0.735483323672044,
      "grad_norm": 0.0,
      "learning_rate": 2.649125535494635e-06,
      "loss": 0.0,
      "step": 52020
    },
    {
      "epoch": 0.735624708394011,
      "grad_norm": 1.119431431106932e-06,
      "learning_rate": 2.6477116882749652e-06,
      "loss": 0.0,
      "step": 52030
    },
    {
      "epoch": 0.7357660931159778,
      "grad_norm": 1.6901081800460815,
      "learning_rate": 2.6462978410552957e-06,
      "loss": 0.0003,
      "step": 52040
    },
    {
      "epoch": 0.7359074778379449,
      "grad_norm": 5.752651486545801e-05,
      "learning_rate": 2.6448839938356265e-06,
      "loss": 0.0,
      "step": 52050
    },
    {
      "epoch": 0.7360488625599118,
      "grad_norm": 0.0,
      "learning_rate": 2.643470146615957e-06,
      "loss": 0.0,
      "step": 52060
    },
    {
      "epoch": 0.7361902472818788,
      "grad_norm": 0.0,
      "learning_rate": 2.6420562993962874e-06,
      "loss": 0.0,
      "step": 52070
    },
    {
      "epoch": 0.7363316320038457,
      "grad_norm": 9.988518900883037e-10,
      "learning_rate": 2.640642452176618e-06,
      "loss": 0.0,
      "step": 52080
    },
    {
      "epoch": 0.7364730167258126,
      "grad_norm": 0.0,
      "learning_rate": 2.6392286049569487e-06,
      "loss": 0.0037,
      "step": 52090
    },
    {
      "epoch": 0.7366144014477796,
      "grad_norm": 1.9312825827455526e-07,
      "learning_rate": 2.637814757737279e-06,
      "loss": 0.0,
      "step": 52100
    },
    {
      "epoch": 0.7367557861697465,
      "grad_norm": 1.397068860597983e-08,
      "learning_rate": 2.6364009105176095e-06,
      "loss": 0.0,
      "step": 52110
    },
    {
      "epoch": 0.7368971708917135,
      "grad_norm": 0.0,
      "learning_rate": 2.63498706329794e-06,
      "loss": 0.0,
      "step": 52120
    },
    {
      "epoch": 0.7370385556136804,
      "grad_norm": 1.4660489000561938e-07,
      "learning_rate": 2.633573216078271e-06,
      "loss": 0.0,
      "step": 52130
    },
    {
      "epoch": 0.7371799403356474,
      "grad_norm": 9.614846931071952e-06,
      "learning_rate": 2.6321593688586013e-06,
      "loss": 0.0,
      "step": 52140
    },
    {
      "epoch": 0.7373213250576143,
      "grad_norm": 0.0,
      "learning_rate": 2.630745521638932e-06,
      "loss": 0.0,
      "step": 52150
    },
    {
      "epoch": 0.7374627097795812,
      "grad_norm": 0.0,
      "learning_rate": 2.6293316744192626e-06,
      "loss": 0.0,
      "step": 52160
    },
    {
      "epoch": 0.7376040945015482,
      "grad_norm": 0.0,
      "learning_rate": 2.627917827199593e-06,
      "loss": 0.0,
      "step": 52170
    },
    {
      "epoch": 0.7377454792235151,
      "grad_norm": 0.0,
      "learning_rate": 2.6265039799799234e-06,
      "loss": 0.0,
      "step": 52180
    },
    {
      "epoch": 0.7378868639454821,
      "grad_norm": 1.8267033965457813e-06,
      "learning_rate": 2.625090132760254e-06,
      "loss": 0.0,
      "step": 52190
    },
    {
      "epoch": 0.738028248667449,
      "grad_norm": 2.4105736429191893e-07,
      "learning_rate": 2.6236762855405847e-06,
      "loss": 0.0,
      "step": 52200
    },
    {
      "epoch": 0.738169633389416,
      "grad_norm": 0.0,
      "learning_rate": 2.622262438320915e-06,
      "loss": 0.0,
      "step": 52210
    },
    {
      "epoch": 0.7383110181113829,
      "grad_norm": 1.6811041859909892e-05,
      "learning_rate": 2.620848591101246e-06,
      "loss": 0.0002,
      "step": 52220
    },
    {
      "epoch": 0.7384524028333498,
      "grad_norm": 0.0,
      "learning_rate": 2.6194347438815764e-06,
      "loss": 0.0,
      "step": 52230
    },
    {
      "epoch": 0.7385937875553168,
      "grad_norm": 0.0,
      "learning_rate": 2.618020896661907e-06,
      "loss": 0.0,
      "step": 52240
    },
    {
      "epoch": 0.7387351722772837,
      "grad_norm": 0.0,
      "learning_rate": 2.6166070494422373e-06,
      "loss": 0.0,
      "step": 52250
    },
    {
      "epoch": 0.7388765569992507,
      "grad_norm": 0.0,
      "learning_rate": 2.6151932022225677e-06,
      "loss": 0.0002,
      "step": 52260
    },
    {
      "epoch": 0.7390179417212176,
      "grad_norm": 0.0,
      "learning_rate": 2.613779355002898e-06,
      "loss": 0.0,
      "step": 52270
    },
    {
      "epoch": 0.7391593264431845,
      "grad_norm": 0.0,
      "learning_rate": 2.6123655077832294e-06,
      "loss": 0.0025,
      "step": 52280
    },
    {
      "epoch": 0.7393007111651515,
      "grad_norm": 0.0,
      "learning_rate": 2.61095166056356e-06,
      "loss": 0.0036,
      "step": 52290
    },
    {
      "epoch": 0.7394420958871184,
      "grad_norm": 0.0003326218284200877,
      "learning_rate": 2.6095378133438903e-06,
      "loss": 0.0,
      "step": 52300
    },
    {
      "epoch": 0.7395834806090854,
      "grad_norm": 0.0,
      "learning_rate": 2.6081239661242207e-06,
      "loss": 0.0,
      "step": 52310
    },
    {
      "epoch": 0.7397248653310523,
      "grad_norm": 4.769757651956752e-06,
      "learning_rate": 2.606710118904551e-06,
      "loss": 0.0,
      "step": 52320
    },
    {
      "epoch": 0.7398662500530193,
      "grad_norm": 0.0,
      "learning_rate": 2.6052962716848816e-06,
      "loss": 0.0,
      "step": 52330
    },
    {
      "epoch": 0.7400076347749862,
      "grad_norm": 3.2792922866065055e-05,
      "learning_rate": 2.603882424465213e-06,
      "loss": 0.0001,
      "step": 52340
    },
    {
      "epoch": 0.7401490194969531,
      "grad_norm": 0.00011544657172635198,
      "learning_rate": 2.6024685772455433e-06,
      "loss": 0.0,
      "step": 52350
    },
    {
      "epoch": 0.7402904042189201,
      "grad_norm": 0.0,
      "learning_rate": 2.6010547300258738e-06,
      "loss": 0.0,
      "step": 52360
    },
    {
      "epoch": 0.740431788940887,
      "grad_norm": 0.0,
      "learning_rate": 2.599640882806204e-06,
      "loss": 0.0137,
      "step": 52370
    },
    {
      "epoch": 0.740573173662854,
      "grad_norm": 0.003548892680555582,
      "learning_rate": 2.5982270355865346e-06,
      "loss": 0.0,
      "step": 52380
    },
    {
      "epoch": 0.7407145583848209,
      "grad_norm": 0.0,
      "learning_rate": 2.596813188366865e-06,
      "loss": 0.0,
      "step": 52390
    },
    {
      "epoch": 0.7408559431067879,
      "grad_norm": 0.0,
      "learning_rate": 2.5953993411471955e-06,
      "loss": 0.0,
      "step": 52400
    },
    {
      "epoch": 0.7409973278287548,
      "grad_norm": 0.00013885751832276583,
      "learning_rate": 2.5939854939275268e-06,
      "loss": 0.0,
      "step": 52410
    },
    {
      "epoch": 0.7411387125507217,
      "grad_norm": 0.0,
      "learning_rate": 2.592571646707857e-06,
      "loss": 0.0,
      "step": 52420
    },
    {
      "epoch": 0.7412800972726887,
      "grad_norm": 0.00041466212132945657,
      "learning_rate": 2.5911577994881876e-06,
      "loss": 0.0,
      "step": 52430
    },
    {
      "epoch": 0.7414214819946556,
      "grad_norm": 0.0,
      "learning_rate": 2.589743952268518e-06,
      "loss": 0.0,
      "step": 52440
    },
    {
      "epoch": 0.7415628667166226,
      "grad_norm": 0.0,
      "learning_rate": 2.5883301050488485e-06,
      "loss": 0.2975,
      "step": 52450
    },
    {
      "epoch": 0.7417042514385895,
      "grad_norm": 2.8711736987929726e-08,
      "learning_rate": 2.586916257829179e-06,
      "loss": 0.0,
      "step": 52460
    },
    {
      "epoch": 0.7418456361605565,
      "grad_norm": 0.0,
      "learning_rate": 2.58550241060951e-06,
      "loss": 0.0,
      "step": 52470
    },
    {
      "epoch": 0.7419870208825234,
      "grad_norm": 0.0,
      "learning_rate": 2.5840885633898406e-06,
      "loss": 0.0,
      "step": 52480
    },
    {
      "epoch": 0.7421284056044903,
      "grad_norm": 9.173732905765064e-06,
      "learning_rate": 2.582674716170171e-06,
      "loss": 0.0,
      "step": 52490
    },
    {
      "epoch": 0.7422697903264573,
      "grad_norm": 0.0,
      "learning_rate": 2.5812608689505015e-06,
      "loss": 0.0024,
      "step": 52500
    },
    {
      "epoch": 0.7424111750484242,
      "grad_norm": 8.655733907403373e-10,
      "learning_rate": 2.579847021730832e-06,
      "loss": 0.0,
      "step": 52510
    },
    {
      "epoch": 0.7425525597703913,
      "grad_norm": 0.0,
      "learning_rate": 2.5784331745111624e-06,
      "loss": 0.0,
      "step": 52520
    },
    {
      "epoch": 0.7426939444923581,
      "grad_norm": 0.0,
      "learning_rate": 2.577019327291493e-06,
      "loss": 0.048,
      "step": 52530
    },
    {
      "epoch": 0.742835329214325,
      "grad_norm": 2.6002056983998045e-05,
      "learning_rate": 2.5756054800718237e-06,
      "loss": 0.0052,
      "step": 52540
    },
    {
      "epoch": 0.742976713936292,
      "grad_norm": 8.482705015921965e-05,
      "learning_rate": 2.5741916328521545e-06,
      "loss": 0.0,
      "step": 52550
    },
    {
      "epoch": 0.743118098658259,
      "grad_norm": 0.0,
      "learning_rate": 2.572777785632485e-06,
      "loss": 0.0,
      "step": 52560
    },
    {
      "epoch": 0.743259483380226,
      "grad_norm": 0.007612711749970913,
      "learning_rate": 2.5713639384128154e-06,
      "loss": 0.0,
      "step": 52570
    },
    {
      "epoch": 0.7434008681021929,
      "grad_norm": 0.004144036211073399,
      "learning_rate": 2.569950091193146e-06,
      "loss": 0.0055,
      "step": 52580
    },
    {
      "epoch": 0.7435422528241599,
      "grad_norm": 0.0,
      "learning_rate": 2.5685362439734763e-06,
      "loss": 0.0044,
      "step": 52590
    },
    {
      "epoch": 0.7436836375461268,
      "grad_norm": 0.0,
      "learning_rate": 2.567122396753807e-06,
      "loss": 0.0051,
      "step": 52600
    },
    {
      "epoch": 0.7438250222680937,
      "grad_norm": 0.0,
      "learning_rate": 2.5657085495341375e-06,
      "loss": 0.0,
      "step": 52610
    },
    {
      "epoch": 0.7439664069900607,
      "grad_norm": 0.0,
      "learning_rate": 2.5642947023144684e-06,
      "loss": 0.0,
      "step": 52620
    },
    {
      "epoch": 0.7441077917120276,
      "grad_norm": 0.03716954216361046,
      "learning_rate": 2.562880855094799e-06,
      "loss": 0.0,
      "step": 52630
    },
    {
      "epoch": 0.7442491764339946,
      "grad_norm": 0.0,
      "learning_rate": 2.5614670078751293e-06,
      "loss": 0.0031,
      "step": 52640
    },
    {
      "epoch": 0.7443905611559615,
      "grad_norm": 0.0,
      "learning_rate": 2.5600531606554597e-06,
      "loss": 0.0,
      "step": 52650
    },
    {
      "epoch": 0.7445319458779285,
      "grad_norm": 0.0,
      "learning_rate": 2.55863931343579e-06,
      "loss": 0.0,
      "step": 52660
    },
    {
      "epoch": 0.7446733305998954,
      "grad_norm": 0.0,
      "learning_rate": 2.557225466216121e-06,
      "loss": 0.0,
      "step": 52670
    },
    {
      "epoch": 0.7448147153218623,
      "grad_norm": 0.0,
      "learning_rate": 2.5558116189964514e-06,
      "loss": 0.0029,
      "step": 52680
    },
    {
      "epoch": 0.7449561000438293,
      "grad_norm": 0.0,
      "learning_rate": 2.554397771776782e-06,
      "loss": 0.3379,
      "step": 52690
    },
    {
      "epoch": 0.7450974847657962,
      "grad_norm": 0.038947589695453644,
      "learning_rate": 2.5529839245571127e-06,
      "loss": 0.0,
      "step": 52700
    },
    {
      "epoch": 0.7452388694877632,
      "grad_norm": 0.0,
      "learning_rate": 2.551570077337443e-06,
      "loss": 0.0,
      "step": 52710
    },
    {
      "epoch": 0.7453802542097301,
      "grad_norm": 0.0,
      "learning_rate": 2.5501562301177736e-06,
      "loss": 0.0,
      "step": 52720
    },
    {
      "epoch": 0.7455216389316971,
      "grad_norm": 0.0,
      "learning_rate": 2.5487423828981044e-06,
      "loss": 0.0,
      "step": 52730
    },
    {
      "epoch": 0.745663023653664,
      "grad_norm": 0.0,
      "learning_rate": 2.547328535678435e-06,
      "loss": 0.0,
      "step": 52740
    },
    {
      "epoch": 0.7458044083756309,
      "grad_norm": 0.0,
      "learning_rate": 2.5459146884587653e-06,
      "loss": 0.0,
      "step": 52750
    },
    {
      "epoch": 0.7459457930975979,
      "grad_norm": 0.0,
      "learning_rate": 2.5445008412390957e-06,
      "loss": 0.0,
      "step": 52760
    },
    {
      "epoch": 0.7460871778195648,
      "grad_norm": 0.0,
      "learning_rate": 2.5430869940194266e-06,
      "loss": 0.0001,
      "step": 52770
    },
    {
      "epoch": 0.7462285625415318,
      "grad_norm": 1.541938786431274e-06,
      "learning_rate": 2.541673146799757e-06,
      "loss": 0.0006,
      "step": 52780
    },
    {
      "epoch": 0.7463699472634987,
      "grad_norm": 0.0,
      "learning_rate": 2.5402592995800875e-06,
      "loss": 0.0,
      "step": 52790
    },
    {
      "epoch": 0.7465113319854656,
      "grad_norm": 0.0,
      "learning_rate": 2.5388454523604183e-06,
      "loss": 0.0,
      "step": 52800
    },
    {
      "epoch": 0.7466527167074326,
      "grad_norm": 0.0,
      "learning_rate": 2.5374316051407487e-06,
      "loss": 0.0,
      "step": 52810
    },
    {
      "epoch": 0.7467941014293995,
      "grad_norm": 0.0,
      "learning_rate": 2.536017757921079e-06,
      "loss": 0.0,
      "step": 52820
    },
    {
      "epoch": 0.7469354861513665,
      "grad_norm": 0.0,
      "learning_rate": 2.5346039107014096e-06,
      "loss": 0.0,
      "step": 52830
    },
    {
      "epoch": 0.7470768708733334,
      "grad_norm": 0.0,
      "learning_rate": 2.53319006348174e-06,
      "loss": 0.0,
      "step": 52840
    },
    {
      "epoch": 0.7472182555953004,
      "grad_norm": 0.0,
      "learning_rate": 2.531776216262071e-06,
      "loss": 0.0,
      "step": 52850
    },
    {
      "epoch": 0.7473596403172673,
      "grad_norm": 0.0008564353338442743,
      "learning_rate": 2.5303623690424018e-06,
      "loss": 0.0,
      "step": 52860
    },
    {
      "epoch": 0.7475010250392342,
      "grad_norm": 0.0,
      "learning_rate": 2.528948521822732e-06,
      "loss": 0.0,
      "step": 52870
    },
    {
      "epoch": 0.7476424097612012,
      "grad_norm": 0.0,
      "learning_rate": 2.5275346746030626e-06,
      "loss": 0.0,
      "step": 52880
    },
    {
      "epoch": 0.7477837944831681,
      "grad_norm": 0.0,
      "learning_rate": 2.526120827383393e-06,
      "loss": 0.0,
      "step": 52890
    },
    {
      "epoch": 0.7479251792051351,
      "grad_norm": 0.0,
      "learning_rate": 2.5247069801637235e-06,
      "loss": 0.0,
      "step": 52900
    },
    {
      "epoch": 0.748066563927102,
      "grad_norm": 0.0,
      "learning_rate": 2.523293132944054e-06,
      "loss": 0.0,
      "step": 52910
    },
    {
      "epoch": 0.748207948649069,
      "grad_norm": 0.0,
      "learning_rate": 2.5218792857243844e-06,
      "loss": 0.0,
      "step": 52920
    },
    {
      "epoch": 0.7483493333710359,
      "grad_norm": 565.934814453125,
      "learning_rate": 2.5204654385047156e-06,
      "loss": 0.133,
      "step": 52930
    },
    {
      "epoch": 0.7484907180930028,
      "grad_norm": 0.0,
      "learning_rate": 2.519051591285046e-06,
      "loss": 0.0,
      "step": 52940
    },
    {
      "epoch": 0.7486321028149698,
      "grad_norm": 0.3470814526081085,
      "learning_rate": 2.5176377440653765e-06,
      "loss": 0.0018,
      "step": 52950
    },
    {
      "epoch": 0.7487734875369367,
      "grad_norm": 0.0,
      "learning_rate": 2.516223896845707e-06,
      "loss": 0.0,
      "step": 52960
    },
    {
      "epoch": 0.7489148722589037,
      "grad_norm": 0.0,
      "learning_rate": 2.5148100496260374e-06,
      "loss": 0.0,
      "step": 52970
    },
    {
      "epoch": 0.7490562569808706,
      "grad_norm": 0.0,
      "learning_rate": 2.513396202406368e-06,
      "loss": 0.0,
      "step": 52980
    },
    {
      "epoch": 0.7491976417028376,
      "grad_norm": 0.0,
      "learning_rate": 2.511982355186699e-06,
      "loss": 0.0,
      "step": 52990
    },
    {
      "epoch": 0.7493390264248045,
      "grad_norm": 0.0,
      "learning_rate": 2.5105685079670295e-06,
      "loss": 0.0,
      "step": 53000
    },
    {
      "epoch": 0.7494804111467714,
      "grad_norm": 0.0001281269360333681,
      "learning_rate": 2.50915466074736e-06,
      "loss": 0.0,
      "step": 53010
    },
    {
      "epoch": 0.7496217958687384,
      "grad_norm": 5.104813226353144e-06,
      "learning_rate": 2.5077408135276904e-06,
      "loss": 0.0,
      "step": 53020
    },
    {
      "epoch": 0.7497631805907053,
      "grad_norm": 1.238356350796721e-08,
      "learning_rate": 2.506326966308021e-06,
      "loss": 0.0,
      "step": 53030
    },
    {
      "epoch": 0.7499045653126724,
      "grad_norm": 0.0,
      "learning_rate": 2.5049131190883512e-06,
      "loss": 0.0,
      "step": 53040
    },
    {
      "epoch": 0.7500459500346393,
      "grad_norm": 0.0,
      "learning_rate": 2.5034992718686817e-06,
      "loss": 0.0,
      "step": 53050
    },
    {
      "epoch": 0.7501873347566062,
      "grad_norm": 0.0,
      "learning_rate": 2.502085424649013e-06,
      "loss": 0.0,
      "step": 53060
    },
    {
      "epoch": 0.7503287194785732,
      "grad_norm": 0.0,
      "learning_rate": 2.5006715774293434e-06,
      "loss": 0.0,
      "step": 53070
    },
    {
      "epoch": 0.7504701042005401,
      "grad_norm": 0.0,
      "learning_rate": 2.499257730209674e-06,
      "loss": 0.0001,
      "step": 53080
    },
    {
      "epoch": 0.7506114889225071,
      "grad_norm": 0.0,
      "learning_rate": 2.4978438829900043e-06,
      "loss": 0.0,
      "step": 53090
    },
    {
      "epoch": 0.750752873644474,
      "grad_norm": 4.6929694690334145e-06,
      "learning_rate": 2.496430035770335e-06,
      "loss": 0.0013,
      "step": 53100
    },
    {
      "epoch": 0.750894258366441,
      "grad_norm": 0.0,
      "learning_rate": 2.4950161885506655e-06,
      "loss": 0.0,
      "step": 53110
    },
    {
      "epoch": 0.7510356430884079,
      "grad_norm": 0.0,
      "learning_rate": 2.493602341330996e-06,
      "loss": 0.0,
      "step": 53120
    },
    {
      "epoch": 0.7511770278103748,
      "grad_norm": 0.0,
      "learning_rate": 2.4921884941113264e-06,
      "loss": 0.0016,
      "step": 53130
    },
    {
      "epoch": 0.7513184125323418,
      "grad_norm": 0.0,
      "learning_rate": 2.4907746468916573e-06,
      "loss": 0.0,
      "step": 53140
    },
    {
      "epoch": 0.7514597972543087,
      "grad_norm": 0.0,
      "learning_rate": 2.4893607996719877e-06,
      "loss": 0.0,
      "step": 53150
    },
    {
      "epoch": 0.7516011819762757,
      "grad_norm": 6.679215402982663e-07,
      "learning_rate": 2.487946952452318e-06,
      "loss": 0.0,
      "step": 53160
    },
    {
      "epoch": 0.7517425666982426,
      "grad_norm": 3.305868006009405e-07,
      "learning_rate": 2.486533105232649e-06,
      "loss": 0.0001,
      "step": 53170
    },
    {
      "epoch": 0.7518839514202096,
      "grad_norm": 0.0,
      "learning_rate": 2.4851192580129794e-06,
      "loss": 0.0,
      "step": 53180
    },
    {
      "epoch": 0.7520253361421765,
      "grad_norm": 0.0,
      "learning_rate": 2.48370541079331e-06,
      "loss": 0.0008,
      "step": 53190
    },
    {
      "epoch": 0.7521667208641434,
      "grad_norm": 6.432022559010875e-08,
      "learning_rate": 2.4822915635736407e-06,
      "loss": 0.0,
      "step": 53200
    },
    {
      "epoch": 0.7523081055861104,
      "grad_norm": 0.0,
      "learning_rate": 2.480877716353971e-06,
      "loss": 0.0,
      "step": 53210
    },
    {
      "epoch": 0.7524494903080773,
      "grad_norm": 0.02098793536424637,
      "learning_rate": 2.4794638691343016e-06,
      "loss": 0.0004,
      "step": 53220
    },
    {
      "epoch": 0.7525908750300443,
      "grad_norm": 0.0,
      "learning_rate": 2.4780500219146324e-06,
      "loss": 0.0,
      "step": 53230
    },
    {
      "epoch": 0.7527322597520112,
      "grad_norm": 0.0,
      "learning_rate": 2.476636174694963e-06,
      "loss": 0.0153,
      "step": 53240
    },
    {
      "epoch": 0.7528736444739781,
      "grad_norm": 0.0,
      "learning_rate": 2.4752223274752933e-06,
      "loss": 0.0,
      "step": 53250
    },
    {
      "epoch": 0.7530150291959451,
      "grad_norm": 0.0,
      "learning_rate": 2.4738084802556237e-06,
      "loss": 0.0,
      "step": 53260
    },
    {
      "epoch": 0.753156413917912,
      "grad_norm": 0.0,
      "learning_rate": 2.4723946330359546e-06,
      "loss": 0.0098,
      "step": 53270
    },
    {
      "epoch": 0.753297798639879,
      "grad_norm": 0.0,
      "learning_rate": 2.470980785816285e-06,
      "loss": 0.0,
      "step": 53280
    },
    {
      "epoch": 0.7534391833618459,
      "grad_norm": 0.0,
      "learning_rate": 2.4697083233185823e-06,
      "loss": 0.5848,
      "step": 53290
    },
    {
      "epoch": 0.7535805680838129,
      "grad_norm": 0.0,
      "learning_rate": 2.468294476098913e-06,
      "loss": 0.0,
      "step": 53300
    },
    {
      "epoch": 0.7537219528057798,
      "grad_norm": 0.001146164140664041,
      "learning_rate": 2.4668806288792436e-06,
      "loss": 0.1224,
      "step": 53310
    },
    {
      "epoch": 0.7538633375277467,
      "grad_norm": 0.0,
      "learning_rate": 2.465466781659574e-06,
      "loss": 0.0,
      "step": 53320
    },
    {
      "epoch": 0.7540047222497137,
      "grad_norm": 0.0,
      "learning_rate": 2.464052934439905e-06,
      "loss": 0.0,
      "step": 53330
    },
    {
      "epoch": 0.7541461069716806,
      "grad_norm": 1.1910408659332461e-07,
      "learning_rate": 2.4626390872202353e-06,
      "loss": 0.0,
      "step": 53340
    },
    {
      "epoch": 0.7542874916936476,
      "grad_norm": 0.0007299212738871574,
      "learning_rate": 2.4612252400005657e-06,
      "loss": 0.0,
      "step": 53350
    },
    {
      "epoch": 0.7544288764156145,
      "grad_norm": 0.0,
      "learning_rate": 2.4598113927808966e-06,
      "loss": 0.0,
      "step": 53360
    },
    {
      "epoch": 0.7545702611375815,
      "grad_norm": 0.0,
      "learning_rate": 2.458397545561227e-06,
      "loss": 0.0,
      "step": 53370
    },
    {
      "epoch": 0.7547116458595484,
      "grad_norm": 0.0,
      "learning_rate": 2.4569836983415574e-06,
      "loss": 0.0,
      "step": 53380
    },
    {
      "epoch": 0.7548530305815153,
      "grad_norm": 0.0,
      "learning_rate": 2.455569851121888e-06,
      "loss": 0.0,
      "step": 53390
    },
    {
      "epoch": 0.7549944153034823,
      "grad_norm": 0.0,
      "learning_rate": 2.4541560039022187e-06,
      "loss": 0.0,
      "step": 53400
    },
    {
      "epoch": 0.7551358000254492,
      "grad_norm": 0.0,
      "learning_rate": 2.452742156682549e-06,
      "loss": 0.0,
      "step": 53410
    },
    {
      "epoch": 0.7552771847474162,
      "grad_norm": 0.0,
      "learning_rate": 2.4513283094628796e-06,
      "loss": 0.0,
      "step": 53420
    },
    {
      "epoch": 0.7554185694693831,
      "grad_norm": 0.0,
      "learning_rate": 2.4499144622432105e-06,
      "loss": 0.0,
      "step": 53430
    },
    {
      "epoch": 0.7555599541913501,
      "grad_norm": 0.0,
      "learning_rate": 2.448500615023541e-06,
      "loss": 0.0,
      "step": 53440
    },
    {
      "epoch": 0.755701338913317,
      "grad_norm": 0.0,
      "learning_rate": 2.4470867678038713e-06,
      "loss": 0.0,
      "step": 53450
    },
    {
      "epoch": 0.7558427236352839,
      "grad_norm": 0.0,
      "learning_rate": 2.4456729205842018e-06,
      "loss": 0.0,
      "step": 53460
    },
    {
      "epoch": 0.7559841083572509,
      "grad_norm": 0.0,
      "learning_rate": 2.4442590733645326e-06,
      "loss": 0.0001,
      "step": 53470
    },
    {
      "epoch": 0.7561254930792178,
      "grad_norm": 0.00013888670946471393,
      "learning_rate": 2.442845226144863e-06,
      "loss": 0.0,
      "step": 53480
    },
    {
      "epoch": 0.7562668778011848,
      "grad_norm": 0.0,
      "learning_rate": 2.4414313789251935e-06,
      "loss": 0.0,
      "step": 53490
    },
    {
      "epoch": 0.7564082625231517,
      "grad_norm": 0.0,
      "learning_rate": 2.4400175317055243e-06,
      "loss": 0.0001,
      "step": 53500
    },
    {
      "epoch": 0.7565496472451186,
      "grad_norm": 0.0,
      "learning_rate": 2.4386036844858548e-06,
      "loss": 0.0,
      "step": 53510
    },
    {
      "epoch": 0.7566910319670856,
      "grad_norm": 0.0,
      "learning_rate": 2.437189837266185e-06,
      "loss": 0.0,
      "step": 53520
    },
    {
      "epoch": 0.7568324166890525,
      "grad_norm": 0.0,
      "learning_rate": 2.4357759900465156e-06,
      "loss": 0.0,
      "step": 53530
    },
    {
      "epoch": 0.7569738014110196,
      "grad_norm": 0.0,
      "learning_rate": 2.4343621428268465e-06,
      "loss": 0.0002,
      "step": 53540
    },
    {
      "epoch": 0.7571151861329865,
      "grad_norm": 0.0,
      "learning_rate": 2.432948295607177e-06,
      "loss": 0.0,
      "step": 53550
    },
    {
      "epoch": 0.7572565708549535,
      "grad_norm": 0.0,
      "learning_rate": 2.4315344483875074e-06,
      "loss": 0.0,
      "step": 53560
    },
    {
      "epoch": 0.7573979555769204,
      "grad_norm": 0.0,
      "learning_rate": 2.430120601167838e-06,
      "loss": 0.0,
      "step": 53570
    },
    {
      "epoch": 0.7575393402988873,
      "grad_norm": 0.0,
      "learning_rate": 2.4287067539481686e-06,
      "loss": 0.0,
      "step": 53580
    },
    {
      "epoch": 0.7576807250208543,
      "grad_norm": 0.0,
      "learning_rate": 2.427292906728499e-06,
      "loss": 0.0,
      "step": 53590
    },
    {
      "epoch": 0.7578221097428212,
      "grad_norm": 2.086173844872974e-05,
      "learning_rate": 2.4258790595088295e-06,
      "loss": 0.0,
      "step": 53600
    },
    {
      "epoch": 0.7579634944647882,
      "grad_norm": 0.0,
      "learning_rate": 2.42446521228916e-06,
      "loss": 0.0021,
      "step": 53610
    },
    {
      "epoch": 0.7581048791867551,
      "grad_norm": 0.0,
      "learning_rate": 2.423051365069491e-06,
      "loss": 0.0,
      "step": 53620
    },
    {
      "epoch": 0.7582462639087221,
      "grad_norm": 0.34558430314064026,
      "learning_rate": 2.4216375178498212e-06,
      "loss": 0.0,
      "step": 53630
    },
    {
      "epoch": 0.758387648630689,
      "grad_norm": 0.051287852227687836,
      "learning_rate": 2.4202236706301517e-06,
      "loss": 0.0003,
      "step": 53640
    },
    {
      "epoch": 0.7585290333526559,
      "grad_norm": 4.314307716413168e-06,
      "learning_rate": 2.4188098234104825e-06,
      "loss": 0.0242,
      "step": 53650
    },
    {
      "epoch": 0.7586704180746229,
      "grad_norm": 2.150144100189209,
      "learning_rate": 2.417395976190813e-06,
      "loss": 0.0003,
      "step": 53660
    },
    {
      "epoch": 0.7588118027965898,
      "grad_norm": 0.0,
      "learning_rate": 2.4159821289711434e-06,
      "loss": 0.0,
      "step": 53670
    },
    {
      "epoch": 0.7589531875185568,
      "grad_norm": 0.0,
      "learning_rate": 2.414568281751474e-06,
      "loss": 0.0,
      "step": 53680
    },
    {
      "epoch": 0.7590945722405237,
      "grad_norm": 0.0,
      "learning_rate": 2.4131544345318047e-06,
      "loss": 0.0,
      "step": 53690
    },
    {
      "epoch": 0.7592359569624907,
      "grad_norm": 0.0,
      "learning_rate": 2.411740587312135e-06,
      "loss": 0.0,
      "step": 53700
    },
    {
      "epoch": 0.7593773416844576,
      "grad_norm": 0.0,
      "learning_rate": 2.4103267400924655e-06,
      "loss": 0.0,
      "step": 53710
    },
    {
      "epoch": 0.7595187264064245,
      "grad_norm": 1.747914240013415e-07,
      "learning_rate": 2.4089128928727964e-06,
      "loss": 0.0,
      "step": 53720
    },
    {
      "epoch": 0.7596601111283915,
      "grad_norm": 0.0,
      "learning_rate": 2.407499045653127e-06,
      "loss": 0.1063,
      "step": 53730
    },
    {
      "epoch": 0.7598014958503584,
      "grad_norm": 0.0,
      "learning_rate": 2.4060851984334573e-06,
      "loss": 0.0,
      "step": 53740
    },
    {
      "epoch": 0.7599428805723254,
      "grad_norm": 0.0005665539065375924,
      "learning_rate": 2.404671351213788e-06,
      "loss": 0.0008,
      "step": 53750
    },
    {
      "epoch": 0.7600842652942923,
      "grad_norm": 0.0,
      "learning_rate": 2.4032575039941186e-06,
      "loss": 0.0,
      "step": 53760
    },
    {
      "epoch": 0.7602256500162592,
      "grad_norm": 0.0,
      "learning_rate": 2.401843656774449e-06,
      "loss": 0.0,
      "step": 53770
    },
    {
      "epoch": 0.7603670347382262,
      "grad_norm": 9.808410750267171e-10,
      "learning_rate": 2.40042980955478e-06,
      "loss": 0.0,
      "step": 53780
    },
    {
      "epoch": 0.7605084194601931,
      "grad_norm": 0.00018842350982595235,
      "learning_rate": 2.3990159623351103e-06,
      "loss": 0.0,
      "step": 53790
    },
    {
      "epoch": 0.7606498041821601,
      "grad_norm": 0.0,
      "learning_rate": 2.3976021151154407e-06,
      "loss": 0.0801,
      "step": 53800
    },
    {
      "epoch": 0.760791188904127,
      "grad_norm": 3.1705691071692854e-05,
      "learning_rate": 2.396188267895771e-06,
      "loss": 0.004,
      "step": 53810
    },
    {
      "epoch": 0.760932573626094,
      "grad_norm": 0.0,
      "learning_rate": 2.394774420676102e-06,
      "loss": 0.0,
      "step": 53820
    },
    {
      "epoch": 0.7610739583480609,
      "grad_norm": 0.0,
      "learning_rate": 2.3933605734564324e-06,
      "loss": 0.0,
      "step": 53830
    },
    {
      "epoch": 0.7612153430700278,
      "grad_norm": 0.0,
      "learning_rate": 2.391946726236763e-06,
      "loss": 0.0,
      "step": 53840
    },
    {
      "epoch": 0.7613567277919948,
      "grad_norm": 0.0,
      "learning_rate": 2.3905328790170937e-06,
      "loss": 0.0,
      "step": 53850
    },
    {
      "epoch": 0.7614981125139617,
      "grad_norm": 0.0,
      "learning_rate": 2.389119031797424e-06,
      "loss": 0.0,
      "step": 53860
    },
    {
      "epoch": 0.7616394972359287,
      "grad_norm": 0.00020832645532209426,
      "learning_rate": 2.3877051845777546e-06,
      "loss": 0.2738,
      "step": 53870
    },
    {
      "epoch": 0.7617808819578956,
      "grad_norm": 0.0,
      "learning_rate": 2.3862913373580854e-06,
      "loss": 0.0036,
      "step": 53880
    },
    {
      "epoch": 0.7619222666798626,
      "grad_norm": 1.852160202986397e-08,
      "learning_rate": 2.384877490138416e-06,
      "loss": 0.0313,
      "step": 53890
    },
    {
      "epoch": 0.7620636514018295,
      "grad_norm": 0.0,
      "learning_rate": 2.3834636429187463e-06,
      "loss": 0.0,
      "step": 53900
    },
    {
      "epoch": 0.7622050361237964,
      "grad_norm": 0.0,
      "learning_rate": 2.382049795699077e-06,
      "loss": 0.0001,
      "step": 53910
    },
    {
      "epoch": 0.7623464208457634,
      "grad_norm": 0.0,
      "learning_rate": 2.3806359484794076e-06,
      "loss": 0.0,
      "step": 53920
    },
    {
      "epoch": 0.7624878055677303,
      "grad_norm": 1.932284021677333e-07,
      "learning_rate": 2.379222101259738e-06,
      "loss": 0.165,
      "step": 53930
    },
    {
      "epoch": 0.7626291902896973,
      "grad_norm": 0.0,
      "learning_rate": 2.3778082540400685e-06,
      "loss": 0.0408,
      "step": 53940
    },
    {
      "epoch": 0.7627705750116642,
      "grad_norm": 0.0,
      "learning_rate": 2.3763944068203993e-06,
      "loss": 0.0,
      "step": 53950
    },
    {
      "epoch": 0.7629119597336311,
      "grad_norm": 9.411920132151863e-10,
      "learning_rate": 2.3749805596007298e-06,
      "loss": 0.0,
      "step": 53960
    },
    {
      "epoch": 0.7630533444555981,
      "grad_norm": 0.0,
      "learning_rate": 2.37356671238106e-06,
      "loss": 0.0,
      "step": 53970
    },
    {
      "epoch": 0.763194729177565,
      "grad_norm": 4.293757793760733e-09,
      "learning_rate": 2.372152865161391e-06,
      "loss": 0.0213,
      "step": 53980
    },
    {
      "epoch": 0.763336113899532,
      "grad_norm": 0.0,
      "learning_rate": 2.3707390179417215e-06,
      "loss": 0.0,
      "step": 53990
    },
    {
      "epoch": 0.7634774986214989,
      "grad_norm": 2.3010386485111667e-06,
      "learning_rate": 2.369325170722052e-06,
      "loss": 0.0009,
      "step": 54000
    },
    {
      "epoch": 0.763618883343466,
      "grad_norm": 0.0,
      "learning_rate": 2.3679113235023828e-06,
      "loss": 0.0004,
      "step": 54010
    },
    {
      "epoch": 0.7637602680654328,
      "grad_norm": 0.0,
      "learning_rate": 2.366497476282713e-06,
      "loss": 0.0,
      "step": 54020
    },
    {
      "epoch": 0.7639016527873997,
      "grad_norm": 0.0,
      "learning_rate": 2.3650836290630436e-06,
      "loss": 0.0002,
      "step": 54030
    },
    {
      "epoch": 0.7640430375093668,
      "grad_norm": 510.6013488769531,
      "learning_rate": 2.3636697818433745e-06,
      "loss": 0.8125,
      "step": 54040
    },
    {
      "epoch": 0.7641844222313336,
      "grad_norm": 0.0,
      "learning_rate": 2.362255934623705e-06,
      "loss": 0.0001,
      "step": 54050
    },
    {
      "epoch": 0.7643258069533007,
      "grad_norm": 0.0,
      "learning_rate": 2.3608420874040354e-06,
      "loss": 0.0,
      "step": 54060
    },
    {
      "epoch": 0.7644671916752676,
      "grad_norm": 4.7519140935037285e-05,
      "learning_rate": 2.359428240184366e-06,
      "loss": 0.0,
      "step": 54070
    },
    {
      "epoch": 0.7646085763972346,
      "grad_norm": 1.305250589211937e-05,
      "learning_rate": 2.3580143929646967e-06,
      "loss": 0.0,
      "step": 54080
    },
    {
      "epoch": 0.7647499611192015,
      "grad_norm": 0.0040323021821677685,
      "learning_rate": 2.356600545745027e-06,
      "loss": 0.0,
      "step": 54090
    },
    {
      "epoch": 0.7648913458411684,
      "grad_norm": 0.0028413545805960894,
      "learning_rate": 2.3551866985253575e-06,
      "loss": 0.0,
      "step": 54100
    },
    {
      "epoch": 0.7650327305631354,
      "grad_norm": 0.00034469374804757535,
      "learning_rate": 2.3537728513056884e-06,
      "loss": 0.0,
      "step": 54110
    },
    {
      "epoch": 0.7651741152851023,
      "grad_norm": 0.0,
      "learning_rate": 2.352359004086019e-06,
      "loss": 0.0,
      "step": 54120
    },
    {
      "epoch": 0.7653155000070693,
      "grad_norm": 2.283090225319029e-06,
      "learning_rate": 2.3509451568663492e-06,
      "loss": 0.0,
      "step": 54130
    },
    {
      "epoch": 0.7654568847290362,
      "grad_norm": 0.0,
      "learning_rate": 2.3495313096466797e-06,
      "loss": 0.0,
      "step": 54140
    },
    {
      "epoch": 0.7655982694510032,
      "grad_norm": 0.0,
      "learning_rate": 2.3481174624270105e-06,
      "loss": 0.0002,
      "step": 54150
    },
    {
      "epoch": 0.7657396541729701,
      "grad_norm": 0.0,
      "learning_rate": 2.346703615207341e-06,
      "loss": 0.0,
      "step": 54160
    },
    {
      "epoch": 0.765881038894937,
      "grad_norm": 0.0,
      "learning_rate": 2.3452897679876714e-06,
      "loss": 0.0002,
      "step": 54170
    },
    {
      "epoch": 0.766022423616904,
      "grad_norm": 0.0,
      "learning_rate": 2.343875920768002e-06,
      "loss": 0.0012,
      "step": 54180
    },
    {
      "epoch": 0.7661638083388709,
      "grad_norm": 1.8134149312973022,
      "learning_rate": 2.3424620735483327e-06,
      "loss": 0.0004,
      "step": 54190
    },
    {
      "epoch": 0.7663051930608379,
      "grad_norm": 0.0,
      "learning_rate": 2.341048226328663e-06,
      "loss": 0.0,
      "step": 54200
    },
    {
      "epoch": 0.7664465777828048,
      "grad_norm": 0.0,
      "learning_rate": 2.3396343791089935e-06,
      "loss": 0.0,
      "step": 54210
    },
    {
      "epoch": 0.7665879625047717,
      "grad_norm": 0.0,
      "learning_rate": 2.338220531889324e-06,
      "loss": 0.0001,
      "step": 54220
    },
    {
      "epoch": 0.7667293472267387,
      "grad_norm": 0.0,
      "learning_rate": 2.336806684669655e-06,
      "loss": 0.0,
      "step": 54230
    },
    {
      "epoch": 0.7668707319487056,
      "grad_norm": 0.00011328425171086565,
      "learning_rate": 2.3353928374499853e-06,
      "loss": 0.0001,
      "step": 54240
    },
    {
      "epoch": 0.7670121166706726,
      "grad_norm": 0.0,
      "learning_rate": 2.3339789902303157e-06,
      "loss": 0.0,
      "step": 54250
    },
    {
      "epoch": 0.7671535013926395,
      "grad_norm": 0.0,
      "learning_rate": 2.332565143010646e-06,
      "loss": 0.0,
      "step": 54260
    },
    {
      "epoch": 0.7672948861146065,
      "grad_norm": 2.034648241533432e-05,
      "learning_rate": 2.331151295790977e-06,
      "loss": 0.0,
      "step": 54270
    },
    {
      "epoch": 0.7674362708365734,
      "grad_norm": 0.0,
      "learning_rate": 2.3297374485713074e-06,
      "loss": 0.0,
      "step": 54280
    },
    {
      "epoch": 0.7675776555585403,
      "grad_norm": 0.0,
      "learning_rate": 2.328323601351638e-06,
      "loss": 0.0,
      "step": 54290
    },
    {
      "epoch": 0.7677190402805073,
      "grad_norm": 0.03839162737131119,
      "learning_rate": 2.3269097541319687e-06,
      "loss": 0.0,
      "step": 54300
    },
    {
      "epoch": 0.7678604250024742,
      "grad_norm": 0.0,
      "learning_rate": 2.325495906912299e-06,
      "loss": 0.0,
      "step": 54310
    },
    {
      "epoch": 0.7680018097244412,
      "grad_norm": 7.371943411271786e-06,
      "learning_rate": 2.3240820596926296e-06,
      "loss": 0.0,
      "step": 54320
    },
    {
      "epoch": 0.7681431944464081,
      "grad_norm": 0.0,
      "learning_rate": 2.3226682124729604e-06,
      "loss": 0.0,
      "step": 54330
    },
    {
      "epoch": 0.7682845791683751,
      "grad_norm": 0.0,
      "learning_rate": 2.321254365253291e-06,
      "loss": 0.0,
      "step": 54340
    },
    {
      "epoch": 0.768425963890342,
      "grad_norm": 0.0,
      "learning_rate": 2.3198405180336213e-06,
      "loss": 0.0,
      "step": 54350
    },
    {
      "epoch": 0.7685673486123089,
      "grad_norm": 2.598538451081822e-08,
      "learning_rate": 2.3184266708139517e-06,
      "loss": 0.0,
      "step": 54360
    },
    {
      "epoch": 0.7687087333342759,
      "grad_norm": 1.6107043027877808,
      "learning_rate": 2.3170128235942826e-06,
      "loss": 0.0004,
      "step": 54370
    },
    {
      "epoch": 0.7688501180562428,
      "grad_norm": 0.0,
      "learning_rate": 2.315598976374613e-06,
      "loss": 0.0,
      "step": 54380
    },
    {
      "epoch": 0.7689915027782098,
      "grad_norm": 0.0,
      "learning_rate": 2.3141851291549435e-06,
      "loss": 0.0,
      "step": 54390
    },
    {
      "epoch": 0.7691328875001767,
      "grad_norm": 0.0,
      "learning_rate": 2.3127712819352743e-06,
      "loss": 0.0,
      "step": 54400
    },
    {
      "epoch": 0.7692742722221437,
      "grad_norm": 0.0,
      "learning_rate": 2.3113574347156047e-06,
      "loss": 0.0,
      "step": 54410
    },
    {
      "epoch": 0.7694156569441106,
      "grad_norm": 0.0,
      "learning_rate": 2.309943587495935e-06,
      "loss": 0.0,
      "step": 54420
    },
    {
      "epoch": 0.7695570416660775,
      "grad_norm": 0.0,
      "learning_rate": 2.308529740276266e-06,
      "loss": 0.0,
      "step": 54430
    },
    {
      "epoch": 0.7696984263880445,
      "grad_norm": 0.0,
      "learning_rate": 2.3071158930565965e-06,
      "loss": 0.0,
      "step": 54440
    },
    {
      "epoch": 0.7698398111100114,
      "grad_norm": 0.0,
      "learning_rate": 2.305702045836927e-06,
      "loss": 0.0008,
      "step": 54450
    },
    {
      "epoch": 0.7699811958319784,
      "grad_norm": 0.0,
      "learning_rate": 2.3042881986172578e-06,
      "loss": 0.0,
      "step": 54460
    },
    {
      "epoch": 0.7701225805539453,
      "grad_norm": 0.0,
      "learning_rate": 2.302874351397588e-06,
      "loss": 0.0,
      "step": 54470
    },
    {
      "epoch": 0.7702639652759122,
      "grad_norm": 0.0,
      "learning_rate": 2.3014605041779186e-06,
      "loss": 0.0,
      "step": 54480
    },
    {
      "epoch": 0.7704053499978792,
      "grad_norm": 0.0,
      "learning_rate": 2.3000466569582495e-06,
      "loss": 0.0,
      "step": 54490
    },
    {
      "epoch": 0.7705467347198461,
      "grad_norm": 8.515931767760776e-06,
      "learning_rate": 2.29863280973858e-06,
      "loss": 0.0,
      "step": 54500
    },
    {
      "epoch": 0.7706881194418131,
      "grad_norm": 0.0,
      "learning_rate": 2.2972189625189103e-06,
      "loss": 0.0,
      "step": 54510
    },
    {
      "epoch": 0.77082950416378,
      "grad_norm": 0.0,
      "learning_rate": 2.2958051152992408e-06,
      "loss": 0.0,
      "step": 54520
    },
    {
      "epoch": 0.770970888885747,
      "grad_norm": 6.758104518667096e-06,
      "learning_rate": 2.2943912680795716e-06,
      "loss": 0.0002,
      "step": 54530
    },
    {
      "epoch": 0.771112273607714,
      "grad_norm": 0.0,
      "learning_rate": 2.292977420859902e-06,
      "loss": 0.0,
      "step": 54540
    },
    {
      "epoch": 0.7712536583296808,
      "grad_norm": 0.0,
      "learning_rate": 2.2915635736402325e-06,
      "loss": 0.0,
      "step": 54550
    },
    {
      "epoch": 0.7713950430516479,
      "grad_norm": 3.4585002595122205e-06,
      "learning_rate": 2.2901497264205634e-06,
      "loss": 0.0021,
      "step": 54560
    },
    {
      "epoch": 0.7715364277736148,
      "grad_norm": 1.3298300473252311e-05,
      "learning_rate": 2.288735879200894e-06,
      "loss": 0.0006,
      "step": 54570
    },
    {
      "epoch": 0.7716778124955818,
      "grad_norm": 0.0,
      "learning_rate": 2.2873220319812242e-06,
      "loss": 0.0,
      "step": 54580
    },
    {
      "epoch": 0.7718191972175487,
      "grad_norm": 0.0,
      "learning_rate": 2.285908184761555e-06,
      "loss": 0.0,
      "step": 54590
    },
    {
      "epoch": 0.7719605819395157,
      "grad_norm": 0.0,
      "learning_rate": 2.2844943375418855e-06,
      "loss": 0.0,
      "step": 54600
    },
    {
      "epoch": 0.7721019666614826,
      "grad_norm": 0.0,
      "learning_rate": 2.283080490322216e-06,
      "loss": 0.0,
      "step": 54610
    },
    {
      "epoch": 0.7722433513834495,
      "grad_norm": 0.0,
      "learning_rate": 2.281666643102547e-06,
      "loss": 0.0,
      "step": 54620
    },
    {
      "epoch": 0.7723847361054165,
      "grad_norm": 0.0,
      "learning_rate": 2.2802527958828772e-06,
      "loss": 0.0,
      "step": 54630
    },
    {
      "epoch": 0.7725261208273834,
      "grad_norm": 0.0,
      "learning_rate": 2.2788389486632077e-06,
      "loss": 0.0,
      "step": 54640
    },
    {
      "epoch": 0.7726675055493504,
      "grad_norm": 0.0,
      "learning_rate": 2.277425101443538e-06,
      "loss": 0.0293,
      "step": 54650
    },
    {
      "epoch": 0.7728088902713173,
      "grad_norm": 0.0,
      "learning_rate": 2.276011254223869e-06,
      "loss": 0.3514,
      "step": 54660
    },
    {
      "epoch": 0.7729502749932843,
      "grad_norm": 0.0,
      "learning_rate": 2.2745974070041994e-06,
      "loss": 0.0,
      "step": 54670
    },
    {
      "epoch": 0.7730916597152512,
      "grad_norm": 1.1903634522525408e-08,
      "learning_rate": 2.27318355978453e-06,
      "loss": 0.0,
      "step": 54680
    },
    {
      "epoch": 0.7732330444372181,
      "grad_norm": 8.82906903143521e-08,
      "learning_rate": 2.2717697125648607e-06,
      "loss": 0.0,
      "step": 54690
    },
    {
      "epoch": 0.7733744291591851,
      "grad_norm": 0.0,
      "learning_rate": 2.270355865345191e-06,
      "loss": 0.0,
      "step": 54700
    },
    {
      "epoch": 0.773515813881152,
      "grad_norm": 0.0,
      "learning_rate": 2.2689420181255215e-06,
      "loss": 0.0,
      "step": 54710
    },
    {
      "epoch": 0.773657198603119,
      "grad_norm": 0.0,
      "learning_rate": 2.2675281709058524e-06,
      "loss": 0.0,
      "step": 54720
    },
    {
      "epoch": 0.7737985833250859,
      "grad_norm": 0.0,
      "learning_rate": 2.266114323686183e-06,
      "loss": 0.0005,
      "step": 54730
    },
    {
      "epoch": 0.7739399680470528,
      "grad_norm": 0.0,
      "learning_rate": 2.2647004764665133e-06,
      "loss": 0.0,
      "step": 54740
    },
    {
      "epoch": 0.7740813527690198,
      "grad_norm": 0.0,
      "learning_rate": 2.2632866292468437e-06,
      "loss": 0.0,
      "step": 54750
    },
    {
      "epoch": 0.7742227374909867,
      "grad_norm": 6.59807231073728e-09,
      "learning_rate": 2.2618727820271746e-06,
      "loss": 0.0,
      "step": 54760
    },
    {
      "epoch": 0.7743641222129537,
      "grad_norm": 0.0,
      "learning_rate": 2.260458934807505e-06,
      "loss": 0.0,
      "step": 54770
    },
    {
      "epoch": 0.7745055069349206,
      "grad_norm": 0.0,
      "learning_rate": 2.2590450875878354e-06,
      "loss": 0.0,
      "step": 54780
    },
    {
      "epoch": 0.7746468916568876,
      "grad_norm": 0.0,
      "learning_rate": 2.257631240368166e-06,
      "loss": 0.0,
      "step": 54790
    },
    {
      "epoch": 0.7747882763788545,
      "grad_norm": 281.3460693359375,
      "learning_rate": 2.2562173931484967e-06,
      "loss": 0.0835,
      "step": 54800
    },
    {
      "epoch": 0.7749296611008214,
      "grad_norm": NaN,
      "learning_rate": 2.254944930650794e-06,
      "loss": 0.1753,
      "step": 54810
    },
    {
      "epoch": 0.7750710458227884,
      "grad_norm": 0.0,
      "learning_rate": 2.253531083431125e-06,
      "loss": 0.0,
      "step": 54820
    },
    {
      "epoch": 0.7752124305447553,
      "grad_norm": 0.0,
      "learning_rate": 2.2521172362114553e-06,
      "loss": 0.0576,
      "step": 54830
    },
    {
      "epoch": 0.7753538152667223,
      "grad_norm": 0.0,
      "learning_rate": 2.2507033889917857e-06,
      "loss": 0.0,
      "step": 54840
    },
    {
      "epoch": 0.7754951999886892,
      "grad_norm": 0.0,
      "learning_rate": 2.2492895417721166e-06,
      "loss": 0.0005,
      "step": 54850
    },
    {
      "epoch": 0.7756365847106562,
      "grad_norm": 0.0,
      "learning_rate": 2.247875694552447e-06,
      "loss": 0.0,
      "step": 54860
    },
    {
      "epoch": 0.7757779694326231,
      "grad_norm": 0.0,
      "learning_rate": 2.2464618473327774e-06,
      "loss": 0.0,
      "step": 54870
    },
    {
      "epoch": 0.77591935415459,
      "grad_norm": 0.0,
      "learning_rate": 2.2450480001131083e-06,
      "loss": 0.0,
      "step": 54880
    },
    {
      "epoch": 0.776060738876557,
      "grad_norm": 0.0,
      "learning_rate": 2.2436341528934387e-06,
      "loss": 0.0,
      "step": 54890
    },
    {
      "epoch": 0.7762021235985239,
      "grad_norm": 0.0,
      "learning_rate": 2.242220305673769e-06,
      "loss": 0.0,
      "step": 54900
    },
    {
      "epoch": 0.7763435083204909,
      "grad_norm": 0.0,
      "learning_rate": 2.2408064584540996e-06,
      "loss": 0.0,
      "step": 54910
    },
    {
      "epoch": 0.7764848930424578,
      "grad_norm": 0.0,
      "learning_rate": 2.2393926112344304e-06,
      "loss": 0.0463,
      "step": 54920
    },
    {
      "epoch": 0.7766262777644247,
      "grad_norm": 1.7977863535634242e-05,
      "learning_rate": 2.237978764014761e-06,
      "loss": 0.0008,
      "step": 54930
    },
    {
      "epoch": 0.7767676624863917,
      "grad_norm": 0.0,
      "learning_rate": 2.2365649167950913e-06,
      "loss": 0.0,
      "step": 54940
    },
    {
      "epoch": 0.7769090472083586,
      "grad_norm": 0.0,
      "learning_rate": 2.2351510695754217e-06,
      "loss": 0.0,
      "step": 54950
    },
    {
      "epoch": 0.7770504319303256,
      "grad_norm": 0.8882701992988586,
      "learning_rate": 2.2337372223557526e-06,
      "loss": 0.0006,
      "step": 54960
    },
    {
      "epoch": 0.7771918166522925,
      "grad_norm": 8.781510629773948e-09,
      "learning_rate": 2.232323375136083e-06,
      "loss": 0.0,
      "step": 54970
    },
    {
      "epoch": 0.7773332013742595,
      "grad_norm": 0.0,
      "learning_rate": 2.2309095279164135e-06,
      "loss": 0.0003,
      "step": 54980
    },
    {
      "epoch": 0.7774745860962264,
      "grad_norm": 0.0,
      "learning_rate": 2.2294956806967443e-06,
      "loss": 0.0,
      "step": 54990
    },
    {
      "epoch": 0.7776159708181933,
      "grad_norm": 0.0,
      "learning_rate": 2.2280818334770747e-06,
      "loss": 0.0005,
      "step": 55000
    },
    {
      "epoch": 0.7777573555401603,
      "grad_norm": 0.0,
      "learning_rate": 2.226667986257405e-06,
      "loss": 0.0,
      "step": 55010
    },
    {
      "epoch": 0.7778987402621272,
      "grad_norm": 0.0,
      "learning_rate": 2.2252541390377356e-06,
      "loss": 0.0,
      "step": 55020
    },
    {
      "epoch": 0.7780401249840943,
      "grad_norm": 0.00035243810270912945,
      "learning_rate": 2.2238402918180665e-06,
      "loss": 0.0,
      "step": 55030
    },
    {
      "epoch": 0.7781815097060611,
      "grad_norm": 0.10863689333200455,
      "learning_rate": 2.222426444598397e-06,
      "loss": 0.0,
      "step": 55040
    },
    {
      "epoch": 0.7783228944280282,
      "grad_norm": 0.0,
      "learning_rate": 2.2210125973787273e-06,
      "loss": 0.0,
      "step": 55050
    },
    {
      "epoch": 0.778464279149995,
      "grad_norm": 0.0,
      "learning_rate": 2.2195987501590578e-06,
      "loss": 0.0009,
      "step": 55060
    },
    {
      "epoch": 0.778605663871962,
      "grad_norm": 0.0,
      "learning_rate": 2.2181849029393886e-06,
      "loss": 0.0,
      "step": 55070
    },
    {
      "epoch": 0.778747048593929,
      "grad_norm": 1.8399599621687912e-09,
      "learning_rate": 2.216771055719719e-06,
      "loss": 0.0,
      "step": 55080
    },
    {
      "epoch": 0.7788884333158959,
      "grad_norm": 0.0,
      "learning_rate": 2.2153572085000495e-06,
      "loss": 0.0,
      "step": 55090
    },
    {
      "epoch": 0.7790298180378629,
      "grad_norm": 0.0,
      "learning_rate": 2.21394336128038e-06,
      "loss": 0.0,
      "step": 55100
    },
    {
      "epoch": 0.7791712027598298,
      "grad_norm": 0.0,
      "learning_rate": 2.2125295140607108e-06,
      "loss": 0.0,
      "step": 55110
    },
    {
      "epoch": 0.7793125874817968,
      "grad_norm": 0.002151825465261936,
      "learning_rate": 2.2111156668410412e-06,
      "loss": 0.0,
      "step": 55120
    },
    {
      "epoch": 0.7794539722037637,
      "grad_norm": 0.0,
      "learning_rate": 2.2097018196213716e-06,
      "loss": 0.0,
      "step": 55130
    },
    {
      "epoch": 0.7795953569257306,
      "grad_norm": 0.0,
      "learning_rate": 2.2082879724017025e-06,
      "loss": 0.0,
      "step": 55140
    },
    {
      "epoch": 0.7797367416476976,
      "grad_norm": 0.0,
      "learning_rate": 2.206874125182033e-06,
      "loss": 0.0,
      "step": 55150
    },
    {
      "epoch": 0.7798781263696645,
      "grad_norm": 0.0,
      "learning_rate": 2.2054602779623634e-06,
      "loss": 0.0,
      "step": 55160
    },
    {
      "epoch": 0.7800195110916315,
      "grad_norm": 0.0,
      "learning_rate": 2.2040464307426942e-06,
      "loss": 0.0,
      "step": 55170
    },
    {
      "epoch": 0.7801608958135984,
      "grad_norm": 0.0,
      "learning_rate": 2.2026325835230247e-06,
      "loss": 0.0,
      "step": 55180
    },
    {
      "epoch": 0.7803022805355653,
      "grad_norm": 0.0,
      "learning_rate": 2.201218736303355e-06,
      "loss": 0.009,
      "step": 55190
    },
    {
      "epoch": 0.7804436652575323,
      "grad_norm": 9.88807441899553e-05,
      "learning_rate": 2.1998048890836855e-06,
      "loss": 0.0,
      "step": 55200
    },
    {
      "epoch": 0.7805850499794992,
      "grad_norm": 0.0,
      "learning_rate": 2.1983910418640164e-06,
      "loss": 0.0001,
      "step": 55210
    },
    {
      "epoch": 0.7807264347014662,
      "grad_norm": 0.7022051215171814,
      "learning_rate": 2.196977194644347e-06,
      "loss": 0.0012,
      "step": 55220
    },
    {
      "epoch": 0.7808678194234331,
      "grad_norm": 0.0,
      "learning_rate": 2.1955633474246772e-06,
      "loss": 0.0,
      "step": 55230
    },
    {
      "epoch": 0.7810092041454001,
      "grad_norm": 0.0,
      "learning_rate": 2.194149500205008e-06,
      "loss": 0.0,
      "step": 55240
    },
    {
      "epoch": 0.781150588867367,
      "grad_norm": 0.0,
      "learning_rate": 2.1927356529853385e-06,
      "loss": 0.0,
      "step": 55250
    },
    {
      "epoch": 0.7812919735893339,
      "grad_norm": 8.466823783237487e-05,
      "learning_rate": 2.191321805765669e-06,
      "loss": 0.0,
      "step": 55260
    },
    {
      "epoch": 0.7814333583113009,
      "grad_norm": 0.0,
      "learning_rate": 2.189907958546e-06,
      "loss": 0.0,
      "step": 55270
    },
    {
      "epoch": 0.7815747430332678,
      "grad_norm": 0.0,
      "learning_rate": 2.1884941113263303e-06,
      "loss": 0.0,
      "step": 55280
    },
    {
      "epoch": 0.7817161277552348,
      "grad_norm": 0.0,
      "learning_rate": 2.1870802641066607e-06,
      "loss": 0.0,
      "step": 55290
    },
    {
      "epoch": 0.7818575124772017,
      "grad_norm": 0.0,
      "learning_rate": 2.1856664168869915e-06,
      "loss": 0.4324,
      "step": 55300
    },
    {
      "epoch": 0.7819988971991687,
      "grad_norm": 27.7211971282959,
      "learning_rate": 2.184252569667322e-06,
      "loss": 0.0052,
      "step": 55310
    },
    {
      "epoch": 0.7821402819211356,
      "grad_norm": 42.35307312011719,
      "learning_rate": 2.1828387224476524e-06,
      "loss": 0.0113,
      "step": 55320
    },
    {
      "epoch": 0.7822816666431025,
      "grad_norm": 0.0,
      "learning_rate": 2.181424875227983e-06,
      "loss": 0.0,
      "step": 55330
    },
    {
      "epoch": 0.7824230513650695,
      "grad_norm": 0.0,
      "learning_rate": 2.1800110280083137e-06,
      "loss": 0.0003,
      "step": 55340
    },
    {
      "epoch": 0.7825644360870364,
      "grad_norm": 0.0,
      "learning_rate": 2.178597180788644e-06,
      "loss": 0.0004,
      "step": 55350
    },
    {
      "epoch": 0.7827058208090034,
      "grad_norm": 0.0,
      "learning_rate": 2.1771833335689746e-06,
      "loss": 0.0,
      "step": 55360
    },
    {
      "epoch": 0.7828472055309703,
      "grad_norm": 0.0,
      "learning_rate": 2.1757694863493054e-06,
      "loss": 0.0054,
      "step": 55370
    },
    {
      "epoch": 0.7829885902529373,
      "grad_norm": 0.0,
      "learning_rate": 2.174355639129636e-06,
      "loss": 0.0,
      "step": 55380
    },
    {
      "epoch": 0.7831299749749042,
      "grad_norm": 2.664155474008112e-08,
      "learning_rate": 2.1729417919099663e-06,
      "loss": 0.0,
      "step": 55390
    },
    {
      "epoch": 0.7832713596968711,
      "grad_norm": 0.0,
      "learning_rate": 2.171527944690297e-06,
      "loss": 0.0,
      "step": 55400
    },
    {
      "epoch": 0.7834127444188381,
      "grad_norm": 4.429852040743754e-09,
      "learning_rate": 2.1701140974706276e-06,
      "loss": 0.0,
      "step": 55410
    },
    {
      "epoch": 0.783554129140805,
      "grad_norm": 6.732196197845042e-06,
      "learning_rate": 2.168700250250958e-06,
      "loss": 0.0,
      "step": 55420
    },
    {
      "epoch": 0.783695513862772,
      "grad_norm": 0.0,
      "learning_rate": 2.167286403031289e-06,
      "loss": 0.0009,
      "step": 55430
    },
    {
      "epoch": 0.7838368985847389,
      "grad_norm": 0.0,
      "learning_rate": 2.1658725558116193e-06,
      "loss": 0.0,
      "step": 55440
    },
    {
      "epoch": 0.7839782833067058,
      "grad_norm": 0.0,
      "learning_rate": 2.1644587085919497e-06,
      "loss": 0.0023,
      "step": 55450
    },
    {
      "epoch": 0.7841196680286728,
      "grad_norm": 0.0,
      "learning_rate": 2.1630448613722806e-06,
      "loss": 0.0,
      "step": 55460
    },
    {
      "epoch": 0.7842610527506397,
      "grad_norm": 0.0,
      "learning_rate": 2.161631014152611e-06,
      "loss": 0.0,
      "step": 55470
    },
    {
      "epoch": 0.7844024374726067,
      "grad_norm": 0.0,
      "learning_rate": 2.1602171669329415e-06,
      "loss": 0.0049,
      "step": 55480
    },
    {
      "epoch": 0.7845438221945736,
      "grad_norm": 0.22346943616867065,
      "learning_rate": 2.158803319713272e-06,
      "loss": 0.0001,
      "step": 55490
    },
    {
      "epoch": 0.7846852069165406,
      "grad_norm": 0.0,
      "learning_rate": 2.1573894724936027e-06,
      "loss": 0.8359,
      "step": 55500
    },
    {
      "epoch": 0.7848265916385075,
      "grad_norm": 0.015820926055312157,
      "learning_rate": 2.155975625273933e-06,
      "loss": 0.0,
      "step": 55510
    },
    {
      "epoch": 0.7849679763604744,
      "grad_norm": 0.0006550147081725299,
      "learning_rate": 2.1545617780542636e-06,
      "loss": 0.0262,
      "step": 55520
    },
    {
      "epoch": 0.7851093610824414,
      "grad_norm": 0.0,
      "learning_rate": 2.1531479308345945e-06,
      "loss": 0.0,
      "step": 55530
    },
    {
      "epoch": 0.7852507458044083,
      "grad_norm": 0.0,
      "learning_rate": 2.151734083614925e-06,
      "loss": 0.0,
      "step": 55540
    },
    {
      "epoch": 0.7853921305263754,
      "grad_norm": 0.0028257006779313087,
      "learning_rate": 2.1503202363952553e-06,
      "loss": 0.0,
      "step": 55550
    },
    {
      "epoch": 0.7855335152483423,
      "grad_norm": 0.0,
      "learning_rate": 2.1489063891755858e-06,
      "loss": 0.0001,
      "step": 55560
    },
    {
      "epoch": 0.7856748999703093,
      "grad_norm": 8.553904251584754e-09,
      "learning_rate": 2.1474925419559166e-06,
      "loss": 0.0,
      "step": 55570
    },
    {
      "epoch": 0.7858162846922762,
      "grad_norm": 0.0,
      "learning_rate": 2.146078694736247e-06,
      "loss": 0.0,
      "step": 55580
    },
    {
      "epoch": 0.785957669414243,
      "grad_norm": 3.155468775162262e-08,
      "learning_rate": 2.1446648475165775e-06,
      "loss": 0.0,
      "step": 55590
    },
    {
      "epoch": 0.7860990541362101,
      "grad_norm": 0.0,
      "learning_rate": 2.1432510002969083e-06,
      "loss": 0.0,
      "step": 55600
    },
    {
      "epoch": 0.786240438858177,
      "grad_norm": 1.5357022285461426,
      "learning_rate": 2.1418371530772388e-06,
      "loss": 0.0005,
      "step": 55610
    },
    {
      "epoch": 0.786381823580144,
      "grad_norm": 0.0,
      "learning_rate": 2.1404233058575692e-06,
      "loss": 0.0,
      "step": 55620
    },
    {
      "epoch": 0.7865232083021109,
      "grad_norm": 0.0,
      "learning_rate": 2.1390094586378996e-06,
      "loss": 0.0002,
      "step": 55630
    },
    {
      "epoch": 0.7866645930240778,
      "grad_norm": 0.3212992548942566,
      "learning_rate": 2.1375956114182305e-06,
      "loss": 0.0041,
      "step": 55640
    },
    {
      "epoch": 0.7868059777460448,
      "grad_norm": 0.0,
      "learning_rate": 2.136181764198561e-06,
      "loss": 0.0,
      "step": 55650
    },
    {
      "epoch": 0.7869473624680117,
      "grad_norm": 0.0,
      "learning_rate": 2.1347679169788914e-06,
      "loss": 0.0,
      "step": 55660
    },
    {
      "epoch": 0.7870887471899787,
      "grad_norm": 0.0,
      "learning_rate": 2.133354069759222e-06,
      "loss": 0.0001,
      "step": 55670
    },
    {
      "epoch": 0.7872301319119456,
      "grad_norm": 0.0,
      "learning_rate": 2.1319402225395527e-06,
      "loss": 0.0663,
      "step": 55680
    },
    {
      "epoch": 0.7873715166339126,
      "grad_norm": 0.0,
      "learning_rate": 2.130526375319883e-06,
      "loss": 0.0,
      "step": 55690
    },
    {
      "epoch": 0.7875129013558795,
      "grad_norm": 1.790626993170008e-05,
      "learning_rate": 2.1291125281002135e-06,
      "loss": 0.0,
      "step": 55700
    },
    {
      "epoch": 0.7876542860778464,
      "grad_norm": 0.0,
      "learning_rate": 2.127698680880544e-06,
      "loss": 0.0,
      "step": 55710
    },
    {
      "epoch": 0.7877956707998134,
      "grad_norm": 0.0,
      "learning_rate": 2.126284833660875e-06,
      "loss": 0.0001,
      "step": 55720
    },
    {
      "epoch": 0.7879370555217803,
      "grad_norm": 0.001581016811542213,
      "learning_rate": 2.1248709864412052e-06,
      "loss": 0.0,
      "step": 55730
    },
    {
      "epoch": 0.7880784402437473,
      "grad_norm": 0.0,
      "learning_rate": 2.1234571392215357e-06,
      "loss": 0.0,
      "step": 55740
    },
    {
      "epoch": 0.7882198249657142,
      "grad_norm": 0.02206534892320633,
      "learning_rate": 2.122043292001866e-06,
      "loss": 0.0,
      "step": 55750
    },
    {
      "epoch": 0.7883612096876812,
      "grad_norm": 3.66751109481811e-08,
      "learning_rate": 2.120629444782197e-06,
      "loss": 0.0,
      "step": 55760
    },
    {
      "epoch": 0.7885025944096481,
      "grad_norm": 1.8544078557169996e-05,
      "learning_rate": 2.1192155975625274e-06,
      "loss": 0.0,
      "step": 55770
    },
    {
      "epoch": 0.788643979131615,
      "grad_norm": 1.879352673483936e-09,
      "learning_rate": 2.117801750342858e-06,
      "loss": 0.0,
      "step": 55780
    },
    {
      "epoch": 0.788785363853582,
      "grad_norm": 0.0,
      "learning_rate": 2.1163879031231887e-06,
      "loss": 0.0,
      "step": 55790
    },
    {
      "epoch": 0.7889267485755489,
      "grad_norm": 0.0,
      "learning_rate": 2.114974055903519e-06,
      "loss": 0.0001,
      "step": 55800
    },
    {
      "epoch": 0.7890681332975159,
      "grad_norm": 0.0,
      "learning_rate": 2.1135602086838496e-06,
      "loss": 0.0,
      "step": 55810
    },
    {
      "epoch": 0.7892095180194828,
      "grad_norm": 1.1135673048556782e-05,
      "learning_rate": 2.1121463614641804e-06,
      "loss": 0.0,
      "step": 55820
    },
    {
      "epoch": 0.7893509027414498,
      "grad_norm": 0.0,
      "learning_rate": 2.110732514244511e-06,
      "loss": 0.0,
      "step": 55830
    },
    {
      "epoch": 0.7894922874634167,
      "grad_norm": 0.0,
      "learning_rate": 2.1093186670248413e-06,
      "loss": 0.0187,
      "step": 55840
    },
    {
      "epoch": 0.7896336721853836,
      "grad_norm": 0.0,
      "learning_rate": 2.107904819805172e-06,
      "loss": 0.0,
      "step": 55850
    },
    {
      "epoch": 0.7897750569073506,
      "grad_norm": 0.0,
      "learning_rate": 2.1064909725855026e-06,
      "loss": 0.0115,
      "step": 55860
    },
    {
      "epoch": 0.7899164416293175,
      "grad_norm": 0.0,
      "learning_rate": 2.105077125365833e-06,
      "loss": 0.0,
      "step": 55870
    },
    {
      "epoch": 0.7900578263512845,
      "grad_norm": 0.0,
      "learning_rate": 2.103663278146164e-06,
      "loss": 0.0,
      "step": 55880
    },
    {
      "epoch": 0.7901992110732514,
      "grad_norm": 0.0,
      "learning_rate": 2.1022494309264943e-06,
      "loss": 0.0309,
      "step": 55890
    },
    {
      "epoch": 0.7903405957952183,
      "grad_norm": 0.0,
      "learning_rate": 2.1008355837068247e-06,
      "loss": 0.0001,
      "step": 55900
    },
    {
      "epoch": 0.7904819805171853,
      "grad_norm": 0.0019049644470214844,
      "learning_rate": 2.099421736487155e-06,
      "loss": 0.0,
      "step": 55910
    },
    {
      "epoch": 0.7906233652391522,
      "grad_norm": 0.0,
      "learning_rate": 2.098007889267486e-06,
      "loss": 0.0,
      "step": 55920
    },
    {
      "epoch": 0.7907647499611192,
      "grad_norm": 0.0,
      "learning_rate": 2.0965940420478164e-06,
      "loss": 0.0,
      "step": 55930
    },
    {
      "epoch": 0.7909061346830861,
      "grad_norm": 0.00013590743765234947,
      "learning_rate": 2.095180194828147e-06,
      "loss": 0.0,
      "step": 55940
    },
    {
      "epoch": 0.7910475194050531,
      "grad_norm": 0.0,
      "learning_rate": 2.0937663476084777e-06,
      "loss": 0.0,
      "step": 55950
    },
    {
      "epoch": 0.79118890412702,
      "grad_norm": 0.0,
      "learning_rate": 2.092352500388808e-06,
      "loss": 0.0,
      "step": 55960
    },
    {
      "epoch": 0.7913302888489869,
      "grad_norm": 0.0,
      "learning_rate": 2.0909386531691386e-06,
      "loss": 0.0,
      "step": 55970
    },
    {
      "epoch": 0.7914716735709539,
      "grad_norm": 0.0,
      "learning_rate": 2.0895248059494695e-06,
      "loss": 0.0,
      "step": 55980
    },
    {
      "epoch": 0.7916130582929208,
      "grad_norm": 0.0,
      "learning_rate": 2.0881109587298e-06,
      "loss": 0.0001,
      "step": 55990
    },
    {
      "epoch": 0.7917544430148878,
      "grad_norm": 1.7806852383728256e-06,
      "learning_rate": 2.0866971115101303e-06,
      "loss": 0.0014,
      "step": 56000
    },
    {
      "epoch": 0.7918958277368547,
      "grad_norm": 0.0,
      "learning_rate": 2.085283264290461e-06,
      "loss": 0.0,
      "step": 56010
    },
    {
      "epoch": 0.7920372124588217,
      "grad_norm": 0.0,
      "learning_rate": 2.0838694170707916e-06,
      "loss": 0.0121,
      "step": 56020
    },
    {
      "epoch": 0.7921785971807886,
      "grad_norm": 1.8393634393376601e-09,
      "learning_rate": 2.082455569851122e-06,
      "loss": 0.0,
      "step": 56030
    },
    {
      "epoch": 0.7923199819027555,
      "grad_norm": 1.648563454637042e-07,
      "learning_rate": 2.0810417226314525e-06,
      "loss": 0.0145,
      "step": 56040
    },
    {
      "epoch": 0.7924613666247226,
      "grad_norm": 1.6624400700493425e-07,
      "learning_rate": 2.0796278754117833e-06,
      "loss": 0.5473,
      "step": 56050
    },
    {
      "epoch": 0.7926027513466894,
      "grad_norm": 0.0,
      "learning_rate": 2.0782140281921138e-06,
      "loss": 0.0,
      "step": 56060
    },
    {
      "epoch": 0.7927441360686565,
      "grad_norm": 0.0,
      "learning_rate": 2.076800180972444e-06,
      "loss": 0.0,
      "step": 56070
    },
    {
      "epoch": 0.7928855207906234,
      "grad_norm": 0.0,
      "learning_rate": 2.075386333752775e-06,
      "loss": 0.0015,
      "step": 56080
    },
    {
      "epoch": 0.7930269055125904,
      "grad_norm": 0.0,
      "learning_rate": 2.0739724865331055e-06,
      "loss": 0.0,
      "step": 56090
    },
    {
      "epoch": 0.7931682902345573,
      "grad_norm": 1.1059042662964202e-05,
      "learning_rate": 2.072558639313436e-06,
      "loss": 0.0,
      "step": 56100
    },
    {
      "epoch": 0.7933096749565242,
      "grad_norm": 0.0001454339799238369,
      "learning_rate": 2.0711447920937668e-06,
      "loss": 0.0,
      "step": 56110
    },
    {
      "epoch": 0.7934510596784912,
      "grad_norm": 0.0,
      "learning_rate": 2.0697309448740972e-06,
      "loss": 0.0,
      "step": 56120
    },
    {
      "epoch": 0.7935924444004581,
      "grad_norm": 0.0,
      "learning_rate": 2.0683170976544276e-06,
      "loss": 0.0,
      "step": 56130
    },
    {
      "epoch": 0.7937338291224251,
      "grad_norm": 0.025621376931667328,
      "learning_rate": 2.0669032504347585e-06,
      "loss": 0.0,
      "step": 56140
    },
    {
      "epoch": 0.793875213844392,
      "grad_norm": 0.0,
      "learning_rate": 2.065489403215089e-06,
      "loss": 0.0,
      "step": 56150
    },
    {
      "epoch": 0.7940165985663589,
      "grad_norm": 0.0,
      "learning_rate": 2.0640755559954194e-06,
      "loss": 0.0001,
      "step": 56160
    },
    {
      "epoch": 0.7941579832883259,
      "grad_norm": 0.0,
      "learning_rate": 2.06266170877575e-06,
      "loss": 0.0,
      "step": 56170
    },
    {
      "epoch": 0.7942993680102928,
      "grad_norm": 0.0,
      "learning_rate": 2.0612478615560807e-06,
      "loss": 0.0,
      "step": 56180
    },
    {
      "epoch": 0.7944407527322598,
      "grad_norm": 0.0,
      "learning_rate": 2.059834014336411e-06,
      "loss": 0.0002,
      "step": 56190
    },
    {
      "epoch": 0.7945821374542267,
      "grad_norm": 0.0,
      "learning_rate": 2.0584201671167415e-06,
      "loss": 0.0,
      "step": 56200
    },
    {
      "epoch": 0.7947235221761937,
      "grad_norm": 0.0,
      "learning_rate": 2.057006319897072e-06,
      "loss": 0.0,
      "step": 56210
    },
    {
      "epoch": 0.7948649068981606,
      "grad_norm": 0.0,
      "learning_rate": 2.055592472677403e-06,
      "loss": 0.0,
      "step": 56220
    },
    {
      "epoch": 0.7950062916201275,
      "grad_norm": 1.8639877453097142e-05,
      "learning_rate": 2.0541786254577332e-06,
      "loss": 0.0017,
      "step": 56230
    },
    {
      "epoch": 0.7951476763420945,
      "grad_norm": 0.0,
      "learning_rate": 2.0527647782380637e-06,
      "loss": 0.0,
      "step": 56240
    },
    {
      "epoch": 0.7952890610640614,
      "grad_norm": 0.0,
      "learning_rate": 2.0513509310183945e-06,
      "loss": 0.0001,
      "step": 56250
    },
    {
      "epoch": 0.7954304457860284,
      "grad_norm": 0.055694032460451126,
      "learning_rate": 2.049937083798725e-06,
      "loss": 0.0,
      "step": 56260
    },
    {
      "epoch": 0.7955718305079953,
      "grad_norm": 0.0,
      "learning_rate": 2.0485232365790554e-06,
      "loss": 0.0012,
      "step": 56270
    },
    {
      "epoch": 0.7957132152299623,
      "grad_norm": 0.0,
      "learning_rate": 2.047109389359386e-06,
      "loss": 0.0002,
      "step": 56280
    },
    {
      "epoch": 0.7958545999519292,
      "grad_norm": 0.0,
      "learning_rate": 2.0456955421397167e-06,
      "loss": 0.0007,
      "step": 56290
    },
    {
      "epoch": 0.7959959846738961,
      "grad_norm": 0.0,
      "learning_rate": 2.044281694920047e-06,
      "loss": 0.0,
      "step": 56300
    },
    {
      "epoch": 0.7961373693958631,
      "grad_norm": 0.0,
      "learning_rate": 2.0428678477003776e-06,
      "loss": 0.0009,
      "step": 56310
    },
    {
      "epoch": 0.79627875411783,
      "grad_norm": 0.0,
      "learning_rate": 2.041454000480708e-06,
      "loss": 0.0,
      "step": 56320
    },
    {
      "epoch": 0.796420138839797,
      "grad_norm": 0.0,
      "learning_rate": 2.040040153261039e-06,
      "loss": 0.0,
      "step": 56330
    },
    {
      "epoch": 0.7965615235617639,
      "grad_norm": 0.0200343057513237,
      "learning_rate": 2.0386263060413693e-06,
      "loss": 0.0,
      "step": 56340
    },
    {
      "epoch": 0.7967029082837309,
      "grad_norm": 0.0,
      "learning_rate": 2.0372124588216997e-06,
      "loss": 0.0,
      "step": 56350
    },
    {
      "epoch": 0.7968442930056978,
      "grad_norm": 0.0,
      "learning_rate": 2.03579861160203e-06,
      "loss": 0.0,
      "step": 56360
    },
    {
      "epoch": 0.7969856777276647,
      "grad_norm": 0.0,
      "learning_rate": 2.034384764382361e-06,
      "loss": 0.0028,
      "step": 56370
    },
    {
      "epoch": 0.7971270624496317,
      "grad_norm": 0.0,
      "learning_rate": 2.0329709171626914e-06,
      "loss": 0.0001,
      "step": 56380
    },
    {
      "epoch": 0.7972684471715986,
      "grad_norm": 0.0,
      "learning_rate": 2.031557069943022e-06,
      "loss": 0.0,
      "step": 56390
    },
    {
      "epoch": 0.7974098318935656,
      "grad_norm": 0.0,
      "learning_rate": 2.0301432227233527e-06,
      "loss": 0.0,
      "step": 56400
    },
    {
      "epoch": 0.7975512166155325,
      "grad_norm": 7.92680549621582,
      "learning_rate": 2.028729375503683e-06,
      "loss": 0.0023,
      "step": 56410
    },
    {
      "epoch": 0.7976926013374994,
      "grad_norm": 0.0,
      "learning_rate": 2.0273155282840136e-06,
      "loss": 0.0,
      "step": 56420
    },
    {
      "epoch": 0.7978339860594664,
      "grad_norm": 2.079570293744837e-09,
      "learning_rate": 2.0259016810643444e-06,
      "loss": 0.0,
      "step": 56430
    },
    {
      "epoch": 0.7979753707814333,
      "grad_norm": 0.014509208500385284,
      "learning_rate": 2.024487833844675e-06,
      "loss": 0.0,
      "step": 56440
    },
    {
      "epoch": 0.7981167555034003,
      "grad_norm": 0.0,
      "learning_rate": 2.0230739866250053e-06,
      "loss": 0.0,
      "step": 56450
    },
    {
      "epoch": 0.7982581402253672,
      "grad_norm": 0.013409438543021679,
      "learning_rate": 2.0216601394053357e-06,
      "loss": 0.0,
      "step": 56460
    },
    {
      "epoch": 0.7983995249473342,
      "grad_norm": 0.0,
      "learning_rate": 2.0202462921856666e-06,
      "loss": 0.3797,
      "step": 56470
    },
    {
      "epoch": 0.7985409096693011,
      "grad_norm": 0.0,
      "learning_rate": 2.018832444965997e-06,
      "loss": 0.0,
      "step": 56480
    },
    {
      "epoch": 0.798682294391268,
      "grad_norm": 0.0,
      "learning_rate": 2.0174185977463275e-06,
      "loss": 0.0,
      "step": 56490
    },
    {
      "epoch": 0.798823679113235,
      "grad_norm": 0.0,
      "learning_rate": 2.0160047505266583e-06,
      "loss": 0.0143,
      "step": 56500
    },
    {
      "epoch": 0.7989650638352019,
      "grad_norm": 0.0,
      "learning_rate": 2.0145909033069888e-06,
      "loss": 0.0,
      "step": 56510
    },
    {
      "epoch": 0.799106448557169,
      "grad_norm": 0.0,
      "learning_rate": 2.013177056087319e-06,
      "loss": 0.0,
      "step": 56520
    },
    {
      "epoch": 0.7992478332791358,
      "grad_norm": 1.968562868270851e-09,
      "learning_rate": 2.01176320886765e-06,
      "loss": 0.001,
      "step": 56530
    },
    {
      "epoch": 0.7993892180011029,
      "grad_norm": 0.0,
      "learning_rate": 2.0103493616479805e-06,
      "loss": 0.0011,
      "step": 56540
    },
    {
      "epoch": 0.7995306027230697,
      "grad_norm": 1.7635943549976218e-07,
      "learning_rate": 2.008935514428311e-06,
      "loss": 0.0002,
      "step": 56550
    },
    {
      "epoch": 0.7996719874450366,
      "grad_norm": 0.0,
      "learning_rate": 2.0075216672086418e-06,
      "loss": 0.0,
      "step": 56560
    },
    {
      "epoch": 0.7998133721670037,
      "grad_norm": 2.666755062818993e-05,
      "learning_rate": 2.006107819988972e-06,
      "loss": 0.0,
      "step": 56570
    },
    {
      "epoch": 0.7999547568889706,
      "grad_norm": 2.163501733321027e-07,
      "learning_rate": 2.0046939727693026e-06,
      "loss": 0.0003,
      "step": 56580
    },
    {
      "epoch": 0.8000961416109376,
      "grad_norm": 1.9171770837544955e-09,
      "learning_rate": 2.003280125549633e-06,
      "loss": 0.0005,
      "step": 56590
    },
    {
      "epoch": 0.8002375263329045,
      "grad_norm": 1.0098277698489255e-06,
      "learning_rate": 2.001866278329964e-06,
      "loss": 0.0,
      "step": 56600
    },
    {
      "epoch": 0.8003789110548714,
      "grad_norm": 0.0,
      "learning_rate": 2.0004524311102944e-06,
      "loss": 0.0,
      "step": 56610
    },
    {
      "epoch": 0.8005202957768384,
      "grad_norm": 0.0011685831705108285,
      "learning_rate": 1.999038583890625e-06,
      "loss": 0.0005,
      "step": 56620
    },
    {
      "epoch": 0.8006616804988053,
      "grad_norm": 0.0,
      "learning_rate": 1.9976247366709556e-06,
      "loss": 0.0,
      "step": 56630
    },
    {
      "epoch": 0.8008030652207723,
      "grad_norm": 0.0,
      "learning_rate": 1.996210889451286e-06,
      "loss": 0.6378,
      "step": 56640
    },
    {
      "epoch": 0.8009444499427392,
      "grad_norm": 0.008006746880710125,
      "learning_rate": 1.9947970422316165e-06,
      "loss": 0.0,
      "step": 56650
    },
    {
      "epoch": 0.8010858346647062,
      "grad_norm": 0.0,
      "learning_rate": 1.9933831950119474e-06,
      "loss": 0.0,
      "step": 56660
    },
    {
      "epoch": 0.8012272193866731,
      "grad_norm": 0.0,
      "learning_rate": 1.991969347792278e-06,
      "loss": 0.0,
      "step": 56670
    },
    {
      "epoch": 0.80136860410864,
      "grad_norm": 1.902162427569465e-09,
      "learning_rate": 1.9905555005726082e-06,
      "loss": 0.0,
      "step": 56680
    },
    {
      "epoch": 0.801509988830607,
      "grad_norm": 0.0018245474202558398,
      "learning_rate": 1.989141653352939e-06,
      "loss": 0.0023,
      "step": 56690
    },
    {
      "epoch": 0.8016513735525739,
      "grad_norm": 0.0,
      "learning_rate": 1.9877278061332695e-06,
      "loss": 0.0001,
      "step": 56700
    },
    {
      "epoch": 0.8017927582745409,
      "grad_norm": 0.0,
      "learning_rate": 1.9863139589136e-06,
      "loss": 1.2805,
      "step": 56710
    },
    {
      "epoch": 0.8019341429965078,
      "grad_norm": 0.0,
      "learning_rate": 1.984900111693931e-06,
      "loss": 0.0,
      "step": 56720
    },
    {
      "epoch": 0.8020755277184748,
      "grad_norm": 0.10519493371248245,
      "learning_rate": 1.9834862644742612e-06,
      "loss": 0.0,
      "step": 56730
    },
    {
      "epoch": 0.8022169124404417,
      "grad_norm": 0.0,
      "learning_rate": 1.9820724172545917e-06,
      "loss": 0.0,
      "step": 56740
    },
    {
      "epoch": 0.8023582971624086,
      "grad_norm": 0.0,
      "learning_rate": 1.980658570034922e-06,
      "loss": 0.0,
      "step": 56750
    },
    {
      "epoch": 0.8024996818843756,
      "grad_norm": 0.0,
      "learning_rate": 1.979244722815253e-06,
      "loss": 0.001,
      "step": 56760
    },
    {
      "epoch": 0.8026410666063425,
      "grad_norm": 1.7122230246968684e-06,
      "learning_rate": 1.9778308755955834e-06,
      "loss": 0.0001,
      "step": 56770
    },
    {
      "epoch": 0.8027824513283095,
      "grad_norm": 0.0004065245157107711,
      "learning_rate": 1.976417028375914e-06,
      "loss": 0.0005,
      "step": 56780
    },
    {
      "epoch": 0.8029238360502764,
      "grad_norm": 1.1477389335632324,
      "learning_rate": 1.9750031811562447e-06,
      "loss": 0.0003,
      "step": 56790
    },
    {
      "epoch": 0.8030652207722434,
      "grad_norm": 0.0,
      "learning_rate": 1.973589333936575e-06,
      "loss": 0.0,
      "step": 56800
    },
    {
      "epoch": 0.8032066054942103,
      "grad_norm": 0.003136452054604888,
      "learning_rate": 1.9721754867169056e-06,
      "loss": 0.0,
      "step": 56810
    },
    {
      "epoch": 0.8033479902161772,
      "grad_norm": 0.0,
      "learning_rate": 1.970761639497236e-06,
      "loss": 0.0,
      "step": 56820
    },
    {
      "epoch": 0.8034893749381442,
      "grad_norm": 0.00027016192325390875,
      "learning_rate": 1.969347792277567e-06,
      "loss": 0.0,
      "step": 56830
    },
    {
      "epoch": 0.8036307596601111,
      "grad_norm": 0.0,
      "learning_rate": 1.9679339450578973e-06,
      "loss": 0.0,
      "step": 56840
    },
    {
      "epoch": 0.8037721443820781,
      "grad_norm": 0.015812229365110397,
      "learning_rate": 1.9665200978382277e-06,
      "loss": 0.0018,
      "step": 56850
    },
    {
      "epoch": 0.803913529104045,
      "grad_norm": 0.013638166710734367,
      "learning_rate": 1.9651062506185586e-06,
      "loss": 0.0,
      "step": 56860
    },
    {
      "epoch": 0.8040549138260119,
      "grad_norm": 0.0,
      "learning_rate": 1.963692403398889e-06,
      "loss": 0.0,
      "step": 56870
    },
    {
      "epoch": 0.8041962985479789,
      "grad_norm": 0.0,
      "learning_rate": 1.9622785561792194e-06,
      "loss": 0.0005,
      "step": 56880
    },
    {
      "epoch": 0.8043376832699458,
      "grad_norm": 0.0,
      "learning_rate": 1.96086470895955e-06,
      "loss": 0.0008,
      "step": 56890
    },
    {
      "epoch": 0.8044790679919128,
      "grad_norm": 0.0,
      "learning_rate": 1.9594508617398807e-06,
      "loss": 0.0,
      "step": 56900
    },
    {
      "epoch": 0.8046204527138797,
      "grad_norm": 0.0,
      "learning_rate": 1.958037014520211e-06,
      "loss": 0.0,
      "step": 56910
    },
    {
      "epoch": 0.8047618374358467,
      "grad_norm": 0.0,
      "learning_rate": 1.9566231673005416e-06,
      "loss": 0.0,
      "step": 56920
    },
    {
      "epoch": 0.8049032221578136,
      "grad_norm": 0.0,
      "learning_rate": 1.955209320080872e-06,
      "loss": 0.0,
      "step": 56930
    },
    {
      "epoch": 0.8050446068797805,
      "grad_norm": 0.0,
      "learning_rate": 1.953795472861203e-06,
      "loss": 0.0,
      "step": 56940
    },
    {
      "epoch": 0.8051859916017475,
      "grad_norm": 495.9982604980469,
      "learning_rate": 1.9523816256415333e-06,
      "loss": 0.0964,
      "step": 56950
    },
    {
      "epoch": 0.8053273763237144,
      "grad_norm": 0.0,
      "learning_rate": 1.9509677784218637e-06,
      "loss": 0.0,
      "step": 56960
    },
    {
      "epoch": 0.8054687610456814,
      "grad_norm": 0.0,
      "learning_rate": 1.949553931202194e-06,
      "loss": 0.0,
      "step": 56970
    },
    {
      "epoch": 0.8056101457676483,
      "grad_norm": 9.819475232930586e-10,
      "learning_rate": 1.948140083982525e-06,
      "loss": 0.0,
      "step": 56980
    },
    {
      "epoch": 0.8057515304896153,
      "grad_norm": 0.0,
      "learning_rate": 1.9467262367628555e-06,
      "loss": 0.0003,
      "step": 56990
    },
    {
      "epoch": 0.8058929152115822,
      "grad_norm": 0.0,
      "learning_rate": 1.945312389543186e-06,
      "loss": 0.0,
      "step": 57000
    },
    {
      "epoch": 0.8060342999335491,
      "grad_norm": 0.0,
      "learning_rate": 1.9438985423235168e-06,
      "loss": 0.0066,
      "step": 57010
    },
    {
      "epoch": 0.8061756846555161,
      "grad_norm": 0.0,
      "learning_rate": 1.942484695103847e-06,
      "loss": 0.0,
      "step": 57020
    },
    {
      "epoch": 0.806317069377483,
      "grad_norm": 0.0,
      "learning_rate": 1.9410708478841776e-06,
      "loss": 0.0,
      "step": 57030
    },
    {
      "epoch": 0.80645845409945,
      "grad_norm": 1.8990220951309311e-06,
      "learning_rate": 1.939657000664508e-06,
      "loss": 0.0,
      "step": 57040
    },
    {
      "epoch": 0.806599838821417,
      "grad_norm": 1.7944063074537553e-05,
      "learning_rate": 1.938243153444839e-06,
      "loss": 0.0,
      "step": 57050
    },
    {
      "epoch": 0.806741223543384,
      "grad_norm": 0.001454932731576264,
      "learning_rate": 1.9368293062251693e-06,
      "loss": 0.0,
      "step": 57060
    },
    {
      "epoch": 0.8068826082653509,
      "grad_norm": 0.0,
      "learning_rate": 1.9354154590054998e-06,
      "loss": 0.0008,
      "step": 57070
    },
    {
      "epoch": 0.8070239929873178,
      "grad_norm": 0.0,
      "learning_rate": 1.9340016117858306e-06,
      "loss": 0.0,
      "step": 57080
    },
    {
      "epoch": 0.8071653777092848,
      "grad_norm": 0.0,
      "learning_rate": 1.932587764566161e-06,
      "loss": 0.0001,
      "step": 57090
    },
    {
      "epoch": 0.8073067624312517,
      "grad_norm": 0.0,
      "learning_rate": 1.9311739173464915e-06,
      "loss": 0.0,
      "step": 57100
    },
    {
      "epoch": 0.8074481471532187,
      "grad_norm": 2.1748926428699633e-06,
      "learning_rate": 1.9297600701268224e-06,
      "loss": 0.0,
      "step": 57110
    },
    {
      "epoch": 0.8075895318751856,
      "grad_norm": 0.0,
      "learning_rate": 1.928346222907153e-06,
      "loss": 0.0,
      "step": 57120
    },
    {
      "epoch": 0.8077309165971525,
      "grad_norm": 0.0,
      "learning_rate": 1.9269323756874832e-06,
      "loss": 0.0,
      "step": 57130
    },
    {
      "epoch": 0.8078723013191195,
      "grad_norm": 0.0,
      "learning_rate": 1.925518528467814e-06,
      "loss": 0.0,
      "step": 57140
    },
    {
      "epoch": 0.8080136860410864,
      "grad_norm": 0.0,
      "learning_rate": 1.9241046812481445e-06,
      "loss": 0.0,
      "step": 57150
    },
    {
      "epoch": 0.8081550707630534,
      "grad_norm": 7.936906740724226e-07,
      "learning_rate": 1.922690834028475e-06,
      "loss": 0.0,
      "step": 57160
    },
    {
      "epoch": 0.8082964554850203,
      "grad_norm": 0.0,
      "learning_rate": 1.9212769868088054e-06,
      "loss": 0.0002,
      "step": 57170
    },
    {
      "epoch": 0.8084378402069873,
      "grad_norm": 0.0,
      "learning_rate": 1.9198631395891362e-06,
      "loss": 0.0,
      "step": 57180
    },
    {
      "epoch": 0.8085792249289542,
      "grad_norm": 0.0,
      "learning_rate": 1.9184492923694667e-06,
      "loss": 0.1033,
      "step": 57190
    },
    {
      "epoch": 0.8087206096509211,
      "grad_norm": 0.00011989491758868098,
      "learning_rate": 1.917035445149797e-06,
      "loss": 0.0,
      "step": 57200
    },
    {
      "epoch": 0.8088619943728881,
      "grad_norm": 0.0001361441391054541,
      "learning_rate": 1.915621597930128e-06,
      "loss": 0.0,
      "step": 57210
    },
    {
      "epoch": 0.809003379094855,
      "grad_norm": 0.0,
      "learning_rate": 1.9142077507104584e-06,
      "loss": 0.0,
      "step": 57220
    },
    {
      "epoch": 0.809144763816822,
      "grad_norm": 0.0,
      "learning_rate": 1.912793903490789e-06,
      "loss": 0.0001,
      "step": 57230
    },
    {
      "epoch": 0.8092861485387889,
      "grad_norm": 0.0,
      "learning_rate": 1.9113800562711197e-06,
      "loss": 0.0,
      "step": 57240
    },
    {
      "epoch": 0.8094275332607559,
      "grad_norm": 0.0,
      "learning_rate": 1.90996620905145e-06,
      "loss": 0.0001,
      "step": 57250
    },
    {
      "epoch": 0.8095689179827228,
      "grad_norm": 0.0028128938283771276,
      "learning_rate": 1.9085523618317805e-06,
      "loss": 0.0,
      "step": 57260
    },
    {
      "epoch": 0.8097103027046897,
      "grad_norm": 0.0,
      "learning_rate": 1.9071385146121114e-06,
      "loss": 0.0,
      "step": 57270
    },
    {
      "epoch": 0.8098516874266567,
      "grad_norm": 0.0,
      "learning_rate": 1.9057246673924418e-06,
      "loss": 0.0,
      "step": 57280
    },
    {
      "epoch": 0.8099930721486236,
      "grad_norm": 0.0,
      "learning_rate": 1.9043108201727723e-06,
      "loss": 0.0,
      "step": 57290
    },
    {
      "epoch": 0.8101344568705906,
      "grad_norm": 0.0,
      "learning_rate": 1.9028969729531027e-06,
      "loss": 0.0,
      "step": 57300
    },
    {
      "epoch": 0.8102758415925575,
      "grad_norm": 8.685453940415755e-05,
      "learning_rate": 1.9014831257334336e-06,
      "loss": 0.0176,
      "step": 57310
    },
    {
      "epoch": 0.8104172263145244,
      "grad_norm": 9.8399817943573e-05,
      "learning_rate": 1.900069278513764e-06,
      "loss": 0.0,
      "step": 57320
    },
    {
      "epoch": 0.8105586110364914,
      "grad_norm": 2.549334077173171e-08,
      "learning_rate": 1.8986554312940944e-06,
      "loss": 0.0,
      "step": 57330
    },
    {
      "epoch": 0.8106999957584583,
      "grad_norm": 0.0,
      "learning_rate": 1.897241584074425e-06,
      "loss": 0.0,
      "step": 57340
    },
    {
      "epoch": 0.8108413804804253,
      "grad_norm": 9.792146011022851e-05,
      "learning_rate": 1.8958277368547557e-06,
      "loss": 0.0,
      "step": 57350
    },
    {
      "epoch": 0.8109827652023922,
      "grad_norm": 0.0,
      "learning_rate": 1.8944138896350861e-06,
      "loss": 0.0,
      "step": 57360
    },
    {
      "epoch": 0.8111241499243592,
      "grad_norm": 0.0,
      "learning_rate": 1.8930000424154168e-06,
      "loss": 0.0043,
      "step": 57370
    },
    {
      "epoch": 0.8112655346463261,
      "grad_norm": 0.0,
      "learning_rate": 1.8915861951957472e-06,
      "loss": 0.0002,
      "step": 57380
    },
    {
      "epoch": 0.811406919368293,
      "grad_norm": 2.9124653339385986,
      "learning_rate": 1.8901723479760779e-06,
      "loss": 0.0188,
      "step": 57390
    },
    {
      "epoch": 0.81154830409026,
      "grad_norm": 1.7563718301971676e-07,
      "learning_rate": 1.8887585007564085e-06,
      "loss": 0.0,
      "step": 57400
    },
    {
      "epoch": 0.8116896888122269,
      "grad_norm": 0.0,
      "learning_rate": 1.887344653536739e-06,
      "loss": 0.0,
      "step": 57410
    },
    {
      "epoch": 0.8118310735341939,
      "grad_norm": 0.0,
      "learning_rate": 1.8859308063170694e-06,
      "loss": 0.0,
      "step": 57420
    },
    {
      "epoch": 0.8119724582561608,
      "grad_norm": 0.0,
      "learning_rate": 1.8845169590974e-06,
      "loss": 0.0,
      "step": 57430
    },
    {
      "epoch": 0.8121138429781278,
      "grad_norm": 0.0,
      "learning_rate": 1.8831031118777307e-06,
      "loss": 0.0,
      "step": 57440
    },
    {
      "epoch": 0.8122552277000947,
      "grad_norm": 9.42820599370009e-10,
      "learning_rate": 1.881689264658061e-06,
      "loss": 0.0,
      "step": 57450
    },
    {
      "epoch": 0.8123966124220616,
      "grad_norm": 0.00010077191836899146,
      "learning_rate": 1.8802754174383917e-06,
      "loss": 0.0,
      "step": 57460
    },
    {
      "epoch": 0.8125379971440286,
      "grad_norm": 0.0,
      "learning_rate": 1.8788615702187224e-06,
      "loss": 0.0029,
      "step": 57470
    },
    {
      "epoch": 0.8126793818659955,
      "grad_norm": 0.0,
      "learning_rate": 1.8774477229990528e-06,
      "loss": 0.0,
      "step": 57480
    },
    {
      "epoch": 0.8128207665879625,
      "grad_norm": 0.0,
      "learning_rate": 1.8760338757793833e-06,
      "loss": 0.0,
      "step": 57490
    },
    {
      "epoch": 0.8129621513099294,
      "grad_norm": 0.0,
      "learning_rate": 1.8746200285597141e-06,
      "loss": 0.0,
      "step": 57500
    },
    {
      "epoch": 0.8131035360318964,
      "grad_norm": 8.087459377748019e-07,
      "learning_rate": 1.8732061813400445e-06,
      "loss": 0.0,
      "step": 57510
    },
    {
      "epoch": 0.8132449207538633,
      "grad_norm": 2.848661461030133e-05,
      "learning_rate": 1.871792334120375e-06,
      "loss": 0.0,
      "step": 57520
    },
    {
      "epoch": 0.8133863054758302,
      "grad_norm": 0.0,
      "learning_rate": 1.8703784869007058e-06,
      "loss": 0.0,
      "step": 57530
    },
    {
      "epoch": 0.8135276901977972,
      "grad_norm": 0.0,
      "learning_rate": 1.8689646396810363e-06,
      "loss": 0.0,
      "step": 57540
    },
    {
      "epoch": 0.8136690749197641,
      "grad_norm": 0.008026099763810635,
      "learning_rate": 1.8675507924613667e-06,
      "loss": 0.0002,
      "step": 57550
    },
    {
      "epoch": 0.8138104596417312,
      "grad_norm": 0.003650215920060873,
      "learning_rate": 1.8661369452416976e-06,
      "loss": 0.0,
      "step": 57560
    },
    {
      "epoch": 0.813951844363698,
      "grad_norm": 0.0,
      "learning_rate": 1.864723098022028e-06,
      "loss": 0.0,
      "step": 57570
    },
    {
      "epoch": 0.814093229085665,
      "grad_norm": 0.0,
      "learning_rate": 1.8633092508023584e-06,
      "loss": 0.9985,
      "step": 57580
    },
    {
      "epoch": 0.814234613807632,
      "grad_norm": 6.263718605041504,
      "learning_rate": 1.8618954035826889e-06,
      "loss": 0.0007,
      "step": 57590
    },
    {
      "epoch": 0.8143759985295989,
      "grad_norm": 3.885321930852115e-08,
      "learning_rate": 1.8604815563630197e-06,
      "loss": 0.0,
      "step": 57600
    },
    {
      "epoch": 0.8145173832515659,
      "grad_norm": 3.993118752987357e-06,
      "learning_rate": 1.8590677091433501e-06,
      "loss": 0.0,
      "step": 57610
    },
    {
      "epoch": 0.8146587679735328,
      "grad_norm": 0.0,
      "learning_rate": 1.8576538619236806e-06,
      "loss": 0.0001,
      "step": 57620
    },
    {
      "epoch": 0.8148001526954998,
      "grad_norm": 0.0,
      "learning_rate": 1.8562400147040112e-06,
      "loss": 0.0,
      "step": 57630
    },
    {
      "epoch": 0.8149415374174667,
      "grad_norm": 0.0,
      "learning_rate": 1.8548261674843419e-06,
      "loss": 0.0081,
      "step": 57640
    },
    {
      "epoch": 0.8150829221394336,
      "grad_norm": 0.0,
      "learning_rate": 1.8534123202646723e-06,
      "loss": 0.0008,
      "step": 57650
    },
    {
      "epoch": 0.8152243068614006,
      "grad_norm": 0.00046559094334952533,
      "learning_rate": 1.851998473045003e-06,
      "loss": 0.0,
      "step": 57660
    },
    {
      "epoch": 0.8153656915833675,
      "grad_norm": 0.0,
      "learning_rate": 1.8505846258253336e-06,
      "loss": 0.0,
      "step": 57670
    },
    {
      "epoch": 0.8155070763053345,
      "grad_norm": 0.0,
      "learning_rate": 1.849170778605664e-06,
      "loss": 0.0,
      "step": 57680
    },
    {
      "epoch": 0.8156484610273014,
      "grad_norm": 0.0,
      "learning_rate": 1.8477569313859947e-06,
      "loss": 0.0002,
      "step": 57690
    },
    {
      "epoch": 0.8157898457492684,
      "grad_norm": 0.0,
      "learning_rate": 1.846343084166325e-06,
      "loss": 0.0,
      "step": 57700
    },
    {
      "epoch": 0.8159312304712353,
      "grad_norm": 0.0,
      "learning_rate": 1.8449292369466557e-06,
      "loss": 0.0,
      "step": 57710
    },
    {
      "epoch": 0.8160726151932022,
      "grad_norm": 0.0,
      "learning_rate": 1.8435153897269862e-06,
      "loss": 0.0097,
      "step": 57720
    },
    {
      "epoch": 0.8162139999151692,
      "grad_norm": 0.0,
      "learning_rate": 1.8421015425073168e-06,
      "loss": 0.0,
      "step": 57730
    },
    {
      "epoch": 0.8163553846371361,
      "grad_norm": 1.1455943584442139,
      "learning_rate": 1.8406876952876473e-06,
      "loss": 0.0002,
      "step": 57740
    },
    {
      "epoch": 0.8164967693591031,
      "grad_norm": 6.946247594896704e-05,
      "learning_rate": 1.839273848067978e-06,
      "loss": 0.0009,
      "step": 57750
    },
    {
      "epoch": 0.81663815408107,
      "grad_norm": 20.69721031188965,
      "learning_rate": 1.8378600008483085e-06,
      "loss": 0.003,
      "step": 57760
    },
    {
      "epoch": 0.816779538803037,
      "grad_norm": 0.0,
      "learning_rate": 1.836446153628639e-06,
      "loss": 0.0,
      "step": 57770
    },
    {
      "epoch": 0.8169209235250039,
      "grad_norm": 0.0,
      "learning_rate": 1.8350323064089694e-06,
      "loss": 0.0021,
      "step": 57780
    },
    {
      "epoch": 0.8170623082469708,
      "grad_norm": 0.0,
      "learning_rate": 1.8336184591893003e-06,
      "loss": 0.3929,
      "step": 57790
    },
    {
      "epoch": 0.8172036929689378,
      "grad_norm": 6.197142798924915e-09,
      "learning_rate": 1.8322046119696307e-06,
      "loss": 0.0,
      "step": 57800
    },
    {
      "epoch": 0.8173450776909047,
      "grad_norm": 5.390252681536367e-06,
      "learning_rate": 1.8307907647499611e-06,
      "loss": 0.0004,
      "step": 57810
    },
    {
      "epoch": 0.8174864624128717,
      "grad_norm": 0.0,
      "learning_rate": 1.829376917530292e-06,
      "loss": 0.0,
      "step": 57820
    },
    {
      "epoch": 0.8176278471348386,
      "grad_norm": 0.0,
      "learning_rate": 1.8279630703106224e-06,
      "loss": 0.1245,
      "step": 57830
    },
    {
      "epoch": 0.8177692318568055,
      "grad_norm": 0.0,
      "learning_rate": 1.8265492230909529e-06,
      "loss": 0.0,
      "step": 57840
    },
    {
      "epoch": 0.8179106165787725,
      "grad_norm": 8.941343221380293e-10,
      "learning_rate": 1.8251353758712833e-06,
      "loss": 0.0,
      "step": 57850
    },
    {
      "epoch": 0.8180520013007394,
      "grad_norm": 0.0,
      "learning_rate": 1.8237215286516141e-06,
      "loss": 0.0,
      "step": 57860
    },
    {
      "epoch": 0.8181933860227064,
      "grad_norm": 0.0,
      "learning_rate": 1.8223076814319446e-06,
      "loss": 0.0,
      "step": 57870
    },
    {
      "epoch": 0.8183347707446733,
      "grad_norm": 5.2680821681860834e-05,
      "learning_rate": 1.820893834212275e-06,
      "loss": 0.0,
      "step": 57880
    },
    {
      "epoch": 0.8184761554666403,
      "grad_norm": 0.00016783701721578836,
      "learning_rate": 1.8194799869926059e-06,
      "loss": 0.0002,
      "step": 57890
    },
    {
      "epoch": 0.8186175401886072,
      "grad_norm": 0.0,
      "learning_rate": 1.8180661397729363e-06,
      "loss": 0.0006,
      "step": 57900
    },
    {
      "epoch": 0.8187589249105741,
      "grad_norm": 4.791697261197214e-09,
      "learning_rate": 1.8166522925532667e-06,
      "loss": 0.0004,
      "step": 57910
    },
    {
      "epoch": 0.8189003096325411,
      "grad_norm": 1.0901429092768922e-08,
      "learning_rate": 1.8152384453335976e-06,
      "loss": 0.0001,
      "step": 57920
    },
    {
      "epoch": 0.819041694354508,
      "grad_norm": 0.0,
      "learning_rate": 1.813824598113928e-06,
      "loss": 0.0,
      "step": 57930
    },
    {
      "epoch": 0.819183079076475,
      "grad_norm": 6.720982437258272e-09,
      "learning_rate": 1.8124107508942585e-06,
      "loss": 0.2322,
      "step": 57940
    },
    {
      "epoch": 0.8193244637984419,
      "grad_norm": 0.0,
      "learning_rate": 1.810996903674589e-06,
      "loss": 0.0,
      "step": 57950
    },
    {
      "epoch": 0.8194658485204089,
      "grad_norm": 0.0,
      "learning_rate": 1.8095830564549197e-06,
      "loss": 0.0,
      "step": 57960
    },
    {
      "epoch": 0.8196072332423758,
      "grad_norm": 0.0,
      "learning_rate": 1.8081692092352502e-06,
      "loss": 0.0004,
      "step": 57970
    },
    {
      "epoch": 0.8197486179643427,
      "grad_norm": 0.0,
      "learning_rate": 1.8067553620155808e-06,
      "loss": 0.0028,
      "step": 57980
    },
    {
      "epoch": 0.8198900026863097,
      "grad_norm": 0.0,
      "learning_rate": 1.8053415147959113e-06,
      "loss": 0.0,
      "step": 57990
    },
    {
      "epoch": 0.8200313874082766,
      "grad_norm": 3.860830929625081e-06,
      "learning_rate": 1.803927667576242e-06,
      "loss": 0.0004,
      "step": 58000
    },
    {
      "epoch": 0.8201727721302436,
      "grad_norm": 0.0,
      "learning_rate": 1.8025138203565723e-06,
      "loss": 0.0,
      "step": 58010
    },
    {
      "epoch": 0.8203141568522105,
      "grad_norm": 0.0,
      "learning_rate": 1.801099973136903e-06,
      "loss": 0.0001,
      "step": 58020
    },
    {
      "epoch": 0.8204555415741775,
      "grad_norm": 0.0,
      "learning_rate": 1.7996861259172334e-06,
      "loss": 0.0,
      "step": 58030
    },
    {
      "epoch": 0.8205969262961444,
      "grad_norm": 5.156067004463694e-07,
      "learning_rate": 1.798272278697564e-06,
      "loss": 0.0008,
      "step": 58040
    },
    {
      "epoch": 0.8207383110181113,
      "grad_norm": 4.354296834208071e-07,
      "learning_rate": 1.7968584314778947e-06,
      "loss": 0.0,
      "step": 58050
    },
    {
      "epoch": 0.8208796957400784,
      "grad_norm": 0.0007378815789707005,
      "learning_rate": 1.7954445842582251e-06,
      "loss": 0.0,
      "step": 58060
    },
    {
      "epoch": 0.8210210804620452,
      "grad_norm": 0.0,
      "learning_rate": 1.7940307370385558e-06,
      "loss": 0.0,
      "step": 58070
    },
    {
      "epoch": 0.8211624651840123,
      "grad_norm": 0.0,
      "learning_rate": 1.7926168898188864e-06,
      "loss": 0.0,
      "step": 58080
    },
    {
      "epoch": 0.8213038499059792,
      "grad_norm": 0.0,
      "learning_rate": 1.7912030425992169e-06,
      "loss": 0.0,
      "step": 58090
    },
    {
      "epoch": 0.821445234627946,
      "grad_norm": 8.38177656987682e-05,
      "learning_rate": 1.7897891953795473e-06,
      "loss": 0.0,
      "step": 58100
    },
    {
      "epoch": 0.8215866193499131,
      "grad_norm": 0.0,
      "learning_rate": 1.7883753481598781e-06,
      "loss": 0.0,
      "step": 58110
    },
    {
      "epoch": 0.82172800407188,
      "grad_norm": 5.086302046208857e-09,
      "learning_rate": 1.7869615009402086e-06,
      "loss": 0.0,
      "step": 58120
    },
    {
      "epoch": 0.821869388793847,
      "grad_norm": 0.0,
      "learning_rate": 1.785547653720539e-06,
      "loss": 0.0,
      "step": 58130
    },
    {
      "epoch": 0.8220107735158139,
      "grad_norm": 0.0,
      "learning_rate": 1.7841338065008694e-06,
      "loss": 0.0,
      "step": 58140
    },
    {
      "epoch": 0.8221521582377809,
      "grad_norm": 0.0,
      "learning_rate": 1.7827199592812003e-06,
      "loss": 0.0,
      "step": 58150
    },
    {
      "epoch": 0.8222935429597478,
      "grad_norm": 0.0,
      "learning_rate": 1.7813061120615307e-06,
      "loss": 0.0,
      "step": 58160
    },
    {
      "epoch": 0.8224349276817147,
      "grad_norm": 0.0,
      "learning_rate": 1.7798922648418612e-06,
      "loss": 0.0001,
      "step": 58170
    },
    {
      "epoch": 0.8225763124036817,
      "grad_norm": 0.0,
      "learning_rate": 1.778478417622192e-06,
      "loss": 0.0,
      "step": 58180
    },
    {
      "epoch": 0.8227176971256486,
      "grad_norm": 0.0006972315604798496,
      "learning_rate": 1.7770645704025225e-06,
      "loss": 0.0,
      "step": 58190
    },
    {
      "epoch": 0.8228590818476156,
      "grad_norm": 0.0,
      "learning_rate": 1.7756507231828529e-06,
      "loss": 0.0,
      "step": 58200
    },
    {
      "epoch": 0.8230004665695825,
      "grad_norm": 0.0,
      "learning_rate": 1.7742368759631837e-06,
      "loss": 0.0,
      "step": 58210
    },
    {
      "epoch": 0.8231418512915495,
      "grad_norm": 0.0,
      "learning_rate": 1.7728230287435142e-06,
      "loss": 0.0,
      "step": 58220
    },
    {
      "epoch": 0.8232832360135164,
      "grad_norm": 0.0006459114956669509,
      "learning_rate": 1.7714091815238446e-06,
      "loss": 0.0001,
      "step": 58230
    },
    {
      "epoch": 0.8234246207354833,
      "grad_norm": 0.0008144766907207668,
      "learning_rate": 1.7699953343041753e-06,
      "loss": 0.0053,
      "step": 58240
    },
    {
      "epoch": 0.8235660054574503,
      "grad_norm": 0.0,
      "learning_rate": 1.768581487084506e-06,
      "loss": 0.0,
      "step": 58250
    },
    {
      "epoch": 0.8237073901794172,
      "grad_norm": 0.0,
      "learning_rate": 1.7671676398648363e-06,
      "loss": 0.0,
      "step": 58260
    },
    {
      "epoch": 0.8238487749013842,
      "grad_norm": 9.217778762504736e-10,
      "learning_rate": 1.7657537926451668e-06,
      "loss": 0.0,
      "step": 58270
    },
    {
      "epoch": 0.8239901596233511,
      "grad_norm": 0.0,
      "learning_rate": 1.7643399454254976e-06,
      "loss": 0.0003,
      "step": 58280
    },
    {
      "epoch": 0.824131544345318,
      "grad_norm": 0.0,
      "learning_rate": 1.762926098205828e-06,
      "loss": 0.0,
      "step": 58290
    },
    {
      "epoch": 0.824272929067285,
      "grad_norm": 0.0,
      "learning_rate": 1.7615122509861585e-06,
      "loss": 0.0004,
      "step": 58300
    },
    {
      "epoch": 0.8244143137892519,
      "grad_norm": 0.0,
      "learning_rate": 1.7600984037664891e-06,
      "loss": 0.0,
      "step": 58310
    },
    {
      "epoch": 0.8245556985112189,
      "grad_norm": 1.7780466805561446e-06,
      "learning_rate": 1.7586845565468198e-06,
      "loss": 0.0,
      "step": 58320
    },
    {
      "epoch": 0.8246970832331858,
      "grad_norm": 1.0871747690543998e-05,
      "learning_rate": 1.7572707093271502e-06,
      "loss": 0.0,
      "step": 58330
    },
    {
      "epoch": 0.8248384679551528,
      "grad_norm": 116.58293914794922,
      "learning_rate": 1.7558568621074809e-06,
      "loss": 0.0158,
      "step": 58340
    },
    {
      "epoch": 0.8249798526771197,
      "grad_norm": 0.0,
      "learning_rate": 1.7544430148878113e-06,
      "loss": 0.0,
      "step": 58350
    },
    {
      "epoch": 0.8251212373990866,
      "grad_norm": 0.0,
      "learning_rate": 1.753029167668142e-06,
      "loss": 0.0,
      "step": 58360
    },
    {
      "epoch": 0.8252626221210536,
      "grad_norm": 0.0,
      "learning_rate": 1.7516153204484726e-06,
      "loss": 0.0,
      "step": 58370
    },
    {
      "epoch": 0.8254040068430205,
      "grad_norm": 0.0,
      "learning_rate": 1.750201473228803e-06,
      "loss": 0.0,
      "step": 58380
    },
    {
      "epoch": 0.8255453915649875,
      "grad_norm": 0.0,
      "learning_rate": 1.7487876260091334e-06,
      "loss": 0.0,
      "step": 58390
    },
    {
      "epoch": 0.8256867762869544,
      "grad_norm": 0.0,
      "learning_rate": 1.747373778789464e-06,
      "loss": 0.0,
      "step": 58400
    },
    {
      "epoch": 0.8258281610089214,
      "grad_norm": 0.0,
      "learning_rate": 1.7459599315697947e-06,
      "loss": 0.1858,
      "step": 58410
    },
    {
      "epoch": 0.8259695457308883,
      "grad_norm": 0.00046419849968515337,
      "learning_rate": 1.7445460843501252e-06,
      "loss": 0.0,
      "step": 58420
    },
    {
      "epoch": 0.8261109304528552,
      "grad_norm": 0.0,
      "learning_rate": 1.7431322371304556e-06,
      "loss": 0.0016,
      "step": 58430
    },
    {
      "epoch": 0.8262523151748222,
      "grad_norm": 0.0,
      "learning_rate": 1.7417183899107865e-06,
      "loss": 0.0,
      "step": 58440
    },
    {
      "epoch": 0.8263936998967891,
      "grad_norm": 0.0,
      "learning_rate": 1.7403045426911169e-06,
      "loss": 0.0576,
      "step": 58450
    },
    {
      "epoch": 0.8265350846187561,
      "grad_norm": 0.0,
      "learning_rate": 1.7388906954714473e-06,
      "loss": 0.0037,
      "step": 58460
    },
    {
      "epoch": 0.826676469340723,
      "grad_norm": 0.0,
      "learning_rate": 1.7374768482517782e-06,
      "loss": 0.0,
      "step": 58470
    },
    {
      "epoch": 0.82681785406269,
      "grad_norm": 8.280057954834774e-06,
      "learning_rate": 1.7360630010321086e-06,
      "loss": 0.0,
      "step": 58480
    },
    {
      "epoch": 0.8269592387846569,
      "grad_norm": 0.0,
      "learning_rate": 1.734649153812439e-06,
      "loss": 0.0,
      "step": 58490
    },
    {
      "epoch": 0.8271006235066238,
      "grad_norm": 0.0,
      "learning_rate": 1.73323530659277e-06,
      "loss": 0.0,
      "step": 58500
    },
    {
      "epoch": 0.8272420082285908,
      "grad_norm": 0.0,
      "learning_rate": 1.7318214593731003e-06,
      "loss": 0.0,
      "step": 58510
    },
    {
      "epoch": 0.8273833929505577,
      "grad_norm": 0.0,
      "learning_rate": 1.7304076121534308e-06,
      "loss": 0.0001,
      "step": 58520
    },
    {
      "epoch": 0.8275247776725247,
      "grad_norm": 9.065903028293576e-10,
      "learning_rate": 1.7289937649337616e-06,
      "loss": 0.0,
      "step": 58530
    },
    {
      "epoch": 0.8276661623944916,
      "grad_norm": 0.004569395445287228,
      "learning_rate": 1.727579917714092e-06,
      "loss": 0.048,
      "step": 58540
    },
    {
      "epoch": 0.8278075471164585,
      "grad_norm": 0.0,
      "learning_rate": 1.7261660704944225e-06,
      "loss": 0.0,
      "step": 58550
    },
    {
      "epoch": 0.8279489318384256,
      "grad_norm": 0.0,
      "learning_rate": 1.724752223274753e-06,
      "loss": 0.0,
      "step": 58560
    },
    {
      "epoch": 0.8280903165603924,
      "grad_norm": 0.0,
      "learning_rate": 1.7233383760550838e-06,
      "loss": 0.0,
      "step": 58570
    },
    {
      "epoch": 0.8282317012823595,
      "grad_norm": 9.408416268286146e-10,
      "learning_rate": 1.7219245288354142e-06,
      "loss": 0.0,
      "step": 58580
    },
    {
      "epoch": 0.8283730860043264,
      "grad_norm": 0.0,
      "learning_rate": 1.7205106816157446e-06,
      "loss": 0.0,
      "step": 58590
    },
    {
      "epoch": 0.8285144707262934,
      "grad_norm": 1.3712727309211914e-07,
      "learning_rate": 1.7190968343960753e-06,
      "loss": 0.0,
      "step": 58600
    },
    {
      "epoch": 0.8286558554482603,
      "grad_norm": 5.9274243540130556e-06,
      "learning_rate": 1.717682987176406e-06,
      "loss": 0.0,
      "step": 58610
    },
    {
      "epoch": 0.8287972401702272,
      "grad_norm": 0.0,
      "learning_rate": 1.7162691399567364e-06,
      "loss": 0.0,
      "step": 58620
    },
    {
      "epoch": 0.8289386248921942,
      "grad_norm": 0.024868478998541832,
      "learning_rate": 1.714855292737067e-06,
      "loss": 0.0,
      "step": 58630
    },
    {
      "epoch": 0.8290800096141611,
      "grad_norm": 0.0,
      "learning_rate": 1.7134414455173974e-06,
      "loss": 0.0097,
      "step": 58640
    },
    {
      "epoch": 0.8292213943361281,
      "grad_norm": 0.0,
      "learning_rate": 1.712027598297728e-06,
      "loss": 0.0,
      "step": 58650
    },
    {
      "epoch": 0.829362779058095,
      "grad_norm": 9.512315045867581e-06,
      "learning_rate": 1.7106137510780587e-06,
      "loss": 0.2812,
      "step": 58660
    },
    {
      "epoch": 0.829504163780062,
      "grad_norm": 0.0,
      "learning_rate": 1.7091999038583892e-06,
      "loss": 0.0001,
      "step": 58670
    },
    {
      "epoch": 0.8296455485020289,
      "grad_norm": 0.41516923904418945,
      "learning_rate": 1.7077860566387198e-06,
      "loss": 0.0001,
      "step": 58680
    },
    {
      "epoch": 0.8297869332239958,
      "grad_norm": 0.0,
      "learning_rate": 1.7063722094190502e-06,
      "loss": 0.0,
      "step": 58690
    },
    {
      "epoch": 0.8299283179459628,
      "grad_norm": 0.0,
      "learning_rate": 1.7049583621993809e-06,
      "loss": 0.0,
      "step": 58700
    },
    {
      "epoch": 0.8300697026679297,
      "grad_norm": 0.0,
      "learning_rate": 1.7035445149797113e-06,
      "loss": 0.0,
      "step": 58710
    },
    {
      "epoch": 0.8302110873898967,
      "grad_norm": 0.0,
      "learning_rate": 1.702130667760042e-06,
      "loss": 0.0,
      "step": 58720
    },
    {
      "epoch": 0.8303524721118636,
      "grad_norm": 0.00028005815693177283,
      "learning_rate": 1.7007168205403726e-06,
      "loss": 0.0,
      "step": 58730
    },
    {
      "epoch": 0.8304938568338306,
      "grad_norm": 0.0,
      "learning_rate": 1.699302973320703e-06,
      "loss": 0.0,
      "step": 58740
    },
    {
      "epoch": 0.8306352415557975,
      "grad_norm": 0.0,
      "learning_rate": 1.6978891261010335e-06,
      "loss": 0.0,
      "step": 58750
    },
    {
      "epoch": 0.8307766262777644,
      "grad_norm": 0.0,
      "learning_rate": 1.6964752788813643e-06,
      "loss": 0.0,
      "step": 58760
    },
    {
      "epoch": 0.8309180109997314,
      "grad_norm": 0.0,
      "learning_rate": 1.6950614316616948e-06,
      "loss": 0.0,
      "step": 58770
    },
    {
      "epoch": 0.8310593957216983,
      "grad_norm": 0.0,
      "learning_rate": 1.6936475844420252e-06,
      "loss": 0.0,
      "step": 58780
    },
    {
      "epoch": 0.8312007804436653,
      "grad_norm": 0.0,
      "learning_rate": 1.692233737222356e-06,
      "loss": 0.0,
      "step": 58790
    },
    {
      "epoch": 0.8313421651656322,
      "grad_norm": 0.0,
      "learning_rate": 1.6908198900026865e-06,
      "loss": 0.0002,
      "step": 58800
    },
    {
      "epoch": 0.8314835498875991,
      "grad_norm": 0.0007372050313279033,
      "learning_rate": 1.689406042783017e-06,
      "loss": 0.0,
      "step": 58810
    },
    {
      "epoch": 0.8316249346095661,
      "grad_norm": 0.0,
      "learning_rate": 1.6879921955633474e-06,
      "loss": 0.0213,
      "step": 58820
    },
    {
      "epoch": 0.831766319331533,
      "grad_norm": 0.18817979097366333,
      "learning_rate": 1.6865783483436782e-06,
      "loss": 0.0,
      "step": 58830
    },
    {
      "epoch": 0.8319077040535,
      "grad_norm": 0.0,
      "learning_rate": 1.6851645011240086e-06,
      "loss": 0.0,
      "step": 58840
    },
    {
      "epoch": 0.8320490887754669,
      "grad_norm": 0.0,
      "learning_rate": 1.683750653904339e-06,
      "loss": 0.0,
      "step": 58850
    },
    {
      "epoch": 0.8321904734974339,
      "grad_norm": 0.0017602643929421902,
      "learning_rate": 1.68233680668467e-06,
      "loss": 0.0,
      "step": 58860
    },
    {
      "epoch": 0.8323318582194008,
      "grad_norm": 0.0,
      "learning_rate": 1.6809229594650004e-06,
      "loss": 0.0,
      "step": 58870
    },
    {
      "epoch": 0.8324732429413677,
      "grad_norm": 0.3537278473377228,
      "learning_rate": 1.6795091122453308e-06,
      "loss": 0.0,
      "step": 58880
    },
    {
      "epoch": 0.8326146276633347,
      "grad_norm": 0.000614759570453316,
      "learning_rate": 1.6780952650256617e-06,
      "loss": 0.0,
      "step": 58890
    },
    {
      "epoch": 0.8327560123853016,
      "grad_norm": 4.765448591292909e-10,
      "learning_rate": 1.676681417805992e-06,
      "loss": 0.0008,
      "step": 58900
    },
    {
      "epoch": 0.8328973971072686,
      "grad_norm": 0.0,
      "learning_rate": 1.6752675705863225e-06,
      "loss": 0.0,
      "step": 58910
    },
    {
      "epoch": 0.8330387818292355,
      "grad_norm": 0.0,
      "learning_rate": 1.6738537233666532e-06,
      "loss": 0.0,
      "step": 58920
    },
    {
      "epoch": 0.8331801665512025,
      "grad_norm": 5.437291727972138e-10,
      "learning_rate": 1.6724398761469838e-06,
      "loss": 0.0074,
      "step": 58930
    },
    {
      "epoch": 0.8333215512731694,
      "grad_norm": 0.0,
      "learning_rate": 1.6710260289273142e-06,
      "loss": 0.0,
      "step": 58940
    },
    {
      "epoch": 0.8334629359951363,
      "grad_norm": 9.920484433934007e-10,
      "learning_rate": 1.6696121817076449e-06,
      "loss": 0.0007,
      "step": 58950
    },
    {
      "epoch": 0.8336043207171033,
      "grad_norm": 0.0,
      "learning_rate": 1.6681983344879753e-06,
      "loss": 0.0,
      "step": 58960
    },
    {
      "epoch": 0.8337457054390702,
      "grad_norm": 2.4388980818912387e-05,
      "learning_rate": 1.666784487268306e-06,
      "loss": 0.0,
      "step": 58970
    },
    {
      "epoch": 0.8338870901610372,
      "grad_norm": 0.0,
      "learning_rate": 1.6653706400486364e-06,
      "loss": 0.0,
      "step": 58980
    },
    {
      "epoch": 0.8340284748830041,
      "grad_norm": 1.072339816943213e-08,
      "learning_rate": 1.663956792828967e-06,
      "loss": 0.0,
      "step": 58990
    },
    {
      "epoch": 0.8341698596049711,
      "grad_norm": 4.753006876967447e-10,
      "learning_rate": 1.6625429456092975e-06,
      "loss": 0.0,
      "step": 59000
    },
    {
      "epoch": 0.834311244326938,
      "grad_norm": 0.0,
      "learning_rate": 1.6611290983896281e-06,
      "loss": 0.0008,
      "step": 59010
    },
    {
      "epoch": 0.8344526290489049,
      "grad_norm": 0.0,
      "learning_rate": 1.6597152511699588e-06,
      "loss": 0.3408,
      "step": 59020
    },
    {
      "epoch": 0.834594013770872,
      "grad_norm": 0.0,
      "learning_rate": 1.6583014039502892e-06,
      "loss": 0.0,
      "step": 59030
    },
    {
      "epoch": 0.8347353984928388,
      "grad_norm": 2.1653517251252197e-05,
      "learning_rate": 1.6568875567306196e-06,
      "loss": 0.0,
      "step": 59040
    },
    {
      "epoch": 0.8348767832148059,
      "grad_norm": 5.074214937117461e-10,
      "learning_rate": 1.6554737095109505e-06,
      "loss": 0.0,
      "step": 59050
    },
    {
      "epoch": 0.8350181679367727,
      "grad_norm": 0.0,
      "learning_rate": 1.654059862291281e-06,
      "loss": 0.0,
      "step": 59060
    },
    {
      "epoch": 0.8351595526587396,
      "grad_norm": 0.0,
      "learning_rate": 1.6526460150716114e-06,
      "loss": 0.0,
      "step": 59070
    },
    {
      "epoch": 0.8353009373807067,
      "grad_norm": 0.0,
      "learning_rate": 1.6512321678519422e-06,
      "loss": 0.0,
      "step": 59080
    },
    {
      "epoch": 0.8354423221026736,
      "grad_norm": 9.735103958519176e-05,
      "learning_rate": 1.6498183206322726e-06,
      "loss": 0.0001,
      "step": 59090
    },
    {
      "epoch": 0.8355837068246406,
      "grad_norm": 0.0,
      "learning_rate": 1.648404473412603e-06,
      "loss": 0.0,
      "step": 59100
    },
    {
      "epoch": 0.8357250915466075,
      "grad_norm": 0.0,
      "learning_rate": 1.6469906261929335e-06,
      "loss": 0.0,
      "step": 59110
    },
    {
      "epoch": 0.8358664762685745,
      "grad_norm": 0.0,
      "learning_rate": 1.6455767789732644e-06,
      "loss": 0.0,
      "step": 59120
    },
    {
      "epoch": 0.8360078609905414,
      "grad_norm": 0.0,
      "learning_rate": 1.6441629317535948e-06,
      "loss": 0.0001,
      "step": 59130
    },
    {
      "epoch": 0.8361492457125083,
      "grad_norm": 0.0,
      "learning_rate": 1.6427490845339252e-06,
      "loss": 0.0,
      "step": 59140
    },
    {
      "epoch": 0.8362906304344753,
      "grad_norm": 4.265582331797191e-10,
      "learning_rate": 1.641335237314256e-06,
      "loss": 0.0,
      "step": 59150
    },
    {
      "epoch": 0.8364320151564422,
      "grad_norm": 0.0,
      "learning_rate": 1.6399213900945865e-06,
      "loss": 0.0,
      "step": 59160
    },
    {
      "epoch": 0.8365733998784092,
      "grad_norm": 4.806035462934233e-07,
      "learning_rate": 1.638507542874917e-06,
      "loss": 0.0,
      "step": 59170
    },
    {
      "epoch": 0.8367147846003761,
      "grad_norm": 0.0,
      "learning_rate": 1.6370936956552478e-06,
      "loss": 0.0,
      "step": 59180
    },
    {
      "epoch": 0.8368561693223431,
      "grad_norm": 0.0,
      "learning_rate": 1.6356798484355782e-06,
      "loss": 0.0004,
      "step": 59190
    },
    {
      "epoch": 0.83699755404431,
      "grad_norm": 0.0,
      "learning_rate": 1.6342660012159087e-06,
      "loss": 0.0006,
      "step": 59200
    },
    {
      "epoch": 0.8371389387662769,
      "grad_norm": 0.0,
      "learning_rate": 1.6328521539962393e-06,
      "loss": 0.0,
      "step": 59210
    },
    {
      "epoch": 0.8372803234882439,
      "grad_norm": 1.3159124456763038e-08,
      "learning_rate": 1.63143830677657e-06,
      "loss": 0.0,
      "step": 59220
    },
    {
      "epoch": 0.8374217082102108,
      "grad_norm": 0.0,
      "learning_rate": 1.6300244595569004e-06,
      "loss": 0.0,
      "step": 59230
    },
    {
      "epoch": 0.8375630929321778,
      "grad_norm": 0.0,
      "learning_rate": 1.6286106123372308e-06,
      "loss": 0.0026,
      "step": 59240
    },
    {
      "epoch": 0.8377044776541447,
      "grad_norm": 0.0,
      "learning_rate": 1.6271967651175615e-06,
      "loss": 0.5582,
      "step": 59250
    },
    {
      "epoch": 0.8378458623761116,
      "grad_norm": 7.783813593675859e-09,
      "learning_rate": 1.6257829178978921e-06,
      "loss": 0.0013,
      "step": 59260
    },
    {
      "epoch": 0.8379872470980786,
      "grad_norm": 0.0,
      "learning_rate": 1.6243690706782226e-06,
      "loss": 0.0,
      "step": 59270
    },
    {
      "epoch": 0.8381286318200455,
      "grad_norm": 0.0,
      "learning_rate": 1.6229552234585532e-06,
      "loss": 0.0,
      "step": 59280
    },
    {
      "epoch": 0.8382700165420125,
      "grad_norm": 4.10046467891334e-08,
      "learning_rate": 1.6215413762388838e-06,
      "loss": 0.0,
      "step": 59290
    },
    {
      "epoch": 0.8384114012639794,
      "grad_norm": 0.0,
      "learning_rate": 1.6201275290192143e-06,
      "loss": 0.0024,
      "step": 59300
    },
    {
      "epoch": 0.8385527859859464,
      "grad_norm": 1.1693658828735352,
      "learning_rate": 1.618713681799545e-06,
      "loss": 0.0001,
      "step": 59310
    },
    {
      "epoch": 0.8386941707079133,
      "grad_norm": 1.0156513452529907,
      "learning_rate": 1.6172998345798754e-06,
      "loss": 0.0002,
      "step": 59320
    },
    {
      "epoch": 0.8388355554298802,
      "grad_norm": 1.0114704629415883e-09,
      "learning_rate": 1.615885987360206e-06,
      "loss": 0.0,
      "step": 59330
    },
    {
      "epoch": 0.8389769401518472,
      "grad_norm": 0.0,
      "learning_rate": 1.6144721401405366e-06,
      "loss": 0.0005,
      "step": 59340
    },
    {
      "epoch": 0.8391183248738141,
      "grad_norm": 2.363014006689923e-09,
      "learning_rate": 1.613058292920867e-06,
      "loss": 0.0,
      "step": 59350
    },
    {
      "epoch": 0.8392597095957811,
      "grad_norm": 0.0,
      "learning_rate": 1.6116444457011975e-06,
      "loss": 0.0037,
      "step": 59360
    },
    {
      "epoch": 0.839401094317748,
      "grad_norm": 7.365576948359376e-08,
      "learning_rate": 1.6102305984815284e-06,
      "loss": 0.0,
      "step": 59370
    },
    {
      "epoch": 0.839542479039715,
      "grad_norm": 0.0,
      "learning_rate": 1.6088167512618588e-06,
      "loss": 0.0005,
      "step": 59380
    },
    {
      "epoch": 0.8396838637616819,
      "grad_norm": 0.0,
      "learning_rate": 1.6074029040421892e-06,
      "loss": 0.0,
      "step": 59390
    },
    {
      "epoch": 0.8398252484836488,
      "grad_norm": 0.0,
      "learning_rate": 1.6059890568225197e-06,
      "loss": 0.0,
      "step": 59400
    },
    {
      "epoch": 0.8399666332056158,
      "grad_norm": 3.2264342308044434,
      "learning_rate": 1.6045752096028505e-06,
      "loss": 0.0007,
      "step": 59410
    },
    {
      "epoch": 0.8401080179275827,
      "grad_norm": 0.0041073583997786045,
      "learning_rate": 1.603161362383181e-06,
      "loss": 0.0,
      "step": 59420
    },
    {
      "epoch": 0.8402494026495497,
      "grad_norm": 0.06583143025636673,
      "learning_rate": 1.6017475151635114e-06,
      "loss": 0.0,
      "step": 59430
    },
    {
      "epoch": 0.8403907873715166,
      "grad_norm": 4.517002050175023e-10,
      "learning_rate": 1.6003336679438422e-06,
      "loss": 0.0,
      "step": 59440
    },
    {
      "epoch": 0.8405321720934836,
      "grad_norm": 2.440379830659367e-05,
      "learning_rate": 1.5989198207241727e-06,
      "loss": 0.0,
      "step": 59450
    },
    {
      "epoch": 0.8406735568154505,
      "grad_norm": 1.3183482394651946e-07,
      "learning_rate": 1.5975059735045031e-06,
      "loss": 0.0,
      "step": 59460
    },
    {
      "epoch": 0.8408149415374174,
      "grad_norm": 0.0,
      "learning_rate": 1.596092126284834e-06,
      "loss": 0.0006,
      "step": 59470
    },
    {
      "epoch": 0.8409563262593844,
      "grad_norm": 0.0,
      "learning_rate": 1.5946782790651644e-06,
      "loss": 0.0,
      "step": 59480
    },
    {
      "epoch": 0.8410977109813513,
      "grad_norm": 0.0,
      "learning_rate": 1.5932644318454948e-06,
      "loss": 0.0,
      "step": 59490
    },
    {
      "epoch": 0.8412390957033183,
      "grad_norm": 0.0,
      "learning_rate": 1.5918505846258257e-06,
      "loss": 0.0,
      "step": 59500
    },
    {
      "epoch": 0.8413804804252852,
      "grad_norm": 33.69254684448242,
      "learning_rate": 1.5904367374061561e-06,
      "loss": 0.0039,
      "step": 59510
    },
    {
      "epoch": 0.8415218651472521,
      "grad_norm": 0.0,
      "learning_rate": 1.5890228901864866e-06,
      "loss": 0.0,
      "step": 59520
    },
    {
      "epoch": 0.8416632498692191,
      "grad_norm": 0.0,
      "learning_rate": 1.587609042966817e-06,
      "loss": 0.0,
      "step": 59530
    },
    {
      "epoch": 0.841804634591186,
      "grad_norm": 0.0,
      "learning_rate": 1.5861951957471478e-06,
      "loss": 0.0,
      "step": 59540
    },
    {
      "epoch": 0.841946019313153,
      "grad_norm": 0.0,
      "learning_rate": 1.5847813485274783e-06,
      "loss": 0.0,
      "step": 59550
    },
    {
      "epoch": 0.84208740403512,
      "grad_norm": 3.435683026964398e-08,
      "learning_rate": 1.5833675013078087e-06,
      "loss": 0.0,
      "step": 59560
    },
    {
      "epoch": 0.842228788757087,
      "grad_norm": 1.974430396955995e-09,
      "learning_rate": 1.5820950388101064e-06,
      "loss": 0.2696,
      "step": 59570
    },
    {
      "epoch": 0.8423701734790539,
      "grad_norm": 0.0,
      "learning_rate": 1.5806811915904368e-06,
      "loss": 0.0,
      "step": 59580
    },
    {
      "epoch": 0.8425115582010207,
      "grad_norm": 0.3585696816444397,
      "learning_rate": 1.5792673443707673e-06,
      "loss": 0.0001,
      "step": 59590
    },
    {
      "epoch": 0.8426529429229878,
      "grad_norm": 0.0,
      "learning_rate": 1.5778534971510981e-06,
      "loss": 0.0,
      "step": 59600
    },
    {
      "epoch": 0.8427943276449547,
      "grad_norm": 0.0,
      "learning_rate": 1.5764396499314286e-06,
      "loss": 0.0,
      "step": 59610
    },
    {
      "epoch": 0.8429357123669217,
      "grad_norm": 7.426008892252867e-07,
      "learning_rate": 1.575025802711759e-06,
      "loss": 0.0172,
      "step": 59620
    },
    {
      "epoch": 0.8430770970888886,
      "grad_norm": 0.0,
      "learning_rate": 1.5736119554920898e-06,
      "loss": 0.0,
      "step": 59630
    },
    {
      "epoch": 0.8432184818108556,
      "grad_norm": 0.0,
      "learning_rate": 1.5721981082724203e-06,
      "loss": 1.1367,
      "step": 59640
    },
    {
      "epoch": 0.8433598665328225,
      "grad_norm": 0.002290101954713464,
      "learning_rate": 1.5707842610527507e-06,
      "loss": 0.0,
      "step": 59650
    },
    {
      "epoch": 0.8435012512547894,
      "grad_norm": 0.0,
      "learning_rate": 1.5693704138330811e-06,
      "loss": 0.0,
      "step": 59660
    },
    {
      "epoch": 0.8436426359767564,
      "grad_norm": 6.41602400719421e-06,
      "learning_rate": 1.567956566613412e-06,
      "loss": 0.0,
      "step": 59670
    },
    {
      "epoch": 0.8437840206987233,
      "grad_norm": 0.0,
      "learning_rate": 1.5665427193937424e-06,
      "loss": 0.0,
      "step": 59680
    },
    {
      "epoch": 0.8439254054206903,
      "grad_norm": 0.0,
      "learning_rate": 1.5651288721740729e-06,
      "loss": 0.0,
      "step": 59690
    },
    {
      "epoch": 0.8440667901426572,
      "grad_norm": 0.0,
      "learning_rate": 1.5637150249544037e-06,
      "loss": 0.0,
      "step": 59700
    },
    {
      "epoch": 0.8442081748646242,
      "grad_norm": 310.57861328125,
      "learning_rate": 1.5623011777347342e-06,
      "loss": 0.0514,
      "step": 59710
    },
    {
      "epoch": 0.8443495595865911,
      "grad_norm": 0.0,
      "learning_rate": 1.5608873305150646e-06,
      "loss": 0.0,
      "step": 59720
    },
    {
      "epoch": 0.844490944308558,
      "grad_norm": 0.0,
      "learning_rate": 1.5594734832953952e-06,
      "loss": 0.0,
      "step": 59730
    },
    {
      "epoch": 0.844632329030525,
      "grad_norm": 9.098541340790689e-05,
      "learning_rate": 1.5580596360757259e-06,
      "loss": 0.0,
      "step": 59740
    },
    {
      "epoch": 0.8447737137524919,
      "grad_norm": 0.0001453860168112442,
      "learning_rate": 1.5566457888560563e-06,
      "loss": 0.6793,
      "step": 59750
    },
    {
      "epoch": 0.8449150984744589,
      "grad_norm": 0.0,
      "learning_rate": 1.555231941636387e-06,
      "loss": 0.0003,
      "step": 59760
    },
    {
      "epoch": 0.8450564831964258,
      "grad_norm": 0.0,
      "learning_rate": 1.5538180944167174e-06,
      "loss": 0.0,
      "step": 59770
    },
    {
      "epoch": 0.8451978679183927,
      "grad_norm": 0.0,
      "learning_rate": 1.552404247197048e-06,
      "loss": 0.0,
      "step": 59780
    },
    {
      "epoch": 0.8453392526403597,
      "grad_norm": 0.0,
      "learning_rate": 1.5509903999773785e-06,
      "loss": 0.0,
      "step": 59790
    },
    {
      "epoch": 0.8454806373623266,
      "grad_norm": 0.0,
      "learning_rate": 1.5495765527577091e-06,
      "loss": 0.0,
      "step": 59800
    },
    {
      "epoch": 0.8456220220842936,
      "grad_norm": 0.0003330135368742049,
      "learning_rate": 1.5481627055380398e-06,
      "loss": 0.0031,
      "step": 59810
    },
    {
      "epoch": 0.8457634068062605,
      "grad_norm": 0.0,
      "learning_rate": 1.5467488583183702e-06,
      "loss": 0.0,
      "step": 59820
    },
    {
      "epoch": 0.8459047915282275,
      "grad_norm": 0.0,
      "learning_rate": 1.5453350110987008e-06,
      "loss": 0.0,
      "step": 59830
    },
    {
      "epoch": 0.8460461762501944,
      "grad_norm": 9.283539492699333e-10,
      "learning_rate": 1.5439211638790313e-06,
      "loss": 0.0,
      "step": 59840
    },
    {
      "epoch": 0.8461875609721613,
      "grad_norm": 1.8560021999292076e-05,
      "learning_rate": 1.542507316659362e-06,
      "loss": 0.0,
      "step": 59850
    },
    {
      "epoch": 0.8463289456941283,
      "grad_norm": 0.632575511932373,
      "learning_rate": 1.5410934694396926e-06,
      "loss": 0.0003,
      "step": 59860
    },
    {
      "epoch": 0.8464703304160952,
      "grad_norm": 0.0,
      "learning_rate": 1.539679622220023e-06,
      "loss": 0.0001,
      "step": 59870
    },
    {
      "epoch": 0.8466117151380622,
      "grad_norm": 0.0,
      "learning_rate": 1.5382657750003534e-06,
      "loss": 0.0039,
      "step": 59880
    },
    {
      "epoch": 0.8467530998600291,
      "grad_norm": 0.0,
      "learning_rate": 1.5368519277806843e-06,
      "loss": 0.0,
      "step": 59890
    },
    {
      "epoch": 0.8468944845819961,
      "grad_norm": 0.0007768034702166915,
      "learning_rate": 1.5354380805610147e-06,
      "loss": 0.0,
      "step": 59900
    },
    {
      "epoch": 0.847035869303963,
      "grad_norm": 0.0,
      "learning_rate": 1.5340242333413451e-06,
      "loss": 0.402,
      "step": 59910
    },
    {
      "epoch": 0.8471772540259299,
      "grad_norm": 0.0,
      "learning_rate": 1.532610386121676e-06,
      "loss": 0.0041,
      "step": 59920
    },
    {
      "epoch": 0.8473186387478969,
      "grad_norm": 0.12937107682228088,
      "learning_rate": 1.5311965389020064e-06,
      "loss": 0.0,
      "step": 59930
    },
    {
      "epoch": 0.8474600234698638,
      "grad_norm": 1.8310369886975764e-09,
      "learning_rate": 1.5297826916823369e-06,
      "loss": 0.0,
      "step": 59940
    },
    {
      "epoch": 0.8476014081918308,
      "grad_norm": 11.345158576965332,
      "learning_rate": 1.5283688444626673e-06,
      "loss": 0.0012,
      "step": 59950
    },
    {
      "epoch": 0.8477427929137977,
      "grad_norm": 1.124482995606968e-08,
      "learning_rate": 1.5269549972429982e-06,
      "loss": 0.0001,
      "step": 59960
    },
    {
      "epoch": 0.8478841776357646,
      "grad_norm": 0.0,
      "learning_rate": 1.5255411500233286e-06,
      "loss": 0.0,
      "step": 59970
    },
    {
      "epoch": 0.8480255623577316,
      "grad_norm": 0.0,
      "learning_rate": 1.524127302803659e-06,
      "loss": 0.0,
      "step": 59980
    },
    {
      "epoch": 0.8481669470796985,
      "grad_norm": 3.7536077002187085e-07,
      "learning_rate": 1.5227134555839899e-06,
      "loss": 0.0,
      "step": 59990
    },
    {
      "epoch": 0.8483083318016655,
      "grad_norm": 0.0,
      "learning_rate": 1.5212996083643203e-06,
      "loss": 0.0,
      "step": 60000
    },
    {
      "epoch": 0.8484497165236324,
      "grad_norm": 0.0,
      "learning_rate": 1.5198857611446507e-06,
      "loss": 0.0,
      "step": 60010
    },
    {
      "epoch": 0.8485911012455994,
      "grad_norm": 7.267588080139831e-05,
      "learning_rate": 1.5184719139249816e-06,
      "loss": 0.0,
      "step": 60020
    },
    {
      "epoch": 0.8487324859675663,
      "grad_norm": 0.0,
      "learning_rate": 1.517058066705312e-06,
      "loss": 0.0,
      "step": 60030
    },
    {
      "epoch": 0.8488738706895332,
      "grad_norm": 0.0,
      "learning_rate": 1.5156442194856425e-06,
      "loss": 0.0031,
      "step": 60040
    },
    {
      "epoch": 0.8490152554115002,
      "grad_norm": 0.0,
      "learning_rate": 1.5142303722659731e-06,
      "loss": 0.0001,
      "step": 60050
    },
    {
      "epoch": 0.8491566401334671,
      "grad_norm": 0.0,
      "learning_rate": 1.5128165250463038e-06,
      "loss": 0.0,
      "step": 60060
    },
    {
      "epoch": 0.8492980248554342,
      "grad_norm": 0.0,
      "learning_rate": 1.5114026778266342e-06,
      "loss": 0.0,
      "step": 60070
    },
    {
      "epoch": 0.849439409577401,
      "grad_norm": 0.003410595702007413,
      "learning_rate": 1.5099888306069646e-06,
      "loss": 0.0,
      "step": 60080
    },
    {
      "epoch": 0.8495807942993681,
      "grad_norm": 9.087188664125279e-05,
      "learning_rate": 1.5085749833872953e-06,
      "loss": 0.0,
      "step": 60090
    },
    {
      "epoch": 0.849722179021335,
      "grad_norm": 0.0,
      "learning_rate": 1.507161136167626e-06,
      "loss": 0.0,
      "step": 60100
    },
    {
      "epoch": 0.8498635637433019,
      "grad_norm": 0.0,
      "learning_rate": 1.5057472889479563e-06,
      "loss": 0.0006,
      "step": 60110
    },
    {
      "epoch": 0.8500049484652689,
      "grad_norm": 0.0,
      "learning_rate": 1.504333441728287e-06,
      "loss": 0.0,
      "step": 60120
    },
    {
      "epoch": 0.8501463331872358,
      "grad_norm": 0.0,
      "learning_rate": 1.5029195945086174e-06,
      "loss": 0.0,
      "step": 60130
    },
    {
      "epoch": 0.8502877179092028,
      "grad_norm": 0.0,
      "learning_rate": 1.501505747288948e-06,
      "loss": 0.0,
      "step": 60140
    },
    {
      "epoch": 0.8504291026311697,
      "grad_norm": 7.092523446772248e-05,
      "learning_rate": 1.5000919000692787e-06,
      "loss": 0.0,
      "step": 60150
    },
    {
      "epoch": 0.8505704873531367,
      "grad_norm": 9.560123803709075e-10,
      "learning_rate": 1.4986780528496091e-06,
      "loss": 0.026,
      "step": 60160
    },
    {
      "epoch": 0.8507118720751036,
      "grad_norm": 0.022457847371697426,
      "learning_rate": 1.4972642056299398e-06,
      "loss": 0.0003,
      "step": 60170
    },
    {
      "epoch": 0.8508532567970705,
      "grad_norm": 0.7309329509735107,
      "learning_rate": 1.4958503584102704e-06,
      "loss": 0.1992,
      "step": 60180
    },
    {
      "epoch": 0.8509946415190375,
      "grad_norm": 0.0,
      "learning_rate": 1.4944365111906009e-06,
      "loss": 0.2293,
      "step": 60190
    },
    {
      "epoch": 0.8511360262410044,
      "grad_norm": 6.983079856581753e-07,
      "learning_rate": 1.4930226639709313e-06,
      "loss": 0.0,
      "step": 60200
    },
    {
      "epoch": 0.8512774109629714,
      "grad_norm": 8.704988431418315e-06,
      "learning_rate": 1.491608816751262e-06,
      "loss": 0.0003,
      "step": 60210
    },
    {
      "epoch": 0.8514187956849383,
      "grad_norm": 0.0,
      "learning_rate": 1.4901949695315926e-06,
      "loss": 0.0,
      "step": 60220
    },
    {
      "epoch": 0.8515601804069052,
      "grad_norm": 0.0,
      "learning_rate": 1.488781122311923e-06,
      "loss": 0.0,
      "step": 60230
    },
    {
      "epoch": 0.8517015651288722,
      "grad_norm": 0.0,
      "learning_rate": 1.4873672750922535e-06,
      "loss": 0.0,
      "step": 60240
    },
    {
      "epoch": 0.8518429498508391,
      "grad_norm": 0.00010547247802605852,
      "learning_rate": 1.4859534278725843e-06,
      "loss": 0.0,
      "step": 60250
    },
    {
      "epoch": 0.8519843345728061,
      "grad_norm": 0.0,
      "learning_rate": 1.4845395806529147e-06,
      "loss": 0.0,
      "step": 60260
    },
    {
      "epoch": 0.852125719294773,
      "grad_norm": 0.17988795042037964,
      "learning_rate": 1.4831257334332452e-06,
      "loss": 0.0001,
      "step": 60270
    },
    {
      "epoch": 0.85226710401674,
      "grad_norm": 0.0,
      "learning_rate": 1.481711886213576e-06,
      "loss": 0.0,
      "step": 60280
    },
    {
      "epoch": 0.8524084887387069,
      "grad_norm": 0.0,
      "learning_rate": 1.4802980389939065e-06,
      "loss": 0.0,
      "step": 60290
    },
    {
      "epoch": 0.8525498734606738,
      "grad_norm": 0.0,
      "learning_rate": 1.478884191774237e-06,
      "loss": 0.0,
      "step": 60300
    },
    {
      "epoch": 0.8526912581826408,
      "grad_norm": 0.00013275191304273903,
      "learning_rate": 1.4774703445545678e-06,
      "loss": 0.0001,
      "step": 60310
    },
    {
      "epoch": 0.8528326429046077,
      "grad_norm": 0.0,
      "learning_rate": 1.4760564973348982e-06,
      "loss": 0.0,
      "step": 60320
    },
    {
      "epoch": 0.8529740276265747,
      "grad_norm": 1.4538008144882042e-05,
      "learning_rate": 1.4746426501152286e-06,
      "loss": 0.568,
      "step": 60330
    },
    {
      "epoch": 0.8531154123485416,
      "grad_norm": 0.0,
      "learning_rate": 1.4732288028955593e-06,
      "loss": 0.0,
      "step": 60340
    },
    {
      "epoch": 0.8532567970705086,
      "grad_norm": 0.0,
      "learning_rate": 1.47181495567589e-06,
      "loss": 0.0,
      "step": 60350
    },
    {
      "epoch": 0.8533981817924755,
      "grad_norm": 0.0,
      "learning_rate": 1.4704011084562203e-06,
      "loss": 0.0,
      "step": 60360
    },
    {
      "epoch": 0.8535395665144424,
      "grad_norm": 9.858236444415525e-05,
      "learning_rate": 1.4689872612365508e-06,
      "loss": 0.0,
      "step": 60370
    },
    {
      "epoch": 0.8536809512364094,
      "grad_norm": 7.458527306880569e-07,
      "learning_rate": 1.4675734140168814e-06,
      "loss": 0.0,
      "step": 60380
    },
    {
      "epoch": 0.8538223359583763,
      "grad_norm": 9.592390215473756e-10,
      "learning_rate": 1.466159566797212e-06,
      "loss": 0.0,
      "step": 60390
    },
    {
      "epoch": 0.8539637206803433,
      "grad_norm": 0.0,
      "learning_rate": 1.4647457195775425e-06,
      "loss": 0.0,
      "step": 60400
    },
    {
      "epoch": 0.8541051054023102,
      "grad_norm": 0.0,
      "learning_rate": 1.4633318723578731e-06,
      "loss": 0.0,
      "step": 60410
    },
    {
      "epoch": 0.8542464901242772,
      "grad_norm": 8.495126166963018e-06,
      "learning_rate": 1.4619180251382038e-06,
      "loss": 0.0,
      "step": 60420
    },
    {
      "epoch": 0.8543878748462441,
      "grad_norm": 0.2793201208114624,
      "learning_rate": 1.4605041779185342e-06,
      "loss": 0.0002,
      "step": 60430
    },
    {
      "epoch": 0.854529259568211,
      "grad_norm": 0.0,
      "learning_rate": 1.4590903306988649e-06,
      "loss": 0.0,
      "step": 60440
    },
    {
      "epoch": 0.854670644290178,
      "grad_norm": 0.0,
      "learning_rate": 1.4576764834791953e-06,
      "loss": 0.011,
      "step": 60450
    },
    {
      "epoch": 0.8548120290121449,
      "grad_norm": 0.0,
      "learning_rate": 1.456262636259526e-06,
      "loss": 0.0,
      "step": 60460
    },
    {
      "epoch": 0.8549534137341119,
      "grad_norm": 0.0,
      "learning_rate": 1.4548487890398566e-06,
      "loss": 0.0,
      "step": 60470
    },
    {
      "epoch": 0.8550947984560788,
      "grad_norm": 0.0,
      "learning_rate": 1.453434941820187e-06,
      "loss": 0.0001,
      "step": 60480
    },
    {
      "epoch": 0.8552361831780457,
      "grad_norm": 0.0,
      "learning_rate": 1.4520210946005175e-06,
      "loss": 0.0001,
      "step": 60490
    },
    {
      "epoch": 0.8553775679000127,
      "grad_norm": 0.0,
      "learning_rate": 1.450607247380848e-06,
      "loss": 0.0,
      "step": 60500
    },
    {
      "epoch": 0.8555189526219796,
      "grad_norm": 0.0002708866377361119,
      "learning_rate": 1.4491934001611787e-06,
      "loss": 0.0,
      "step": 60510
    },
    {
      "epoch": 0.8556603373439466,
      "grad_norm": 0.0005744308000430465,
      "learning_rate": 1.4477795529415092e-06,
      "loss": 0.0,
      "step": 60520
    },
    {
      "epoch": 0.8558017220659135,
      "grad_norm": 1.3612275123596191,
      "learning_rate": 1.4463657057218396e-06,
      "loss": 0.0003,
      "step": 60530
    },
    {
      "epoch": 0.8559431067878805,
      "grad_norm": 0.0,
      "learning_rate": 1.4449518585021705e-06,
      "loss": 0.0,
      "step": 60540
    },
    {
      "epoch": 0.8560844915098474,
      "grad_norm": 0.0,
      "learning_rate": 1.443538011282501e-06,
      "loss": 0.2096,
      "step": 60550
    },
    {
      "epoch": 0.8562258762318143,
      "grad_norm": 0.0,
      "learning_rate": 1.4421241640628313e-06,
      "loss": 0.0,
      "step": 60560
    },
    {
      "epoch": 0.8563672609537814,
      "grad_norm": 1.336435252596857e-05,
      "learning_rate": 1.4407103168431622e-06,
      "loss": 0.0151,
      "step": 60570
    },
    {
      "epoch": 0.8565086456757482,
      "grad_norm": 0.0,
      "learning_rate": 1.4392964696234926e-06,
      "loss": 1.3469,
      "step": 60580
    },
    {
      "epoch": 0.8566500303977153,
      "grad_norm": 0.00021095889678690583,
      "learning_rate": 1.437882622403823e-06,
      "loss": 0.1006,
      "step": 60590
    },
    {
      "epoch": 0.8567914151196822,
      "grad_norm": 0.0,
      "learning_rate": 1.436468775184154e-06,
      "loss": 0.0,
      "step": 60600
    },
    {
      "epoch": 0.8569327998416492,
      "grad_norm": 0.0,
      "learning_rate": 1.4350549279644843e-06,
      "loss": 0.0024,
      "step": 60610
    },
    {
      "epoch": 0.8570741845636161,
      "grad_norm": 0.0,
      "learning_rate": 1.4336410807448148e-06,
      "loss": 0.0,
      "step": 60620
    },
    {
      "epoch": 0.857215569285583,
      "grad_norm": 0.0,
      "learning_rate": 1.4322272335251452e-06,
      "loss": 0.0,
      "step": 60630
    },
    {
      "epoch": 0.85735695400755,
      "grad_norm": 0.0,
      "learning_rate": 1.430813386305476e-06,
      "loss": 0.0074,
      "step": 60640
    },
    {
      "epoch": 0.8574983387295169,
      "grad_norm": 4.127262087649797e-08,
      "learning_rate": 1.4293995390858065e-06,
      "loss": 0.0,
      "step": 60650
    },
    {
      "epoch": 0.8576397234514839,
      "grad_norm": 0.0,
      "learning_rate": 1.427985691866137e-06,
      "loss": 0.0489,
      "step": 60660
    },
    {
      "epoch": 0.8577811081734508,
      "grad_norm": 2.5054887373698875e-05,
      "learning_rate": 1.4265718446464678e-06,
      "loss": 0.0028,
      "step": 60670
    },
    {
      "epoch": 0.8579224928954178,
      "grad_norm": 0.0,
      "learning_rate": 1.4251579974267982e-06,
      "loss": 0.0,
      "step": 60680
    },
    {
      "epoch": 0.8580638776173847,
      "grad_norm": 0.0,
      "learning_rate": 1.4237441502071287e-06,
      "loss": 0.0,
      "step": 60690
    },
    {
      "epoch": 0.8582052623393516,
      "grad_norm": 0.0,
      "learning_rate": 1.4223303029874593e-06,
      "loss": 0.0,
      "step": 60700
    },
    {
      "epoch": 0.8583466470613186,
      "grad_norm": 0.0,
      "learning_rate": 1.42091645576779e-06,
      "loss": 0.0,
      "step": 60710
    },
    {
      "epoch": 0.8584880317832855,
      "grad_norm": 0.0,
      "learning_rate": 1.4195026085481204e-06,
      "loss": 0.0,
      "step": 60720
    },
    {
      "epoch": 0.8586294165052525,
      "grad_norm": 0.0,
      "learning_rate": 1.418088761328451e-06,
      "loss": 0.6176,
      "step": 60730
    },
    {
      "epoch": 0.8587708012272194,
      "grad_norm": 6.115677388152108e-05,
      "learning_rate": 1.4166749141087815e-06,
      "loss": 0.0,
      "step": 60740
    },
    {
      "epoch": 0.8589121859491863,
      "grad_norm": 4.228827510388555e-09,
      "learning_rate": 1.415261066889112e-06,
      "loss": 0.027,
      "step": 60750
    },
    {
      "epoch": 0.8590535706711533,
      "grad_norm": 0.0,
      "learning_rate": 1.4138472196694427e-06,
      "loss": 0.0,
      "step": 60760
    },
    {
      "epoch": 0.8591949553931202,
      "grad_norm": 0.0,
      "learning_rate": 1.4124333724497732e-06,
      "loss": 0.0486,
      "step": 60770
    },
    {
      "epoch": 0.8593363401150872,
      "grad_norm": 214.9087371826172,
      "learning_rate": 1.4110195252301036e-06,
      "loss": 0.0536,
      "step": 60780
    },
    {
      "epoch": 0.8594777248370541,
      "grad_norm": 0.0,
      "learning_rate": 1.4096056780104343e-06,
      "loss": 0.0,
      "step": 60790
    },
    {
      "epoch": 0.8596191095590211,
      "grad_norm": 3.954736484956811e-07,
      "learning_rate": 1.408191830790765e-06,
      "loss": 0.0,
      "step": 60800
    },
    {
      "epoch": 0.859760494280988,
      "grad_norm": 0.0,
      "learning_rate": 1.4067779835710953e-06,
      "loss": 0.0007,
      "step": 60810
    },
    {
      "epoch": 0.8599018790029549,
      "grad_norm": 0.0,
      "learning_rate": 1.405364136351426e-06,
      "loss": 0.0,
      "step": 60820
    },
    {
      "epoch": 0.8600432637249219,
      "grad_norm": 0.0,
      "learning_rate": 1.4039502891317566e-06,
      "loss": 0.0,
      "step": 60830
    },
    {
      "epoch": 0.8601846484468888,
      "grad_norm": 0.0,
      "learning_rate": 1.402536441912087e-06,
      "loss": 0.0,
      "step": 60840
    },
    {
      "epoch": 0.8603260331688558,
      "grad_norm": 0.0,
      "learning_rate": 1.4011225946924175e-06,
      "loss": 0.0002,
      "step": 60850
    },
    {
      "epoch": 0.8604674178908227,
      "grad_norm": 7.646115409443155e-05,
      "learning_rate": 1.3997087474727483e-06,
      "loss": 0.0001,
      "step": 60860
    },
    {
      "epoch": 0.8606088026127897,
      "grad_norm": 0.0,
      "learning_rate": 1.3982949002530788e-06,
      "loss": 0.0001,
      "step": 60870
    },
    {
      "epoch": 0.8607501873347566,
      "grad_norm": 0.00017808692064136267,
      "learning_rate": 1.3968810530334092e-06,
      "loss": 0.0,
      "step": 60880
    },
    {
      "epoch": 0.8608915720567235,
      "grad_norm": 0.00528461579233408,
      "learning_rate": 1.39546720581374e-06,
      "loss": 0.0,
      "step": 60890
    },
    {
      "epoch": 0.8610329567786905,
      "grad_norm": 0.0,
      "learning_rate": 1.3940533585940705e-06,
      "loss": 0.0,
      "step": 60900
    },
    {
      "epoch": 0.8611743415006574,
      "grad_norm": 0.0,
      "learning_rate": 1.392639511374401e-06,
      "loss": 0.0,
      "step": 60910
    },
    {
      "epoch": 0.8613157262226244,
      "grad_norm": 0.0,
      "learning_rate": 1.3912256641547314e-06,
      "loss": 0.0,
      "step": 60920
    },
    {
      "epoch": 0.8614571109445913,
      "grad_norm": 0.0,
      "learning_rate": 1.3898118169350622e-06,
      "loss": 0.0,
      "step": 60930
    },
    {
      "epoch": 0.8615984956665582,
      "grad_norm": 0.006928971502929926,
      "learning_rate": 1.3883979697153927e-06,
      "loss": 0.0,
      "step": 60940
    },
    {
      "epoch": 0.8617398803885252,
      "grad_norm": 0.0,
      "learning_rate": 1.386984122495723e-06,
      "loss": 0.0,
      "step": 60950
    },
    {
      "epoch": 0.8618812651104921,
      "grad_norm": 0.0,
      "learning_rate": 1.385570275276054e-06,
      "loss": 0.0,
      "step": 60960
    },
    {
      "epoch": 0.8620226498324591,
      "grad_norm": 0.0,
      "learning_rate": 1.3841564280563844e-06,
      "loss": 0.0,
      "step": 60970
    },
    {
      "epoch": 0.862164034554426,
      "grad_norm": 0.0,
      "learning_rate": 1.3827425808367148e-06,
      "loss": 0.0004,
      "step": 60980
    },
    {
      "epoch": 0.862305419276393,
      "grad_norm": 2.715669324970804e-06,
      "learning_rate": 1.3813287336170455e-06,
      "loss": 0.0,
      "step": 60990
    },
    {
      "epoch": 0.8624468039983599,
      "grad_norm": 0.0,
      "learning_rate": 1.379914886397376e-06,
      "loss": 0.0001,
      "step": 61000
    },
    {
      "epoch": 0.8625881887203268,
      "grad_norm": 0.0,
      "learning_rate": 1.3785010391777065e-06,
      "loss": 0.0006,
      "step": 61010
    },
    {
      "epoch": 0.8627295734422938,
      "grad_norm": 0.0,
      "learning_rate": 1.3770871919580372e-06,
      "loss": 0.0,
      "step": 61020
    },
    {
      "epoch": 0.8628709581642607,
      "grad_norm": 0.0,
      "learning_rate": 1.3756733447383678e-06,
      "loss": 0.0002,
      "step": 61030
    },
    {
      "epoch": 0.8630123428862277,
      "grad_norm": 5.70763086216175e-06,
      "learning_rate": 1.3742594975186983e-06,
      "loss": 0.0,
      "step": 61040
    },
    {
      "epoch": 0.8631537276081946,
      "grad_norm": 0.0014372093137353659,
      "learning_rate": 1.3728456502990287e-06,
      "loss": 0.0,
      "step": 61050
    },
    {
      "epoch": 0.8632951123301617,
      "grad_norm": 0.0,
      "learning_rate": 1.3714318030793593e-06,
      "loss": 0.0,
      "step": 61060
    },
    {
      "epoch": 0.8634364970521285,
      "grad_norm": 0.0,
      "learning_rate": 1.37001795585969e-06,
      "loss": 0.0,
      "step": 61070
    },
    {
      "epoch": 0.8635778817740954,
      "grad_norm": 0.0,
      "learning_rate": 1.3686041086400204e-06,
      "loss": 0.0,
      "step": 61080
    },
    {
      "epoch": 0.8637192664960625,
      "grad_norm": 0.0,
      "learning_rate": 1.367190261420351e-06,
      "loss": 0.0,
      "step": 61090
    },
    {
      "epoch": 0.8638606512180294,
      "grad_norm": 0.0,
      "learning_rate": 1.3657764142006815e-06,
      "loss": 0.0,
      "step": 61100
    },
    {
      "epoch": 0.8640020359399964,
      "grad_norm": 0.0002709691470954567,
      "learning_rate": 1.3643625669810121e-06,
      "loss": 0.0009,
      "step": 61110
    },
    {
      "epoch": 0.8641434206619633,
      "grad_norm": 0.0,
      "learning_rate": 1.3629487197613428e-06,
      "loss": 0.0,
      "step": 61120
    },
    {
      "epoch": 0.8642848053839303,
      "grad_norm": 0.0,
      "learning_rate": 1.3615348725416732e-06,
      "loss": 0.0,
      "step": 61130
    },
    {
      "epoch": 0.8644261901058972,
      "grad_norm": 0.0,
      "learning_rate": 1.3601210253220036e-06,
      "loss": 0.0003,
      "step": 61140
    },
    {
      "epoch": 0.8645675748278641,
      "grad_norm": 0.0,
      "learning_rate": 1.3587071781023345e-06,
      "loss": 0.0,
      "step": 61150
    },
    {
      "epoch": 0.8647089595498311,
      "grad_norm": 0.0,
      "learning_rate": 1.357293330882665e-06,
      "loss": 0.0002,
      "step": 61160
    },
    {
      "epoch": 0.864850344271798,
      "grad_norm": 0.0,
      "learning_rate": 1.3558794836629954e-06,
      "loss": 0.0,
      "step": 61170
    },
    {
      "epoch": 0.864991728993765,
      "grad_norm": 0.0,
      "learning_rate": 1.3544656364433262e-06,
      "loss": 0.0,
      "step": 61180
    },
    {
      "epoch": 0.8651331137157319,
      "grad_norm": 0.0,
      "learning_rate": 1.3530517892236567e-06,
      "loss": 0.0,
      "step": 61190
    },
    {
      "epoch": 0.8652744984376988,
      "grad_norm": 0.0,
      "learning_rate": 1.351637942003987e-06,
      "loss": 0.0,
      "step": 61200
    },
    {
      "epoch": 0.8654158831596658,
      "grad_norm": 9.209639717511209e-10,
      "learning_rate": 1.3502240947843175e-06,
      "loss": 0.0001,
      "step": 61210
    },
    {
      "epoch": 0.8655572678816327,
      "grad_norm": 4.407429969432997e-06,
      "learning_rate": 1.3488102475646484e-06,
      "loss": 0.0259,
      "step": 61220
    },
    {
      "epoch": 0.8656986526035997,
      "grad_norm": 0.5683178901672363,
      "learning_rate": 1.3473964003449788e-06,
      "loss": 0.0001,
      "step": 61230
    },
    {
      "epoch": 0.8658400373255666,
      "grad_norm": 4.829787485505221e-07,
      "learning_rate": 1.3459825531253092e-06,
      "loss": 0.0,
      "step": 61240
    },
    {
      "epoch": 0.8659814220475336,
      "grad_norm": 0.0,
      "learning_rate": 1.34456870590564e-06,
      "loss": 0.0003,
      "step": 61250
    },
    {
      "epoch": 0.8661228067695005,
      "grad_norm": 0.0,
      "learning_rate": 1.3431548586859705e-06,
      "loss": 0.0,
      "step": 61260
    },
    {
      "epoch": 0.8662641914914674,
      "grad_norm": 0.0,
      "learning_rate": 1.341741011466301e-06,
      "loss": 0.0,
      "step": 61270
    },
    {
      "epoch": 0.8664055762134344,
      "grad_norm": 0.0,
      "learning_rate": 1.3403271642466318e-06,
      "loss": 0.0,
      "step": 61280
    },
    {
      "epoch": 0.8665469609354013,
      "grad_norm": 0.0,
      "learning_rate": 1.3389133170269623e-06,
      "loss": 0.0,
      "step": 61290
    },
    {
      "epoch": 0.8666883456573683,
      "grad_norm": 0.0,
      "learning_rate": 1.3374994698072927e-06,
      "loss": 0.0042,
      "step": 61300
    },
    {
      "epoch": 0.8668297303793352,
      "grad_norm": 0.0,
      "learning_rate": 1.3360856225876233e-06,
      "loss": 0.0001,
      "step": 61310
    },
    {
      "epoch": 0.8669711151013022,
      "grad_norm": 0.0,
      "learning_rate": 1.334671775367954e-06,
      "loss": 0.0,
      "step": 61320
    },
    {
      "epoch": 0.8671124998232691,
      "grad_norm": 0.0,
      "learning_rate": 1.3332579281482844e-06,
      "loss": 0.0,
      "step": 61330
    },
    {
      "epoch": 0.867253884545236,
      "grad_norm": 0.0,
      "learning_rate": 1.3318440809286148e-06,
      "loss": 0.0,
      "step": 61340
    },
    {
      "epoch": 0.867395269267203,
      "grad_norm": 9.360888730824968e-10,
      "learning_rate": 1.3304302337089455e-06,
      "loss": 0.0,
      "step": 61350
    },
    {
      "epoch": 0.8675366539891699,
      "grad_norm": 0.0,
      "learning_rate": 1.3290163864892761e-06,
      "loss": 0.0,
      "step": 61360
    },
    {
      "epoch": 0.8676780387111369,
      "grad_norm": 3.9617825677851215e-06,
      "learning_rate": 1.3276025392696066e-06,
      "loss": 0.0,
      "step": 61370
    },
    {
      "epoch": 0.8678194234331038,
      "grad_norm": 0.0,
      "learning_rate": 1.3261886920499372e-06,
      "loss": 0.0,
      "step": 61380
    },
    {
      "epoch": 0.8679608081550708,
      "grad_norm": 9.170575410166748e-10,
      "learning_rate": 1.3247748448302676e-06,
      "loss": 0.0,
      "step": 61390
    },
    {
      "epoch": 0.8681021928770377,
      "grad_norm": 0.0,
      "learning_rate": 1.3233609976105983e-06,
      "loss": 0.0,
      "step": 61400
    },
    {
      "epoch": 0.8682435775990046,
      "grad_norm": 0.0,
      "learning_rate": 1.321947150390929e-06,
      "loss": 0.0,
      "step": 61410
    },
    {
      "epoch": 0.8683849623209716,
      "grad_norm": 8.67865446174676e-10,
      "learning_rate": 1.3205333031712594e-06,
      "loss": 0.0,
      "step": 61420
    },
    {
      "epoch": 0.8685263470429385,
      "grad_norm": 9.100597497813112e-10,
      "learning_rate": 1.31911945595159e-06,
      "loss": 0.0457,
      "step": 61430
    },
    {
      "epoch": 0.8686677317649055,
      "grad_norm": 1.1730968196843605e-07,
      "learning_rate": 1.3177056087319207e-06,
      "loss": 0.0,
      "step": 61440
    },
    {
      "epoch": 0.8688091164868724,
      "grad_norm": 0.0,
      "learning_rate": 1.316291761512251e-06,
      "loss": 0.0,
      "step": 61450
    },
    {
      "epoch": 0.8689505012088393,
      "grad_norm": 0.0,
      "learning_rate": 1.3148779142925815e-06,
      "loss": 0.0445,
      "step": 61460
    },
    {
      "epoch": 0.8690918859308063,
      "grad_norm": 0.0,
      "learning_rate": 1.3134640670729122e-06,
      "loss": 0.3545,
      "step": 61470
    },
    {
      "epoch": 0.8692332706527732,
      "grad_norm": 0.0,
      "learning_rate": 1.3120502198532428e-06,
      "loss": 0.0,
      "step": 61480
    },
    {
      "epoch": 0.8693746553747402,
      "grad_norm": 0.0,
      "learning_rate": 1.3106363726335732e-06,
      "loss": 0.0,
      "step": 61490
    },
    {
      "epoch": 0.8695160400967071,
      "grad_norm": 0.005440279841423035,
      "learning_rate": 1.3092225254139037e-06,
      "loss": 0.0,
      "step": 61500
    },
    {
      "epoch": 0.8696574248186741,
      "grad_norm": 0.0,
      "learning_rate": 1.3078086781942345e-06,
      "loss": 0.0,
      "step": 61510
    },
    {
      "epoch": 0.869798809540641,
      "grad_norm": 0.0,
      "learning_rate": 1.306394830974565e-06,
      "loss": 0.0,
      "step": 61520
    },
    {
      "epoch": 0.8699401942626079,
      "grad_norm": 0.0,
      "learning_rate": 1.3049809837548954e-06,
      "loss": 0.5281,
      "step": 61530
    },
    {
      "epoch": 0.8700815789845749,
      "grad_norm": 0.0,
      "learning_rate": 1.3035671365352263e-06,
      "loss": 0.0,
      "step": 61540
    },
    {
      "epoch": 0.8702229637065418,
      "grad_norm": 0.0,
      "learning_rate": 1.3021532893155567e-06,
      "loss": 0.0002,
      "step": 61550
    },
    {
      "epoch": 0.8703643484285088,
      "grad_norm": 0.0,
      "learning_rate": 1.3007394420958871e-06,
      "loss": 0.0,
      "step": 61560
    },
    {
      "epoch": 0.8705057331504757,
      "grad_norm": 0.0,
      "learning_rate": 1.299325594876218e-06,
      "loss": 0.0,
      "step": 61570
    },
    {
      "epoch": 0.8706471178724428,
      "grad_norm": 0.0,
      "learning_rate": 1.2979117476565484e-06,
      "loss": 0.0,
      "step": 61580
    },
    {
      "epoch": 0.8707885025944097,
      "grad_norm": 1.4931664793493837e-07,
      "learning_rate": 1.2964979004368788e-06,
      "loss": 0.0,
      "step": 61590
    },
    {
      "epoch": 0.8709298873163765,
      "grad_norm": 0.005764758680015802,
      "learning_rate": 1.2950840532172095e-06,
      "loss": 0.0,
      "step": 61600
    },
    {
      "epoch": 0.8710712720383436,
      "grad_norm": 0.0,
      "learning_rate": 1.2936702059975401e-06,
      "loss": 0.0,
      "step": 61610
    },
    {
      "epoch": 0.8712126567603105,
      "grad_norm": 0.0,
      "learning_rate": 1.2922563587778706e-06,
      "loss": 0.0,
      "step": 61620
    },
    {
      "epoch": 0.8713540414822775,
      "grad_norm": 2.8324951273361876e-09,
      "learning_rate": 1.290842511558201e-06,
      "loss": 0.0,
      "step": 61630
    },
    {
      "epoch": 0.8714954262042444,
      "grad_norm": 0.0,
      "learning_rate": 1.2894286643385319e-06,
      "loss": 0.0,
      "step": 61640
    },
    {
      "epoch": 0.8716368109262113,
      "grad_norm": 0.0,
      "learning_rate": 1.2880148171188623e-06,
      "loss": 0.0,
      "step": 61650
    },
    {
      "epoch": 0.8717781956481783,
      "grad_norm": 0.0,
      "learning_rate": 1.2866009698991927e-06,
      "loss": 0.0009,
      "step": 61660
    },
    {
      "epoch": 0.8719195803701452,
      "grad_norm": 8.561446884414181e-05,
      "learning_rate": 1.2851871226795234e-06,
      "loss": 0.0,
      "step": 61670
    },
    {
      "epoch": 0.8720609650921122,
      "grad_norm": 0.0,
      "learning_rate": 1.283773275459854e-06,
      "loss": 0.0,
      "step": 61680
    },
    {
      "epoch": 0.8722023498140791,
      "grad_norm": 0.0,
      "learning_rate": 1.2823594282401844e-06,
      "loss": 0.0,
      "step": 61690
    },
    {
      "epoch": 0.8723437345360461,
      "grad_norm": 0.0,
      "learning_rate": 1.280945581020515e-06,
      "loss": 0.0,
      "step": 61700
    },
    {
      "epoch": 0.872485119258013,
      "grad_norm": 2.6433608013576304e-07,
      "learning_rate": 1.2795317338008455e-06,
      "loss": 0.0,
      "step": 61710
    },
    {
      "epoch": 0.8726265039799799,
      "grad_norm": 0.7336905598640442,
      "learning_rate": 1.2781178865811762e-06,
      "loss": 0.0002,
      "step": 61720
    },
    {
      "epoch": 0.8727678887019469,
      "grad_norm": 0.0,
      "learning_rate": 1.2767040393615068e-06,
      "loss": 0.0,
      "step": 61730
    },
    {
      "epoch": 0.8729092734239138,
      "grad_norm": 0.0,
      "learning_rate": 1.2752901921418372e-06,
      "loss": 0.0,
      "step": 61740
    },
    {
      "epoch": 0.8730506581458808,
      "grad_norm": 0.0,
      "learning_rate": 1.2738763449221677e-06,
      "loss": 0.0,
      "step": 61750
    },
    {
      "epoch": 0.8731920428678477,
      "grad_norm": 0.0,
      "learning_rate": 1.2724624977024983e-06,
      "loss": 0.0,
      "step": 61760
    },
    {
      "epoch": 0.8733334275898147,
      "grad_norm": 0.0,
      "learning_rate": 1.271048650482829e-06,
      "loss": 0.0,
      "step": 61770
    },
    {
      "epoch": 0.8734748123117816,
      "grad_norm": 5.868432850775207e-08,
      "learning_rate": 1.2696348032631594e-06,
      "loss": 0.0008,
      "step": 61780
    },
    {
      "epoch": 0.8736161970337485,
      "grad_norm": 0.0,
      "learning_rate": 1.2682209560434898e-06,
      "loss": 0.0001,
      "step": 61790
    },
    {
      "epoch": 0.8737575817557155,
      "grad_norm": 0.0,
      "learning_rate": 1.2668071088238207e-06,
      "loss": 0.0,
      "step": 61800
    },
    {
      "epoch": 0.8738989664776824,
      "grad_norm": 0.0,
      "learning_rate": 1.2653932616041511e-06,
      "loss": 0.0,
      "step": 61810
    },
    {
      "epoch": 0.8740403511996494,
      "grad_norm": 0.0,
      "learning_rate": 1.2639794143844816e-06,
      "loss": 0.0,
      "step": 61820
    },
    {
      "epoch": 0.8741817359216163,
      "grad_norm": 0.0,
      "learning_rate": 1.2625655671648124e-06,
      "loss": 0.0,
      "step": 61830
    },
    {
      "epoch": 0.8743231206435833,
      "grad_norm": 1.8428743715048768e-05,
      "learning_rate": 1.2611517199451428e-06,
      "loss": 0.0,
      "step": 61840
    },
    {
      "epoch": 0.8744645053655502,
      "grad_norm": 0.0,
      "learning_rate": 1.2597378727254733e-06,
      "loss": 0.0659,
      "step": 61850
    },
    {
      "epoch": 0.8746058900875171,
      "grad_norm": 0.0,
      "learning_rate": 1.2583240255058041e-06,
      "loss": 0.4172,
      "step": 61860
    },
    {
      "epoch": 0.8747472748094841,
      "grad_norm": 0.0,
      "learning_rate": 1.2569101782861346e-06,
      "loss": 0.3713,
      "step": 61870
    },
    {
      "epoch": 0.874888659531451,
      "grad_norm": 0.0,
      "learning_rate": 1.255496331066465e-06,
      "loss": 0.0,
      "step": 61880
    },
    {
      "epoch": 0.875030044253418,
      "grad_norm": 0.0,
      "learning_rate": 1.2540824838467954e-06,
      "loss": 0.0,
      "step": 61890
    },
    {
      "epoch": 0.8751714289753849,
      "grad_norm": 0.0,
      "learning_rate": 1.2526686366271263e-06,
      "loss": 0.0,
      "step": 61900
    },
    {
      "epoch": 0.8753128136973518,
      "grad_norm": 0.0,
      "learning_rate": 1.2512547894074567e-06,
      "loss": 0.0,
      "step": 61910
    },
    {
      "epoch": 0.8754541984193188,
      "grad_norm": 2.6441431266732707e-09,
      "learning_rate": 1.2498409421877874e-06,
      "loss": 0.0,
      "step": 61920
    },
    {
      "epoch": 0.8755955831412857,
      "grad_norm": 0.0,
      "learning_rate": 1.2484270949681178e-06,
      "loss": 0.0001,
      "step": 61930
    },
    {
      "epoch": 0.8757369678632527,
      "grad_norm": 0.0,
      "learning_rate": 1.2470132477484484e-06,
      "loss": 0.0,
      "step": 61940
    },
    {
      "epoch": 0.8758783525852196,
      "grad_norm": 0.0,
      "learning_rate": 1.245599400528779e-06,
      "loss": 0.3529,
      "step": 61950
    },
    {
      "epoch": 0.8760197373071866,
      "grad_norm": 0.006037389859557152,
      "learning_rate": 1.2441855533091095e-06,
      "loss": 0.0,
      "step": 61960
    },
    {
      "epoch": 0.8761611220291535,
      "grad_norm": 0.0,
      "learning_rate": 1.2427717060894402e-06,
      "loss": 0.0,
      "step": 61970
    },
    {
      "epoch": 0.8763025067511204,
      "grad_norm": 0.0,
      "learning_rate": 1.2413578588697706e-06,
      "loss": 0.0,
      "step": 61980
    },
    {
      "epoch": 0.8764438914730874,
      "grad_norm": 0.0022870642133057117,
      "learning_rate": 1.2399440116501012e-06,
      "loss": 0.0005,
      "step": 61990
    },
    {
      "epoch": 0.8765852761950543,
      "grad_norm": 0.0,
      "learning_rate": 1.2385301644304317e-06,
      "loss": 0.0,
      "step": 62000
    },
    {
      "epoch": 0.8767266609170213,
      "grad_norm": 0.0,
      "learning_rate": 1.2371163172107623e-06,
      "loss": 0.0,
      "step": 62010
    },
    {
      "epoch": 0.8768680456389882,
      "grad_norm": 0.0,
      "learning_rate": 1.235702469991093e-06,
      "loss": 0.0,
      "step": 62020
    },
    {
      "epoch": 0.8770094303609552,
      "grad_norm": 7.625012443668311e-08,
      "learning_rate": 1.2342886227714234e-06,
      "loss": 0.0,
      "step": 62030
    },
    {
      "epoch": 0.8771508150829221,
      "grad_norm": 0.0,
      "learning_rate": 1.232874775551754e-06,
      "loss": 0.0,
      "step": 62040
    },
    {
      "epoch": 0.877292199804889,
      "grad_norm": 0.0,
      "learning_rate": 1.2314609283320845e-06,
      "loss": 0.0007,
      "step": 62050
    },
    {
      "epoch": 0.877433584526856,
      "grad_norm": 5.359868282539537e-08,
      "learning_rate": 1.2300470811124151e-06,
      "loss": 0.0,
      "step": 62060
    },
    {
      "epoch": 0.877574969248823,
      "grad_norm": 0.0,
      "learning_rate": 1.2286332338927456e-06,
      "loss": 0.0388,
      "step": 62070
    },
    {
      "epoch": 0.87771635397079,
      "grad_norm": 2.3148616037360625e-06,
      "learning_rate": 1.2272193866730762e-06,
      "loss": 0.0,
      "step": 62080
    },
    {
      "epoch": 0.8778577386927569,
      "grad_norm": 0.21143995225429535,
      "learning_rate": 1.2258055394534066e-06,
      "loss": 0.0,
      "step": 62090
    },
    {
      "epoch": 0.8779991234147239,
      "grad_norm": 0.0,
      "learning_rate": 1.2243916922337373e-06,
      "loss": 0.0,
      "step": 62100
    },
    {
      "epoch": 0.8781405081366908,
      "grad_norm": 1.2186118364334106,
      "learning_rate": 1.222977845014068e-06,
      "loss": 0.0004,
      "step": 62110
    },
    {
      "epoch": 0.8782818928586577,
      "grad_norm": 0.0,
      "learning_rate": 1.2215639977943984e-06,
      "loss": 0.0,
      "step": 62120
    },
    {
      "epoch": 0.8784232775806247,
      "grad_norm": 0.0,
      "learning_rate": 1.220150150574729e-06,
      "loss": 0.0,
      "step": 62130
    },
    {
      "epoch": 0.8785646623025916,
      "grad_norm": 0.0,
      "learning_rate": 1.2187363033550594e-06,
      "loss": 0.0067,
      "step": 62140
    },
    {
      "epoch": 0.8787060470245586,
      "grad_norm": 1.0727807975285941e-08,
      "learning_rate": 1.21732245613539e-06,
      "loss": 0.0,
      "step": 62150
    },
    {
      "epoch": 0.8788474317465255,
      "grad_norm": 0.0,
      "learning_rate": 1.2159086089157207e-06,
      "loss": 0.0,
      "step": 62160
    },
    {
      "epoch": 0.8789888164684924,
      "grad_norm": 0.0,
      "learning_rate": 1.2144947616960512e-06,
      "loss": 0.0,
      "step": 62170
    },
    {
      "epoch": 0.8791302011904594,
      "grad_norm": 0.0,
      "learning_rate": 1.2130809144763818e-06,
      "loss": 0.0,
      "step": 62180
    },
    {
      "epoch": 0.8792715859124263,
      "grad_norm": 0.0,
      "learning_rate": 1.2116670672567124e-06,
      "loss": 0.4797,
      "step": 62190
    },
    {
      "epoch": 0.8794129706343933,
      "grad_norm": 4.288727950552129e-08,
      "learning_rate": 1.2102532200370429e-06,
      "loss": 0.3576,
      "step": 62200
    },
    {
      "epoch": 0.8795543553563602,
      "grad_norm": 0.0,
      "learning_rate": 1.2088393728173735e-06,
      "loss": 0.0,
      "step": 62210
    },
    {
      "epoch": 0.8796957400783272,
      "grad_norm": 0.0,
      "learning_rate": 1.207425525597704e-06,
      "loss": 0.0,
      "step": 62220
    },
    {
      "epoch": 0.8798371248002941,
      "grad_norm": 0.0,
      "learning_rate": 1.2060116783780346e-06,
      "loss": 0.0,
      "step": 62230
    },
    {
      "epoch": 0.879978509522261,
      "grad_norm": 0.0,
      "learning_rate": 1.2045978311583652e-06,
      "loss": 0.0,
      "step": 62240
    },
    {
      "epoch": 0.880119894244228,
      "grad_norm": 0.0008409265428781509,
      "learning_rate": 1.2031839839386957e-06,
      "loss": 0.0,
      "step": 62250
    },
    {
      "epoch": 0.8802612789661949,
      "grad_norm": 1.106930582750465e-07,
      "learning_rate": 1.2017701367190263e-06,
      "loss": 0.0,
      "step": 62260
    },
    {
      "epoch": 0.8804026636881619,
      "grad_norm": 1.7573066202203336e-07,
      "learning_rate": 1.2003562894993568e-06,
      "loss": 0.0,
      "step": 62270
    },
    {
      "epoch": 0.8805440484101288,
      "grad_norm": 1.868758954515215e-05,
      "learning_rate": 1.1989424422796874e-06,
      "loss": 0.0,
      "step": 62280
    },
    {
      "epoch": 0.8806854331320958,
      "grad_norm": 0.0,
      "learning_rate": 1.197528595060018e-06,
      "loss": 0.0004,
      "step": 62290
    },
    {
      "epoch": 0.8808268178540627,
      "grad_norm": 0.0,
      "learning_rate": 1.1961147478403485e-06,
      "loss": 0.0,
      "step": 62300
    },
    {
      "epoch": 0.8809682025760296,
      "grad_norm": 0.0,
      "learning_rate": 1.1947009006206791e-06,
      "loss": 0.0,
      "step": 62310
    },
    {
      "epoch": 0.8811095872979966,
      "grad_norm": 0.01821734383702278,
      "learning_rate": 1.1932870534010096e-06,
      "loss": 0.0,
      "step": 62320
    },
    {
      "epoch": 0.8812509720199635,
      "grad_norm": 1.1638403520919383e-05,
      "learning_rate": 1.1918732061813402e-06,
      "loss": 0.0,
      "step": 62330
    },
    {
      "epoch": 0.8813923567419305,
      "grad_norm": 0.0,
      "learning_rate": 1.1904593589616706e-06,
      "loss": 0.0,
      "step": 62340
    },
    {
      "epoch": 0.8815337414638974,
      "grad_norm": 0.0,
      "learning_rate": 1.1890455117420013e-06,
      "loss": 0.0,
      "step": 62350
    },
    {
      "epoch": 0.8816751261858644,
      "grad_norm": 0.0,
      "learning_rate": 1.1876316645223317e-06,
      "loss": 0.0,
      "step": 62360
    },
    {
      "epoch": 0.8818165109078313,
      "grad_norm": 7.549329438916175e-06,
      "learning_rate": 1.1862178173026624e-06,
      "loss": 0.0001,
      "step": 62370
    },
    {
      "epoch": 0.8819578956297982,
      "grad_norm": 0.0,
      "learning_rate": 1.1848039700829928e-06,
      "loss": 0.0,
      "step": 62380
    },
    {
      "epoch": 0.8820992803517652,
      "grad_norm": 0.0,
      "learning_rate": 1.1833901228633234e-06,
      "loss": 0.0,
      "step": 62390
    },
    {
      "epoch": 0.8822406650737321,
      "grad_norm": 0.0,
      "learning_rate": 1.181976275643654e-06,
      "loss": 0.0,
      "step": 62400
    },
    {
      "epoch": 0.8823820497956991,
      "grad_norm": 0.0,
      "learning_rate": 1.1805624284239845e-06,
      "loss": 0.0,
      "step": 62410
    },
    {
      "epoch": 0.882523434517666,
      "grad_norm": 2.3179780317406085e-09,
      "learning_rate": 1.1791485812043152e-06,
      "loss": 0.0,
      "step": 62420
    },
    {
      "epoch": 0.8826648192396329,
      "grad_norm": 0.0,
      "learning_rate": 1.1777347339846456e-06,
      "loss": 0.0063,
      "step": 62430
    },
    {
      "epoch": 0.8828062039615999,
      "grad_norm": 0.0,
      "learning_rate": 1.1763208867649762e-06,
      "loss": 0.0004,
      "step": 62440
    },
    {
      "epoch": 0.8829475886835668,
      "grad_norm": 1.3159836775855638e-08,
      "learning_rate": 1.1749070395453069e-06,
      "loss": 0.0,
      "step": 62450
    },
    {
      "epoch": 0.8830889734055338,
      "grad_norm": 0.0,
      "learning_rate": 1.1734931923256373e-06,
      "loss": 0.0,
      "step": 62460
    },
    {
      "epoch": 0.8832303581275007,
      "grad_norm": 5.0114285841118544e-05,
      "learning_rate": 1.172079345105968e-06,
      "loss": 0.0,
      "step": 62470
    },
    {
      "epoch": 0.8833717428494677,
      "grad_norm": 0.0,
      "learning_rate": 1.1706654978862984e-06,
      "loss": 0.0004,
      "step": 62480
    },
    {
      "epoch": 0.8835131275714346,
      "grad_norm": 0.00033716714824549854,
      "learning_rate": 1.169251650666629e-06,
      "loss": 0.0,
      "step": 62490
    },
    {
      "epoch": 0.8836545122934015,
      "grad_norm": 0.013475687243044376,
      "learning_rate": 1.1678378034469597e-06,
      "loss": 0.0,
      "step": 62500
    },
    {
      "epoch": 0.8837958970153685,
      "grad_norm": 0.0,
      "learning_rate": 1.1664239562272901e-06,
      "loss": 0.0163,
      "step": 62510
    },
    {
      "epoch": 0.8839372817373354,
      "grad_norm": 0.0,
      "learning_rate": 1.1650101090076208e-06,
      "loss": 0.0,
      "step": 62520
    },
    {
      "epoch": 0.8840786664593024,
      "grad_norm": 0.0001520878286100924,
      "learning_rate": 1.1635962617879514e-06,
      "loss": 0.0,
      "step": 62530
    },
    {
      "epoch": 0.8842200511812693,
      "grad_norm": 0.0,
      "learning_rate": 1.1621824145682818e-06,
      "loss": 0.0,
      "step": 62540
    },
    {
      "epoch": 0.8843614359032363,
      "grad_norm": 0.0,
      "learning_rate": 1.1607685673486125e-06,
      "loss": 0.0,
      "step": 62550
    },
    {
      "epoch": 0.8845028206252032,
      "grad_norm": 0.0,
      "learning_rate": 1.159354720128943e-06,
      "loss": 0.1006,
      "step": 62560
    },
    {
      "epoch": 0.8846442053471701,
      "grad_norm": 0.0,
      "learning_rate": 1.1579408729092736e-06,
      "loss": 0.0004,
      "step": 62570
    },
    {
      "epoch": 0.8847855900691372,
      "grad_norm": 0.0,
      "learning_rate": 1.1565270256896042e-06,
      "loss": 0.0,
      "step": 62580
    },
    {
      "epoch": 0.884926974791104,
      "grad_norm": 0.0,
      "learning_rate": 1.1551131784699346e-06,
      "loss": 0.0,
      "step": 62590
    },
    {
      "epoch": 0.8850683595130711,
      "grad_norm": 0.0,
      "learning_rate": 1.1536993312502653e-06,
      "loss": 0.0027,
      "step": 62600
    },
    {
      "epoch": 0.885209744235038,
      "grad_norm": 0.0,
      "learning_rate": 1.1522854840305957e-06,
      "loss": 0.0003,
      "step": 62610
    },
    {
      "epoch": 0.8853511289570049,
      "grad_norm": 0.0,
      "learning_rate": 1.1508716368109264e-06,
      "loss": 0.0,
      "step": 62620
    },
    {
      "epoch": 0.8854925136789719,
      "grad_norm": 0.0,
      "learning_rate": 1.149457789591257e-06,
      "loss": 0.0016,
      "step": 62630
    },
    {
      "epoch": 0.8856338984009388,
      "grad_norm": 0.0,
      "learning_rate": 1.1480439423715874e-06,
      "loss": 0.0,
      "step": 62640
    },
    {
      "epoch": 0.8857752831229058,
      "grad_norm": 0.0,
      "learning_rate": 1.146630095151918e-06,
      "loss": 0.0,
      "step": 62650
    },
    {
      "epoch": 0.8859166678448727,
      "grad_norm": 0.0,
      "learning_rate": 1.1453576326542156e-06,
      "loss": 1.1234,
      "step": 62660
    },
    {
      "epoch": 0.8860580525668397,
      "grad_norm": 0.0,
      "learning_rate": 1.143943785434546e-06,
      "loss": 0.0,
      "step": 62670
    },
    {
      "epoch": 0.8861994372888066,
      "grad_norm": 0.0,
      "learning_rate": 1.1425299382148766e-06,
      "loss": 0.0001,
      "step": 62680
    },
    {
      "epoch": 0.8863408220107735,
      "grad_norm": 0.0,
      "learning_rate": 1.141116090995207e-06,
      "loss": 0.0041,
      "step": 62690
    },
    {
      "epoch": 0.8864822067327405,
      "grad_norm": 0.0,
      "learning_rate": 1.1397022437755377e-06,
      "loss": 0.0,
      "step": 62700
    },
    {
      "epoch": 0.8866235914547074,
      "grad_norm": 0.0,
      "learning_rate": 1.1382883965558684e-06,
      "loss": 0.0559,
      "step": 62710
    },
    {
      "epoch": 0.8867649761766744,
      "grad_norm": 0.006352083291858435,
      "learning_rate": 1.1368745493361988e-06,
      "loss": 0.1095,
      "step": 62720
    },
    {
      "epoch": 0.8869063608986413,
      "grad_norm": 6.936499175935751e-06,
      "learning_rate": 1.1354607021165294e-06,
      "loss": 0.0034,
      "step": 62730
    },
    {
      "epoch": 0.8870477456206083,
      "grad_norm": 0.0,
      "learning_rate": 1.13404685489686e-06,
      "loss": 0.0,
      "step": 62740
    },
    {
      "epoch": 0.8871891303425752,
      "grad_norm": 0.0,
      "learning_rate": 1.1326330076771905e-06,
      "loss": 0.0,
      "step": 62750
    },
    {
      "epoch": 0.8873305150645421,
      "grad_norm": 0.0,
      "learning_rate": 1.1312191604575212e-06,
      "loss": 1.4451,
      "step": 62760
    },
    {
      "epoch": 0.8874718997865091,
      "grad_norm": 0.0,
      "learning_rate": 1.1298053132378516e-06,
      "loss": 0.0,
      "step": 62770
    },
    {
      "epoch": 0.887613284508476,
      "grad_norm": 0.0,
      "learning_rate": 1.1283914660181822e-06,
      "loss": 0.0,
      "step": 62780
    },
    {
      "epoch": 0.887754669230443,
      "grad_norm": 8.426589920418337e-05,
      "learning_rate": 1.1269776187985129e-06,
      "loss": 0.0001,
      "step": 62790
    },
    {
      "epoch": 0.8878960539524099,
      "grad_norm": 0.3931179642677307,
      "learning_rate": 1.1255637715788433e-06,
      "loss": 0.0001,
      "step": 62800
    },
    {
      "epoch": 0.8880374386743769,
      "grad_norm": 0.0,
      "learning_rate": 1.124149924359174e-06,
      "loss": 0.0,
      "step": 62810
    },
    {
      "epoch": 0.8881788233963438,
      "grad_norm": 339.44293212890625,
      "learning_rate": 1.1227360771395044e-06,
      "loss": 0.2113,
      "step": 62820
    },
    {
      "epoch": 0.8883202081183107,
      "grad_norm": 0.0,
      "learning_rate": 1.121322229919835e-06,
      "loss": 0.0,
      "step": 62830
    },
    {
      "epoch": 0.8884615928402777,
      "grad_norm": 0.0,
      "learning_rate": 1.1199083827001655e-06,
      "loss": 0.0,
      "step": 62840
    },
    {
      "epoch": 0.8886029775622446,
      "grad_norm": 0.0,
      "learning_rate": 1.1184945354804961e-06,
      "loss": 0.0,
      "step": 62850
    },
    {
      "epoch": 0.8887443622842116,
      "grad_norm": 0.0,
      "learning_rate": 1.1170806882608265e-06,
      "loss": 0.0,
      "step": 62860
    },
    {
      "epoch": 0.8888857470061785,
      "grad_norm": 9.094561050915217e-07,
      "learning_rate": 1.1156668410411572e-06,
      "loss": 0.0031,
      "step": 62870
    },
    {
      "epoch": 0.8890271317281454,
      "grad_norm": 0.0,
      "learning_rate": 1.1142529938214876e-06,
      "loss": 0.0,
      "step": 62880
    },
    {
      "epoch": 0.8891685164501124,
      "grad_norm": 0.0,
      "learning_rate": 1.1128391466018183e-06,
      "loss": 0.0,
      "step": 62890
    },
    {
      "epoch": 0.8893099011720793,
      "grad_norm": 0.0,
      "learning_rate": 1.1114252993821487e-06,
      "loss": 0.0,
      "step": 62900
    },
    {
      "epoch": 0.8894512858940463,
      "grad_norm": 2.9740010631940095e-07,
      "learning_rate": 1.1100114521624793e-06,
      "loss": 0.0002,
      "step": 62910
    },
    {
      "epoch": 0.8895926706160132,
      "grad_norm": 0.0,
      "learning_rate": 1.10859760494281e-06,
      "loss": 0.0001,
      "step": 62920
    },
    {
      "epoch": 0.8897340553379802,
      "grad_norm": 0.00019653208437375724,
      "learning_rate": 1.1071837577231404e-06,
      "loss": 0.0,
      "step": 62930
    },
    {
      "epoch": 0.8898754400599471,
      "grad_norm": 5.88414707181073e-07,
      "learning_rate": 1.105769910503471e-06,
      "loss": 0.0,
      "step": 62940
    },
    {
      "epoch": 0.890016824781914,
      "grad_norm": 0.0,
      "learning_rate": 1.1043560632838017e-06,
      "loss": 0.0002,
      "step": 62950
    },
    {
      "epoch": 0.890158209503881,
      "grad_norm": 0.0,
      "learning_rate": 1.1029422160641321e-06,
      "loss": 0.0,
      "step": 62960
    },
    {
      "epoch": 0.8902995942258479,
      "grad_norm": 3.069965532631613e-05,
      "learning_rate": 1.1015283688444628e-06,
      "loss": 0.0,
      "step": 62970
    },
    {
      "epoch": 0.8904409789478149,
      "grad_norm": 0.0,
      "learning_rate": 1.1001145216247932e-06,
      "loss": 0.0,
      "step": 62980
    },
    {
      "epoch": 0.8905823636697818,
      "grad_norm": 1.4032328181201592e-05,
      "learning_rate": 1.0987006744051239e-06,
      "loss": 0.0,
      "step": 62990
    },
    {
      "epoch": 0.8907237483917488,
      "grad_norm": 0.0,
      "learning_rate": 1.0972868271854545e-06,
      "loss": 0.0,
      "step": 63000
    },
    {
      "epoch": 0.8908651331137157,
      "grad_norm": 0.0,
      "learning_rate": 1.095872979965785e-06,
      "loss": 0.0,
      "step": 63010
    },
    {
      "epoch": 0.8910065178356826,
      "grad_norm": 0.0,
      "learning_rate": 1.0944591327461156e-06,
      "loss": 0.0,
      "step": 63020
    },
    {
      "epoch": 0.8911479025576496,
      "grad_norm": 0.0,
      "learning_rate": 1.093045285526446e-06,
      "loss": 0.0,
      "step": 63030
    },
    {
      "epoch": 0.8912892872796165,
      "grad_norm": 0.0,
      "learning_rate": 1.0916314383067767e-06,
      "loss": 0.0001,
      "step": 63040
    },
    {
      "epoch": 0.8914306720015835,
      "grad_norm": 0.0,
      "learning_rate": 1.0902175910871073e-06,
      "loss": 0.0,
      "step": 63050
    },
    {
      "epoch": 0.8915720567235504,
      "grad_norm": 0.0,
      "learning_rate": 1.0888037438674377e-06,
      "loss": 0.0,
      "step": 63060
    },
    {
      "epoch": 0.8917134414455175,
      "grad_norm": 0.0,
      "learning_rate": 1.0873898966477684e-06,
      "loss": 0.0,
      "step": 63070
    },
    {
      "epoch": 0.8918548261674843,
      "grad_norm": 0.0,
      "learning_rate": 1.085976049428099e-06,
      "loss": 0.0,
      "step": 63080
    },
    {
      "epoch": 0.8919962108894512,
      "grad_norm": 0.0,
      "learning_rate": 1.0845622022084295e-06,
      "loss": 0.0,
      "step": 63090
    },
    {
      "epoch": 0.8921375956114183,
      "grad_norm": 9.890592991723679e-06,
      "learning_rate": 1.0831483549887601e-06,
      "loss": 0.0,
      "step": 63100
    },
    {
      "epoch": 0.8922789803333852,
      "grad_norm": 0.0,
      "learning_rate": 1.0817345077690905e-06,
      "loss": 0.0,
      "step": 63110
    },
    {
      "epoch": 0.8924203650553522,
      "grad_norm": 0.0,
      "learning_rate": 1.0803206605494212e-06,
      "loss": 0.0,
      "step": 63120
    },
    {
      "epoch": 0.8925617497773191,
      "grad_norm": 0.0,
      "learning_rate": 1.0789068133297516e-06,
      "loss": 0.0009,
      "step": 63130
    },
    {
      "epoch": 0.892703134499286,
      "grad_norm": 0.0,
      "learning_rate": 1.0774929661100823e-06,
      "loss": 0.0,
      "step": 63140
    },
    {
      "epoch": 0.892844519221253,
      "grad_norm": 8.726078748466648e-10,
      "learning_rate": 1.076079118890413e-06,
      "loss": 0.0,
      "step": 63150
    },
    {
      "epoch": 0.8929859039432199,
      "grad_norm": 3.827494310826296e-06,
      "learning_rate": 1.0746652716707433e-06,
      "loss": 0.5801,
      "step": 63160
    },
    {
      "epoch": 0.8931272886651869,
      "grad_norm": 0.0,
      "learning_rate": 1.073251424451074e-06,
      "loss": 0.0,
      "step": 63170
    },
    {
      "epoch": 0.8932686733871538,
      "grad_norm": 0.0,
      "learning_rate": 1.0718375772314044e-06,
      "loss": 0.0003,
      "step": 63180
    },
    {
      "epoch": 0.8934100581091208,
      "grad_norm": 8.085342395602879e-10,
      "learning_rate": 1.070423730011735e-06,
      "loss": 0.0006,
      "step": 63190
    },
    {
      "epoch": 0.8935514428310877,
      "grad_norm": 0.0,
      "learning_rate": 1.0690098827920655e-06,
      "loss": 0.0,
      "step": 63200
    },
    {
      "epoch": 0.8936928275530546,
      "grad_norm": 0.0,
      "learning_rate": 1.0675960355723961e-06,
      "loss": 0.0,
      "step": 63210
    },
    {
      "epoch": 0.8938342122750216,
      "grad_norm": 0.0,
      "learning_rate": 1.0661821883527266e-06,
      "loss": 0.0,
      "step": 63220
    },
    {
      "epoch": 0.8939755969969885,
      "grad_norm": 0.0,
      "learning_rate": 1.0647683411330572e-06,
      "loss": 0.0,
      "step": 63230
    },
    {
      "epoch": 0.8941169817189555,
      "grad_norm": 0.23568136990070343,
      "learning_rate": 1.0633544939133877e-06,
      "loss": 0.0001,
      "step": 63240
    },
    {
      "epoch": 0.8942583664409224,
      "grad_norm": 4.064059734344482,
      "learning_rate": 1.0619406466937183e-06,
      "loss": 0.0004,
      "step": 63250
    },
    {
      "epoch": 0.8943997511628894,
      "grad_norm": 1.1602071026572958e-05,
      "learning_rate": 1.060526799474049e-06,
      "loss": 0.0,
      "step": 63260
    },
    {
      "epoch": 0.8945411358848563,
      "grad_norm": 0.0,
      "learning_rate": 1.0591129522543794e-06,
      "loss": 0.0012,
      "step": 63270
    },
    {
      "epoch": 0.8946825206068232,
      "grad_norm": 0.0,
      "learning_rate": 1.05769910503471e-06,
      "loss": 0.0,
      "step": 63280
    },
    {
      "epoch": 0.8948239053287902,
      "grad_norm": 0.0,
      "learning_rate": 1.0562852578150407e-06,
      "loss": 0.0,
      "step": 63290
    },
    {
      "epoch": 0.8949652900507571,
      "grad_norm": 0.0,
      "learning_rate": 1.054871410595371e-06,
      "loss": 0.0,
      "step": 63300
    },
    {
      "epoch": 0.8951066747727241,
      "grad_norm": 0.0,
      "learning_rate": 1.0534575633757017e-06,
      "loss": 0.0,
      "step": 63310
    },
    {
      "epoch": 0.895248059494691,
      "grad_norm": 0.0,
      "learning_rate": 1.0520437161560322e-06,
      "loss": 0.0,
      "step": 63320
    },
    {
      "epoch": 0.8953894442166579,
      "grad_norm": 0.0,
      "learning_rate": 1.0506298689363628e-06,
      "loss": 0.0,
      "step": 63330
    },
    {
      "epoch": 0.8955308289386249,
      "grad_norm": 0.0,
      "learning_rate": 1.0492160217166935e-06,
      "loss": 0.0,
      "step": 63340
    },
    {
      "epoch": 0.8956722136605918,
      "grad_norm": 5.992168098600814e-06,
      "learning_rate": 1.047802174497024e-06,
      "loss": 0.0,
      "step": 63350
    },
    {
      "epoch": 0.8958135983825588,
      "grad_norm": 0.0,
      "learning_rate": 1.0463883272773545e-06,
      "loss": 0.0013,
      "step": 63360
    },
    {
      "epoch": 0.8959549831045257,
      "grad_norm": 0.0,
      "learning_rate": 1.0449744800576852e-06,
      "loss": 0.0,
      "step": 63370
    },
    {
      "epoch": 0.8960963678264927,
      "grad_norm": 0.0008816074696369469,
      "learning_rate": 1.0435606328380156e-06,
      "loss": 0.0,
      "step": 63380
    },
    {
      "epoch": 0.8962377525484596,
      "grad_norm": 0.0,
      "learning_rate": 1.0421467856183463e-06,
      "loss": 0.0147,
      "step": 63390
    },
    {
      "epoch": 0.8963791372704265,
      "grad_norm": 0.0,
      "learning_rate": 1.0407329383986767e-06,
      "loss": 0.0,
      "step": 63400
    },
    {
      "epoch": 0.8965205219923935,
      "grad_norm": 0.0,
      "learning_rate": 1.0393190911790073e-06,
      "loss": 0.0313,
      "step": 63410
    },
    {
      "epoch": 0.8966619067143604,
      "grad_norm": 0.0,
      "learning_rate": 1.037905243959338e-06,
      "loss": 0.0,
      "step": 63420
    },
    {
      "epoch": 0.8968032914363274,
      "grad_norm": 0.2047688513994217,
      "learning_rate": 1.0364913967396684e-06,
      "loss": 0.0001,
      "step": 63430
    },
    {
      "epoch": 0.8969446761582943,
      "grad_norm": 0.0,
      "learning_rate": 1.035077549519999e-06,
      "loss": 0.0,
      "step": 63440
    },
    {
      "epoch": 0.8970860608802613,
      "grad_norm": 0.0,
      "learning_rate": 1.0336637023003295e-06,
      "loss": 0.2658,
      "step": 63450
    },
    {
      "epoch": 0.8972274456022282,
      "grad_norm": 0.0,
      "learning_rate": 1.0322498550806601e-06,
      "loss": 0.0,
      "step": 63460
    },
    {
      "epoch": 0.8973688303241951,
      "grad_norm": 0.0,
      "learning_rate": 1.0308360078609906e-06,
      "loss": 0.0,
      "step": 63470
    },
    {
      "epoch": 0.8975102150461621,
      "grad_norm": 6.457394761127944e-07,
      "learning_rate": 1.0294221606413212e-06,
      "loss": 0.0042,
      "step": 63480
    },
    {
      "epoch": 0.897651599768129,
      "grad_norm": 5.644588327413658e-06,
      "learning_rate": 1.0280083134216517e-06,
      "loss": 0.0,
      "step": 63490
    },
    {
      "epoch": 0.897792984490096,
      "grad_norm": 0.0,
      "learning_rate": 1.0265944662019823e-06,
      "loss": 0.0,
      "step": 63500
    },
    {
      "epoch": 0.8979343692120629,
      "grad_norm": 8.091565746326523e-07,
      "learning_rate": 1.0251806189823127e-06,
      "loss": 0.0068,
      "step": 63510
    },
    {
      "epoch": 0.8980757539340299,
      "grad_norm": 11.827506065368652,
      "learning_rate": 1.0237667717626434e-06,
      "loss": 0.0034,
      "step": 63520
    },
    {
      "epoch": 0.8982171386559968,
      "grad_norm": 0.0,
      "learning_rate": 1.022352924542974e-06,
      "loss": 0.0,
      "step": 63530
    },
    {
      "epoch": 0.8983585233779637,
      "grad_norm": 0.0,
      "learning_rate": 1.0209390773233045e-06,
      "loss": 0.0,
      "step": 63540
    },
    {
      "epoch": 0.8984999080999307,
      "grad_norm": 0.0,
      "learning_rate": 1.019525230103635e-06,
      "loss": 0.0,
      "step": 63550
    },
    {
      "epoch": 0.8986412928218976,
      "grad_norm": 0.0,
      "learning_rate": 1.0181113828839655e-06,
      "loss": 0.0024,
      "step": 63560
    },
    {
      "epoch": 0.8987826775438646,
      "grad_norm": 0.0,
      "learning_rate": 1.0166975356642962e-06,
      "loss": 0.0091,
      "step": 63570
    },
    {
      "epoch": 0.8989240622658315,
      "grad_norm": 0.0,
      "learning_rate": 1.0152836884446268e-06,
      "loss": 0.0045,
      "step": 63580
    },
    {
      "epoch": 0.8990654469877984,
      "grad_norm": 0.0,
      "learning_rate": 1.0138698412249573e-06,
      "loss": 0.1663,
      "step": 63590
    },
    {
      "epoch": 0.8992068317097655,
      "grad_norm": 0.0,
      "learning_rate": 1.012455994005288e-06,
      "loss": 0.0,
      "step": 63600
    },
    {
      "epoch": 0.8993482164317324,
      "grad_norm": 0.0,
      "learning_rate": 1.0110421467856183e-06,
      "loss": 0.0,
      "step": 63610
    },
    {
      "epoch": 0.8994896011536994,
      "grad_norm": 2.8419742648111423e-07,
      "learning_rate": 1.009628299565949e-06,
      "loss": 0.0611,
      "step": 63620
    },
    {
      "epoch": 0.8996309858756663,
      "grad_norm": 0.0,
      "learning_rate": 1.0082144523462796e-06,
      "loss": 0.0693,
      "step": 63630
    },
    {
      "epoch": 0.8997723705976333,
      "grad_norm": 0.0,
      "learning_rate": 1.00680060512661e-06,
      "loss": 0.0012,
      "step": 63640
    },
    {
      "epoch": 0.8999137553196002,
      "grad_norm": 0.0,
      "learning_rate": 1.0053867579069407e-06,
      "loss": 0.0,
      "step": 63650
    },
    {
      "epoch": 0.9000551400415671,
      "grad_norm": 0.0,
      "learning_rate": 1.0039729106872711e-06,
      "loss": 0.0,
      "step": 63660
    },
    {
      "epoch": 0.9001965247635341,
      "grad_norm": 0.0,
      "learning_rate": 1.0025590634676018e-06,
      "loss": 0.0,
      "step": 63670
    },
    {
      "epoch": 0.900337909485501,
      "grad_norm": 0.0,
      "learning_rate": 1.0011452162479324e-06,
      "loss": 0.0,
      "step": 63680
    },
    {
      "epoch": 0.900479294207468,
      "grad_norm": 0.0,
      "learning_rate": 9.997313690282629e-07,
      "loss": 0.0,
      "step": 63690
    },
    {
      "epoch": 0.9006206789294349,
      "grad_norm": 0.0,
      "learning_rate": 9.983175218085935e-07,
      "loss": 0.0,
      "step": 63700
    },
    {
      "epoch": 0.9007620636514019,
      "grad_norm": 0.0,
      "learning_rate": 9.969036745889241e-07,
      "loss": 0.0,
      "step": 63710
    },
    {
      "epoch": 0.9009034483733688,
      "grad_norm": 0.0,
      "learning_rate": 9.954898273692546e-07,
      "loss": 0.0,
      "step": 63720
    },
    {
      "epoch": 0.9010448330953357,
      "grad_norm": 1.77677051738101e-07,
      "learning_rate": 9.940759801495852e-07,
      "loss": 0.0039,
      "step": 63730
    },
    {
      "epoch": 0.9011862178173027,
      "grad_norm": 0.0,
      "learning_rate": 9.926621329299157e-07,
      "loss": 0.0,
      "step": 63740
    },
    {
      "epoch": 0.9013276025392696,
      "grad_norm": 0.0,
      "learning_rate": 9.912482857102463e-07,
      "loss": 0.0001,
      "step": 63750
    },
    {
      "epoch": 0.9014689872612366,
      "grad_norm": 5.292978312354535e-05,
      "learning_rate": 9.89834438490577e-07,
      "loss": 0.0,
      "step": 63760
    },
    {
      "epoch": 0.9016103719832035,
      "grad_norm": 0.0,
      "learning_rate": 9.884205912709074e-07,
      "loss": 0.0004,
      "step": 63770
    },
    {
      "epoch": 0.9017517567051705,
      "grad_norm": 0.0,
      "learning_rate": 9.87006744051238e-07,
      "loss": 0.0,
      "step": 63780
    },
    {
      "epoch": 0.9018931414271374,
      "grad_norm": 0.0,
      "learning_rate": 9.855928968315685e-07,
      "loss": 0.0,
      "step": 63790
    },
    {
      "epoch": 0.9020345261491043,
      "grad_norm": 0.0,
      "learning_rate": 9.84179049611899e-07,
      "loss": 0.3004,
      "step": 63800
    },
    {
      "epoch": 0.9021759108710713,
      "grad_norm": 0.0,
      "learning_rate": 9.827652023922295e-07,
      "loss": 0.0,
      "step": 63810
    },
    {
      "epoch": 0.9023172955930382,
      "grad_norm": 4.3906597966270056e-06,
      "learning_rate": 9.813513551725602e-07,
      "loss": 0.0,
      "step": 63820
    },
    {
      "epoch": 0.9024586803150052,
      "grad_norm": 1.2686006130024907e-06,
      "learning_rate": 9.799375079528906e-07,
      "loss": 0.0,
      "step": 63830
    },
    {
      "epoch": 0.9026000650369721,
      "grad_norm": 0.0,
      "learning_rate": 9.785236607332213e-07,
      "loss": 0.0,
      "step": 63840
    },
    {
      "epoch": 0.902741449758939,
      "grad_norm": 0.0,
      "learning_rate": 9.771098135135517e-07,
      "loss": 0.0,
      "step": 63850
    },
    {
      "epoch": 0.902882834480906,
      "grad_norm": 0.0,
      "learning_rate": 9.756959662938823e-07,
      "loss": 0.0,
      "step": 63860
    },
    {
      "epoch": 0.9030242192028729,
      "grad_norm": 0.0,
      "learning_rate": 9.742821190742128e-07,
      "loss": 0.0,
      "step": 63870
    },
    {
      "epoch": 0.9031656039248399,
      "grad_norm": 0.0,
      "learning_rate": 9.728682718545434e-07,
      "loss": 0.0,
      "step": 63880
    },
    {
      "epoch": 0.9033069886468068,
      "grad_norm": 0.0,
      "learning_rate": 9.71454424634874e-07,
      "loss": 0.0,
      "step": 63890
    },
    {
      "epoch": 0.9034483733687738,
      "grad_norm": 0.0,
      "learning_rate": 9.700405774152045e-07,
      "loss": 0.0,
      "step": 63900
    },
    {
      "epoch": 0.9035897580907407,
      "grad_norm": 0.0,
      "learning_rate": 9.686267301955351e-07,
      "loss": 0.0008,
      "step": 63910
    },
    {
      "epoch": 0.9037311428127076,
      "grad_norm": 0.0,
      "learning_rate": 9.672128829758658e-07,
      "loss": 0.0003,
      "step": 63920
    },
    {
      "epoch": 0.9038725275346746,
      "grad_norm": 0.0,
      "learning_rate": 9.657990357561962e-07,
      "loss": 0.0,
      "step": 63930
    },
    {
      "epoch": 0.9040139122566415,
      "grad_norm": 0.0,
      "learning_rate": 9.643851885365269e-07,
      "loss": 0.0,
      "step": 63940
    },
    {
      "epoch": 0.9041552969786085,
      "grad_norm": 3.007471832461306e-06,
      "learning_rate": 9.629713413168573e-07,
      "loss": 0.0003,
      "step": 63950
    },
    {
      "epoch": 0.9042966817005754,
      "grad_norm": 0.0,
      "learning_rate": 9.61557494097188e-07,
      "loss": 0.0004,
      "step": 63960
    },
    {
      "epoch": 0.9044380664225424,
      "grad_norm": 0.0,
      "learning_rate": 9.601436468775186e-07,
      "loss": 0.0,
      "step": 63970
    },
    {
      "epoch": 0.9045794511445093,
      "grad_norm": 0.0,
      "learning_rate": 9.58729799657849e-07,
      "loss": 0.0,
      "step": 63980
    },
    {
      "epoch": 0.9047208358664762,
      "grad_norm": 0.0,
      "learning_rate": 9.573159524381797e-07,
      "loss": 0.8672,
      "step": 63990
    },
    {
      "epoch": 0.9048622205884432,
      "grad_norm": 5.924911405941202e-08,
      "learning_rate": 9.559021052185103e-07,
      "loss": 0.0,
      "step": 64000
    },
    {
      "epoch": 0.9050036053104101,
      "grad_norm": 0.0,
      "learning_rate": 9.544882579988407e-07,
      "loss": 0.0,
      "step": 64010
    },
    {
      "epoch": 0.9051449900323771,
      "grad_norm": 0.0,
      "learning_rate": 9.530744107791713e-07,
      "loss": 0.0,
      "step": 64020
    },
    {
      "epoch": 0.905286374754344,
      "grad_norm": 0.0,
      "learning_rate": 9.516605635595018e-07,
      "loss": 0.0,
      "step": 64030
    },
    {
      "epoch": 0.905427759476311,
      "grad_norm": 0.0,
      "learning_rate": 9.502467163398323e-07,
      "loss": 0.0,
      "step": 64040
    },
    {
      "epoch": 0.9055691441982779,
      "grad_norm": 0.0,
      "learning_rate": 9.48832869120163e-07,
      "loss": 0.0,
      "step": 64050
    },
    {
      "epoch": 0.9057105289202448,
      "grad_norm": 0.0003184677043464035,
      "learning_rate": 9.474190219004935e-07,
      "loss": 0.0,
      "step": 64060
    },
    {
      "epoch": 0.9058519136422118,
      "grad_norm": 0.0,
      "learning_rate": 9.460051746808241e-07,
      "loss": 0.0,
      "step": 64070
    },
    {
      "epoch": 0.9059932983641787,
      "grad_norm": 0.0,
      "learning_rate": 9.445913274611546e-07,
      "loss": 0.0,
      "step": 64080
    },
    {
      "epoch": 0.9061346830861458,
      "grad_norm": 0.0,
      "learning_rate": 9.431774802414851e-07,
      "loss": 0.0,
      "step": 64090
    },
    {
      "epoch": 0.9062760678081127,
      "grad_norm": 0.0,
      "learning_rate": 9.417636330218158e-07,
      "loss": 0.0,
      "step": 64100
    },
    {
      "epoch": 0.9064174525300795,
      "grad_norm": 0.0,
      "learning_rate": 9.403497858021462e-07,
      "loss": 0.0,
      "step": 64110
    },
    {
      "epoch": 0.9065588372520466,
      "grad_norm": 0.0,
      "learning_rate": 9.389359385824769e-07,
      "loss": 0.0,
      "step": 64120
    },
    {
      "epoch": 0.9067002219740135,
      "grad_norm": 0.0,
      "learning_rate": 9.375220913628075e-07,
      "loss": 0.001,
      "step": 64130
    },
    {
      "epoch": 0.9068416066959805,
      "grad_norm": 0.0,
      "learning_rate": 9.361082441431379e-07,
      "loss": 0.0,
      "step": 64140
    },
    {
      "epoch": 0.9069829914179474,
      "grad_norm": 0.0,
      "learning_rate": 9.346943969234686e-07,
      "loss": 0.0,
      "step": 64150
    },
    {
      "epoch": 0.9071243761399144,
      "grad_norm": 0.0,
      "learning_rate": 9.33280549703799e-07,
      "loss": 0.0051,
      "step": 64160
    },
    {
      "epoch": 0.9072657608618813,
      "grad_norm": 0.0,
      "learning_rate": 9.318667024841297e-07,
      "loss": 0.0,
      "step": 64170
    },
    {
      "epoch": 0.9074071455838482,
      "grad_norm": 0.0008609672077000141,
      "learning_rate": 9.304528552644602e-07,
      "loss": 0.0,
      "step": 64180
    },
    {
      "epoch": 0.9075485303058152,
      "grad_norm": 0.0,
      "learning_rate": 9.290390080447907e-07,
      "loss": 0.0,
      "step": 64190
    },
    {
      "epoch": 0.9076899150277821,
      "grad_norm": 9.100342146517448e-10,
      "learning_rate": 9.276251608251213e-07,
      "loss": 0.0002,
      "step": 64200
    },
    {
      "epoch": 0.9078312997497491,
      "grad_norm": 0.0,
      "learning_rate": 9.262113136054519e-07,
      "loss": 0.0015,
      "step": 64210
    },
    {
      "epoch": 0.907972684471716,
      "grad_norm": 0.0,
      "learning_rate": 9.247974663857824e-07,
      "loss": 0.0,
      "step": 64220
    },
    {
      "epoch": 0.908114069193683,
      "grad_norm": 0.0,
      "learning_rate": 9.23383619166113e-07,
      "loss": 0.0,
      "step": 64230
    },
    {
      "epoch": 0.9082554539156499,
      "grad_norm": 0.0,
      "learning_rate": 9.219697719464434e-07,
      "loss": 0.0,
      "step": 64240
    },
    {
      "epoch": 0.9083968386376168,
      "grad_norm": 0.0,
      "learning_rate": 9.205559247267741e-07,
      "loss": 0.0026,
      "step": 64250
    },
    {
      "epoch": 0.9085382233595838,
      "grad_norm": 0.0,
      "learning_rate": 9.191420775071047e-07,
      "loss": 0.0,
      "step": 64260
    },
    {
      "epoch": 0.9086796080815507,
      "grad_norm": 0.0,
      "learning_rate": 9.177282302874352e-07,
      "loss": 0.0,
      "step": 64270
    },
    {
      "epoch": 0.9088209928035177,
      "grad_norm": 6.230398139450699e-05,
      "learning_rate": 9.163143830677658e-07,
      "loss": 0.0001,
      "step": 64280
    },
    {
      "epoch": 0.9089623775254846,
      "grad_norm": 0.0,
      "learning_rate": 9.149005358480962e-07,
      "loss": 0.053,
      "step": 64290
    },
    {
      "epoch": 0.9091037622474515,
      "grad_norm": 0.0,
      "learning_rate": 9.134866886284269e-07,
      "loss": 0.0,
      "step": 64300
    },
    {
      "epoch": 0.9092451469694185,
      "grad_norm": 0.00030123861506581306,
      "learning_rate": 9.120728414087575e-07,
      "loss": 0.0511,
      "step": 64310
    },
    {
      "epoch": 0.9093865316913854,
      "grad_norm": 0.0,
      "learning_rate": 9.10658994189088e-07,
      "loss": 0.0335,
      "step": 64320
    },
    {
      "epoch": 0.9095279164133524,
      "grad_norm": 0.0,
      "learning_rate": 9.092451469694186e-07,
      "loss": 0.0,
      "step": 64330
    },
    {
      "epoch": 0.9096693011353193,
      "grad_norm": 0.0,
      "learning_rate": 9.078312997497491e-07,
      "loss": 0.0014,
      "step": 64340
    },
    {
      "epoch": 0.9098106858572863,
      "grad_norm": 0.0,
      "learning_rate": 9.064174525300797e-07,
      "loss": 0.0001,
      "step": 64350
    },
    {
      "epoch": 0.9099520705792532,
      "grad_norm": 0.0,
      "learning_rate": 9.050036053104102e-07,
      "loss": 0.0031,
      "step": 64360
    },
    {
      "epoch": 0.9100934553012201,
      "grad_norm": 0.0,
      "learning_rate": 9.035897580907408e-07,
      "loss": 0.0,
      "step": 64370
    },
    {
      "epoch": 0.9102348400231871,
      "grad_norm": 0.05227500945329666,
      "learning_rate": 9.021759108710713e-07,
      "loss": 0.0,
      "step": 64380
    },
    {
      "epoch": 0.910376224745154,
      "grad_norm": 0.0,
      "learning_rate": 9.007620636514019e-07,
      "loss": 0.0002,
      "step": 64390
    },
    {
      "epoch": 0.910517609467121,
      "grad_norm": 1.7713244915285031e-06,
      "learning_rate": 8.993482164317324e-07,
      "loss": 0.0002,
      "step": 64400
    },
    {
      "epoch": 0.9106589941890879,
      "grad_norm": 1.6940565616963e-09,
      "learning_rate": 8.97934369212063e-07,
      "loss": 0.0,
      "step": 64410
    },
    {
      "epoch": 0.9108003789110549,
      "grad_norm": 0.0,
      "learning_rate": 8.965205219923937e-07,
      "loss": 0.0,
      "step": 64420
    },
    {
      "epoch": 0.9109417636330218,
      "grad_norm": 1.3000429532894486e-07,
      "learning_rate": 8.951066747727241e-07,
      "loss": 0.0,
      "step": 64430
    },
    {
      "epoch": 0.9110831483549887,
      "grad_norm": 0.0,
      "learning_rate": 8.936928275530547e-07,
      "loss": 0.0,
      "step": 64440
    },
    {
      "epoch": 0.9112245330769557,
      "grad_norm": 6.122167542343959e-05,
      "learning_rate": 8.922789803333852e-07,
      "loss": 0.0,
      "step": 64450
    },
    {
      "epoch": 0.9113659177989226,
      "grad_norm": 0.0,
      "learning_rate": 8.908651331137158e-07,
      "loss": 0.0,
      "step": 64460
    },
    {
      "epoch": 0.9115073025208896,
      "grad_norm": 0.0,
      "learning_rate": 8.894512858940465e-07,
      "loss": 0.0,
      "step": 64470
    },
    {
      "epoch": 0.9116486872428565,
      "grad_norm": 1.7186376908284728e-06,
      "learning_rate": 8.880374386743769e-07,
      "loss": 0.0,
      "step": 64480
    },
    {
      "epoch": 0.9117900719648235,
      "grad_norm": 0.0,
      "learning_rate": 8.866235914547075e-07,
      "loss": 0.0,
      "step": 64490
    },
    {
      "epoch": 0.9119314566867904,
      "grad_norm": 8.007618816918693e-07,
      "learning_rate": 8.85209744235038e-07,
      "loss": 0.0,
      "step": 64500
    },
    {
      "epoch": 0.9120728414087573,
      "grad_norm": 0.0,
      "learning_rate": 8.837958970153686e-07,
      "loss": 0.0,
      "step": 64510
    },
    {
      "epoch": 0.9122142261307243,
      "grad_norm": 0.0,
      "learning_rate": 8.823820497956992e-07,
      "loss": 0.0,
      "step": 64520
    },
    {
      "epoch": 0.9123556108526912,
      "grad_norm": 0.0,
      "learning_rate": 8.809682025760297e-07,
      "loss": 0.0,
      "step": 64530
    },
    {
      "epoch": 0.9124969955746582,
      "grad_norm": 0.001002797856926918,
      "learning_rate": 8.795543553563602e-07,
      "loss": 0.0,
      "step": 64540
    },
    {
      "epoch": 0.9126383802966251,
      "grad_norm": 0.0,
      "learning_rate": 8.781405081366909e-07,
      "loss": 0.0,
      "step": 64550
    },
    {
      "epoch": 0.912779765018592,
      "grad_norm": 0.0,
      "learning_rate": 8.767266609170213e-07,
      "loss": 0.0,
      "step": 64560
    },
    {
      "epoch": 0.912921149740559,
      "grad_norm": 2.422286016212638e-08,
      "learning_rate": 8.75312813697352e-07,
      "loss": 0.0,
      "step": 64570
    },
    {
      "epoch": 0.9130625344625259,
      "grad_norm": 0.0,
      "learning_rate": 8.738989664776824e-07,
      "loss": 0.0,
      "step": 64580
    },
    {
      "epoch": 0.913203919184493,
      "grad_norm": 0.0,
      "learning_rate": 8.72485119258013e-07,
      "loss": 0.0,
      "step": 64590
    },
    {
      "epoch": 0.9133453039064598,
      "grad_norm": 0.0,
      "learning_rate": 8.710712720383437e-07,
      "loss": 0.0,
      "step": 64600
    },
    {
      "epoch": 0.9134866886284269,
      "grad_norm": 1.3745482192462077e-06,
      "learning_rate": 8.696574248186741e-07,
      "loss": 0.0,
      "step": 64610
    },
    {
      "epoch": 0.9136280733503938,
      "grad_norm": 0.0,
      "learning_rate": 8.682435775990048e-07,
      "loss": 0.0,
      "step": 64620
    },
    {
      "epoch": 0.9137694580723607,
      "grad_norm": 2.3649190552532673e-06,
      "learning_rate": 8.668297303793353e-07,
      "loss": 0.0017,
      "step": 64630
    },
    {
      "epoch": 0.9139108427943277,
      "grad_norm": 0.0,
      "learning_rate": 8.654158831596658e-07,
      "loss": 0.3686,
      "step": 64640
    },
    {
      "epoch": 0.9140522275162946,
      "grad_norm": 0.0,
      "learning_rate": 8.640020359399964e-07,
      "loss": 0.0,
      "step": 64650
    },
    {
      "epoch": 0.9141936122382616,
      "grad_norm": 0.2710137367248535,
      "learning_rate": 8.625881887203269e-07,
      "loss": 0.0,
      "step": 64660
    },
    {
      "epoch": 0.9143349969602285,
      "grad_norm": 0.0001383611815981567,
      "learning_rate": 8.611743415006576e-07,
      "loss": 0.0,
      "step": 64670
    },
    {
      "epoch": 0.9144763816821955,
      "grad_norm": 0.0,
      "learning_rate": 8.597604942809881e-07,
      "loss": 0.0,
      "step": 64680
    },
    {
      "epoch": 0.9146177664041624,
      "grad_norm": 3.332107371534221e-05,
      "learning_rate": 8.583466470613186e-07,
      "loss": 0.0,
      "step": 64690
    },
    {
      "epoch": 0.9147591511261293,
      "grad_norm": 0.0,
      "learning_rate": 8.569327998416492e-07,
      "loss": 0.0,
      "step": 64700
    },
    {
      "epoch": 0.9149005358480963,
      "grad_norm": 0.0,
      "learning_rate": 8.555189526219797e-07,
      "loss": 0.0,
      "step": 64710
    },
    {
      "epoch": 0.9150419205700632,
      "grad_norm": 0.0014777284814044833,
      "learning_rate": 8.541051054023103e-07,
      "loss": 0.0,
      "step": 64720
    },
    {
      "epoch": 0.9151833052920302,
      "grad_norm": 0.006107899826020002,
      "learning_rate": 8.526912581826409e-07,
      "loss": 0.0002,
      "step": 64730
    },
    {
      "epoch": 0.9153246900139971,
      "grad_norm": 3.948978701373562e-06,
      "learning_rate": 8.512774109629713e-07,
      "loss": 0.0001,
      "step": 64740
    },
    {
      "epoch": 0.9154660747359641,
      "grad_norm": 0.0,
      "learning_rate": 8.49863563743302e-07,
      "loss": 0.0,
      "step": 64750
    },
    {
      "epoch": 0.915607459457931,
      "grad_norm": 0.0,
      "learning_rate": 8.484497165236326e-07,
      "loss": 0.0,
      "step": 64760
    },
    {
      "epoch": 0.9157488441798979,
      "grad_norm": 0.0,
      "learning_rate": 8.470358693039631e-07,
      "loss": 0.0,
      "step": 64770
    },
    {
      "epoch": 0.9158902289018649,
      "grad_norm": 0.00012065026385243982,
      "learning_rate": 8.456220220842937e-07,
      "loss": 0.0,
      "step": 64780
    },
    {
      "epoch": 0.9160316136238318,
      "grad_norm": 5.2881266032756e-10,
      "learning_rate": 8.442081748646241e-07,
      "loss": 0.0,
      "step": 64790
    },
    {
      "epoch": 0.9161729983457988,
      "grad_norm": 0.0,
      "learning_rate": 8.427943276449548e-07,
      "loss": 0.0,
      "step": 64800
    },
    {
      "epoch": 0.9163143830677657,
      "grad_norm": 1.2554202157843974e-06,
      "learning_rate": 8.413804804252853e-07,
      "loss": 0.0,
      "step": 64810
    },
    {
      "epoch": 0.9164557677897326,
      "grad_norm": 0.0,
      "learning_rate": 8.399666332056159e-07,
      "loss": 0.0,
      "step": 64820
    },
    {
      "epoch": 0.9165971525116996,
      "grad_norm": 0.0,
      "learning_rate": 8.385527859859464e-07,
      "loss": 0.0,
      "step": 64830
    },
    {
      "epoch": 0.9167385372336665,
      "grad_norm": 0.0,
      "learning_rate": 8.37138938766277e-07,
      "loss": 0.0,
      "step": 64840
    },
    {
      "epoch": 0.9168799219556335,
      "grad_norm": 0.0,
      "learning_rate": 8.357250915466075e-07,
      "loss": 0.0693,
      "step": 64850
    },
    {
      "epoch": 0.9170213066776004,
      "grad_norm": 0.011980553157627583,
      "learning_rate": 8.343112443269381e-07,
      "loss": 0.0,
      "step": 64860
    },
    {
      "epoch": 0.9171626913995674,
      "grad_norm": 0.0,
      "learning_rate": 8.328973971072687e-07,
      "loss": 0.0004,
      "step": 64870
    },
    {
      "epoch": 0.9173040761215343,
      "grad_norm": 0.0,
      "learning_rate": 8.314835498875992e-07,
      "loss": 0.0,
      "step": 64880
    },
    {
      "epoch": 0.9174454608435012,
      "grad_norm": 8.965763299784157e-06,
      "learning_rate": 8.300697026679298e-07,
      "loss": 0.0,
      "step": 64890
    },
    {
      "epoch": 0.9175868455654682,
      "grad_norm": 0.0,
      "learning_rate": 8.286558554482603e-07,
      "loss": 0.0,
      "step": 64900
    },
    {
      "epoch": 0.9177282302874351,
      "grad_norm": 0.0,
      "learning_rate": 8.272420082285909e-07,
      "loss": 0.0,
      "step": 64910
    },
    {
      "epoch": 0.9178696150094021,
      "grad_norm": 0.0,
      "learning_rate": 8.258281610089214e-07,
      "loss": 0.7328,
      "step": 64920
    },
    {
      "epoch": 0.918010999731369,
      "grad_norm": 0.0,
      "learning_rate": 8.24414313789252e-07,
      "loss": 0.0,
      "step": 64930
    },
    {
      "epoch": 0.918152384453336,
      "grad_norm": 0.0,
      "learning_rate": 8.230004665695826e-07,
      "loss": 0.0,
      "step": 64940
    },
    {
      "epoch": 0.9182937691753029,
      "grad_norm": 6.918133294675499e-05,
      "learning_rate": 8.215866193499131e-07,
      "loss": 0.0,
      "step": 64950
    },
    {
      "epoch": 0.9184351538972698,
      "grad_norm": 2.0285775462980382e-05,
      "learning_rate": 8.201727721302437e-07,
      "loss": 0.0,
      "step": 64960
    },
    {
      "epoch": 0.9185765386192368,
      "grad_norm": 0.0,
      "learning_rate": 8.187589249105743e-07,
      "loss": 0.0,
      "step": 64970
    },
    {
      "epoch": 0.9187179233412037,
      "grad_norm": 0.0,
      "learning_rate": 8.173450776909048e-07,
      "loss": 0.0,
      "step": 64980
    },
    {
      "epoch": 0.9188593080631707,
      "grad_norm": 0.0,
      "learning_rate": 8.159312304712353e-07,
      "loss": 0.0002,
      "step": 64990
    },
    {
      "epoch": 0.9190006927851376,
      "grad_norm": 0.0,
      "learning_rate": 8.145173832515659e-07,
      "loss": 0.0,
      "step": 65000
    },
    {
      "epoch": 0.9191420775071045,
      "grad_norm": 4.721889546033253e-10,
      "learning_rate": 8.131035360318964e-07,
      "loss": 0.0,
      "step": 65010
    },
    {
      "epoch": 0.9192834622290715,
      "grad_norm": 0.0,
      "learning_rate": 8.116896888122271e-07,
      "loss": 0.0,
      "step": 65020
    },
    {
      "epoch": 0.9194248469510384,
      "grad_norm": 0.0,
      "learning_rate": 8.102758415925575e-07,
      "loss": 0.0,
      "step": 65030
    },
    {
      "epoch": 0.9195662316730054,
      "grad_norm": 0.0,
      "learning_rate": 8.088619943728881e-07,
      "loss": 0.0,
      "step": 65040
    },
    {
      "epoch": 0.9197076163949723,
      "grad_norm": 0.012311670929193497,
      "learning_rate": 8.074481471532186e-07,
      "loss": 0.0,
      "step": 65050
    },
    {
      "epoch": 0.9198490011169393,
      "grad_norm": 0.0,
      "learning_rate": 8.060342999335492e-07,
      "loss": 0.0,
      "step": 65060
    },
    {
      "epoch": 0.9199903858389062,
      "grad_norm": 1.7452210158808157e-05,
      "learning_rate": 8.046204527138799e-07,
      "loss": 0.0,
      "step": 65070
    },
    {
      "epoch": 0.9201317705608731,
      "grad_norm": 2.101933631593056e-07,
      "learning_rate": 8.032066054942103e-07,
      "loss": 0.0001,
      "step": 65080
    },
    {
      "epoch": 0.9202731552828401,
      "grad_norm": 0.0,
      "learning_rate": 8.017927582745409e-07,
      "loss": 0.0,
      "step": 65090
    },
    {
      "epoch": 0.920414540004807,
      "grad_norm": 0.0,
      "learning_rate": 8.003789110548716e-07,
      "loss": 0.0,
      "step": 65100
    },
    {
      "epoch": 0.920555924726774,
      "grad_norm": 0.0,
      "learning_rate": 7.98965063835202e-07,
      "loss": 0.0,
      "step": 65110
    },
    {
      "epoch": 0.920697309448741,
      "grad_norm": 5.458708756123087e-07,
      "learning_rate": 7.975512166155327e-07,
      "loss": 0.0,
      "step": 65120
    },
    {
      "epoch": 0.920838694170708,
      "grad_norm": 0.0023455247282981873,
      "learning_rate": 7.961373693958631e-07,
      "loss": 0.0,
      "step": 65130
    },
    {
      "epoch": 0.9209800788926749,
      "grad_norm": 4.67350247390641e-07,
      "learning_rate": 7.947235221761937e-07,
      "loss": 0.0,
      "step": 65140
    },
    {
      "epoch": 0.9211214636146418,
      "grad_norm": 9.402654654877551e-08,
      "learning_rate": 7.933096749565243e-07,
      "loss": 0.0,
      "step": 65150
    },
    {
      "epoch": 0.9212628483366088,
      "grad_norm": 22.923847198486328,
      "learning_rate": 7.918958277368548e-07,
      "loss": 0.0053,
      "step": 65160
    },
    {
      "epoch": 0.9214042330585757,
      "grad_norm": 0.0,
      "learning_rate": 7.904819805171854e-07,
      "loss": 0.0,
      "step": 65170
    },
    {
      "epoch": 0.9215456177805427,
      "grad_norm": 0.0,
      "learning_rate": 7.89068133297516e-07,
      "loss": 0.0,
      "step": 65180
    },
    {
      "epoch": 0.9216870025025096,
      "grad_norm": 0.0,
      "learning_rate": 7.876542860778464e-07,
      "loss": 0.0,
      "step": 65190
    },
    {
      "epoch": 0.9218283872244766,
      "grad_norm": 4.45355749130249,
      "learning_rate": 7.862404388581771e-07,
      "loss": 0.0006,
      "step": 65200
    },
    {
      "epoch": 0.9219697719464435,
      "grad_norm": 0.0017777030589058995,
      "learning_rate": 7.848265916385075e-07,
      "loss": 0.0,
      "step": 65210
    },
    {
      "epoch": 0.9221111566684104,
      "grad_norm": 0.0,
      "learning_rate": 7.834127444188382e-07,
      "loss": 0.0026,
      "step": 65220
    },
    {
      "epoch": 0.9222525413903774,
      "grad_norm": 0.0038052944000810385,
      "learning_rate": 7.819988971991688e-07,
      "loss": 0.0,
      "step": 65230
    },
    {
      "epoch": 0.9223939261123443,
      "grad_norm": 0.0,
      "learning_rate": 7.805850499794992e-07,
      "loss": 0.0,
      "step": 65240
    },
    {
      "epoch": 0.9225353108343113,
      "grad_norm": 618.8894653320312,
      "learning_rate": 7.791712027598299e-07,
      "loss": 0.1964,
      "step": 65250
    },
    {
      "epoch": 0.9226766955562782,
      "grad_norm": 7.007310731665939e-09,
      "learning_rate": 7.777573555401603e-07,
      "loss": 0.0,
      "step": 65260
    },
    {
      "epoch": 0.9228180802782451,
      "grad_norm": 0.0,
      "learning_rate": 7.76343508320491e-07,
      "loss": 0.0,
      "step": 65270
    },
    {
      "epoch": 0.9229594650002121,
      "grad_norm": 0.0,
      "learning_rate": 7.749296611008216e-07,
      "loss": 0.0,
      "step": 65280
    },
    {
      "epoch": 0.923100849722179,
      "grad_norm": 0.0,
      "learning_rate": 7.73515813881152e-07,
      "loss": 0.0,
      "step": 65290
    },
    {
      "epoch": 0.923242234444146,
      "grad_norm": 0.0017248557414859533,
      "learning_rate": 7.721019666614827e-07,
      "loss": 0.0015,
      "step": 65300
    },
    {
      "epoch": 0.9233836191661129,
      "grad_norm": 0.0,
      "learning_rate": 7.706881194418132e-07,
      "loss": 0.0,
      "step": 65310
    },
    {
      "epoch": 0.9235250038880799,
      "grad_norm": 0.0,
      "learning_rate": 7.692742722221438e-07,
      "loss": 0.0,
      "step": 65320
    },
    {
      "epoch": 0.9236663886100468,
      "grad_norm": 1.5044019222259521,
      "learning_rate": 7.678604250024743e-07,
      "loss": 0.0002,
      "step": 65330
    },
    {
      "epoch": 0.9238077733320137,
      "grad_norm": 0.0,
      "learning_rate": 7.664465777828048e-07,
      "loss": 0.0,
      "step": 65340
    },
    {
      "epoch": 0.9239491580539807,
      "grad_norm": 0.0,
      "learning_rate": 7.650327305631354e-07,
      "loss": 0.0,
      "step": 65350
    },
    {
      "epoch": 0.9240905427759476,
      "grad_norm": 0.0,
      "learning_rate": 7.63618883343466e-07,
      "loss": 0.0,
      "step": 65360
    },
    {
      "epoch": 0.9242319274979146,
      "grad_norm": 6.463208990226121e-08,
      "learning_rate": 7.622050361237964e-07,
      "loss": 0.0,
      "step": 65370
    },
    {
      "epoch": 0.9243733122198815,
      "grad_norm": 0.0,
      "learning_rate": 7.607911889041271e-07,
      "loss": 0.0,
      "step": 65380
    },
    {
      "epoch": 0.9245146969418485,
      "grad_norm": 0.0,
      "learning_rate": 7.593773416844577e-07,
      "loss": 0.0,
      "step": 65390
    },
    {
      "epoch": 0.9246560816638154,
      "grad_norm": 0.0,
      "learning_rate": 7.579634944647882e-07,
      "loss": 0.0,
      "step": 65400
    },
    {
      "epoch": 0.9247974663857823,
      "grad_norm": 8.814939064905047e-05,
      "learning_rate": 7.565496472451188e-07,
      "loss": 0.0,
      "step": 65410
    },
    {
      "epoch": 0.9249388511077493,
      "grad_norm": 0.0,
      "learning_rate": 7.551358000254492e-07,
      "loss": 0.0001,
      "step": 65420
    },
    {
      "epoch": 0.9250802358297162,
      "grad_norm": 0.00041892268927767873,
      "learning_rate": 7.537219528057799e-07,
      "loss": 0.0,
      "step": 65430
    },
    {
      "epoch": 0.9252216205516832,
      "grad_norm": 0.0,
      "learning_rate": 7.523081055861104e-07,
      "loss": 0.0,
      "step": 65440
    },
    {
      "epoch": 0.9253630052736501,
      "grad_norm": 0.0,
      "learning_rate": 7.50894258366441e-07,
      "loss": 0.0001,
      "step": 65450
    },
    {
      "epoch": 0.9255043899956171,
      "grad_norm": 0.0,
      "learning_rate": 7.494804111467715e-07,
      "loss": 0.0196,
      "step": 65460
    },
    {
      "epoch": 0.925645774717584,
      "grad_norm": 1.8573564908308526e-09,
      "learning_rate": 7.48066563927102e-07,
      "loss": 0.0027,
      "step": 65470
    },
    {
      "epoch": 0.9257871594395509,
      "grad_norm": 2.0530862911982695e-06,
      "learning_rate": 7.466527167074327e-07,
      "loss": 0.0,
      "step": 65480
    },
    {
      "epoch": 0.9259285441615179,
      "grad_norm": 0.0,
      "learning_rate": 7.452388694877632e-07,
      "loss": 0.0,
      "step": 65490
    },
    {
      "epoch": 0.9260699288834848,
      "grad_norm": 0.0,
      "learning_rate": 7.438250222680938e-07,
      "loss": 0.0,
      "step": 65500
    },
    {
      "epoch": 0.9262113136054518,
      "grad_norm": 0.0,
      "learning_rate": 7.424111750484243e-07,
      "loss": 0.0,
      "step": 65510
    },
    {
      "epoch": 0.9263526983274187,
      "grad_norm": 0.0,
      "learning_rate": 7.40997327828755e-07,
      "loss": 0.0,
      "step": 65520
    },
    {
      "epoch": 0.9264940830493856,
      "grad_norm": 0.0,
      "learning_rate": 7.395834806090854e-07,
      "loss": 0.0,
      "step": 65530
    },
    {
      "epoch": 0.9266354677713526,
      "grad_norm": 1.0127251925950986e-06,
      "learning_rate": 7.38169633389416e-07,
      "loss": 0.0001,
      "step": 65540
    },
    {
      "epoch": 0.9267768524933195,
      "grad_norm": 0.22779493033885956,
      "learning_rate": 7.367557861697465e-07,
      "loss": 0.0001,
      "step": 65550
    },
    {
      "epoch": 0.9269182372152865,
      "grad_norm": 0.0,
      "learning_rate": 7.353419389500771e-07,
      "loss": 0.0,
      "step": 65560
    },
    {
      "epoch": 0.9270596219372534,
      "grad_norm": 0.0,
      "learning_rate": 7.339280917304078e-07,
      "loss": 0.0,
      "step": 65570
    },
    {
      "epoch": 0.9272010066592205,
      "grad_norm": 0.0,
      "learning_rate": 7.325142445107382e-07,
      "loss": 0.0193,
      "step": 65580
    },
    {
      "epoch": 0.9273423913811873,
      "grad_norm": 5.428842086985242e-07,
      "learning_rate": 7.311003972910688e-07,
      "loss": 0.0001,
      "step": 65590
    },
    {
      "epoch": 0.9274837761031542,
      "grad_norm": 9.051174254537386e-10,
      "learning_rate": 7.296865500713994e-07,
      "loss": 0.0,
      "step": 65600
    },
    {
      "epoch": 0.9276251608251213,
      "grad_norm": 0.0,
      "learning_rate": 7.282727028517299e-07,
      "loss": 0.0,
      "step": 65610
    },
    {
      "epoch": 0.9277665455470882,
      "grad_norm": 0.0,
      "learning_rate": 7.268588556320604e-07,
      "loss": 0.0,
      "step": 65620
    },
    {
      "epoch": 0.9279079302690552,
      "grad_norm": 0.0,
      "learning_rate": 7.25445008412391e-07,
      "loss": 0.0,
      "step": 65630
    },
    {
      "epoch": 0.9280493149910221,
      "grad_norm": 0.0,
      "learning_rate": 7.240311611927215e-07,
      "loss": 0.0322,
      "step": 65640
    },
    {
      "epoch": 0.9281906997129891,
      "grad_norm": 0.0,
      "learning_rate": 7.226173139730522e-07,
      "loss": 0.0,
      "step": 65650
    },
    {
      "epoch": 0.928332084434956,
      "grad_norm": 0.0,
      "learning_rate": 7.212034667533826e-07,
      "loss": 0.0004,
      "step": 65660
    },
    {
      "epoch": 0.9284734691569229,
      "grad_norm": 0.0,
      "learning_rate": 7.197896195337132e-07,
      "loss": 0.0018,
      "step": 65670
    },
    {
      "epoch": 0.9286148538788899,
      "grad_norm": 0.0,
      "learning_rate": 7.183757723140438e-07,
      "loss": 0.0,
      "step": 65680
    },
    {
      "epoch": 0.9287562386008568,
      "grad_norm": 1.053405776474392e-05,
      "learning_rate": 7.169619250943743e-07,
      "loss": 0.0001,
      "step": 65690
    },
    {
      "epoch": 0.9288976233228238,
      "grad_norm": 0.0,
      "learning_rate": 7.15548077874705e-07,
      "loss": 0.0,
      "step": 65700
    },
    {
      "epoch": 0.9290390080447907,
      "grad_norm": 0.0,
      "learning_rate": 7.141342306550354e-07,
      "loss": 0.0,
      "step": 65710
    },
    {
      "epoch": 0.9291803927667577,
      "grad_norm": 0.00041425463859923184,
      "learning_rate": 7.12720383435366e-07,
      "loss": 0.0,
      "step": 65720
    },
    {
      "epoch": 0.9293217774887246,
      "grad_norm": 0.0,
      "learning_rate": 7.113065362156967e-07,
      "loss": 0.0,
      "step": 65730
    },
    {
      "epoch": 0.9294631622106915,
      "grad_norm": 0.0,
      "learning_rate": 7.098926889960271e-07,
      "loss": 0.0,
      "step": 65740
    },
    {
      "epoch": 0.9296045469326585,
      "grad_norm": 0.0,
      "learning_rate": 7.084788417763578e-07,
      "loss": 0.0,
      "step": 65750
    },
    {
      "epoch": 0.9297459316546254,
      "grad_norm": 5.023412654736603e-07,
      "learning_rate": 7.070649945566882e-07,
      "loss": 0.0023,
      "step": 65760
    },
    {
      "epoch": 0.9298873163765924,
      "grad_norm": 1.4207691911849452e-08,
      "learning_rate": 7.056511473370188e-07,
      "loss": 0.0,
      "step": 65770
    },
    {
      "epoch": 0.9300287010985593,
      "grad_norm": 0.0,
      "learning_rate": 7.042373001173494e-07,
      "loss": 0.0002,
      "step": 65780
    },
    {
      "epoch": 0.9301700858205262,
      "grad_norm": 0.0,
      "learning_rate": 7.028234528976799e-07,
      "loss": 0.0,
      "step": 65790
    },
    {
      "epoch": 0.9303114705424932,
      "grad_norm": 0.0,
      "learning_rate": 7.014096056780105e-07,
      "loss": 0.0001,
      "step": 65800
    },
    {
      "epoch": 0.9304528552644601,
      "grad_norm": 0.003218909492716193,
      "learning_rate": 6.999957584583411e-07,
      "loss": 0.0,
      "step": 65810
    },
    {
      "epoch": 0.9305942399864271,
      "grad_norm": 0.0,
      "learning_rate": 6.985819112386715e-07,
      "loss": 0.0,
      "step": 65820
    },
    {
      "epoch": 0.930735624708394,
      "grad_norm": 0.0,
      "learning_rate": 6.971680640190022e-07,
      "loss": 0.0,
      "step": 65830
    },
    {
      "epoch": 0.930877009430361,
      "grad_norm": 0.0,
      "learning_rate": 6.957542167993326e-07,
      "loss": 0.0,
      "step": 65840
    },
    {
      "epoch": 0.9310183941523279,
      "grad_norm": 0.0,
      "learning_rate": 6.943403695796633e-07,
      "loss": 0.0,
      "step": 65850
    },
    {
      "epoch": 0.9311597788742948,
      "grad_norm": 0.0,
      "learning_rate": 6.929265223599939e-07,
      "loss": 0.0,
      "step": 65860
    },
    {
      "epoch": 0.9313011635962618,
      "grad_norm": 0.00012595705629792064,
      "learning_rate": 6.915126751403243e-07,
      "loss": 0.0,
      "step": 65870
    },
    {
      "epoch": 0.9314425483182287,
      "grad_norm": 1.2844994671468157e-05,
      "learning_rate": 6.90098827920655e-07,
      "loss": 0.0028,
      "step": 65880
    },
    {
      "epoch": 0.9315839330401957,
      "grad_norm": 0.0,
      "learning_rate": 6.886849807009854e-07,
      "loss": 0.0,
      "step": 65890
    },
    {
      "epoch": 0.9317253177621626,
      "grad_norm": 0.0,
      "learning_rate": 6.872711334813161e-07,
      "loss": 0.0,
      "step": 65900
    },
    {
      "epoch": 0.9318667024841296,
      "grad_norm": 0.0,
      "learning_rate": 6.858572862616467e-07,
      "loss": 0.0,
      "step": 65910
    },
    {
      "epoch": 0.9320080872060965,
      "grad_norm": 8.017162326723337e-05,
      "learning_rate": 6.844434390419771e-07,
      "loss": 0.0,
      "step": 65920
    },
    {
      "epoch": 0.9321494719280634,
      "grad_norm": 0.0,
      "learning_rate": 6.830295918223078e-07,
      "loss": 0.0,
      "step": 65930
    },
    {
      "epoch": 0.9322908566500304,
      "grad_norm": 0.0,
      "learning_rate": 6.816157446026383e-07,
      "loss": 0.0,
      "step": 65940
    },
    {
      "epoch": 0.9324322413719973,
      "grad_norm": 0.017339162528514862,
      "learning_rate": 6.802018973829689e-07,
      "loss": 0.0,
      "step": 65950
    },
    {
      "epoch": 0.9325736260939643,
      "grad_norm": 0.0,
      "learning_rate": 6.787880501632994e-07,
      "loss": 0.0,
      "step": 65960
    },
    {
      "epoch": 0.9327150108159312,
      "grad_norm": 0.0,
      "learning_rate": 6.773742029436299e-07,
      "loss": 0.0031,
      "step": 65970
    },
    {
      "epoch": 0.9328563955378981,
      "grad_norm": 1.7856697013485245e-05,
      "learning_rate": 6.759603557239605e-07,
      "loss": 0.0,
      "step": 65980
    },
    {
      "epoch": 0.9329977802598651,
      "grad_norm": 6.8708549960661e-09,
      "learning_rate": 6.745465085042911e-07,
      "loss": 0.0,
      "step": 65990
    },
    {
      "epoch": 0.933139164981832,
      "grad_norm": 0.005364372860640287,
      "learning_rate": 6.731326612846216e-07,
      "loss": 0.0,
      "step": 66000
    },
    {
      "epoch": 0.933280549703799,
      "grad_norm": 0.0,
      "learning_rate": 6.717188140649522e-07,
      "loss": 0.0,
      "step": 66010
    },
    {
      "epoch": 0.9334219344257659,
      "grad_norm": 0.0,
      "learning_rate": 6.703049668452828e-07,
      "loss": 0.0,
      "step": 66020
    },
    {
      "epoch": 0.9335633191477329,
      "grad_norm": 0.0,
      "learning_rate": 6.688911196256133e-07,
      "loss": 0.0,
      "step": 66030
    },
    {
      "epoch": 0.9337047038696998,
      "grad_norm": 0.0,
      "learning_rate": 6.674772724059439e-07,
      "loss": 0.0,
      "step": 66040
    },
    {
      "epoch": 0.9338460885916667,
      "grad_norm": 0.0,
      "learning_rate": 6.660634251862744e-07,
      "loss": 0.0,
      "step": 66050
    },
    {
      "epoch": 0.9339874733136337,
      "grad_norm": 0.002718220232054591,
      "learning_rate": 6.64649577966605e-07,
      "loss": 0.0,
      "step": 66060
    },
    {
      "epoch": 0.9341288580356006,
      "grad_norm": 0.0,
      "learning_rate": 6.632357307469355e-07,
      "loss": 0.0132,
      "step": 66070
    },
    {
      "epoch": 0.9342702427575676,
      "grad_norm": 0.0,
      "learning_rate": 6.618218835272661e-07,
      "loss": 0.0418,
      "step": 66080
    },
    {
      "epoch": 0.9344116274795345,
      "grad_norm": 0.0,
      "learning_rate": 6.604080363075967e-07,
      "loss": 0.0,
      "step": 66090
    },
    {
      "epoch": 0.9345530122015016,
      "grad_norm": 2.5858798835542984e-05,
      "learning_rate": 6.589941890879272e-07,
      "loss": 0.0,
      "step": 66100
    },
    {
      "epoch": 0.9346943969234685,
      "grad_norm": 0.0,
      "learning_rate": 6.575803418682578e-07,
      "loss": 0.002,
      "step": 66110
    },
    {
      "epoch": 0.9348357816454353,
      "grad_norm": 0.0,
      "learning_rate": 6.561664946485883e-07,
      "loss": 0.0,
      "step": 66120
    },
    {
      "epoch": 0.9349771663674024,
      "grad_norm": 0.0,
      "learning_rate": 6.547526474289189e-07,
      "loss": 0.3287,
      "step": 66130
    },
    {
      "epoch": 0.9351185510893693,
      "grad_norm": 3.841260642900579e-09,
      "learning_rate": 6.533388002092494e-07,
      "loss": 0.0,
      "step": 66140
    },
    {
      "epoch": 0.9352599358113363,
      "grad_norm": 0.0,
      "learning_rate": 6.519249529895801e-07,
      "loss": 0.0,
      "step": 66150
    },
    {
      "epoch": 0.9354013205333032,
      "grad_norm": 0.0,
      "learning_rate": 6.505111057699105e-07,
      "loss": 0.0001,
      "step": 66160
    },
    {
      "epoch": 0.9355427052552702,
      "grad_norm": 0.0,
      "learning_rate": 6.490972585502411e-07,
      "loss": 0.0,
      "step": 66170
    },
    {
      "epoch": 0.9356840899772371,
      "grad_norm": 0.0,
      "learning_rate": 6.476834113305716e-07,
      "loss": 0.0,
      "step": 66180
    },
    {
      "epoch": 0.935825474699204,
      "grad_norm": 0.0,
      "learning_rate": 6.462695641109022e-07,
      "loss": 0.0,
      "step": 66190
    },
    {
      "epoch": 0.935966859421171,
      "grad_norm": 0.0,
      "learning_rate": 6.448557168912329e-07,
      "loss": 0.0,
      "step": 66200
    },
    {
      "epoch": 0.9361082441431379,
      "grad_norm": 3.319797059475604e-08,
      "learning_rate": 6.434418696715633e-07,
      "loss": 0.0,
      "step": 66210
    },
    {
      "epoch": 0.9362496288651049,
      "grad_norm": 0.27301326394081116,
      "learning_rate": 6.420280224518939e-07,
      "loss": 0.0004,
      "step": 66220
    },
    {
      "epoch": 0.9363910135870718,
      "grad_norm": 0.0965820774435997,
      "learning_rate": 6.406141752322245e-07,
      "loss": 0.0,
      "step": 66230
    },
    {
      "epoch": 0.9365323983090387,
      "grad_norm": 2.583896957730758e-06,
      "learning_rate": 6.39200328012555e-07,
      "loss": 0.0451,
      "step": 66240
    },
    {
      "epoch": 0.9366737830310057,
      "grad_norm": 0.0,
      "learning_rate": 6.377864807928856e-07,
      "loss": 0.0,
      "step": 66250
    },
    {
      "epoch": 0.9368151677529726,
      "grad_norm": 0.11750032752752304,
      "learning_rate": 6.363726335732161e-07,
      "loss": 0.0001,
      "step": 66260
    },
    {
      "epoch": 0.9369565524749396,
      "grad_norm": 0.0,
      "learning_rate": 6.349587863535466e-07,
      "loss": 0.0,
      "step": 66270
    },
    {
      "epoch": 0.9370979371969065,
      "grad_norm": 4.946675291606084e-10,
      "learning_rate": 6.335449391338773e-07,
      "loss": 0.0,
      "step": 66280
    },
    {
      "epoch": 0.9372393219188735,
      "grad_norm": 0.0,
      "learning_rate": 6.321310919142078e-07,
      "loss": 0.0004,
      "step": 66290
    },
    {
      "epoch": 0.9373807066408404,
      "grad_norm": 3.168265294561934e-08,
      "learning_rate": 6.307172446945384e-07,
      "loss": 0.0,
      "step": 66300
    },
    {
      "epoch": 0.9375220913628073,
      "grad_norm": 0.0,
      "learning_rate": 6.293033974748689e-07,
      "loss": 0.0,
      "step": 66310
    },
    {
      "epoch": 0.9376634760847743,
      "grad_norm": 0.0,
      "learning_rate": 6.278895502551994e-07,
      "loss": 0.0005,
      "step": 66320
    },
    {
      "epoch": 0.9378048608067412,
      "grad_norm": 0.0,
      "learning_rate": 6.264757030355301e-07,
      "loss": 0.0,
      "step": 66330
    },
    {
      "epoch": 0.9379462455287082,
      "grad_norm": 0.0,
      "learning_rate": 6.250618558158605e-07,
      "loss": 0.0,
      "step": 66340
    },
    {
      "epoch": 0.9380876302506751,
      "grad_norm": 688.8370361328125,
      "learning_rate": 6.236480085961912e-07,
      "loss": 0.9875,
      "step": 66350
    },
    {
      "epoch": 0.9382290149726421,
      "grad_norm": 0.0,
      "learning_rate": 6.222341613765217e-07,
      "loss": 0.0,
      "step": 66360
    },
    {
      "epoch": 0.938370399694609,
      "grad_norm": 0.0,
      "learning_rate": 6.208203141568522e-07,
      "loss": 0.0,
      "step": 66370
    },
    {
      "epoch": 0.9385117844165759,
      "grad_norm": 0.0,
      "learning_rate": 6.194064669371829e-07,
      "loss": 0.0,
      "step": 66380
    },
    {
      "epoch": 0.9386531691385429,
      "grad_norm": 0.0,
      "learning_rate": 6.179926197175134e-07,
      "loss": 0.0,
      "step": 66390
    },
    {
      "epoch": 0.9387945538605098,
      "grad_norm": 0.0,
      "learning_rate": 6.16578772497844e-07,
      "loss": 0.0,
      "step": 66400
    },
    {
      "epoch": 0.9389359385824768,
      "grad_norm": 3.39009659455769e-07,
      "learning_rate": 6.151649252781745e-07,
      "loss": 0.0,
      "step": 66410
    },
    {
      "epoch": 0.9390773233044437,
      "grad_norm": 0.0,
      "learning_rate": 6.13751078058505e-07,
      "loss": 0.0,
      "step": 66420
    },
    {
      "epoch": 0.9392187080264107,
      "grad_norm": 0.0,
      "learning_rate": 6.123372308388356e-07,
      "loss": 0.0,
      "step": 66430
    },
    {
      "epoch": 0.9393600927483776,
      "grad_norm": 82.58609008789062,
      "learning_rate": 6.109233836191661e-07,
      "loss": 0.018,
      "step": 66440
    },
    {
      "epoch": 0.9395014774703445,
      "grad_norm": 0.0,
      "learning_rate": 6.095095363994967e-07,
      "loss": 0.0,
      "step": 66450
    },
    {
      "epoch": 0.9396428621923115,
      "grad_norm": 0.0,
      "learning_rate": 6.080956891798273e-07,
      "loss": 0.0106,
      "step": 66460
    },
    {
      "epoch": 0.9397842469142784,
      "grad_norm": 0.0,
      "learning_rate": 6.066818419601578e-07,
      "loss": 0.0,
      "step": 66470
    },
    {
      "epoch": 0.9399256316362454,
      "grad_norm": 0.0,
      "learning_rate": 6.052679947404884e-07,
      "loss": 0.0,
      "step": 66480
    },
    {
      "epoch": 0.9400670163582123,
      "grad_norm": 0.0,
      "learning_rate": 6.038541475208189e-07,
      "loss": 0.0,
      "step": 66490
    },
    {
      "epoch": 0.9402084010801792,
      "grad_norm": 0.0,
      "learning_rate": 6.024403003011496e-07,
      "loss": 0.0,
      "step": 66500
    },
    {
      "epoch": 0.9403497858021462,
      "grad_norm": 0.0,
      "learning_rate": 6.010264530814801e-07,
      "loss": 0.0,
      "step": 66510
    },
    {
      "epoch": 0.9404911705241131,
      "grad_norm": 0.0,
      "learning_rate": 5.996126058618106e-07,
      "loss": 0.0,
      "step": 66520
    },
    {
      "epoch": 0.9406325552460801,
      "grad_norm": 0.0,
      "learning_rate": 5.981987586421412e-07,
      "loss": 0.0,
      "step": 66530
    },
    {
      "epoch": 0.940773939968047,
      "grad_norm": 0.0,
      "learning_rate": 5.967849114224717e-07,
      "loss": 0.6844,
      "step": 66540
    },
    {
      "epoch": 0.940915324690014,
      "grad_norm": 0.0,
      "learning_rate": 5.953710642028024e-07,
      "loss": 0.1069,
      "step": 66550
    },
    {
      "epoch": 0.9410567094119809,
      "grad_norm": 0.011856979690492153,
      "learning_rate": 5.939572169831329e-07,
      "loss": 0.0,
      "step": 66560
    },
    {
      "epoch": 0.9411980941339478,
      "grad_norm": 9.318730008089915e-05,
      "learning_rate": 5.925433697634634e-07,
      "loss": 0.0,
      "step": 66570
    },
    {
      "epoch": 0.9413394788559148,
      "grad_norm": 0.0,
      "learning_rate": 5.91129522543794e-07,
      "loss": 0.0,
      "step": 66580
    },
    {
      "epoch": 0.9414808635778817,
      "grad_norm": 0.0,
      "learning_rate": 5.897156753241245e-07,
      "loss": 0.0,
      "step": 66590
    },
    {
      "epoch": 0.9416222482998488,
      "grad_norm": 0.0,
      "learning_rate": 5.88301828104455e-07,
      "loss": 0.0,
      "step": 66600
    },
    {
      "epoch": 0.9417636330218156,
      "grad_norm": 9.631655473185674e-10,
      "learning_rate": 5.868879808847856e-07,
      "loss": 0.0,
      "step": 66610
    },
    {
      "epoch": 0.9419050177437827,
      "grad_norm": 0.0,
      "learning_rate": 5.854741336651161e-07,
      "loss": 0.0,
      "step": 66620
    },
    {
      "epoch": 0.9420464024657496,
      "grad_norm": 0.0,
      "learning_rate": 5.840602864454468e-07,
      "loss": 0.0,
      "step": 66630
    },
    {
      "epoch": 0.9421877871877165,
      "grad_norm": 0.0,
      "learning_rate": 5.826464392257773e-07,
      "loss": 0.0,
      "step": 66640
    },
    {
      "epoch": 0.9423291719096835,
      "grad_norm": 0.0,
      "learning_rate": 5.812325920061079e-07,
      "loss": 0.0,
      "step": 66650
    },
    {
      "epoch": 0.9424705566316504,
      "grad_norm": 0.0009102761396206915,
      "learning_rate": 5.798187447864384e-07,
      "loss": 0.0,
      "step": 66660
    },
    {
      "epoch": 0.9426119413536174,
      "grad_norm": 0.0,
      "learning_rate": 5.78404897566769e-07,
      "loss": 0.0013,
      "step": 66670
    },
    {
      "epoch": 0.9427533260755843,
      "grad_norm": 1.2549669747841108e-07,
      "learning_rate": 5.769910503470996e-07,
      "loss": 0.0,
      "step": 66680
    },
    {
      "epoch": 0.9428947107975513,
      "grad_norm": 0.0,
      "learning_rate": 5.755772031274301e-07,
      "loss": 0.0,
      "step": 66690
    },
    {
      "epoch": 0.9430360955195182,
      "grad_norm": 5.212418273004005e-06,
      "learning_rate": 5.741633559077607e-07,
      "loss": 0.0,
      "step": 66700
    },
    {
      "epoch": 0.9431774802414851,
      "grad_norm": 0.0,
      "learning_rate": 5.727495086880913e-07,
      "loss": 0.0,
      "step": 66710
    },
    {
      "epoch": 0.9433188649634521,
      "grad_norm": 3.688540458679199,
      "learning_rate": 5.713356614684218e-07,
      "loss": 0.0007,
      "step": 66720
    },
    {
      "epoch": 0.943460249685419,
      "grad_norm": 0.0,
      "learning_rate": 5.699218142487524e-07,
      "loss": 0.0,
      "step": 66730
    },
    {
      "epoch": 0.943601634407386,
      "grad_norm": NaN,
      "learning_rate": 5.686493517510498e-07,
      "loss": 0.1483,
      "step": 66740
    },
    {
      "epoch": 0.9437430191293529,
      "grad_norm": 0.00014347850810736418,
      "learning_rate": 5.672355045313804e-07,
      "loss": 0.003,
      "step": 66750
    },
    {
      "epoch": 0.9438844038513198,
      "grad_norm": 0.0,
      "learning_rate": 5.658216573117109e-07,
      "loss": 0.0,
      "step": 66760
    },
    {
      "epoch": 0.9440257885732868,
      "grad_norm": 0.0,
      "learning_rate": 5.644078100920415e-07,
      "loss": 0.0,
      "step": 66770
    },
    {
      "epoch": 0.9441671732952537,
      "grad_norm": 0.0,
      "learning_rate": 5.62993962872372e-07,
      "loss": 0.0,
      "step": 66780
    },
    {
      "epoch": 0.9443085580172207,
      "grad_norm": 0.0,
      "learning_rate": 5.615801156527025e-07,
      "loss": 0.0,
      "step": 66790
    },
    {
      "epoch": 0.9444499427391876,
      "grad_norm": 0.0,
      "learning_rate": 5.601662684330332e-07,
      "loss": 0.0,
      "step": 66800
    },
    {
      "epoch": 0.9445913274611546,
      "grad_norm": 0.0,
      "learning_rate": 5.587524212133637e-07,
      "loss": 0.0,
      "step": 66810
    },
    {
      "epoch": 0.9447327121831215,
      "grad_norm": 0.0,
      "learning_rate": 5.573385739936943e-07,
      "loss": 0.0,
      "step": 66820
    },
    {
      "epoch": 0.9448740969050884,
      "grad_norm": 0.00033792154863476753,
      "learning_rate": 5.559247267740248e-07,
      "loss": 0.0,
      "step": 66830
    },
    {
      "epoch": 0.9450154816270554,
      "grad_norm": 0.0,
      "learning_rate": 5.545108795543553e-07,
      "loss": 0.0,
      "step": 66840
    },
    {
      "epoch": 0.9451568663490223,
      "grad_norm": 0.0,
      "learning_rate": 5.53097032334686e-07,
      "loss": 0.0,
      "step": 66850
    },
    {
      "epoch": 0.9452982510709893,
      "grad_norm": 0.0,
      "learning_rate": 5.516831851150165e-07,
      "loss": 0.017,
      "step": 66860
    },
    {
      "epoch": 0.9454396357929562,
      "grad_norm": 0.0,
      "learning_rate": 5.502693378953471e-07,
      "loss": 0.0,
      "step": 66870
    },
    {
      "epoch": 0.9455810205149232,
      "grad_norm": 0.0,
      "learning_rate": 5.488554906756776e-07,
      "loss": 0.0,
      "step": 66880
    },
    {
      "epoch": 0.9457224052368901,
      "grad_norm": 0.0,
      "learning_rate": 5.474416434560082e-07,
      "loss": 0.0,
      "step": 66890
    },
    {
      "epoch": 0.945863789958857,
      "grad_norm": 817.080322265625,
      "learning_rate": 5.460277962363388e-07,
      "loss": 0.9607,
      "step": 66900
    },
    {
      "epoch": 0.946005174680824,
      "grad_norm": 6.23624146101065e-05,
      "learning_rate": 5.446139490166693e-07,
      "loss": 0.0,
      "step": 66910
    },
    {
      "epoch": 0.9461465594027909,
      "grad_norm": 0.0,
      "learning_rate": 5.432001017969999e-07,
      "loss": 0.0,
      "step": 66920
    },
    {
      "epoch": 0.9462879441247579,
      "grad_norm": 0.0,
      "learning_rate": 5.417862545773304e-07,
      "loss": 0.0,
      "step": 66930
    },
    {
      "epoch": 0.9464293288467248,
      "grad_norm": 0.06059373542666435,
      "learning_rate": 5.403724073576609e-07,
      "loss": 0.0,
      "step": 66940
    },
    {
      "epoch": 0.9465707135686917,
      "grad_norm": 1.9778368368861265e-05,
      "learning_rate": 5.389585601379915e-07,
      "loss": 0.0,
      "step": 66950
    },
    {
      "epoch": 0.9467120982906587,
      "grad_norm": 0.0,
      "learning_rate": 5.37544712918322e-07,
      "loss": 0.0,
      "step": 66960
    },
    {
      "epoch": 0.9468534830126256,
      "grad_norm": 0.0,
      "learning_rate": 5.361308656986527e-07,
      "loss": 0.0,
      "step": 66970
    },
    {
      "epoch": 0.9469948677345926,
      "grad_norm": 0.0,
      "learning_rate": 5.347170184789832e-07,
      "loss": 0.0,
      "step": 66980
    },
    {
      "epoch": 0.9471362524565595,
      "grad_norm": 0.0,
      "learning_rate": 5.333031712593137e-07,
      "loss": 0.0,
      "step": 66990
    },
    {
      "epoch": 0.9472776371785265,
      "grad_norm": 0.0,
      "learning_rate": 5.318893240396443e-07,
      "loss": 0.0,
      "step": 67000
    },
    {
      "epoch": 0.9474190219004934,
      "grad_norm": 0.0,
      "learning_rate": 5.304754768199749e-07,
      "loss": 0.0,
      "step": 67010
    },
    {
      "epoch": 0.9475604066224603,
      "grad_norm": 6.311957258731127e-05,
      "learning_rate": 5.290616296003055e-07,
      "loss": 0.0,
      "step": 67020
    },
    {
      "epoch": 0.9477017913444273,
      "grad_norm": 0.0,
      "learning_rate": 5.27647782380636e-07,
      "loss": 0.0,
      "step": 67030
    },
    {
      "epoch": 0.9478431760663942,
      "grad_norm": 0.0,
      "learning_rate": 5.262339351609665e-07,
      "loss": 0.0,
      "step": 67040
    },
    {
      "epoch": 0.9479845607883612,
      "grad_norm": 0.0,
      "learning_rate": 5.248200879412971e-07,
      "loss": 0.0,
      "step": 67050
    },
    {
      "epoch": 0.9481259455103281,
      "grad_norm": 0.0,
      "learning_rate": 5.234062407216277e-07,
      "loss": 0.0,
      "step": 67060
    },
    {
      "epoch": 0.9482673302322951,
      "grad_norm": 0.0,
      "learning_rate": 5.219923935019583e-07,
      "loss": 0.0,
      "step": 67070
    },
    {
      "epoch": 0.948408714954262,
      "grad_norm": 0.0,
      "learning_rate": 5.205785462822888e-07,
      "loss": 0.0,
      "step": 67080
    },
    {
      "epoch": 0.9485500996762289,
      "grad_norm": 4.2855526771745645e-06,
      "learning_rate": 5.191646990626193e-07,
      "loss": 0.0035,
      "step": 67090
    },
    {
      "epoch": 0.948691484398196,
      "grad_norm": 3.432682049719915e-08,
      "learning_rate": 5.177508518429499e-07,
      "loss": 0.0,
      "step": 67100
    },
    {
      "epoch": 0.9488328691201628,
      "grad_norm": 0.0,
      "learning_rate": 5.163370046232804e-07,
      "loss": 0.0,
      "step": 67110
    },
    {
      "epoch": 0.9489742538421299,
      "grad_norm": 8.275601430796087e-07,
      "learning_rate": 5.14923157403611e-07,
      "loss": 0.0,
      "step": 67120
    },
    {
      "epoch": 0.9491156385640968,
      "grad_norm": 0.0,
      "learning_rate": 5.135093101839415e-07,
      "loss": 0.0,
      "step": 67130
    },
    {
      "epoch": 0.9492570232860638,
      "grad_norm": 0.0,
      "learning_rate": 5.120954629642721e-07,
      "loss": 0.0,
      "step": 67140
    },
    {
      "epoch": 0.9493984080080307,
      "grad_norm": 4.440753936767578,
      "learning_rate": 5.106816157446027e-07,
      "loss": 0.0747,
      "step": 67150
    },
    {
      "epoch": 0.9495397927299976,
      "grad_norm": 0.0,
      "learning_rate": 5.092677685249332e-07,
      "loss": 0.0,
      "step": 67160
    },
    {
      "epoch": 0.9496811774519646,
      "grad_norm": 0.0,
      "learning_rate": 5.078539213052638e-07,
      "loss": 0.0,
      "step": 67170
    },
    {
      "epoch": 0.9498225621739315,
      "grad_norm": 0.0,
      "learning_rate": 5.064400740855944e-07,
      "loss": 0.0,
      "step": 67180
    },
    {
      "epoch": 0.9499639468958985,
      "grad_norm": 0.0,
      "learning_rate": 5.050262268659249e-07,
      "loss": 0.0,
      "step": 67190
    },
    {
      "epoch": 0.9501053316178654,
      "grad_norm": 0.0,
      "learning_rate": 5.036123796462555e-07,
      "loss": 0.0001,
      "step": 67200
    },
    {
      "epoch": 0.9502467163398323,
      "grad_norm": 0.0,
      "learning_rate": 5.02198532426586e-07,
      "loss": 0.0,
      "step": 67210
    },
    {
      "epoch": 0.9503881010617993,
      "grad_norm": 0.00024434173246845603,
      "learning_rate": 5.007846852069167e-07,
      "loss": 0.0,
      "step": 67220
    },
    {
      "epoch": 0.9505294857837662,
      "grad_norm": 0.0,
      "learning_rate": 4.993708379872472e-07,
      "loss": 0.0,
      "step": 67230
    },
    {
      "epoch": 0.9506708705057332,
      "grad_norm": 1.4783702226850437e-06,
      "learning_rate": 4.979569907675777e-07,
      "loss": 0.0,
      "step": 67240
    },
    {
      "epoch": 0.9508122552277001,
      "grad_norm": 0.0,
      "learning_rate": 4.965431435479083e-07,
      "loss": 0.0105,
      "step": 67250
    },
    {
      "epoch": 0.9509536399496671,
      "grad_norm": 0.0,
      "learning_rate": 4.951292963282388e-07,
      "loss": 0.0,
      "step": 67260
    },
    {
      "epoch": 0.951095024671634,
      "grad_norm": 0.0,
      "learning_rate": 4.937154491085694e-07,
      "loss": 0.0,
      "step": 67270
    },
    {
      "epoch": 0.9512364093936009,
      "grad_norm": 0.0001313764660153538,
      "learning_rate": 4.923016018888999e-07,
      "loss": 0.0,
      "step": 67280
    },
    {
      "epoch": 0.9513777941155679,
      "grad_norm": 0.0,
      "learning_rate": 4.908877546692304e-07,
      "loss": 0.0,
      "step": 67290
    },
    {
      "epoch": 0.9515191788375348,
      "grad_norm": 0.0,
      "learning_rate": 4.89473907449561e-07,
      "loss": 0.0037,
      "step": 67300
    },
    {
      "epoch": 0.9516605635595018,
      "grad_norm": 0.0,
      "learning_rate": 4.880600602298916e-07,
      "loss": 0.0,
      "step": 67310
    },
    {
      "epoch": 0.9518019482814687,
      "grad_norm": 0.0,
      "learning_rate": 4.866462130102222e-07,
      "loss": 0.0,
      "step": 67320
    },
    {
      "epoch": 0.9519433330034357,
      "grad_norm": 0.0,
      "learning_rate": 4.852323657905527e-07,
      "loss": 0.0,
      "step": 67330
    },
    {
      "epoch": 0.9520847177254026,
      "grad_norm": 0.0,
      "learning_rate": 4.838185185708832e-07,
      "loss": 0.0,
      "step": 67340
    },
    {
      "epoch": 0.9522261024473695,
      "grad_norm": 5.101021272047035e-10,
      "learning_rate": 4.824046713512139e-07,
      "loss": 0.0,
      "step": 67350
    },
    {
      "epoch": 0.9523674871693365,
      "grad_norm": 0.0,
      "learning_rate": 4.809908241315444e-07,
      "loss": 0.0,
      "step": 67360
    },
    {
      "epoch": 0.9525088718913034,
      "grad_norm": 0.0,
      "learning_rate": 4.79576976911875e-07,
      "loss": 0.0,
      "step": 67370
    },
    {
      "epoch": 0.9526502566132704,
      "grad_norm": 0.0,
      "learning_rate": 4.781631296922055e-07,
      "loss": 0.5973,
      "step": 67380
    },
    {
      "epoch": 0.9527916413352373,
      "grad_norm": 0.026002176105976105,
      "learning_rate": 4.767492824725361e-07,
      "loss": 0.0,
      "step": 67390
    },
    {
      "epoch": 0.9529330260572043,
      "grad_norm": 0.0,
      "learning_rate": 4.7533543525286663e-07,
      "loss": 0.0358,
      "step": 67400
    },
    {
      "epoch": 0.9530744107791712,
      "grad_norm": 1.700232701296045e-07,
      "learning_rate": 4.7392158803319717e-07,
      "loss": 0.0,
      "step": 67410
    },
    {
      "epoch": 0.9532157955011381,
      "grad_norm": 4.5509822577116665e-09,
      "learning_rate": 4.725077408135277e-07,
      "loss": 0.0,
      "step": 67420
    },
    {
      "epoch": 0.9533571802231051,
      "grad_norm": 0.12837886810302734,
      "learning_rate": 4.710938935938583e-07,
      "loss": 0.0,
      "step": 67430
    },
    {
      "epoch": 0.953498564945072,
      "grad_norm": 4.710210554925709e-10,
      "learning_rate": 4.6968004637418884e-07,
      "loss": 0.0,
      "step": 67440
    },
    {
      "epoch": 0.953639949667039,
      "grad_norm": 0.0,
      "learning_rate": 4.682661991545194e-07,
      "loss": 0.0,
      "step": 67450
    },
    {
      "epoch": 0.9537813343890059,
      "grad_norm": 0.006233518011868,
      "learning_rate": 4.6685235193484997e-07,
      "loss": 0.0,
      "step": 67460
    },
    {
      "epoch": 0.9539227191109728,
      "grad_norm": 4.4140138766302073e-10,
      "learning_rate": 4.654385047151805e-07,
      "loss": 0.0,
      "step": 67470
    },
    {
      "epoch": 0.9540641038329398,
      "grad_norm": 0.0,
      "learning_rate": 4.640246574955111e-07,
      "loss": 0.0,
      "step": 67480
    },
    {
      "epoch": 0.9542054885549067,
      "grad_norm": 0.0,
      "learning_rate": 4.6261081027584164e-07,
      "loss": 0.0,
      "step": 67490
    },
    {
      "epoch": 0.9543468732768737,
      "grad_norm": 3.534246229008886e-09,
      "learning_rate": 4.611969630561722e-07,
      "loss": 0.0,
      "step": 67500
    },
    {
      "epoch": 0.9544882579988406,
      "grad_norm": 0.0,
      "learning_rate": 4.597831158365027e-07,
      "loss": 0.0,
      "step": 67510
    },
    {
      "epoch": 0.9546296427208076,
      "grad_norm": 0.0,
      "learning_rate": 4.583692686168333e-07,
      "loss": 0.0,
      "step": 67520
    },
    {
      "epoch": 0.9547710274427745,
      "grad_norm": 4.87495988021891e-10,
      "learning_rate": 4.5695542139716385e-07,
      "loss": 0.0,
      "step": 67530
    },
    {
      "epoch": 0.9549124121647414,
      "grad_norm": 2.9631356923687235e-08,
      "learning_rate": 4.555415741774944e-07,
      "loss": 0.0925,
      "step": 67540
    },
    {
      "epoch": 0.9550537968867084,
      "grad_norm": 0.0,
      "learning_rate": 4.541277269578249e-07,
      "loss": 0.0,
      "step": 67550
    },
    {
      "epoch": 0.9551951816086753,
      "grad_norm": 0.0,
      "learning_rate": 4.5271387973815557e-07,
      "loss": 0.0001,
      "step": 67560
    },
    {
      "epoch": 0.9553365663306423,
      "grad_norm": 2.877506403820007e-06,
      "learning_rate": 4.513000325184861e-07,
      "loss": 0.0,
      "step": 67570
    },
    {
      "epoch": 0.9554779510526092,
      "grad_norm": 0.0,
      "learning_rate": 4.4988618529881665e-07,
      "loss": 0.0002,
      "step": 67580
    },
    {
      "epoch": 0.9556193357745763,
      "grad_norm": 0.0,
      "learning_rate": 4.484723380791472e-07,
      "loss": 0.0,
      "step": 67590
    },
    {
      "epoch": 0.9557607204965431,
      "grad_norm": 0.0,
      "learning_rate": 4.470584908594778e-07,
      "loss": 0.0,
      "step": 67600
    },
    {
      "epoch": 0.95590210521851,
      "grad_norm": 0.0,
      "learning_rate": 4.456446436398083e-07,
      "loss": 0.0,
      "step": 67610
    },
    {
      "epoch": 0.956043489940477,
      "grad_norm": 0.0,
      "learning_rate": 4.4423079642013885e-07,
      "loss": 0.0,
      "step": 67620
    },
    {
      "epoch": 0.956184874662444,
      "grad_norm": 10.502239227294922,
      "learning_rate": 4.428169492004694e-07,
      "loss": 0.0025,
      "step": 67630
    },
    {
      "epoch": 0.956326259384411,
      "grad_norm": 0.0,
      "learning_rate": 4.4140310198080004e-07,
      "loss": 0.0002,
      "step": 67640
    },
    {
      "epoch": 0.9564676441063779,
      "grad_norm": 0.0,
      "learning_rate": 4.399892547611306e-07,
      "loss": 0.0,
      "step": 67650
    },
    {
      "epoch": 0.9566090288283448,
      "grad_norm": 8.982792842893161e-10,
      "learning_rate": 4.385754075414611e-07,
      "loss": 0.0,
      "step": 67660
    },
    {
      "epoch": 0.9567504135503118,
      "grad_norm": 0.0,
      "learning_rate": 4.3716156032179165e-07,
      "loss": 0.0,
      "step": 67670
    },
    {
      "epoch": 0.9568917982722787,
      "grad_norm": 0.0,
      "learning_rate": 4.357477131021222e-07,
      "loss": 0.0,
      "step": 67680
    },
    {
      "epoch": 0.9570331829942457,
      "grad_norm": 0.0,
      "learning_rate": 4.343338658824528e-07,
      "loss": 0.0368,
      "step": 67690
    },
    {
      "epoch": 0.9571745677162126,
      "grad_norm": 0.0,
      "learning_rate": 4.329200186627833e-07,
      "loss": 0.0,
      "step": 67700
    },
    {
      "epoch": 0.9573159524381796,
      "grad_norm": 0.1375178098678589,
      "learning_rate": 4.3150617144311386e-07,
      "loss": 0.0,
      "step": 67710
    },
    {
      "epoch": 0.9574573371601465,
      "grad_norm": 0.0,
      "learning_rate": 4.300923242234444e-07,
      "loss": 0.0,
      "step": 67720
    },
    {
      "epoch": 0.9575987218821134,
      "grad_norm": 0.0,
      "learning_rate": 4.2867847700377505e-07,
      "loss": 0.0139,
      "step": 67730
    },
    {
      "epoch": 0.9577401066040804,
      "grad_norm": 0.0,
      "learning_rate": 4.272646297841056e-07,
      "loss": 0.0,
      "step": 67740
    },
    {
      "epoch": 0.9578814913260473,
      "grad_norm": 0.0,
      "learning_rate": 4.258507825644361e-07,
      "loss": 0.0,
      "step": 67750
    },
    {
      "epoch": 0.9580228760480143,
      "grad_norm": 6.817040411988273e-05,
      "learning_rate": 4.2443693534476666e-07,
      "loss": 0.0,
      "step": 67760
    },
    {
      "epoch": 0.9581642607699812,
      "grad_norm": 0.0,
      "learning_rate": 4.2302308812509725e-07,
      "loss": 0.0,
      "step": 67770
    },
    {
      "epoch": 0.9583056454919482,
      "grad_norm": 0.0,
      "learning_rate": 4.216092409054278e-07,
      "loss": 0.0048,
      "step": 67780
    },
    {
      "epoch": 0.9584470302139151,
      "grad_norm": 0.0,
      "learning_rate": 4.2019539368575833e-07,
      "loss": 0.0,
      "step": 67790
    },
    {
      "epoch": 0.958588414935882,
      "grad_norm": 0.0,
      "learning_rate": 4.1878154646608887e-07,
      "loss": 0.0,
      "step": 67800
    },
    {
      "epoch": 0.958729799657849,
      "grad_norm": 0.0,
      "learning_rate": 4.173676992464195e-07,
      "loss": 0.0,
      "step": 67810
    },
    {
      "epoch": 0.9588711843798159,
      "grad_norm": 0.0,
      "learning_rate": 4.1595385202675005e-07,
      "loss": 0.0,
      "step": 67820
    },
    {
      "epoch": 0.9590125691017829,
      "grad_norm": 0.0,
      "learning_rate": 4.145400048070806e-07,
      "loss": 0.0,
      "step": 67830
    },
    {
      "epoch": 0.9591539538237498,
      "grad_norm": 0.0,
      "learning_rate": 4.1312615758741113e-07,
      "loss": 0.0,
      "step": 67840
    },
    {
      "epoch": 0.9592953385457168,
      "grad_norm": 0.0,
      "learning_rate": 4.117123103677417e-07,
      "loss": 0.0,
      "step": 67850
    },
    {
      "epoch": 0.9594367232676837,
      "grad_norm": 0.0,
      "learning_rate": 4.1029846314807226e-07,
      "loss": 0.0,
      "step": 67860
    },
    {
      "epoch": 0.9595781079896506,
      "grad_norm": 0.0,
      "learning_rate": 4.088846159284028e-07,
      "loss": 0.0,
      "step": 67870
    },
    {
      "epoch": 0.9597194927116176,
      "grad_norm": 0.0,
      "learning_rate": 4.0747076870873334e-07,
      "loss": 0.0,
      "step": 67880
    },
    {
      "epoch": 0.9598608774335845,
      "grad_norm": 0.0,
      "learning_rate": 4.060569214890639e-07,
      "loss": 0.0,
      "step": 67890
    },
    {
      "epoch": 0.9600022621555515,
      "grad_norm": 1.247615077915043e-08,
      "learning_rate": 4.046430742693945e-07,
      "loss": 0.004,
      "step": 67900
    },
    {
      "epoch": 0.9601436468775184,
      "grad_norm": 75.21918487548828,
      "learning_rate": 4.0322922704972506e-07,
      "loss": 0.0225,
      "step": 67910
    },
    {
      "epoch": 0.9602850315994853,
      "grad_norm": 0.0,
      "learning_rate": 4.018153798300556e-07,
      "loss": 0.0001,
      "step": 67920
    },
    {
      "epoch": 0.9604264163214523,
      "grad_norm": 0.0,
      "learning_rate": 4.0040153261038614e-07,
      "loss": 0.0,
      "step": 67930
    },
    {
      "epoch": 0.9605678010434192,
      "grad_norm": 0.0,
      "learning_rate": 3.9898768539071673e-07,
      "loss": 0.0,
      "step": 67940
    },
    {
      "epoch": 0.9607091857653862,
      "grad_norm": 0.0,
      "learning_rate": 3.9757383817104727e-07,
      "loss": 0.0,
      "step": 67950
    },
    {
      "epoch": 0.9608505704873531,
      "grad_norm": 0.0,
      "learning_rate": 3.961599909513778e-07,
      "loss": 0.0,
      "step": 67960
    },
    {
      "epoch": 0.9609919552093201,
      "grad_norm": 0.0,
      "learning_rate": 3.9474614373170835e-07,
      "loss": 0.0052,
      "step": 67970
    },
    {
      "epoch": 0.961133339931287,
      "grad_norm": 0.0,
      "learning_rate": 3.93332296512039e-07,
      "loss": 0.0,
      "step": 67980
    },
    {
      "epoch": 0.9612747246532539,
      "grad_norm": 0.0,
      "learning_rate": 3.9191844929236953e-07,
      "loss": 0.0005,
      "step": 67990
    },
    {
      "epoch": 0.9614161093752209,
      "grad_norm": 0.0,
      "learning_rate": 3.9050460207270007e-07,
      "loss": 0.0647,
      "step": 68000
    },
    {
      "epoch": 0.9615574940971878,
      "grad_norm": 0.07341185212135315,
      "learning_rate": 3.890907548530306e-07,
      "loss": 0.0,
      "step": 68010
    },
    {
      "epoch": 0.9616988788191548,
      "grad_norm": 0.19064635038375854,
      "learning_rate": 3.876769076333612e-07,
      "loss": 0.0001,
      "step": 68020
    },
    {
      "epoch": 0.9618402635411217,
      "grad_norm": 4.414837384558723e-10,
      "learning_rate": 3.8626306041369174e-07,
      "loss": 0.0,
      "step": 68030
    },
    {
      "epoch": 0.9619816482630887,
      "grad_norm": 0.0,
      "learning_rate": 3.848492131940223e-07,
      "loss": 0.0001,
      "step": 68040
    },
    {
      "epoch": 0.9621230329850556,
      "grad_norm": 0.0,
      "learning_rate": 3.834353659743528e-07,
      "loss": 0.0,
      "step": 68050
    },
    {
      "epoch": 0.9622644177070225,
      "grad_norm": 1.0448632110637845e-06,
      "learning_rate": 3.820215187546834e-07,
      "loss": 0.0,
      "step": 68060
    },
    {
      "epoch": 0.9624058024289895,
      "grad_norm": 0.0,
      "learning_rate": 3.80607671535014e-07,
      "loss": 0.0002,
      "step": 68070
    },
    {
      "epoch": 0.9625471871509564,
      "grad_norm": 1.6676109453328536e-06,
      "learning_rate": 3.7919382431534454e-07,
      "loss": 0.0,
      "step": 68080
    },
    {
      "epoch": 0.9626885718729234,
      "grad_norm": 0.0,
      "learning_rate": 3.777799770956751e-07,
      "loss": 0.0,
      "step": 68090
    },
    {
      "epoch": 0.9628299565948903,
      "grad_norm": 0.0,
      "learning_rate": 3.763661298760056e-07,
      "loss": 0.0,
      "step": 68100
    },
    {
      "epoch": 0.9629713413168574,
      "grad_norm": 0.0,
      "learning_rate": 3.749522826563362e-07,
      "loss": 0.0,
      "step": 68110
    },
    {
      "epoch": 0.9631127260388243,
      "grad_norm": 0.0,
      "learning_rate": 3.7353843543666675e-07,
      "loss": 0.0004,
      "step": 68120
    },
    {
      "epoch": 0.9632541107607911,
      "grad_norm": 0.0,
      "learning_rate": 3.721245882169973e-07,
      "loss": 0.0,
      "step": 68130
    },
    {
      "epoch": 0.9633954954827582,
      "grad_norm": 0.0,
      "learning_rate": 3.707107409973278e-07,
      "loss": 0.0,
      "step": 68140
    },
    {
      "epoch": 0.963536880204725,
      "grad_norm": 0.0,
      "learning_rate": 3.692968937776584e-07,
      "loss": 0.0,
      "step": 68150
    },
    {
      "epoch": 0.9636782649266921,
      "grad_norm": 1.8460049489021912e-08,
      "learning_rate": 3.6788304655798896e-07,
      "loss": 0.0,
      "step": 68160
    },
    {
      "epoch": 0.963819649648659,
      "grad_norm": 0.0,
      "learning_rate": 3.6646919933831955e-07,
      "loss": 0.0,
      "step": 68170
    },
    {
      "epoch": 0.9639610343706259,
      "grad_norm": 0.0,
      "learning_rate": 3.650553521186501e-07,
      "loss": 0.0,
      "step": 68180
    },
    {
      "epoch": 0.9641024190925929,
      "grad_norm": 0.0,
      "learning_rate": 3.636415048989807e-07,
      "loss": 0.0001,
      "step": 68190
    },
    {
      "epoch": 0.9642438038145598,
      "grad_norm": 0.01560512837022543,
      "learning_rate": 3.622276576793112e-07,
      "loss": 0.0003,
      "step": 68200
    },
    {
      "epoch": 0.9643851885365268,
      "grad_norm": 0.0,
      "learning_rate": 3.6081381045964176e-07,
      "loss": 0.0,
      "step": 68210
    },
    {
      "epoch": 0.9645265732584937,
      "grad_norm": 0.0,
      "learning_rate": 3.593999632399723e-07,
      "loss": 0.0,
      "step": 68220
    },
    {
      "epoch": 0.9646679579804607,
      "grad_norm": 0.0,
      "learning_rate": 3.579861160203029e-07,
      "loss": 0.0,
      "step": 68230
    },
    {
      "epoch": 0.9648093427024276,
      "grad_norm": 0.0,
      "learning_rate": 3.565722688006334e-07,
      "loss": 0.0,
      "step": 68240
    },
    {
      "epoch": 0.9649507274243945,
      "grad_norm": 0.0002916215162258595,
      "learning_rate": 3.5515842158096397e-07,
      "loss": 0.0,
      "step": 68250
    },
    {
      "epoch": 0.9650921121463615,
      "grad_norm": 8.171749868779443e-06,
      "learning_rate": 3.537445743612945e-07,
      "loss": 0.0,
      "step": 68260
    },
    {
      "epoch": 0.9652334968683284,
      "grad_norm": 0.0,
      "learning_rate": 3.523307271416251e-07,
      "loss": 0.0,
      "step": 68270
    },
    {
      "epoch": 0.9653748815902954,
      "grad_norm": 0.00028899512835778296,
      "learning_rate": 3.509168799219557e-07,
      "loss": 0.0,
      "step": 68280
    },
    {
      "epoch": 0.9655162663122623,
      "grad_norm": 5.032824788031576e-07,
      "learning_rate": 3.495030327022862e-07,
      "loss": 0.0,
      "step": 68290
    },
    {
      "epoch": 0.9656576510342293,
      "grad_norm": 0.0,
      "learning_rate": 3.4808918548261677e-07,
      "loss": 0.0,
      "step": 68300
    },
    {
      "epoch": 0.9657990357561962,
      "grad_norm": 0.0,
      "learning_rate": 3.466753382629473e-07,
      "loss": 0.0,
      "step": 68310
    },
    {
      "epoch": 0.9659404204781631,
      "grad_norm": 0.0,
      "learning_rate": 3.452614910432779e-07,
      "loss": 0.0,
      "step": 68320
    },
    {
      "epoch": 0.9660818052001301,
      "grad_norm": 1.222361305508457e-07,
      "learning_rate": 3.4384764382360843e-07,
      "loss": 0.0,
      "step": 68330
    },
    {
      "epoch": 0.966223189922097,
      "grad_norm": 0.0,
      "learning_rate": 3.42433796603939e-07,
      "loss": 0.0,
      "step": 68340
    },
    {
      "epoch": 0.966364574644064,
      "grad_norm": 55.3059196472168,
      "learning_rate": 3.410199493842695e-07,
      "loss": 0.0116,
      "step": 68350
    },
    {
      "epoch": 0.9665059593660309,
      "grad_norm": 0.0,
      "learning_rate": 3.3960610216460016e-07,
      "loss": 0.0,
      "step": 68360
    },
    {
      "epoch": 0.9666473440879979,
      "grad_norm": 0.0,
      "learning_rate": 3.381922549449307e-07,
      "loss": 0.0,
      "step": 68370
    },
    {
      "epoch": 0.9667887288099648,
      "grad_norm": 0.0,
      "learning_rate": 3.3677840772526123e-07,
      "loss": 0.0,
      "step": 68380
    },
    {
      "epoch": 0.9669301135319317,
      "grad_norm": 0.0,
      "learning_rate": 3.353645605055918e-07,
      "loss": 0.0,
      "step": 68390
    },
    {
      "epoch": 0.9670714982538987,
      "grad_norm": 0.0,
      "learning_rate": 3.3395071328592237e-07,
      "loss": 0.0,
      "step": 68400
    },
    {
      "epoch": 0.9672128829758656,
      "grad_norm": 9.558602798165339e-10,
      "learning_rate": 3.325368660662529e-07,
      "loss": 0.0,
      "step": 68410
    },
    {
      "epoch": 0.9673542676978326,
      "grad_norm": 0.0,
      "learning_rate": 3.3112301884658344e-07,
      "loss": 0.0,
      "step": 68420
    },
    {
      "epoch": 0.9674956524197995,
      "grad_norm": 0.0,
      "learning_rate": 3.29709171626914e-07,
      "loss": 0.0002,
      "step": 68430
    },
    {
      "epoch": 0.9676370371417664,
      "grad_norm": 0.0,
      "learning_rate": 3.2829532440724463e-07,
      "loss": 0.0,
      "step": 68440
    },
    {
      "epoch": 0.9677784218637334,
      "grad_norm": 0.0,
      "learning_rate": 3.2688147718757517e-07,
      "loss": 0.0,
      "step": 68450
    },
    {
      "epoch": 0.9679198065857003,
      "grad_norm": 0.0,
      "learning_rate": 3.254676299679057e-07,
      "loss": 0.0,
      "step": 68460
    },
    {
      "epoch": 0.9680611913076673,
      "grad_norm": 0.0,
      "learning_rate": 3.2405378274823624e-07,
      "loss": 0.0,
      "step": 68470
    },
    {
      "epoch": 0.9682025760296342,
      "grad_norm": 5.049143880775375e-10,
      "learning_rate": 3.226399355285668e-07,
      "loss": 0.0,
      "step": 68480
    },
    {
      "epoch": 0.9683439607516012,
      "grad_norm": 0.0,
      "learning_rate": 3.212260883088974e-07,
      "loss": 0.0002,
      "step": 68490
    },
    {
      "epoch": 0.9684853454735681,
      "grad_norm": 0.0,
      "learning_rate": 3.198122410892279e-07,
      "loss": 0.0004,
      "step": 68500
    },
    {
      "epoch": 0.968626730195535,
      "grad_norm": 0.0,
      "learning_rate": 3.1839839386955845e-07,
      "loss": 0.0002,
      "step": 68510
    },
    {
      "epoch": 0.968768114917502,
      "grad_norm": 0.0,
      "learning_rate": 3.16984546649889e-07,
      "loss": 0.0,
      "step": 68520
    },
    {
      "epoch": 0.9689094996394689,
      "grad_norm": 0.0,
      "learning_rate": 3.1557069943021963e-07,
      "loss": 0.0001,
      "step": 68530
    },
    {
      "epoch": 0.9690508843614359,
      "grad_norm": 0.0,
      "learning_rate": 3.141568522105502e-07,
      "loss": 0.0,
      "step": 68540
    },
    {
      "epoch": 0.9691922690834028,
      "grad_norm": 0.0,
      "learning_rate": 3.127430049908807e-07,
      "loss": 0.0,
      "step": 68550
    },
    {
      "epoch": 0.9693336538053698,
      "grad_norm": 0.0,
      "learning_rate": 3.1132915777121125e-07,
      "loss": 0.0,
      "step": 68560
    },
    {
      "epoch": 0.9694750385273367,
      "grad_norm": 7.07127692294307e-05,
      "learning_rate": 3.099153105515418e-07,
      "loss": 0.0162,
      "step": 68570
    },
    {
      "epoch": 0.9696164232493036,
      "grad_norm": 0.0,
      "learning_rate": 3.085014633318724e-07,
      "loss": 0.0,
      "step": 68580
    },
    {
      "epoch": 0.9697578079712706,
      "grad_norm": 0.0,
      "learning_rate": 3.070876161122029e-07,
      "loss": 0.0,
      "step": 68590
    },
    {
      "epoch": 0.9698991926932375,
      "grad_norm": 930.4056396484375,
      "learning_rate": 3.056737688925335e-07,
      "loss": 1.1063,
      "step": 68600
    },
    {
      "epoch": 0.9700405774152046,
      "grad_norm": 0.0,
      "learning_rate": 3.0425992167286405e-07,
      "loss": 0.0,
      "step": 68610
    },
    {
      "epoch": 0.9701819621371714,
      "grad_norm": 0.0,
      "learning_rate": 3.0284607445319464e-07,
      "loss": 0.0001,
      "step": 68620
    },
    {
      "epoch": 0.9703233468591383,
      "grad_norm": 0.0,
      "learning_rate": 3.014322272335252e-07,
      "loss": 0.0,
      "step": 68630
    },
    {
      "epoch": 0.9704647315811054,
      "grad_norm": 0.0,
      "learning_rate": 3.000183800138557e-07,
      "loss": 0.0,
      "step": 68640
    },
    {
      "epoch": 0.9706061163030723,
      "grad_norm": 0.0,
      "learning_rate": 2.9860453279418626e-07,
      "loss": 0.0,
      "step": 68650
    },
    {
      "epoch": 0.9707475010250393,
      "grad_norm": 0.0,
      "learning_rate": 2.9719068557451685e-07,
      "loss": 0.0,
      "step": 68660
    },
    {
      "epoch": 0.9708888857470062,
      "grad_norm": 0.0,
      "learning_rate": 2.957768383548474e-07,
      "loss": 0.0,
      "step": 68670
    },
    {
      "epoch": 0.9710302704689732,
      "grad_norm": 0.0,
      "learning_rate": 2.94362991135178e-07,
      "loss": 0.0,
      "step": 68680
    },
    {
      "epoch": 0.9711716551909401,
      "grad_norm": 0.0,
      "learning_rate": 2.929491439155085e-07,
      "loss": 0.0,
      "step": 68690
    },
    {
      "epoch": 0.971313039912907,
      "grad_norm": 0.0,
      "learning_rate": 2.9153529669583906e-07,
      "loss": 0.0,
      "step": 68700
    },
    {
      "epoch": 0.971454424634874,
      "grad_norm": 0.0,
      "learning_rate": 2.9012144947616965e-07,
      "loss": 0.0,
      "step": 68710
    },
    {
      "epoch": 0.9715958093568409,
      "grad_norm": 1.1162166657641137e-07,
      "learning_rate": 2.887076022565002e-07,
      "loss": 0.0,
      "step": 68720
    },
    {
      "epoch": 0.9717371940788079,
      "grad_norm": 0.0,
      "learning_rate": 2.8729375503683073e-07,
      "loss": 0.0,
      "step": 68730
    },
    {
      "epoch": 0.9718785788007748,
      "grad_norm": 0.0,
      "learning_rate": 2.8587990781716127e-07,
      "loss": 0.0248,
      "step": 68740
    },
    {
      "epoch": 0.9720199635227418,
      "grad_norm": 0.04366862401366234,
      "learning_rate": 2.8446606059749186e-07,
      "loss": 0.0,
      "step": 68750
    },
    {
      "epoch": 0.9721613482447087,
      "grad_norm": 0.0,
      "learning_rate": 2.830522133778224e-07,
      "loss": 0.0,
      "step": 68760
    },
    {
      "epoch": 0.9723027329666756,
      "grad_norm": 0.0,
      "learning_rate": 2.81638366158153e-07,
      "loss": 0.0,
      "step": 68770
    },
    {
      "epoch": 0.9724441176886426,
      "grad_norm": 0.6387181878089905,
      "learning_rate": 2.8022451893848353e-07,
      "loss": 0.0002,
      "step": 68780
    },
    {
      "epoch": 0.9725855024106095,
      "grad_norm": 0.0,
      "learning_rate": 2.788106717188141e-07,
      "loss": 0.0,
      "step": 68790
    },
    {
      "epoch": 0.9727268871325765,
      "grad_norm": 0.0,
      "learning_rate": 2.7739682449914466e-07,
      "loss": 0.0,
      "step": 68800
    },
    {
      "epoch": 0.9728682718545434,
      "grad_norm": 0.008222338743507862,
      "learning_rate": 2.759829772794752e-07,
      "loss": 0.0,
      "step": 68810
    },
    {
      "epoch": 0.9730096565765104,
      "grad_norm": 0.0,
      "learning_rate": 2.7456913005980574e-07,
      "loss": 0.0,
      "step": 68820
    },
    {
      "epoch": 0.9731510412984773,
      "grad_norm": 2.1246257801976753e-06,
      "learning_rate": 2.7315528284013633e-07,
      "loss": 0.0,
      "step": 68830
    },
    {
      "epoch": 0.9732924260204442,
      "grad_norm": 0.0,
      "learning_rate": 2.7174143562046687e-07,
      "loss": 0.0,
      "step": 68840
    },
    {
      "epoch": 0.9734338107424112,
      "grad_norm": 0.0,
      "learning_rate": 2.7032758840079746e-07,
      "loss": 0.0193,
      "step": 68850
    },
    {
      "epoch": 0.9735751954643781,
      "grad_norm": 0.0,
      "learning_rate": 2.68913741181128e-07,
      "loss": 0.0,
      "step": 68860
    },
    {
      "epoch": 0.9737165801863451,
      "grad_norm": 0.0,
      "learning_rate": 2.6749989396145854e-07,
      "loss": 0.0,
      "step": 68870
    },
    {
      "epoch": 0.973857964908312,
      "grad_norm": 0.0,
      "learning_rate": 2.6622743146375606e-07,
      "loss": 0.1478,
      "step": 68880
    },
    {
      "epoch": 0.9739993496302789,
      "grad_norm": 0.5266693830490112,
      "learning_rate": 2.648135842440866e-07,
      "loss": 0.0001,
      "step": 68890
    },
    {
      "epoch": 0.9741407343522459,
      "grad_norm": 0.0,
      "learning_rate": 2.633997370244172e-07,
      "loss": 0.0,
      "step": 68900
    },
    {
      "epoch": 0.9742821190742128,
      "grad_norm": 0.0,
      "learning_rate": 2.6198588980474773e-07,
      "loss": 0.2838,
      "step": 68910
    },
    {
      "epoch": 0.9744235037961798,
      "grad_norm": 0.0,
      "learning_rate": 2.6057204258507827e-07,
      "loss": 0.0,
      "step": 68920
    },
    {
      "epoch": 0.9745648885181467,
      "grad_norm": 0.0,
      "learning_rate": 2.591581953654088e-07,
      "loss": 0.0,
      "step": 68930
    },
    {
      "epoch": 0.9747062732401137,
      "grad_norm": 4.583836921145945e-10,
      "learning_rate": 2.577443481457394e-07,
      "loss": 0.0,
      "step": 68940
    },
    {
      "epoch": 0.9748476579620806,
      "grad_norm": 0.0,
      "learning_rate": 2.5633050092606994e-07,
      "loss": 0.0,
      "step": 68950
    },
    {
      "epoch": 0.9749890426840475,
      "grad_norm": 0.0,
      "learning_rate": 2.5491665370640053e-07,
      "loss": 0.0,
      "step": 68960
    },
    {
      "epoch": 0.9751304274060145,
      "grad_norm": 0.0,
      "learning_rate": 2.5350280648673107e-07,
      "loss": 0.0,
      "step": 68970
    },
    {
      "epoch": 0.9752718121279814,
      "grad_norm": 0.0,
      "learning_rate": 2.520889592670616e-07,
      "loss": 0.0054,
      "step": 68980
    },
    {
      "epoch": 0.9754131968499484,
      "grad_norm": 3.208410134902806e-06,
      "learning_rate": 2.506751120473922e-07,
      "loss": 0.0001,
      "step": 68990
    },
    {
      "epoch": 0.9755545815719153,
      "grad_norm": 0.0,
      "learning_rate": 2.4926126482772274e-07,
      "loss": 0.1255,
      "step": 69000
    },
    {
      "epoch": 0.9756959662938823,
      "grad_norm": 0.0,
      "learning_rate": 2.478474176080533e-07,
      "loss": 0.0,
      "step": 69010
    },
    {
      "epoch": 0.9758373510158492,
      "grad_norm": 0.014637814834713936,
      "learning_rate": 2.464335703883838e-07,
      "loss": 0.0,
      "step": 69020
    },
    {
      "epoch": 0.9759787357378161,
      "grad_norm": 2.0976145267486572,
      "learning_rate": 2.450197231687144e-07,
      "loss": 0.0002,
      "step": 69030
    },
    {
      "epoch": 0.9761201204597831,
      "grad_norm": 0.0,
      "learning_rate": 2.4360587594904495e-07,
      "loss": 0.0,
      "step": 69040
    },
    {
      "epoch": 0.97626150518175,
      "grad_norm": 0.13249047100543976,
      "learning_rate": 2.4219202872937554e-07,
      "loss": 0.003,
      "step": 69050
    },
    {
      "epoch": 0.976402889903717,
      "grad_norm": 0.0,
      "learning_rate": 2.407781815097061e-07,
      "loss": 0.0001,
      "step": 69060
    },
    {
      "epoch": 0.9765442746256839,
      "grad_norm": 0.0,
      "learning_rate": 2.393643342900366e-07,
      "loss": 0.0,
      "step": 69070
    },
    {
      "epoch": 0.976685659347651,
      "grad_norm": 0.0,
      "learning_rate": 2.3795048707036719e-07,
      "loss": 0.0,
      "step": 69080
    },
    {
      "epoch": 0.9768270440696178,
      "grad_norm": 0.00042379539809189737,
      "learning_rate": 2.3653663985069775e-07,
      "loss": 0.0005,
      "step": 69090
    },
    {
      "epoch": 0.9769684287915847,
      "grad_norm": 0.0,
      "learning_rate": 2.351227926310283e-07,
      "loss": 0.0,
      "step": 69100
    },
    {
      "epoch": 0.9771098135135518,
      "grad_norm": 0.0,
      "learning_rate": 2.3370894541135888e-07,
      "loss": 0.0,
      "step": 69110
    },
    {
      "epoch": 0.9772511982355186,
      "grad_norm": 0.0,
      "learning_rate": 2.3229509819168942e-07,
      "loss": 0.0,
      "step": 69120
    },
    {
      "epoch": 0.9773925829574857,
      "grad_norm": 0.11363881081342697,
      "learning_rate": 2.3088125097201999e-07,
      "loss": 0.0,
      "step": 69130
    },
    {
      "epoch": 0.9775339676794526,
      "grad_norm": 0.0,
      "learning_rate": 2.2946740375235052e-07,
      "loss": 0.0,
      "step": 69140
    },
    {
      "epoch": 0.9776753524014195,
      "grad_norm": 0.0,
      "learning_rate": 2.2805355653268112e-07,
      "loss": 0.0,
      "step": 69150
    },
    {
      "epoch": 0.9778167371233865,
      "grad_norm": 0.0,
      "learning_rate": 2.2663970931301166e-07,
      "loss": 0.0043,
      "step": 69160
    },
    {
      "epoch": 0.9779581218453534,
      "grad_norm": 0.0,
      "learning_rate": 2.2522586209334222e-07,
      "loss": 0.0,
      "step": 69170
    },
    {
      "epoch": 0.9780995065673204,
      "grad_norm": 1.5762323528178968e-05,
      "learning_rate": 2.2381201487367276e-07,
      "loss": 0.0,
      "step": 69180
    },
    {
      "epoch": 0.9782408912892873,
      "grad_norm": 0.0,
      "learning_rate": 2.2239816765400335e-07,
      "loss": 0.0,
      "step": 69190
    },
    {
      "epoch": 0.9783822760112543,
      "grad_norm": 3.5125225394949666e-08,
      "learning_rate": 2.209843204343339e-07,
      "loss": 0.0028,
      "step": 69200
    },
    {
      "epoch": 0.9785236607332212,
      "grad_norm": 0.0,
      "learning_rate": 2.1957047321466443e-07,
      "loss": 0.0,
      "step": 69210
    },
    {
      "epoch": 0.9786650454551881,
      "grad_norm": 0.0,
      "learning_rate": 2.18156625994995e-07,
      "loss": 0.0001,
      "step": 69220
    },
    {
      "epoch": 0.9788064301771551,
      "grad_norm": 0.0,
      "learning_rate": 2.1674277877532553e-07,
      "loss": 0.0001,
      "step": 69230
    },
    {
      "epoch": 0.978947814899122,
      "grad_norm": 0.0003546463849488646,
      "learning_rate": 2.1532893155565612e-07,
      "loss": 0.1427,
      "step": 69240
    },
    {
      "epoch": 0.979089199621089,
      "grad_norm": 5.017421478292761e-10,
      "learning_rate": 2.1391508433598666e-07,
      "loss": 0.3514,
      "step": 69250
    },
    {
      "epoch": 0.9792305843430559,
      "grad_norm": 11.83515453338623,
      "learning_rate": 2.1250123711631723e-07,
      "loss": 0.002,
      "step": 69260
    },
    {
      "epoch": 0.9793719690650229,
      "grad_norm": 0.0,
      "learning_rate": 2.1108738989664777e-07,
      "loss": 0.0,
      "step": 69270
    },
    {
      "epoch": 0.9795133537869898,
      "grad_norm": 0.0,
      "learning_rate": 2.0967354267697836e-07,
      "loss": 0.0,
      "step": 69280
    },
    {
      "epoch": 0.9796547385089567,
      "grad_norm": 0.0,
      "learning_rate": 2.082596954573089e-07,
      "loss": 0.0,
      "step": 69290
    },
    {
      "epoch": 0.9797961232309237,
      "grad_norm": 1.1165016644554271e-07,
      "learning_rate": 2.0684584823763946e-07,
      "loss": 0.0002,
      "step": 69300
    },
    {
      "epoch": 0.9799375079528906,
      "grad_norm": 1.445029397473263e-06,
      "learning_rate": 2.0543200101797e-07,
      "loss": 0.0,
      "step": 69310
    },
    {
      "epoch": 0.9800788926748576,
      "grad_norm": 0.0,
      "learning_rate": 2.040181537983006e-07,
      "loss": 0.0,
      "step": 69320
    },
    {
      "epoch": 0.9802202773968245,
      "grad_norm": 0.0,
      "learning_rate": 2.0260430657863113e-07,
      "loss": 0.0,
      "step": 69330
    },
    {
      "epoch": 0.9803616621187914,
      "grad_norm": 0.0,
      "learning_rate": 2.011904593589617e-07,
      "loss": 0.0,
      "step": 69340
    },
    {
      "epoch": 0.9805030468407584,
      "grad_norm": 0.0,
      "learning_rate": 1.9977661213929224e-07,
      "loss": 0.0,
      "step": 69350
    },
    {
      "epoch": 0.9806444315627253,
      "grad_norm": 0.0,
      "learning_rate": 1.983627649196228e-07,
      "loss": 0.0,
      "step": 69360
    },
    {
      "epoch": 0.9807858162846923,
      "grad_norm": 0.0,
      "learning_rate": 1.9694891769995337e-07,
      "loss": 0.0,
      "step": 69370
    },
    {
      "epoch": 0.9809272010066592,
      "grad_norm": 0.0004233920481055975,
      "learning_rate": 1.9553507048028393e-07,
      "loss": 0.0001,
      "step": 69380
    },
    {
      "epoch": 0.9810685857286262,
      "grad_norm": 0.0,
      "learning_rate": 1.9412122326061447e-07,
      "loss": 0.0,
      "step": 69390
    },
    {
      "epoch": 0.9812099704505931,
      "grad_norm": 0.0,
      "learning_rate": 1.9270737604094504e-07,
      "loss": 0.0,
      "step": 69400
    },
    {
      "epoch": 0.98135135517256,
      "grad_norm": 4.038593104382926e-08,
      "learning_rate": 1.9129352882127558e-07,
      "loss": 0.0001,
      "step": 69410
    },
    {
      "epoch": 0.981492739894527,
      "grad_norm": 4.229133026001364e-07,
      "learning_rate": 1.8987968160160614e-07,
      "loss": 0.0,
      "step": 69420
    },
    {
      "epoch": 0.9816341246164939,
      "grad_norm": 0.0,
      "learning_rate": 1.884658343819367e-07,
      "loss": 0.0149,
      "step": 69430
    },
    {
      "epoch": 0.9817755093384609,
      "grad_norm": 0.0,
      "learning_rate": 1.8705198716226725e-07,
      "loss": 0.0,
      "step": 69440
    },
    {
      "epoch": 0.9819168940604278,
      "grad_norm": 0.0,
      "learning_rate": 1.856381399425978e-07,
      "loss": 0.0002,
      "step": 69450
    },
    {
      "epoch": 0.9820582787823948,
      "grad_norm": 0.0,
      "learning_rate": 1.8422429272292835e-07,
      "loss": 0.0,
      "step": 69460
    },
    {
      "epoch": 0.9821996635043617,
      "grad_norm": 2.6710631484405667e-09,
      "learning_rate": 1.8281044550325894e-07,
      "loss": 0.0,
      "step": 69470
    },
    {
      "epoch": 0.9823410482263286,
      "grad_norm": 3.818468741201286e-08,
      "learning_rate": 1.8139659828358948e-07,
      "loss": 0.0,
      "step": 69480
    },
    {
      "epoch": 0.9824824329482956,
      "grad_norm": 0.00010185378778260201,
      "learning_rate": 1.7998275106392005e-07,
      "loss": 0.0,
      "step": 69490
    },
    {
      "epoch": 0.9826238176702625,
      "grad_norm": 0.0,
      "learning_rate": 1.7856890384425058e-07,
      "loss": 0.0,
      "step": 69500
    },
    {
      "epoch": 0.9827652023922295,
      "grad_norm": 0.0,
      "learning_rate": 1.7715505662458118e-07,
      "loss": 0.0,
      "step": 69510
    },
    {
      "epoch": 0.9829065871141964,
      "grad_norm": 0.0,
      "learning_rate": 1.7574120940491171e-07,
      "loss": 0.001,
      "step": 69520
    },
    {
      "epoch": 0.9830479718361634,
      "grad_norm": 0.0,
      "learning_rate": 1.7432736218524228e-07,
      "loss": 0.0001,
      "step": 69530
    },
    {
      "epoch": 0.9831893565581303,
      "grad_norm": 8.33864974975586,
      "learning_rate": 1.7291351496557282e-07,
      "loss": 0.0015,
      "step": 69540
    },
    {
      "epoch": 0.9833307412800972,
      "grad_norm": 1.1984972161371843e-06,
      "learning_rate": 1.714996677459034e-07,
      "loss": 0.0,
      "step": 69550
    },
    {
      "epoch": 0.9834721260020642,
      "grad_norm": 0.0,
      "learning_rate": 1.7008582052623395e-07,
      "loss": 0.0,
      "step": 69560
    },
    {
      "epoch": 0.9836135107240311,
      "grad_norm": 0.0,
      "learning_rate": 1.6867197330656451e-07,
      "loss": 0.0,
      "step": 69570
    },
    {
      "epoch": 0.9837548954459981,
      "grad_norm": 0.00015332532348111272,
      "learning_rate": 1.6725812608689505e-07,
      "loss": 0.0,
      "step": 69580
    },
    {
      "epoch": 0.983896280167965,
      "grad_norm": 0.0,
      "learning_rate": 1.6584427886722565e-07,
      "loss": 0.002,
      "step": 69590
    },
    {
      "epoch": 0.9840376648899319,
      "grad_norm": 0.04279113560914993,
      "learning_rate": 1.6443043164755618e-07,
      "loss": 0.0,
      "step": 69600
    },
    {
      "epoch": 0.984179049611899,
      "grad_norm": 0.0,
      "learning_rate": 1.6301658442788675e-07,
      "loss": 0.0,
      "step": 69610
    },
    {
      "epoch": 0.9843204343338658,
      "grad_norm": 4.87633045054281e-10,
      "learning_rate": 1.616027372082173e-07,
      "loss": 0.0,
      "step": 69620
    },
    {
      "epoch": 0.9844618190558329,
      "grad_norm": 4.1756184145924635e-06,
      "learning_rate": 1.6018888998854783e-07,
      "loss": 0.0,
      "step": 69630
    },
    {
      "epoch": 0.9846032037777998,
      "grad_norm": 0.0,
      "learning_rate": 1.5877504276887842e-07,
      "loss": 0.0,
      "step": 69640
    },
    {
      "epoch": 0.9847445884997668,
      "grad_norm": 0.0,
      "learning_rate": 1.5736119554920896e-07,
      "loss": 0.0036,
      "step": 69650
    },
    {
      "epoch": 0.9848859732217337,
      "grad_norm": 0.0,
      "learning_rate": 1.5594734832953952e-07,
      "loss": 0.0002,
      "step": 69660
    },
    {
      "epoch": 0.9850273579437006,
      "grad_norm": 0.0,
      "learning_rate": 1.545335011098701e-07,
      "loss": 0.0,
      "step": 69670
    },
    {
      "epoch": 0.9851687426656676,
      "grad_norm": 0.0,
      "learning_rate": 1.5311965389020065e-07,
      "loss": 0.0,
      "step": 69680
    },
    {
      "epoch": 0.9853101273876345,
      "grad_norm": 0.0,
      "learning_rate": 1.517058066705312e-07,
      "loss": 0.0,
      "step": 69690
    },
    {
      "epoch": 0.9854515121096015,
      "grad_norm": 0.0,
      "learning_rate": 1.5029195945086176e-07,
      "loss": 0.0017,
      "step": 69700
    },
    {
      "epoch": 0.9855928968315684,
      "grad_norm": 0.002652229741215706,
      "learning_rate": 1.4887811223119232e-07,
      "loss": 0.007,
      "step": 69710
    },
    {
      "epoch": 0.9857342815535354,
      "grad_norm": 3.682407907490415e-07,
      "learning_rate": 1.4746426501152286e-07,
      "loss": 0.0,
      "step": 69720
    },
    {
      "epoch": 0.9858756662755023,
      "grad_norm": 0.0,
      "learning_rate": 1.4605041779185343e-07,
      "loss": 0.0,
      "step": 69730
    },
    {
      "epoch": 0.9860170509974692,
      "grad_norm": 4.559692901029422e-10,
      "learning_rate": 1.4463657057218397e-07,
      "loss": 0.0,
      "step": 69740
    },
    {
      "epoch": 0.9861584357194362,
      "grad_norm": 0.0,
      "learning_rate": 1.4322272335251453e-07,
      "loss": 0.0,
      "step": 69750
    },
    {
      "epoch": 0.9862998204414031,
      "grad_norm": 0.0,
      "learning_rate": 1.418088761328451e-07,
      "loss": 0.0001,
      "step": 69760
    },
    {
      "epoch": 0.9864412051633701,
      "grad_norm": 0.0,
      "learning_rate": 1.4039502891317564e-07,
      "loss": 0.0,
      "step": 69770
    },
    {
      "epoch": 0.986582589885337,
      "grad_norm": 0.0,
      "learning_rate": 1.389811816935062e-07,
      "loss": 0.0,
      "step": 69780
    },
    {
      "epoch": 0.986723974607304,
      "grad_norm": 0.0,
      "learning_rate": 1.3756733447383677e-07,
      "loss": 0.0,
      "step": 69790
    },
    {
      "epoch": 0.9868653593292709,
      "grad_norm": 0.0,
      "learning_rate": 1.3615348725416733e-07,
      "loss": 0.0,
      "step": 69800
    },
    {
      "epoch": 0.9870067440512378,
      "grad_norm": 0.0,
      "learning_rate": 1.3473964003449787e-07,
      "loss": 0.0001,
      "step": 69810
    },
    {
      "epoch": 0.9871481287732048,
      "grad_norm": 0.0,
      "learning_rate": 1.3332579281482844e-07,
      "loss": 0.0,
      "step": 69820
    },
    {
      "epoch": 0.9872895134951717,
      "grad_norm": 0.0,
      "learning_rate": 1.31911945595159e-07,
      "loss": 0.0,
      "step": 69830
    },
    {
      "epoch": 0.9874308982171387,
      "grad_norm": 0.004862419329583645,
      "learning_rate": 1.3049809837548957e-07,
      "loss": 0.0012,
      "step": 69840
    },
    {
      "epoch": 0.9875722829391056,
      "grad_norm": 0.0,
      "learning_rate": 1.290842511558201e-07,
      "loss": 0.0,
      "step": 69850
    },
    {
      "epoch": 0.9877136676610725,
      "grad_norm": 0.0,
      "learning_rate": 1.2767040393615067e-07,
      "loss": 0.0,
      "step": 69860
    },
    {
      "epoch": 0.9878550523830395,
      "grad_norm": 0.0,
      "learning_rate": 1.2625655671648124e-07,
      "loss": 0.0,
      "step": 69870
    },
    {
      "epoch": 0.9879964371050064,
      "grad_norm": 0.0,
      "learning_rate": 1.248427094968118e-07,
      "loss": 0.0,
      "step": 69880
    },
    {
      "epoch": 0.9881378218269734,
      "grad_norm": 0.0,
      "learning_rate": 1.2342886227714234e-07,
      "loss": 0.0165,
      "step": 69890
    },
    {
      "epoch": 0.9882792065489403,
      "grad_norm": 0.0,
      "learning_rate": 1.220150150574729e-07,
      "loss": 0.0001,
      "step": 69900
    },
    {
      "epoch": 0.9884205912709073,
      "grad_norm": 0.0,
      "learning_rate": 1.2060116783780347e-07,
      "loss": 0.0,
      "step": 69910
    },
    {
      "epoch": 0.9885619759928742,
      "grad_norm": 0.00010074312012875453,
      "learning_rate": 1.1918732061813402e-07,
      "loss": 0.0,
      "step": 69920
    },
    {
      "epoch": 0.9887033607148411,
      "grad_norm": 0.0,
      "learning_rate": 1.1777347339846457e-07,
      "loss": 0.0,
      "step": 69930
    },
    {
      "epoch": 0.9888447454368081,
      "grad_norm": 0.0,
      "learning_rate": 1.1635962617879513e-07,
      "loss": 0.0,
      "step": 69940
    },
    {
      "epoch": 0.988986130158775,
      "grad_norm": 0.00039001915138214827,
      "learning_rate": 1.1494577895912568e-07,
      "loss": 0.0,
      "step": 69950
    },
    {
      "epoch": 0.989127514880742,
      "grad_norm": 0.0,
      "learning_rate": 1.1353193173945623e-07,
      "loss": 0.0,
      "step": 69960
    },
    {
      "epoch": 0.9892688996027089,
      "grad_norm": 0.0,
      "learning_rate": 1.121180845197868e-07,
      "loss": 0.0894,
      "step": 69970
    },
    {
      "epoch": 0.9894102843246759,
      "grad_norm": 0.0,
      "learning_rate": 1.1070423730011735e-07,
      "loss": 0.0,
      "step": 69980
    },
    {
      "epoch": 0.9895516690466428,
      "grad_norm": 0.0,
      "learning_rate": 1.0929039008044791e-07,
      "loss": 0.0011,
      "step": 69990
    },
    {
      "epoch": 0.9896930537686097,
      "grad_norm": 0.004068499431014061,
      "learning_rate": 1.0787654286077847e-07,
      "loss": 0.0069,
      "step": 70000
    },
    {
      "epoch": 0.9898344384905767,
      "grad_norm": 0.0,
      "learning_rate": 1.0646269564110903e-07,
      "loss": 0.0,
      "step": 70010
    },
    {
      "epoch": 0.9899758232125436,
      "grad_norm": 0.0,
      "learning_rate": 1.0504884842143958e-07,
      "loss": 0.0,
      "step": 70020
    },
    {
      "epoch": 0.9901172079345106,
      "grad_norm": 0.0,
      "learning_rate": 1.0363500120177015e-07,
      "loss": 0.0,
      "step": 70030
    },
    {
      "epoch": 0.9902585926564775,
      "grad_norm": 0.0,
      "learning_rate": 1.022211539821007e-07,
      "loss": 0.0002,
      "step": 70040
    },
    {
      "epoch": 0.9903999773784445,
      "grad_norm": 5.014186288399003e-10,
      "learning_rate": 1.0080730676243127e-07,
      "loss": 0.0,
      "step": 70050
    },
    {
      "epoch": 0.9905413621004114,
      "grad_norm": 0.0,
      "learning_rate": 9.939345954276182e-08,
      "loss": 0.0,
      "step": 70060
    },
    {
      "epoch": 0.9906827468223783,
      "grad_norm": 0.0,
      "learning_rate": 9.797961232309238e-08,
      "loss": 0.0,
      "step": 70070
    },
    {
      "epoch": 0.9908241315443453,
      "grad_norm": 0.0,
      "learning_rate": 9.656576510342294e-08,
      "loss": 0.0012,
      "step": 70080
    },
    {
      "epoch": 0.9909655162663122,
      "grad_norm": 0.0,
      "learning_rate": 9.51519178837535e-08,
      "loss": 0.0,
      "step": 70090
    },
    {
      "epoch": 0.9911069009882792,
      "grad_norm": 4.111716862098547e-06,
      "learning_rate": 9.373807066408405e-08,
      "loss": 0.0,
      "step": 70100
    },
    {
      "epoch": 0.9912482857102461,
      "grad_norm": 0.0,
      "learning_rate": 9.23242234444146e-08,
      "loss": 0.0,
      "step": 70110
    },
    {
      "epoch": 0.991389670432213,
      "grad_norm": 0.0,
      "learning_rate": 9.091037622474517e-08,
      "loss": 0.0002,
      "step": 70120
    },
    {
      "epoch": 0.99153105515418,
      "grad_norm": 0.0,
      "learning_rate": 8.949652900507572e-08,
      "loss": 0.0,
      "step": 70130
    },
    {
      "epoch": 0.991672439876147,
      "grad_norm": 0.0,
      "learning_rate": 8.808268178540627e-08,
      "loss": 0.0,
      "step": 70140
    },
    {
      "epoch": 0.991813824598114,
      "grad_norm": 0.0,
      "learning_rate": 8.666883456573683e-08,
      "loss": 0.0,
      "step": 70150
    },
    {
      "epoch": 0.9919552093200809,
      "grad_norm": 0.0,
      "learning_rate": 8.525498734606738e-08,
      "loss": 0.0,
      "step": 70160
    },
    {
      "epoch": 0.9920965940420479,
      "grad_norm": 0.0,
      "learning_rate": 8.384114012639794e-08,
      "loss": 0.0,
      "step": 70170
    },
    {
      "epoch": 0.9922379787640148,
      "grad_norm": 0.0,
      "learning_rate": 8.24272929067285e-08,
      "loss": 0.0,
      "step": 70180
    },
    {
      "epoch": 0.9923793634859817,
      "grad_norm": 0.0,
      "learning_rate": 8.101344568705906e-08,
      "loss": 0.0,
      "step": 70190
    },
    {
      "epoch": 0.9925207482079487,
      "grad_norm": 0.0,
      "learning_rate": 7.959959846738961e-08,
      "loss": 0.0,
      "step": 70200
    },
    {
      "epoch": 0.9926621329299156,
      "grad_norm": 0.0,
      "learning_rate": 7.818575124772018e-08,
      "loss": 0.0,
      "step": 70210
    },
    {
      "epoch": 0.9928035176518826,
      "grad_norm": 0.0,
      "learning_rate": 7.677190402805073e-08,
      "loss": 0.0016,
      "step": 70220
    },
    {
      "epoch": 0.9929449023738495,
      "grad_norm": 0.0,
      "learning_rate": 7.53580568083813e-08,
      "loss": 0.0,
      "step": 70230
    },
    {
      "epoch": 0.9930862870958165,
      "grad_norm": 0.0013831807300448418,
      "learning_rate": 7.394420958871185e-08,
      "loss": 0.0052,
      "step": 70240
    },
    {
      "epoch": 0.9932276718177834,
      "grad_norm": 0.0,
      "learning_rate": 7.253036236904241e-08,
      "loss": 0.0,
      "step": 70250
    },
    {
      "epoch": 0.9933690565397503,
      "grad_norm": 0.0,
      "learning_rate": 7.111651514937296e-08,
      "loss": 0.0004,
      "step": 70260
    },
    {
      "epoch": 0.9935104412617173,
      "grad_norm": 0.0009331972105428576,
      "learning_rate": 6.970266792970353e-08,
      "loss": 0.0,
      "step": 70270
    },
    {
      "epoch": 0.9936518259836842,
      "grad_norm": 0.0,
      "learning_rate": 6.828882071003408e-08,
      "loss": 0.0,
      "step": 70280
    },
    {
      "epoch": 0.9937932107056512,
      "grad_norm": 0.0,
      "learning_rate": 6.687497349036463e-08,
      "loss": 0.0,
      "step": 70290
    },
    {
      "epoch": 0.9939345954276181,
      "grad_norm": 9.945045167114586e-05,
      "learning_rate": 6.546112627069519e-08,
      "loss": 0.0,
      "step": 70300
    },
    {
      "epoch": 0.994075980149585,
      "grad_norm": 0.0,
      "learning_rate": 6.404727905102575e-08,
      "loss": 0.0026,
      "step": 70310
    },
    {
      "epoch": 0.994217364871552,
      "grad_norm": 0.00012103597691748291,
      "learning_rate": 6.26334318313563e-08,
      "loss": 0.0,
      "step": 70320
    },
    {
      "epoch": 0.9943587495935189,
      "grad_norm": 0.0,
      "learning_rate": 6.121958461168687e-08,
      "loss": 0.0,
      "step": 70330
    },
    {
      "epoch": 0.9945001343154859,
      "grad_norm": 0.0,
      "learning_rate": 5.980573739201742e-08,
      "loss": 0.0,
      "step": 70340
    },
    {
      "epoch": 0.9946415190374528,
      "grad_norm": 0.0,
      "learning_rate": 5.839189017234798e-08,
      "loss": 0.0,
      "step": 70350
    },
    {
      "epoch": 0.9947829037594198,
      "grad_norm": 3.344097137451172,
      "learning_rate": 5.697804295267854e-08,
      "loss": 0.0007,
      "step": 70360
    },
    {
      "epoch": 0.9949242884813867,
      "grad_norm": 0.0,
      "learning_rate": 5.55641957330091e-08,
      "loss": 0.0,
      "step": 70370
    },
    {
      "epoch": 0.9950656732033536,
      "grad_norm": 3.196501711499877e-05,
      "learning_rate": 5.4150348513339656e-08,
      "loss": 0.0,
      "step": 70380
    },
    {
      "epoch": 0.9952070579253206,
      "grad_norm": 0.0,
      "learning_rate": 5.2736501293670215e-08,
      "loss": 0.003,
      "step": 70390
    },
    {
      "epoch": 0.9953484426472875,
      "grad_norm": 0.0,
      "learning_rate": 5.132265407400076e-08,
      "loss": 0.0,
      "step": 70400
    },
    {
      "epoch": 0.9954898273692545,
      "grad_norm": 0.0,
      "learning_rate": 4.990880685433132e-08,
      "loss": 0.0001,
      "step": 70410
    },
    {
      "epoch": 0.9956312120912214,
      "grad_norm": 0.0,
      "learning_rate": 4.849495963466188e-08,
      "loss": 0.0,
      "step": 70420
    },
    {
      "epoch": 0.9957725968131884,
      "grad_norm": 0.0,
      "learning_rate": 4.7081112414992436e-08,
      "loss": 0.0,
      "step": 70430
    },
    {
      "epoch": 0.9959139815351553,
      "grad_norm": 9.771362385890825e-08,
      "learning_rate": 4.5667265195322995e-08,
      "loss": 0.0001,
      "step": 70440
    },
    {
      "epoch": 0.9960553662571222,
      "grad_norm": 0.0,
      "learning_rate": 4.4253417975653554e-08,
      "loss": 0.0,
      "step": 70450
    },
    {
      "epoch": 0.9961967509790892,
      "grad_norm": 1.0725819947765558e-06,
      "learning_rate": 4.283957075598411e-08,
      "loss": 0.0,
      "step": 70460
    },
    {
      "epoch": 0.9963381357010561,
      "grad_norm": 0.0,
      "learning_rate": 4.142572353631467e-08,
      "loss": 0.0,
      "step": 70470
    },
    {
      "epoch": 0.9964795204230231,
      "grad_norm": 0.004065412096679211,
      "learning_rate": 4.001187631664523e-08,
      "loss": 0.0,
      "step": 70480
    },
    {
      "epoch": 0.99662090514499,
      "grad_norm": 0.0,
      "learning_rate": 3.859802909697578e-08,
      "loss": 0.0,
      "step": 70490
    },
    {
      "epoch": 0.996762289866957,
      "grad_norm": 0.0,
      "learning_rate": 3.718418187730634e-08,
      "loss": 0.0,
      "step": 70500
    },
    {
      "epoch": 0.9969036745889239,
      "grad_norm": 0.0,
      "learning_rate": 3.57703346576369e-08,
      "loss": 0.0,
      "step": 70510
    },
    {
      "epoch": 0.9970450593108908,
      "grad_norm": 7.760577318549622e-07,
      "learning_rate": 3.435648743796746e-08,
      "loss": 0.0,
      "step": 70520
    },
    {
      "epoch": 0.9971864440328578,
      "grad_norm": 0.0,
      "learning_rate": 3.294264021829801e-08,
      "loss": 0.0,
      "step": 70530
    },
    {
      "epoch": 0.9973278287548247,
      "grad_norm": 2.057885103567969e-05,
      "learning_rate": 3.152879299862857e-08,
      "loss": 0.0,
      "step": 70540
    },
    {
      "epoch": 0.9974692134767917,
      "grad_norm": 0.0,
      "learning_rate": 3.011494577895913e-08,
      "loss": 0.0012,
      "step": 70550
    },
    {
      "epoch": 0.9976105981987586,
      "grad_norm": 0.0,
      "learning_rate": 2.8701098559289686e-08,
      "loss": 0.019,
      "step": 70560
    },
    {
      "epoch": 0.9977519829207255,
      "grad_norm": 0.0,
      "learning_rate": 2.7287251339620244e-08,
      "loss": 0.001,
      "step": 70570
    },
    {
      "epoch": 0.9978933676426925,
      "grad_norm": 0.0,
      "learning_rate": 2.5873404119950803e-08,
      "loss": 0.0,
      "step": 70580
    },
    {
      "epoch": 0.9980347523646594,
      "grad_norm": 1.8746630914279194e-09,
      "learning_rate": 2.4459556900281355e-08,
      "loss": 0.0,
      "step": 70590
    },
    {
      "epoch": 0.9981761370866264,
      "grad_norm": 0.0,
      "learning_rate": 2.3045709680611914e-08,
      "loss": 0.0,
      "step": 70600
    },
    {
      "epoch": 0.9983175218085933,
      "grad_norm": 0.0,
      "learning_rate": 2.1631862460942473e-08,
      "loss": 0.0,
      "step": 70610
    },
    {
      "epoch": 0.9984589065305604,
      "grad_norm": 0.0,
      "learning_rate": 2.021801524127303e-08,
      "loss": 0.0,
      "step": 70620
    },
    {
      "epoch": 0.9986002912525273,
      "grad_norm": 2.225066886296645e-08,
      "learning_rate": 1.8804168021603587e-08,
      "loss": 0.0,
      "step": 70630
    },
    {
      "epoch": 0.9987416759744941,
      "grad_norm": 554.9219970703125,
      "learning_rate": 1.7390320801934145e-08,
      "loss": 0.1201,
      "step": 70640
    },
    {
      "epoch": 0.9988830606964612,
      "grad_norm": 0.007035449612885714,
      "learning_rate": 1.59764735822647e-08,
      "loss": 0.0,
      "step": 70650
    },
    {
      "epoch": 0.999024445418428,
      "grad_norm": 0.0,
      "learning_rate": 1.456262636259526e-08,
      "loss": 0.0,
      "step": 70660
    },
    {
      "epoch": 0.9991658301403951,
      "grad_norm": 0.0,
      "learning_rate": 1.3148779142925818e-08,
      "loss": 0.0,
      "step": 70670
    },
    {
      "epoch": 0.999307214862362,
      "grad_norm": 0.0,
      "learning_rate": 1.1734931923256373e-08,
      "loss": 0.0004,
      "step": 70680
    },
    {
      "epoch": 0.999448599584329,
      "grad_norm": 2.0479092199821025e-07,
      "learning_rate": 1.0321084703586932e-08,
      "loss": 0.0,
      "step": 70690
    },
    {
      "epoch": 0.9995899843062959,
      "grad_norm": 8.877401705831289e-05,
      "learning_rate": 8.90723748391749e-09,
      "loss": 0.0,
      "step": 70700
    },
    {
      "epoch": 0.9997313690282628,
      "grad_norm": 0.0,
      "learning_rate": 7.493390264248046e-09,
      "loss": 0.0005,
      "step": 70710
    },
    {
      "epoch": 0.9998727537502298,
      "grad_norm": 0.00013561491505242884,
      "learning_rate": 6.079543044578603e-09,
      "loss": 0.001,
      "step": 70720
    }
  ],
  "logging_steps": 10,
  "max_steps": 70729,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 20000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}