{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.41394440726610415,
  "eval_steps": 500,
  "global_step": 60000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.89907345443507e-05,
      "grad_norm": 0.00011222957982681692,
      "learning_rate": 9.999379083389101e-06,
      "loss": 0.0051,
      "step": 10
    },
    {
      "epoch": 0.0001379814690887014,
      "grad_norm": 0.17417871952056885,
      "learning_rate": 9.998827157512746e-06,
      "loss": 0.3817,
      "step": 20
    },
    {
      "epoch": 0.00020697220363305208,
      "grad_norm": 8.138256072998047,
      "learning_rate": 9.998137250167303e-06,
      "loss": 0.0129,
      "step": 30
    },
    {
      "epoch": 0.0002759629381774028,
      "grad_norm": 0.0460987351834774,
      "learning_rate": 9.997516333556404e-06,
      "loss": 0.0364,
      "step": 40
    },
    {
      "epoch": 0.00034495367272175347,
      "grad_norm": 3.265582323074341,
      "learning_rate": 9.99682642621096e-06,
      "loss": 0.0011,
      "step": 50
    },
    {
      "epoch": 0.00041394440726610416,
      "grad_norm": 0.06020026281476021,
      "learning_rate": 9.996136518865516e-06,
      "loss": 0.0328,
      "step": 60
    },
    {
      "epoch": 0.00048293514181045486,
      "grad_norm": 0.0,
      "learning_rate": 9.995446611520073e-06,
      "loss": 0.0025,
      "step": 70
    },
    {
      "epoch": 0.0005519258763548055,
      "grad_norm": 0.0009600712219253182,
      "learning_rate": 9.99475670417463e-06,
      "loss": 0.0001,
      "step": 80
    },
    {
      "epoch": 0.0006209166108991562,
      "grad_norm": 1.0819045215271217e-08,
      "learning_rate": 9.994066796829188e-06,
      "loss": 0.0031,
      "step": 90
    },
    {
      "epoch": 0.0006899073454435069,
      "grad_norm": 0.00022638787049800158,
      "learning_rate": 9.993376889483744e-06,
      "loss": 0.0046,
      "step": 100
    },
    {
      "epoch": 0.0007588980799878576,
      "grad_norm": 1.1406569910832332e-08,
      "learning_rate": 9.9926869821383e-06,
      "loss": 0.005,
      "step": 110
    },
    {
      "epoch": 0.0008278888145322083,
      "grad_norm": 0.0012971981195732951,
      "learning_rate": 9.991997074792857e-06,
      "loss": 0.0018,
      "step": 120
    },
    {
      "epoch": 0.000896879549076559,
      "grad_norm": 1.264641391607313e-09,
      "learning_rate": 9.991307167447412e-06,
      "loss": 0.0,
      "step": 130
    },
    {
      "epoch": 0.0009658702836209097,
      "grad_norm": 4.786876161233522e-05,
      "learning_rate": 9.990617260101968e-06,
      "loss": 0.0001,
      "step": 140
    },
    {
      "epoch": 0.0010348610181652605,
      "grad_norm": 0.001031873282045126,
      "learning_rate": 9.989927352756525e-06,
      "loss": 0.0066,
      "step": 150
    },
    {
      "epoch": 0.001103851752709611,
      "grad_norm": 0.011338403448462486,
      "learning_rate": 9.989237445411081e-06,
      "loss": 0.0002,
      "step": 160
    },
    {
      "epoch": 0.001172842487253962,
      "grad_norm": 2.0826701474518927e-10,
      "learning_rate": 9.988547538065638e-06,
      "loss": 0.0013,
      "step": 170
    },
    {
      "epoch": 0.0012418332217983125,
      "grad_norm": 0.00012302969116717577,
      "learning_rate": 9.987857630720194e-06,
      "loss": 0.0,
      "step": 180
    },
    {
      "epoch": 0.0013108239563426633,
      "grad_norm": 9.173053741455078,
      "learning_rate": 9.987167723374751e-06,
      "loss": 0.0025,
      "step": 190
    },
    {
      "epoch": 0.0013798146908870139,
      "grad_norm": 4.416388037498109e-05,
      "learning_rate": 9.986477816029309e-06,
      "loss": 0.0,
      "step": 200
    },
    {
      "epoch": 0.0014488054254313647,
      "grad_norm": 0.00019524099479895085,
      "learning_rate": 9.985787908683866e-06,
      "loss": 0.0001,
      "step": 210
    },
    {
      "epoch": 0.0015177961599757153,
      "grad_norm": 4.591324806213379,
      "learning_rate": 9.985098001338422e-06,
      "loss": 0.001,
      "step": 220
    },
    {
      "epoch": 0.001586786894520066,
      "grad_norm": 0.0011398019269108772,
      "learning_rate": 9.984408093992979e-06,
      "loss": 0.0001,
      "step": 230
    },
    {
      "epoch": 0.0016557776290644166,
      "grad_norm": 9.500353392866145e-09,
      "learning_rate": 9.983718186647533e-06,
      "loss": 0.001,
      "step": 240
    },
    {
      "epoch": 0.0017247683636087674,
      "grad_norm": 3.374801593736265e-08,
      "learning_rate": 9.98302827930209e-06,
      "loss": 0.008,
      "step": 250
    },
    {
      "epoch": 0.001793759098153118,
      "grad_norm": 1.4565864603355294e-06,
      "learning_rate": 9.98240736269119e-06,
      "loss": 0.0563,
      "step": 260
    },
    {
      "epoch": 0.0018627498326974688,
      "grad_norm": 0.005463605280965567,
      "learning_rate": 9.981717455345747e-06,
      "loss": 0.0018,
      "step": 270
    },
    {
      "epoch": 0.0019317405672418194,
      "grad_norm": 0.0011831159936264157,
      "learning_rate": 9.981027548000303e-06,
      "loss": 0.0313,
      "step": 280
    },
    {
      "epoch": 0.0020007313017861702,
      "grad_norm": 1.5842642824281938e-06,
      "learning_rate": 9.980337640654862e-06,
      "loss": 0.0139,
      "step": 290
    },
    {
      "epoch": 0.002069722036330521,
      "grad_norm": 0.00014429337170440704,
      "learning_rate": 9.979647733309418e-06,
      "loss": 0.0001,
      "step": 300
    },
    {
      "epoch": 0.0021387127708748714,
      "grad_norm": 4.3330678636266384e-06,
      "learning_rate": 9.978957825963975e-06,
      "loss": 0.0002,
      "step": 310
    },
    {
      "epoch": 0.002207703505419222,
      "grad_norm": 3.0802716537436936e-06,
      "learning_rate": 9.978267918618531e-06,
      "loss": 0.0011,
      "step": 320
    },
    {
      "epoch": 0.002276694239963573,
      "grad_norm": 1.4535776376724243,
      "learning_rate": 9.977647002007632e-06,
      "loss": 0.264,
      "step": 330
    },
    {
      "epoch": 0.002345684974507924,
      "grad_norm": 1.2289291589695495e-05,
      "learning_rate": 9.976957094662187e-06,
      "loss": 0.0001,
      "step": 340
    },
    {
      "epoch": 0.002414675709052274,
      "grad_norm": 0.0065452903509140015,
      "learning_rate": 9.976267187316743e-06,
      "loss": 0.0,
      "step": 350
    },
    {
      "epoch": 0.002483666443596625,
      "grad_norm": 0.28994324803352356,
      "learning_rate": 9.9755772799713e-06,
      "loss": 0.0004,
      "step": 360
    },
    {
      "epoch": 0.0025526571781409758,
      "grad_norm": 39.13631057739258,
      "learning_rate": 9.974887372625858e-06,
      "loss": 0.3606,
      "step": 370
    },
    {
      "epoch": 0.0026216479126853266,
      "grad_norm": 0.004843785427510738,
      "learning_rate": 9.974197465280414e-06,
      "loss": 0.0,
      "step": 380
    },
    {
      "epoch": 0.002690638647229677,
      "grad_norm": 0.0035524864215403795,
      "learning_rate": 9.97350755793497e-06,
      "loss": 0.0,
      "step": 390
    },
    {
      "epoch": 0.0027596293817740277,
      "grad_norm": 0.08471328765153885,
      "learning_rate": 9.972817650589527e-06,
      "loss": 0.0001,
      "step": 400
    },
    {
      "epoch": 0.0028286201163183785,
      "grad_norm": 2.6892220006402567e-08,
      "learning_rate": 9.972127743244084e-06,
      "loss": 0.0002,
      "step": 410
    },
    {
      "epoch": 0.0028976108508627293,
      "grad_norm": 0.3459404706954956,
      "learning_rate": 9.97143783589864e-06,
      "loss": 0.0015,
      "step": 420
    },
    {
      "epoch": 0.0029666015854070797,
      "grad_norm": 0.0008964891894720495,
      "learning_rate": 9.970747928553197e-06,
      "loss": 0.0004,
      "step": 430
    },
    {
      "epoch": 0.0030355923199514305,
      "grad_norm": 9.523880180495325e-06,
      "learning_rate": 9.970058021207753e-06,
      "loss": 0.0,
      "step": 440
    },
    {
      "epoch": 0.0031045830544957813,
      "grad_norm": 0.0015833821380510926,
      "learning_rate": 9.96936811386231e-06,
      "loss": 0.0,
      "step": 450
    },
    {
      "epoch": 0.003173573789040132,
      "grad_norm": 6.80257244312088e-06,
      "learning_rate": 9.968678206516864e-06,
      "loss": 0.0061,
      "step": 460
    },
    {
      "epoch": 0.0032425645235844825,
      "grad_norm": 5.697336491472527e-10,
      "learning_rate": 9.967988299171421e-06,
      "loss": 0.003,
      "step": 470
    },
    {
      "epoch": 0.0033115552581288333,
      "grad_norm": 0.0,
      "learning_rate": 9.967298391825979e-06,
      "loss": 0.0,
      "step": 480
    },
    {
      "epoch": 0.003380545992673184,
      "grad_norm": 9.849828508379233e-09,
      "learning_rate": 9.966608484480536e-06,
      "loss": 0.0073,
      "step": 490
    },
    {
      "epoch": 0.003449536727217535,
      "grad_norm": 1.8166218751503038e-07,
      "learning_rate": 9.965918577135092e-06,
      "loss": 0.0004,
      "step": 500
    },
    {
      "epoch": 0.0035185274617618853,
      "grad_norm": 0.0043383194133639336,
      "learning_rate": 9.965228669789649e-06,
      "loss": 0.0,
      "step": 510
    },
    {
      "epoch": 0.003587518196306236,
      "grad_norm": 0.0009673055028542876,
      "learning_rate": 9.964538762444205e-06,
      "loss": 0.0026,
      "step": 520
    },
    {
      "epoch": 0.003656508930850587,
      "grad_norm": 0.0,
      "learning_rate": 9.963848855098762e-06,
      "loss": 0.0122,
      "step": 530
    },
    {
      "epoch": 0.0037254996653949377,
      "grad_norm": 8.976323897513794e-07,
      "learning_rate": 9.963158947753318e-06,
      "loss": 0.0002,
      "step": 540
    },
    {
      "epoch": 0.003794490399939288,
      "grad_norm": 3.026991367340088,
      "learning_rate": 9.962469040407875e-06,
      "loss": 0.0009,
      "step": 550
    },
    {
      "epoch": 0.003863481134483639,
      "grad_norm": 1.153221296235074e-09,
      "learning_rate": 9.961779133062431e-06,
      "loss": 0.0003,
      "step": 560
    },
    {
      "epoch": 0.003932471869027989,
      "grad_norm": 9.897039853967726e-05,
      "learning_rate": 9.961089225716986e-06,
      "loss": 0.2975,
      "step": 570
    },
    {
      "epoch": 0.0040014626035723404,
      "grad_norm": 2.8191266210342292e-08,
      "learning_rate": 9.960399318371542e-06,
      "loss": 0.0013,
      "step": 580
    },
    {
      "epoch": 0.004070453338116691,
      "grad_norm": 0.002458341885358095,
      "learning_rate": 9.9597094110261e-06,
      "loss": 0.0018,
      "step": 590
    },
    {
      "epoch": 0.004139444072661042,
      "grad_norm": 0.01742110401391983,
      "learning_rate": 9.959019503680657e-06,
      "loss": 0.0,
      "step": 600
    },
    {
      "epoch": 0.004208434807205392,
      "grad_norm": 2.6468876512808492e-06,
      "learning_rate": 9.958329596335214e-06,
      "loss": 0.0106,
      "step": 610
    },
    {
      "epoch": 0.004277425541749743,
      "grad_norm": 1.464867409595172e-07,
      "learning_rate": 9.95763968898977e-06,
      "loss": 0.0001,
      "step": 620
    },
    {
      "epoch": 0.004346416276294094,
      "grad_norm": 0.00010619535896694288,
      "learning_rate": 9.956949781644326e-06,
      "loss": 0.0,
      "step": 630
    },
    {
      "epoch": 0.004415407010838444,
      "grad_norm": 1.2944901754963212e-05,
      "learning_rate": 9.956259874298883e-06,
      "loss": 0.0,
      "step": 640
    },
    {
      "epoch": 0.004484397745382795,
      "grad_norm": 6.631881115026772e-05,
      "learning_rate": 9.95556996695344e-06,
      "loss": 0.0001,
      "step": 650
    },
    {
      "epoch": 0.004553388479927146,
      "grad_norm": 0.0,
      "learning_rate": 9.954880059607996e-06,
      "loss": 0.0012,
      "step": 660
    },
    {
      "epoch": 0.004622379214471496,
      "grad_norm": 9.929733641911298e-05,
      "learning_rate": 9.954190152262552e-06,
      "loss": 0.0011,
      "step": 670
    },
    {
      "epoch": 0.004691369949015848,
      "grad_norm": 0.2635310888290405,
      "learning_rate": 9.953500244917107e-06,
      "loss": 0.0,
      "step": 680
    },
    {
      "epoch": 0.004760360683560198,
      "grad_norm": 3.810168323070684e-07,
      "learning_rate": 9.952810337571664e-06,
      "loss": 0.0007,
      "step": 690
    },
    {
      "epoch": 0.004829351418104548,
      "grad_norm": 0.0,
      "learning_rate": 9.952120430226222e-06,
      "loss": 0.0,
      "step": 700
    },
    {
      "epoch": 0.0048983421526489,
      "grad_norm": 5.18747766964367e-10,
      "learning_rate": 9.951430522880778e-06,
      "loss": 0.0019,
      "step": 710
    },
    {
      "epoch": 0.00496733288719325,
      "grad_norm": 1.8703746107462393e-08,
      "learning_rate": 9.950740615535335e-06,
      "loss": 0.0,
      "step": 720
    },
    {
      "epoch": 0.0050363236217376,
      "grad_norm": 0.0,
      "learning_rate": 9.950050708189891e-06,
      "loss": 0.0,
      "step": 730
    },
    {
      "epoch": 0.0051053143562819515,
      "grad_norm": 0.0,
      "learning_rate": 9.949360800844448e-06,
      "loss": 0.5871,
      "step": 740
    },
    {
      "epoch": 0.005174305090826302,
      "grad_norm": 0.0001783966872608289,
      "learning_rate": 9.948670893499004e-06,
      "loss": 0.0,
      "step": 750
    },
    {
      "epoch": 0.005243295825370653,
      "grad_norm": 114.77677917480469,
      "learning_rate": 9.947980986153561e-06,
      "loss": 0.0385,
      "step": 760
    },
    {
      "epoch": 0.0053122865599150035,
      "grad_norm": 7.162038855312858e-07,
      "learning_rate": 9.947291078808117e-06,
      "loss": 0.0001,
      "step": 770
    },
    {
      "epoch": 0.005381277294459354,
      "grad_norm": 0.017689310014247894,
      "learning_rate": 9.946601171462674e-06,
      "loss": 0.0,
      "step": 780
    },
    {
      "epoch": 0.005450268029003705,
      "grad_norm": 5.28181831604968e-10,
      "learning_rate": 9.94591126411723e-06,
      "loss": 0.0,
      "step": 790
    },
    {
      "epoch": 0.0055192587635480555,
      "grad_norm": 3.7083958659422933e-07,
      "learning_rate": 9.945221356771785e-06,
      "loss": 0.0,
      "step": 800
    },
    {
      "epoch": 0.005588249498092406,
      "grad_norm": 1.8758004216579138e-07,
      "learning_rate": 9.944531449426343e-06,
      "loss": 0.0003,
      "step": 810
    },
    {
      "epoch": 0.005657240232636757,
      "grad_norm": 6.176499312005035e-10,
      "learning_rate": 9.9438415420809e-06,
      "loss": 0.0,
      "step": 820
    },
    {
      "epoch": 0.0057262309671811075,
      "grad_norm": 7.015024311840534e-05,
      "learning_rate": 9.943151634735456e-06,
      "loss": 0.0,
      "step": 830
    },
    {
      "epoch": 0.005795221701725459,
      "grad_norm": 2.8280794620513916,
      "learning_rate": 9.942461727390013e-06,
      "loss": 0.0005,
      "step": 840
    },
    {
      "epoch": 0.005864212436269809,
      "grad_norm": 0.0,
      "learning_rate": 9.94177182004457e-06,
      "loss": 0.0001,
      "step": 850
    },
    {
      "epoch": 0.005933203170814159,
      "grad_norm": 0.0,
      "learning_rate": 9.941081912699126e-06,
      "loss": 0.0,
      "step": 860
    },
    {
      "epoch": 0.006002193905358511,
      "grad_norm": 0.00056278525153175,
      "learning_rate": 9.940392005353682e-06,
      "loss": 0.0,
      "step": 870
    },
    {
      "epoch": 0.006071184639902861,
      "grad_norm": 0.0,
      "learning_rate": 9.939702098008239e-06,
      "loss": 0.6328,
      "step": 880
    },
    {
      "epoch": 0.006140175374447211,
      "grad_norm": 0.0,
      "learning_rate": 9.939012190662795e-06,
      "loss": 0.0,
      "step": 890
    },
    {
      "epoch": 0.006209166108991563,
      "grad_norm": 9.825647794059478e-06,
      "learning_rate": 9.938322283317352e-06,
      "loss": 0.0,
      "step": 900
    },
    {
      "epoch": 0.006278156843535913,
      "grad_norm": 0.2298017144203186,
      "learning_rate": 9.937632375971907e-06,
      "loss": 0.0029,
      "step": 910
    },
    {
      "epoch": 0.006347147578080264,
      "grad_norm": 9.138823952525854e-05,
      "learning_rate": 9.936942468626465e-06,
      "loss": 0.0,
      "step": 920
    },
    {
      "epoch": 0.006416138312624615,
      "grad_norm": 0.0308407973498106,
      "learning_rate": 9.936252561281021e-06,
      "loss": 0.0,
      "step": 930
    },
    {
      "epoch": 0.006485129047168965,
      "grad_norm": 4.297233681427315e-05,
      "learning_rate": 9.935562653935578e-06,
      "loss": 0.0011,
      "step": 940
    },
    {
      "epoch": 0.006554119781713316,
      "grad_norm": 2.0112407207489014,
      "learning_rate": 9.934872746590134e-06,
      "loss": 0.0334,
      "step": 950
    },
    {
      "epoch": 0.006623110516257667,
      "grad_norm": 0.31853970885276794,
      "learning_rate": 9.93418283924469e-06,
      "loss": 0.0001,
      "step": 960
    },
    {
      "epoch": 0.006692101250802017,
      "grad_norm": 6.041682154567241e-10,
      "learning_rate": 9.933492931899247e-06,
      "loss": 0.0745,
      "step": 970
    },
    {
      "epoch": 0.006761091985346368,
      "grad_norm": 1.3671727083774954e-09,
      "learning_rate": 9.932803024553804e-06,
      "loss": 0.0029,
      "step": 980
    },
    {
      "epoch": 0.0068300827198907186,
      "grad_norm": 0.00017256668070331216,
      "learning_rate": 9.93211311720836e-06,
      "loss": 0.0,
      "step": 990
    },
    {
      "epoch": 0.00689907345443507,
      "grad_norm": 0.11617961525917053,
      "learning_rate": 9.931423209862917e-06,
      "loss": 0.0004,
      "step": 1000
    },
    {
      "epoch": 0.00696806418897942,
      "grad_norm": 0.0,
      "learning_rate": 9.930733302517473e-06,
      "loss": 0.0002,
      "step": 1010
    },
    {
      "epoch": 0.0070370549235237705,
      "grad_norm": 0.0038849906995892525,
      "learning_rate": 9.930043395172028e-06,
      "loss": 0.3742,
      "step": 1020
    },
    {
      "epoch": 0.007106045658068122,
      "grad_norm": 5.140336489795061e-10,
      "learning_rate": 9.929353487826584e-06,
      "loss": 0.0,
      "step": 1030
    },
    {
      "epoch": 0.007175036392612472,
      "grad_norm": 0.5124749541282654,
      "learning_rate": 9.928663580481143e-06,
      "loss": 0.0001,
      "step": 1040
    },
    {
      "epoch": 0.0072440271271568225,
      "grad_norm": 0.0,
      "learning_rate": 9.927973673135699e-06,
      "loss": 0.0,
      "step": 1050
    },
    {
      "epoch": 0.007313017861701174,
      "grad_norm": 0.00016734229575376958,
      "learning_rate": 9.927283765790256e-06,
      "loss": 0.0,
      "step": 1060
    },
    {
      "epoch": 0.007382008596245524,
      "grad_norm": 9.173207899948466e-07,
      "learning_rate": 9.926593858444812e-06,
      "loss": 0.002,
      "step": 1070
    },
    {
      "epoch": 0.007450999330789875,
      "grad_norm": 0.0,
      "learning_rate": 9.925903951099369e-06,
      "loss": 0.0132,
      "step": 1080
    },
    {
      "epoch": 0.007519990065334226,
      "grad_norm": 3.566971145119169e-07,
      "learning_rate": 9.925214043753925e-06,
      "loss": 0.0,
      "step": 1090
    },
    {
      "epoch": 0.007588980799878576,
      "grad_norm": 0.0,
      "learning_rate": 9.924524136408482e-06,
      "loss": 0.0057,
      "step": 1100
    },
    {
      "epoch": 0.007657971534422927,
      "grad_norm": 0.0016905278898775578,
      "learning_rate": 9.923834229063038e-06,
      "loss": 0.0,
      "step": 1110
    },
    {
      "epoch": 0.007726962268967278,
      "grad_norm": 1.2363362312316895,
      "learning_rate": 9.923144321717595e-06,
      "loss": 0.0064,
      "step": 1120
    },
    {
      "epoch": 0.007795953003511628,
      "grad_norm": 1.342353894706605e-09,
      "learning_rate": 9.922454414372151e-06,
      "loss": 0.1642,
      "step": 1130
    },
    {
      "epoch": 0.007864943738055978,
      "grad_norm": 1.215474185300991e-06,
      "learning_rate": 9.921764507026706e-06,
      "loss": 0.0001,
      "step": 1140
    },
    {
      "epoch": 0.00793393447260033,
      "grad_norm": 0.42875736951828003,
      "learning_rate": 9.921074599681264e-06,
      "loss": 0.0001,
      "step": 1150
    },
    {
      "epoch": 0.008002925207144681,
      "grad_norm": 1.3860927765563247e-06,
      "learning_rate": 9.92038469233582e-06,
      "loss": 0.002,
      "step": 1160
    },
    {
      "epoch": 0.008071915941689031,
      "grad_norm": 1.143737904385489e-07,
      "learning_rate": 9.919694784990377e-06,
      "loss": 0.0,
      "step": 1170
    },
    {
      "epoch": 0.008140906676233382,
      "grad_norm": 0.007383116986602545,
      "learning_rate": 9.919004877644933e-06,
      "loss": 0.0002,
      "step": 1180
    },
    {
      "epoch": 0.008209897410777732,
      "grad_norm": 0.0,
      "learning_rate": 9.91831497029949e-06,
      "loss": 0.0286,
      "step": 1190
    },
    {
      "epoch": 0.008278888145322084,
      "grad_norm": 2.5686742901598336e-06,
      "learning_rate": 9.917625062954046e-06,
      "loss": 0.0,
      "step": 1200
    },
    {
      "epoch": 0.008347878879866434,
      "grad_norm": 5.234279676358256e-10,
      "learning_rate": 9.916935155608603e-06,
      "loss": 0.0,
      "step": 1210
    },
    {
      "epoch": 0.008416869614410785,
      "grad_norm": 5.294443994330322e-08,
      "learning_rate": 9.91624524826316e-06,
      "loss": 0.0,
      "step": 1220
    },
    {
      "epoch": 0.008485860348955135,
      "grad_norm": 2.6260105201458828e-08,
      "learning_rate": 9.915555340917716e-06,
      "loss": 0.0002,
      "step": 1230
    },
    {
      "epoch": 0.008554851083499486,
      "grad_norm": 0.0,
      "learning_rate": 9.914865433572272e-06,
      "loss": 0.0486,
      "step": 1240
    },
    {
      "epoch": 0.008623841818043836,
      "grad_norm": 5.224699998507276e-05,
      "learning_rate": 9.914175526226827e-06,
      "loss": 0.0,
      "step": 1250
    },
    {
      "epoch": 0.008692832552588188,
      "grad_norm": 0.0,
      "learning_rate": 9.913485618881385e-06,
      "loss": 0.0006,
      "step": 1260
    },
    {
      "epoch": 0.008761823287132538,
      "grad_norm": 0.2586248219013214,
      "learning_rate": 9.912795711535942e-06,
      "loss": 0.0,
      "step": 1270
    },
    {
      "epoch": 0.008830814021676889,
      "grad_norm": 0.0006744968122802675,
      "learning_rate": 9.912105804190498e-06,
      "loss": 0.0861,
      "step": 1280
    },
    {
      "epoch": 0.00889980475622124,
      "grad_norm": 0.0,
      "learning_rate": 9.911415896845055e-06,
      "loss": 0.0004,
      "step": 1290
    },
    {
      "epoch": 0.00896879549076559,
      "grad_norm": 0.0036323866806924343,
      "learning_rate": 9.910725989499611e-06,
      "loss": 0.0,
      "step": 1300
    },
    {
      "epoch": 0.009037786225309942,
      "grad_norm": 3.623789072036743,
      "learning_rate": 9.910036082154168e-06,
      "loss": 0.0006,
      "step": 1310
    },
    {
      "epoch": 0.009106776959854292,
      "grad_norm": 1.1184769777017323e-09,
      "learning_rate": 9.909346174808724e-06,
      "loss": 0.0,
      "step": 1320
    },
    {
      "epoch": 0.009175767694398642,
      "grad_norm": 7.058700361994852e-07,
      "learning_rate": 9.90865626746328e-06,
      "loss": 0.0,
      "step": 1330
    },
    {
      "epoch": 0.009244758428942993,
      "grad_norm": 0.0,
      "learning_rate": 9.907966360117837e-06,
      "loss": 0.0013,
      "step": 1340
    },
    {
      "epoch": 0.009313749163487343,
      "grad_norm": 1.382156006002333e-05,
      "learning_rate": 9.907276452772394e-06,
      "loss": 0.0,
      "step": 1350
    },
    {
      "epoch": 0.009382739898031695,
      "grad_norm": 0.0,
      "learning_rate": 9.906586545426949e-06,
      "loss": 0.0,
      "step": 1360
    },
    {
      "epoch": 0.009451730632576046,
      "grad_norm": 1.6501118249578894e-09,
      "learning_rate": 9.905896638081507e-06,
      "loss": 0.0,
      "step": 1370
    },
    {
      "epoch": 0.009520721367120396,
      "grad_norm": 0.0013124076649546623,
      "learning_rate": 9.905206730736063e-06,
      "loss": 0.0034,
      "step": 1380
    },
    {
      "epoch": 0.009589712101664746,
      "grad_norm": 1.388899272569688e-05,
      "learning_rate": 9.90451682339062e-06,
      "loss": 0.0,
      "step": 1390
    },
    {
      "epoch": 0.009658702836209097,
      "grad_norm": 0.0,
      "learning_rate": 9.903826916045176e-06,
      "loss": 0.0002,
      "step": 1400
    },
    {
      "epoch": 0.009727693570753447,
      "grad_norm": 1.2999422871473598e-08,
      "learning_rate": 9.903137008699733e-06,
      "loss": 0.0,
      "step": 1410
    },
    {
      "epoch": 0.0097966843052978,
      "grad_norm": 0.0,
      "learning_rate": 9.90244710135429e-06,
      "loss": 0.0,
      "step": 1420
    },
    {
      "epoch": 0.00986567503984215,
      "grad_norm": 0.0,
      "learning_rate": 9.901757194008846e-06,
      "loss": 0.334,
      "step": 1430
    },
    {
      "epoch": 0.0099346657743865,
      "grad_norm": 0.0,
      "learning_rate": 9.901067286663402e-06,
      "loss": 0.0019,
      "step": 1440
    },
    {
      "epoch": 0.01000365650893085,
      "grad_norm": 0.0,
      "learning_rate": 9.900377379317959e-06,
      "loss": 0.0018,
      "step": 1450
    },
    {
      "epoch": 0.0100726472434752,
      "grad_norm": 9.614233853483256e-10,
      "learning_rate": 9.899687471972515e-06,
      "loss": 0.0,
      "step": 1460
    },
    {
      "epoch": 0.010141637978019553,
      "grad_norm": 1.890835577000871e-08,
      "learning_rate": 9.898997564627072e-06,
      "loss": 0.0039,
      "step": 1470
    },
    {
      "epoch": 0.010210628712563903,
      "grad_norm": 0.0,
      "learning_rate": 9.898307657281628e-06,
      "loss": 0.08,
      "step": 1480
    },
    {
      "epoch": 0.010279619447108253,
      "grad_norm": 0.0,
      "learning_rate": 9.897617749936185e-06,
      "loss": 0.0,
      "step": 1490
    },
    {
      "epoch": 0.010348610181652604,
      "grad_norm": 0.0,
      "learning_rate": 9.896927842590741e-06,
      "loss": 0.0001,
      "step": 1500
    },
    {
      "epoch": 0.010417600916196954,
      "grad_norm": 3.2997333619277924e-05,
      "learning_rate": 9.896237935245298e-06,
      "loss": 0.0,
      "step": 1510
    },
    {
      "epoch": 0.010486591650741306,
      "grad_norm": 1.04883781659737e-06,
      "learning_rate": 9.895548027899854e-06,
      "loss": 0.0155,
      "step": 1520
    },
    {
      "epoch": 0.010555582385285657,
      "grad_norm": 9.53003836912103e-06,
      "learning_rate": 9.89485812055441e-06,
      "loss": 0.0032,
      "step": 1530
    },
    {
      "epoch": 0.010624573119830007,
      "grad_norm": 2.512200891757743e-09,
      "learning_rate": 9.894168213208967e-06,
      "loss": 0.0024,
      "step": 1540
    },
    {
      "epoch": 0.010693563854374357,
      "grad_norm": 0.007814495824277401,
      "learning_rate": 9.893478305863524e-06,
      "loss": 0.0,
      "step": 1550
    },
    {
      "epoch": 0.010762554588918708,
      "grad_norm": 0.00010314479732187465,
      "learning_rate": 9.89278839851808e-06,
      "loss": 0.0001,
      "step": 1560
    },
    {
      "epoch": 0.010831545323463058,
      "grad_norm": 0.0,
      "learning_rate": 9.892098491172637e-06,
      "loss": 0.0,
      "step": 1570
    },
    {
      "epoch": 0.01090053605800741,
      "grad_norm": 0.012568238191306591,
      "learning_rate": 9.891408583827193e-06,
      "loss": 0.6735,
      "step": 1580
    },
    {
      "epoch": 0.01096952679255176,
      "grad_norm": 0.0,
      "learning_rate": 9.89071867648175e-06,
      "loss": 0.0,
      "step": 1590
    },
    {
      "epoch": 0.011038517527096111,
      "grad_norm": 5.0949049182236195e-05,
      "learning_rate": 9.890028769136306e-06,
      "loss": 0.0,
      "step": 1600
    },
    {
      "epoch": 0.011107508261640461,
      "grad_norm": 0.0,
      "learning_rate": 9.889338861790863e-06,
      "loss": 0.0,
      "step": 1610
    },
    {
      "epoch": 0.011176498996184812,
      "grad_norm": 6.443872990757882e-08,
      "learning_rate": 9.888648954445419e-06,
      "loss": 0.0,
      "step": 1620
    },
    {
      "epoch": 0.011245489730729164,
      "grad_norm": 0.0,
      "learning_rate": 9.887959047099976e-06,
      "loss": 0.0,
      "step": 1630
    },
    {
      "epoch": 0.011314480465273514,
      "grad_norm": 0.0,
      "learning_rate": 9.887269139754532e-06,
      "loss": 0.0093,
      "step": 1640
    },
    {
      "epoch": 0.011383471199817865,
      "grad_norm": 0.00022462219931185246,
      "learning_rate": 9.886579232409089e-06,
      "loss": 0.0001,
      "step": 1650
    },
    {
      "epoch": 0.011452461934362215,
      "grad_norm": 0.0007001527701504529,
      "learning_rate": 9.885889325063645e-06,
      "loss": 0.0,
      "step": 1660
    },
    {
      "epoch": 0.011521452668906565,
      "grad_norm": 0.0,
      "learning_rate": 9.885199417718201e-06,
      "loss": 0.0,
      "step": 1670
    },
    {
      "epoch": 0.011590443403450917,
      "grad_norm": 2.575731195975095e-05,
      "learning_rate": 9.884509510372758e-06,
      "loss": 0.0,
      "step": 1680
    },
    {
      "epoch": 0.011659434137995268,
      "grad_norm": 0.0,
      "learning_rate": 9.883819603027314e-06,
      "loss": 0.0046,
      "step": 1690
    },
    {
      "epoch": 0.011728424872539618,
      "grad_norm": 0.0,
      "learning_rate": 9.883129695681871e-06,
      "loss": 0.0,
      "step": 1700
    },
    {
      "epoch": 0.011797415607083968,
      "grad_norm": 4.60377151512148e-07,
      "learning_rate": 9.882439788336427e-06,
      "loss": 0.0,
      "step": 1710
    },
    {
      "epoch": 0.011866406341628319,
      "grad_norm": 0.0,
      "learning_rate": 9.881749880990984e-06,
      "loss": 0.0,
      "step": 1720
    },
    {
      "epoch": 0.01193539707617267,
      "grad_norm": 4.682962351232334e-10,
      "learning_rate": 9.88105997364554e-06,
      "loss": 0.0,
      "step": 1730
    },
    {
      "epoch": 0.012004387810717021,
      "grad_norm": 1.7960427385332878e-06,
      "learning_rate": 9.880370066300097e-06,
      "loss": 0.0001,
      "step": 1740
    },
    {
      "epoch": 0.012073378545261372,
      "grad_norm": 0.0,
      "learning_rate": 9.879680158954653e-06,
      "loss": 0.0,
      "step": 1750
    },
    {
      "epoch": 0.012142369279805722,
      "grad_norm": 0.0,
      "learning_rate": 9.87899025160921e-06,
      "loss": 0.0,
      "step": 1760
    },
    {
      "epoch": 0.012211360014350072,
      "grad_norm": 1.149840682046488e-05,
      "learning_rate": 9.878300344263766e-06,
      "loss": 0.0,
      "step": 1770
    },
    {
      "epoch": 0.012280350748894423,
      "grad_norm": 6.41109954813146e-06,
      "learning_rate": 9.877610436918323e-06,
      "loss": 0.0,
      "step": 1780
    },
    {
      "epoch": 0.012349341483438775,
      "grad_norm": 0.0,
      "learning_rate": 9.87692052957288e-06,
      "loss": 0.0,
      "step": 1790
    },
    {
      "epoch": 0.012418332217983125,
      "grad_norm": 0.0,
      "learning_rate": 9.876230622227436e-06,
      "loss": 0.0,
      "step": 1800
    },
    {
      "epoch": 0.012487322952527476,
      "grad_norm": 0.0,
      "learning_rate": 9.875540714881992e-06,
      "loss": 0.0,
      "step": 1810
    },
    {
      "epoch": 0.012556313687071826,
      "grad_norm": 0.05592973902821541,
      "learning_rate": 9.874850807536549e-06,
      "loss": 0.0003,
      "step": 1820
    },
    {
      "epoch": 0.012625304421616176,
      "grad_norm": 0.0,
      "learning_rate": 9.874160900191105e-06,
      "loss": 0.0,
      "step": 1830
    },
    {
      "epoch": 0.012694295156160528,
      "grad_norm": 8.634918231109623e-06,
      "learning_rate": 9.873470992845662e-06,
      "loss": 0.0,
      "step": 1840
    },
    {
      "epoch": 0.012763285890704879,
      "grad_norm": 2.4053812012425624e-05,
      "learning_rate": 9.872781085500218e-06,
      "loss": 0.0212,
      "step": 1850
    },
    {
      "epoch": 0.01283227662524923,
      "grad_norm": 1.4181620827002916e-05,
      "learning_rate": 9.872091178154775e-06,
      "loss": 0.0003,
      "step": 1860
    },
    {
      "epoch": 0.01290126735979358,
      "grad_norm": 3.870967191232921e-08,
      "learning_rate": 9.871401270809331e-06,
      "loss": 0.0,
      "step": 1870
    },
    {
      "epoch": 0.01297025809433793,
      "grad_norm": 0.2523908019065857,
      "learning_rate": 9.870711363463888e-06,
      "loss": 0.0205,
      "step": 1880
    },
    {
      "epoch": 0.01303924882888228,
      "grad_norm": 23.197324752807617,
      "learning_rate": 9.870021456118444e-06,
      "loss": 0.0069,
      "step": 1890
    },
    {
      "epoch": 0.013108239563426632,
      "grad_norm": 0.00017800922796595842,
      "learning_rate": 9.869331548773e-06,
      "loss": 0.0003,
      "step": 1900
    },
    {
      "epoch": 0.013177230297970983,
      "grad_norm": 0.0,
      "learning_rate": 9.868641641427557e-06,
      "loss": 0.0,
      "step": 1910
    },
    {
      "epoch": 0.013246221032515333,
      "grad_norm": 0.0,
      "learning_rate": 9.867951734082114e-06,
      "loss": 0.0,
      "step": 1920
    },
    {
      "epoch": 0.013315211767059684,
      "grad_norm": 0.0007261246792040765,
      "learning_rate": 9.86726182673667e-06,
      "loss": 0.0004,
      "step": 1930
    },
    {
      "epoch": 0.013384202501604034,
      "grad_norm": 0.007969099096953869,
      "learning_rate": 9.866571919391227e-06,
      "loss": 0.0,
      "step": 1940
    },
    {
      "epoch": 0.013453193236148386,
      "grad_norm": 0.0,
      "learning_rate": 9.865882012045783e-06,
      "loss": 0.0862,
      "step": 1950
    },
    {
      "epoch": 0.013522183970692736,
      "grad_norm": 0.00247814878821373,
      "learning_rate": 9.86519210470034e-06,
      "loss": 0.0,
      "step": 1960
    },
    {
      "epoch": 0.013591174705237087,
      "grad_norm": 0.0,
      "learning_rate": 9.864502197354896e-06,
      "loss": 0.0307,
      "step": 1970
    },
    {
      "epoch": 0.013660165439781437,
      "grad_norm": 0.0,
      "learning_rate": 9.863812290009453e-06,
      "loss": 0.0035,
      "step": 1980
    },
    {
      "epoch": 0.013729156174325787,
      "grad_norm": 177.79531860351562,
      "learning_rate": 9.86312238266401e-06,
      "loss": 0.0829,
      "step": 1990
    },
    {
      "epoch": 0.01379814690887014,
      "grad_norm": 0.0,
      "learning_rate": 9.862432475318566e-06,
      "loss": 0.0057,
      "step": 2000
    },
    {
      "epoch": 0.01386713764341449,
      "grad_norm": 0.0,
      "learning_rate": 9.861742567973122e-06,
      "loss": 0.0003,
      "step": 2010
    },
    {
      "epoch": 0.01393612837795884,
      "grad_norm": 7.60870477733988e-09,
      "learning_rate": 9.861052660627679e-06,
      "loss": 0.1116,
      "step": 2020
    },
    {
      "epoch": 0.01400511911250319,
      "grad_norm": 2.393839849901269e-06,
      "learning_rate": 9.860362753282235e-06,
      "loss": 0.057,
      "step": 2030
    },
    {
      "epoch": 0.014074109847047541,
      "grad_norm": 2.7021491405321285e-05,
      "learning_rate": 9.859672845936792e-06,
      "loss": 0.0002,
      "step": 2040
    },
    {
      "epoch": 0.014143100581591891,
      "grad_norm": 5.14019937725152e-10,
      "learning_rate": 9.858982938591348e-06,
      "loss": 0.0,
      "step": 2050
    },
    {
      "epoch": 0.014212091316136244,
      "grad_norm": 0.0,
      "learning_rate": 9.858293031245905e-06,
      "loss": 0.0,
      "step": 2060
    },
    {
      "epoch": 0.014281082050680594,
      "grad_norm": 0.0,
      "learning_rate": 9.857603123900461e-06,
      "loss": 0.0001,
      "step": 2070
    },
    {
      "epoch": 0.014350072785224944,
      "grad_norm": 0.0026891904417425394,
      "learning_rate": 9.856913216555018e-06,
      "loss": 0.0,
      "step": 2080
    },
    {
      "epoch": 0.014419063519769295,
      "grad_norm": 5.3279332945521674e-08,
      "learning_rate": 9.856223309209574e-06,
      "loss": 0.0,
      "step": 2090
    },
    {
      "epoch": 0.014488054254313645,
      "grad_norm": 0.11718526482582092,
      "learning_rate": 9.85553340186413e-06,
      "loss": 0.084,
      "step": 2100
    },
    {
      "epoch": 0.014557044988857997,
      "grad_norm": 9.937484719557688e-07,
      "learning_rate": 9.854843494518687e-06,
      "loss": 0.0,
      "step": 2110
    },
    {
      "epoch": 0.014626035723402347,
      "grad_norm": 0.007573988754302263,
      "learning_rate": 9.854153587173244e-06,
      "loss": 0.0007,
      "step": 2120
    },
    {
      "epoch": 0.014695026457946698,
      "grad_norm": 0.009642322547733784,
      "learning_rate": 9.8534636798278e-06,
      "loss": 0.0,
      "step": 2130
    },
    {
      "epoch": 0.014764017192491048,
      "grad_norm": 2.301829908901709e-06,
      "learning_rate": 9.852773772482357e-06,
      "loss": 0.0,
      "step": 2140
    },
    {
      "epoch": 0.014833007927035399,
      "grad_norm": 0.0,
      "learning_rate": 9.852083865136913e-06,
      "loss": 0.0,
      "step": 2150
    },
    {
      "epoch": 0.01490199866157975,
      "grad_norm": 0.017099348828196526,
      "learning_rate": 9.85139395779147e-06,
      "loss": 0.0,
      "step": 2160
    },
    {
      "epoch": 0.014970989396124101,
      "grad_norm": 0.6120503544807434,
      "learning_rate": 9.850704050446026e-06,
      "loss": 0.0003,
      "step": 2170
    },
    {
      "epoch": 0.015039980130668451,
      "grad_norm": 2.8864574432373047,
      "learning_rate": 9.850014143100582e-06,
      "loss": 0.0009,
      "step": 2180
    },
    {
      "epoch": 0.015108970865212802,
      "grad_norm": 0.0,
      "learning_rate": 9.849324235755139e-06,
      "loss": 0.0,
      "step": 2190
    },
    {
      "epoch": 0.015177961599757152,
      "grad_norm": 0.0,
      "learning_rate": 9.848634328409695e-06,
      "loss": 0.0,
      "step": 2200
    },
    {
      "epoch": 0.015246952334301503,
      "grad_norm": 0.0,
      "learning_rate": 9.847944421064252e-06,
      "loss": 0.0002,
      "step": 2210
    },
    {
      "epoch": 0.015315943068845855,
      "grad_norm": 0.0,
      "learning_rate": 9.847254513718808e-06,
      "loss": 0.0003,
      "step": 2220
    },
    {
      "epoch": 0.015384933803390205,
      "grad_norm": 246.43312072753906,
      "learning_rate": 9.846564606373365e-06,
      "loss": 0.0809,
      "step": 2230
    },
    {
      "epoch": 0.015453924537934555,
      "grad_norm": 0.0,
      "learning_rate": 9.845874699027921e-06,
      "loss": 0.0,
      "step": 2240
    },
    {
      "epoch": 0.015522915272478906,
      "grad_norm": 0.001674336614087224,
      "learning_rate": 9.845184791682478e-06,
      "loss": 0.0,
      "step": 2250
    },
    {
      "epoch": 0.015591906007023256,
      "grad_norm": 0.6877070665359497,
      "learning_rate": 9.844494884337034e-06,
      "loss": 0.0003,
      "step": 2260
    },
    {
      "epoch": 0.015660896741567606,
      "grad_norm": 0.0,
      "learning_rate": 9.843804976991591e-06,
      "loss": 0.0,
      "step": 2270
    },
    {
      "epoch": 0.015729887476111957,
      "grad_norm": 0.02825876511633396,
      "learning_rate": 9.843115069646147e-06,
      "loss": 0.0,
      "step": 2280
    },
    {
      "epoch": 0.015798878210656307,
      "grad_norm": 6.0335845947265625,
      "learning_rate": 9.842425162300704e-06,
      "loss": 0.0012,
      "step": 2290
    },
    {
      "epoch": 0.01586786894520066,
      "grad_norm": 1.7432723709021047e-08,
      "learning_rate": 9.84173525495526e-06,
      "loss": 0.0,
      "step": 2300
    },
    {
      "epoch": 0.01593685967974501,
      "grad_norm": 0.0,
      "learning_rate": 9.841045347609817e-06,
      "loss": 0.0038,
      "step": 2310
    },
    {
      "epoch": 0.016005850414289362,
      "grad_norm": 0.0,
      "learning_rate": 9.840355440264373e-06,
      "loss": 0.0,
      "step": 2320
    },
    {
      "epoch": 0.016074841148833712,
      "grad_norm": 0.0008846206474117935,
      "learning_rate": 9.83966553291893e-06,
      "loss": 0.0,
      "step": 2330
    },
    {
      "epoch": 0.016143831883378063,
      "grad_norm": 13.107681274414062,
      "learning_rate": 9.838975625573486e-06,
      "loss": 0.0026,
      "step": 2340
    },
    {
      "epoch": 0.016212822617922413,
      "grad_norm": 6.843339974693663e-07,
      "learning_rate": 9.838285718228043e-06,
      "loss": 0.0647,
      "step": 2350
    },
    {
      "epoch": 0.016281813352466763,
      "grad_norm": 0.0,
      "learning_rate": 9.8375958108826e-06,
      "loss": 0.0,
      "step": 2360
    },
    {
      "epoch": 0.016350804087011114,
      "grad_norm": 0.0,
      "learning_rate": 9.836905903537156e-06,
      "loss": 0.0,
      "step": 2370
    },
    {
      "epoch": 0.016419794821555464,
      "grad_norm": 0.0,
      "learning_rate": 9.836215996191712e-06,
      "loss": 0.0007,
      "step": 2380
    },
    {
      "epoch": 0.016488785556099814,
      "grad_norm": 57.82735824584961,
      "learning_rate": 9.835526088846269e-06,
      "loss": 0.0168,
      "step": 2390
    },
    {
      "epoch": 0.016557776290644168,
      "grad_norm": 1.4099719919613563e-05,
      "learning_rate": 9.834836181500825e-06,
      "loss": 0.0001,
      "step": 2400
    },
    {
      "epoch": 0.01662676702518852,
      "grad_norm": 2.7681010261737526e-10,
      "learning_rate": 9.834146274155382e-06,
      "loss": 0.0,
      "step": 2410
    },
    {
      "epoch": 0.01669575775973287,
      "grad_norm": 1.6342320350304362e-06,
      "learning_rate": 9.833456366809938e-06,
      "loss": 0.0002,
      "step": 2420
    },
    {
      "epoch": 0.01676474849427722,
      "grad_norm": 0.0062063587829470634,
      "learning_rate": 9.832766459464495e-06,
      "loss": 0.0,
      "step": 2430
    },
    {
      "epoch": 0.01683373922882157,
      "grad_norm": 0.0,
      "learning_rate": 9.832076552119051e-06,
      "loss": 0.0141,
      "step": 2440
    },
    {
      "epoch": 0.01690272996336592,
      "grad_norm": 7.312730303965509e-05,
      "learning_rate": 9.831386644773608e-06,
      "loss": 0.0,
      "step": 2450
    },
    {
      "epoch": 0.01697172069791027,
      "grad_norm": 0.0,
      "learning_rate": 9.830696737428164e-06,
      "loss": 0.0012,
      "step": 2460
    },
    {
      "epoch": 0.01704071143245462,
      "grad_norm": 2.615517524784394e-10,
      "learning_rate": 9.83000683008272e-06,
      "loss": 0.0,
      "step": 2470
    },
    {
      "epoch": 0.01710970216699897,
      "grad_norm": 3.7908304761913314e-07,
      "learning_rate": 9.829316922737277e-06,
      "loss": 0.0002,
      "step": 2480
    },
    {
      "epoch": 0.01717869290154332,
      "grad_norm": 3.4277763916179538e-06,
      "learning_rate": 9.828627015391834e-06,
      "loss": 0.0,
      "step": 2490
    },
    {
      "epoch": 0.017247683636087672,
      "grad_norm": 4.6413342613682573e-10,
      "learning_rate": 9.82793710804639e-06,
      "loss": 0.0,
      "step": 2500
    },
    {
      "epoch": 0.017316674370632026,
      "grad_norm": 5.556948678986373e-08,
      "learning_rate": 9.827247200700947e-06,
      "loss": 0.0039,
      "step": 2510
    },
    {
      "epoch": 0.017385665105176376,
      "grad_norm": 2.60270555108022e-10,
      "learning_rate": 9.826557293355503e-06,
      "loss": 0.0141,
      "step": 2520
    },
    {
      "epoch": 0.017454655839720726,
      "grad_norm": 0.031119007617235184,
      "learning_rate": 9.82586738601006e-06,
      "loss": 0.0,
      "step": 2530
    },
    {
      "epoch": 0.017523646574265077,
      "grad_norm": 0.0,
      "learning_rate": 9.825177478664616e-06,
      "loss": 0.0,
      "step": 2540
    },
    {
      "epoch": 0.017592637308809427,
      "grad_norm": 5.535717995108769e-10,
      "learning_rate": 9.824556562053717e-06,
      "loss": 0.5637,
      "step": 2550
    },
    {
      "epoch": 0.017661628043353778,
      "grad_norm": 0.0,
      "learning_rate": 9.823866654708273e-06,
      "loss": 0.0,
      "step": 2560
    },
    {
      "epoch": 0.017730618777898128,
      "grad_norm": 0.0,
      "learning_rate": 9.82317674736283e-06,
      "loss": 0.001,
      "step": 2570
    },
    {
      "epoch": 0.01779960951244248,
      "grad_norm": 1.5067802223711624e-06,
      "learning_rate": 9.822486840017386e-06,
      "loss": 0.0658,
      "step": 2580
    },
    {
      "epoch": 0.01786860024698683,
      "grad_norm": 0.001151421689428389,
      "learning_rate": 9.821796932671943e-06,
      "loss": 0.001,
      "step": 2590
    },
    {
      "epoch": 0.01793759098153118,
      "grad_norm": 0.3574353754520416,
      "learning_rate": 9.8211070253265e-06,
      "loss": 0.0001,
      "step": 2600
    },
    {
      "epoch": 0.01800658171607553,
      "grad_norm": 0.0,
      "learning_rate": 9.820417117981056e-06,
      "loss": 0.0,
      "step": 2610
    },
    {
      "epoch": 0.018075572450619883,
      "grad_norm": 0.003819757141172886,
      "learning_rate": 9.819727210635612e-06,
      "loss": 0.0,
      "step": 2620
    },
    {
      "epoch": 0.018144563185164234,
      "grad_norm": 0.0,
      "learning_rate": 9.819037303290169e-06,
      "loss": 0.0,
      "step": 2630
    },
    {
      "epoch": 0.018213553919708584,
      "grad_norm": 0.0005704404320567846,
      "learning_rate": 9.818347395944725e-06,
      "loss": 0.0003,
      "step": 2640
    },
    {
      "epoch": 0.018282544654252934,
      "grad_norm": 3.914003343652439e-08,
      "learning_rate": 9.817657488599282e-06,
      "loss": 0.0,
      "step": 2650
    },
    {
      "epoch": 0.018351535388797285,
      "grad_norm": 0.0,
      "learning_rate": 9.816967581253838e-06,
      "loss": 0.5473,
      "step": 2660
    },
    {
      "epoch": 0.018420526123341635,
      "grad_norm": 0.2177378535270691,
      "learning_rate": 9.816277673908395e-06,
      "loss": 0.0,
      "step": 2670
    },
    {
      "epoch": 0.018489516857885985,
      "grad_norm": 7.915277092251927e-05,
      "learning_rate": 9.815587766562951e-06,
      "loss": 0.0,
      "step": 2680
    },
    {
      "epoch": 0.018558507592430336,
      "grad_norm": 0.10902252793312073,
      "learning_rate": 9.814897859217508e-06,
      "loss": 0.0005,
      "step": 2690
    },
    {
      "epoch": 0.018627498326974686,
      "grad_norm": 8.404355433810906e-09,
      "learning_rate": 9.814207951872064e-06,
      "loss": 0.0,
      "step": 2700
    },
    {
      "epoch": 0.018696489061519037,
      "grad_norm": 4.28242691308256e-10,
      "learning_rate": 9.81351804452662e-06,
      "loss": 0.0,
      "step": 2710
    },
    {
      "epoch": 0.01876547979606339,
      "grad_norm": 0.0,
      "learning_rate": 9.812828137181177e-06,
      "loss": 0.0,
      "step": 2720
    },
    {
      "epoch": 0.01883447053060774,
      "grad_norm": 0.0,
      "learning_rate": 9.812138229835734e-06,
      "loss": 0.0,
      "step": 2730
    },
    {
      "epoch": 0.01890346126515209,
      "grad_norm": 6.023535803478808e-08,
      "learning_rate": 9.81144832249029e-06,
      "loss": 0.0,
      "step": 2740
    },
    {
      "epoch": 0.01897245199969644,
      "grad_norm": 2.1577066945610568e-05,
      "learning_rate": 9.810758415144847e-06,
      "loss": 0.0009,
      "step": 2750
    },
    {
      "epoch": 0.019041442734240792,
      "grad_norm": 0.0007471358985640109,
      "learning_rate": 9.810068507799403e-06,
      "loss": 0.0,
      "step": 2760
    },
    {
      "epoch": 0.019110433468785142,
      "grad_norm": 0.0,
      "learning_rate": 9.80937860045396e-06,
      "loss": 0.0008,
      "step": 2770
    },
    {
      "epoch": 0.019179424203329493,
      "grad_norm": 0.02438054047524929,
      "learning_rate": 9.808688693108516e-06,
      "loss": 0.0001,
      "step": 2780
    },
    {
      "epoch": 0.019248414937873843,
      "grad_norm": 0.0013753767125308514,
      "learning_rate": 9.807998785763073e-06,
      "loss": 0.0475,
      "step": 2790
    },
    {
      "epoch": 0.019317405672418193,
      "grad_norm": 0.0,
      "learning_rate": 9.807308878417629e-06,
      "loss": 0.0,
      "step": 2800
    },
    {
      "epoch": 0.019386396406962544,
      "grad_norm": 0.006645300425589085,
      "learning_rate": 9.806618971072186e-06,
      "loss": 0.0027,
      "step": 2810
    },
    {
      "epoch": 0.019455387141506894,
      "grad_norm": 7.56134976853673e-09,
      "learning_rate": 9.805929063726742e-06,
      "loss": 0.0,
      "step": 2820
    },
    {
      "epoch": 0.019524377876051248,
      "grad_norm": 2.1843302278057308e-08,
      "learning_rate": 9.805239156381299e-06,
      "loss": 0.0,
      "step": 2830
    },
    {
      "epoch": 0.0195933686105956,
      "grad_norm": 0.0,
      "learning_rate": 9.804549249035855e-06,
      "loss": 0.0,
      "step": 2840
    },
    {
      "epoch": 0.01966235934513995,
      "grad_norm": 734.3946533203125,
      "learning_rate": 9.803859341690412e-06,
      "loss": 0.6703,
      "step": 2850
    },
    {
      "epoch": 0.0197313500796843,
      "grad_norm": 0.0,
      "learning_rate": 9.803169434344968e-06,
      "loss": 0.0,
      "step": 2860
    },
    {
      "epoch": 0.01980034081422865,
      "grad_norm": 225.9371337890625,
      "learning_rate": 9.802479526999525e-06,
      "loss": 0.0504,
      "step": 2870
    },
    {
      "epoch": 0.019869331548773,
      "grad_norm": 0.0,
      "learning_rate": 9.801789619654081e-06,
      "loss": 0.0,
      "step": 2880
    },
    {
      "epoch": 0.01993832228331735,
      "grad_norm": 2.3707333873801417e-09,
      "learning_rate": 9.801099712308637e-06,
      "loss": 0.0,
      "step": 2890
    },
    {
      "epoch": 0.0200073130178617,
      "grad_norm": 0.0,
      "learning_rate": 9.800409804963194e-06,
      "loss": 0.0,
      "step": 2900
    },
    {
      "epoch": 0.02007630375240605,
      "grad_norm": 0.0,
      "learning_rate": 9.79971989761775e-06,
      "loss": 0.0,
      "step": 2910
    },
    {
      "epoch": 0.0201452944869504,
      "grad_norm": 0.007832271046936512,
      "learning_rate": 9.799029990272307e-06,
      "loss": 0.012,
      "step": 2920
    },
    {
      "epoch": 0.02021428522149475,
      "grad_norm": 1.1749706345653976e-06,
      "learning_rate": 9.798340082926863e-06,
      "loss": 0.0,
      "step": 2930
    },
    {
      "epoch": 0.020283275956039105,
      "grad_norm": 0.0,
      "learning_rate": 9.79765017558142e-06,
      "loss": 0.0028,
      "step": 2940
    },
    {
      "epoch": 0.020352266690583456,
      "grad_norm": 0.0,
      "learning_rate": 9.796960268235976e-06,
      "loss": 0.0002,
      "step": 2950
    },
    {
      "epoch": 0.020421257425127806,
      "grad_norm": 0.0,
      "learning_rate": 9.796270360890533e-06,
      "loss": 0.0,
      "step": 2960
    },
    {
      "epoch": 0.020490248159672157,
      "grad_norm": 0.0,
      "learning_rate": 9.79558045354509e-06,
      "loss": 0.0,
      "step": 2970
    },
    {
      "epoch": 0.020559238894216507,
      "grad_norm": 0.04446733742952347,
      "learning_rate": 9.794890546199646e-06,
      "loss": 0.0002,
      "step": 2980
    },
    {
      "epoch": 0.020628229628760857,
      "grad_norm": 4.964587629885386e-10,
      "learning_rate": 9.794200638854202e-06,
      "loss": 0.0,
      "step": 2990
    },
    {
      "epoch": 0.020697220363305208,
      "grad_norm": 0.0,
      "learning_rate": 9.793510731508759e-06,
      "loss": 0.0418,
      "step": 3000
    },
    {
      "epoch": 0.020766211097849558,
      "grad_norm": 0.0,
      "learning_rate": 9.792820824163315e-06,
      "loss": 0.0,
      "step": 3010
    },
    {
      "epoch": 0.02083520183239391,
      "grad_norm": 0.0010562562383711338,
      "learning_rate": 9.792130916817872e-06,
      "loss": 0.0043,
      "step": 3020
    },
    {
      "epoch": 0.02090419256693826,
      "grad_norm": 0.0,
      "learning_rate": 9.791441009472428e-06,
      "loss": 0.0,
      "step": 3030
    },
    {
      "epoch": 0.020973183301482613,
      "grad_norm": 1.505286491010338e-06,
      "learning_rate": 9.790751102126985e-06,
      "loss": 0.0012,
      "step": 3040
    },
    {
      "epoch": 0.021042174036026963,
      "grad_norm": 0.0,
      "learning_rate": 9.790061194781541e-06,
      "loss": 0.0073,
      "step": 3050
    },
    {
      "epoch": 0.021111164770571313,
      "grad_norm": 0.0,
      "learning_rate": 9.789371287436098e-06,
      "loss": 0.0004,
      "step": 3060
    },
    {
      "epoch": 0.021180155505115664,
      "grad_norm": 0.0,
      "learning_rate": 9.788681380090654e-06,
      "loss": 0.1625,
      "step": 3070
    },
    {
      "epoch": 0.021249146239660014,
      "grad_norm": 0.0003871853114105761,
      "learning_rate": 9.78799147274521e-06,
      "loss": 0.0002,
      "step": 3080
    },
    {
      "epoch": 0.021318136974204364,
      "grad_norm": 0.0,
      "learning_rate": 9.787301565399767e-06,
      "loss": 0.0,
      "step": 3090
    },
    {
      "epoch": 0.021387127708748715,
      "grad_norm": 7.279188594111474e-07,
      "learning_rate": 9.786611658054324e-06,
      "loss": 0.0151,
      "step": 3100
    },
    {
      "epoch": 0.021456118443293065,
      "grad_norm": 0.0,
      "learning_rate": 9.78592175070888e-06,
      "loss": 0.0,
      "step": 3110
    },
    {
      "epoch": 0.021525109177837416,
      "grad_norm": 75.95635986328125,
      "learning_rate": 9.785231843363437e-06,
      "loss": 0.0326,
      "step": 3120
    },
    {
      "epoch": 0.021594099912381766,
      "grad_norm": 18.149946212768555,
      "learning_rate": 9.784541936017993e-06,
      "loss": 0.0039,
      "step": 3130
    },
    {
      "epoch": 0.021663090646926116,
      "grad_norm": 0.005141737405210733,
      "learning_rate": 9.78385202867255e-06,
      "loss": 0.4755,
      "step": 3140
    },
    {
      "epoch": 0.02173208138147047,
      "grad_norm": 0.0,
      "learning_rate": 9.783162121327106e-06,
      "loss": 0.0,
      "step": 3150
    },
    {
      "epoch": 0.02180107211601482,
      "grad_norm": 1.9227761640649987e-06,
      "learning_rate": 9.782472213981663e-06,
      "loss": 0.0015,
      "step": 3160
    },
    {
      "epoch": 0.02187006285055917,
      "grad_norm": 0.0,
      "learning_rate": 9.78178230663622e-06,
      "loss": 0.0,
      "step": 3170
    },
    {
      "epoch": 0.02193905358510352,
      "grad_norm": 0.0,
      "learning_rate": 9.781092399290776e-06,
      "loss": 0.0,
      "step": 3180
    },
    {
      "epoch": 0.02200804431964787,
      "grad_norm": 0.0,
      "learning_rate": 9.780402491945332e-06,
      "loss": 0.0,
      "step": 3190
    },
    {
      "epoch": 0.022077035054192222,
      "grad_norm": 0.0,
      "learning_rate": 9.779712584599889e-06,
      "loss": 0.0014,
      "step": 3200
    },
    {
      "epoch": 0.022146025788736572,
      "grad_norm": 0.0006382575375027955,
      "learning_rate": 9.779022677254445e-06,
      "loss": 0.0,
      "step": 3210
    },
    {
      "epoch": 0.022215016523280923,
      "grad_norm": 1.0402134656906128,
      "learning_rate": 9.778332769909002e-06,
      "loss": 0.0002,
      "step": 3220
    },
    {
      "epoch": 0.022284007257825273,
      "grad_norm": 5.689083648618976e-10,
      "learning_rate": 9.777642862563558e-06,
      "loss": 0.0,
      "step": 3230
    },
    {
      "epoch": 0.022352997992369623,
      "grad_norm": 0.0,
      "learning_rate": 9.776952955218115e-06,
      "loss": 0.0,
      "step": 3240
    },
    {
      "epoch": 0.022421988726913974,
      "grad_norm": 0.0,
      "learning_rate": 9.776263047872671e-06,
      "loss": 0.0,
      "step": 3250
    },
    {
      "epoch": 0.022490979461458328,
      "grad_norm": 1.0903780101045868e-09,
      "learning_rate": 9.775573140527228e-06,
      "loss": 0.0,
      "step": 3260
    },
    {
      "epoch": 0.022559970196002678,
      "grad_norm": 0.028465013951063156,
      "learning_rate": 9.774883233181784e-06,
      "loss": 0.0345,
      "step": 3270
    },
    {
      "epoch": 0.02262896093054703,
      "grad_norm": 0.0,
      "learning_rate": 9.77419332583634e-06,
      "loss": 0.0,
      "step": 3280
    },
    {
      "epoch": 0.02269795166509138,
      "grad_norm": 6.543196207076107e-09,
      "learning_rate": 9.773503418490897e-06,
      "loss": 0.0,
      "step": 3290
    },
    {
      "epoch": 0.02276694239963573,
      "grad_norm": 0.006624211091548204,
      "learning_rate": 9.772813511145454e-06,
      "loss": 0.0,
      "step": 3300
    },
    {
      "epoch": 0.02283593313418008,
      "grad_norm": 0.0,
      "learning_rate": 9.77212360380001e-06,
      "loss": 0.0,
      "step": 3310
    },
    {
      "epoch": 0.02290492386872443,
      "grad_norm": 0.0,
      "learning_rate": 9.771433696454567e-06,
      "loss": 0.0009,
      "step": 3320
    },
    {
      "epoch": 0.02297391460326878,
      "grad_norm": 0.0,
      "learning_rate": 9.770743789109123e-06,
      "loss": 0.0,
      "step": 3330
    },
    {
      "epoch": 0.02304290533781313,
      "grad_norm": 4.996880687002658e-10,
      "learning_rate": 9.77005388176368e-06,
      "loss": 0.0,
      "step": 3340
    },
    {
      "epoch": 0.02311189607235748,
      "grad_norm": 0.0,
      "learning_rate": 9.769363974418236e-06,
      "loss": 0.0,
      "step": 3350
    },
    {
      "epoch": 0.023180886806901835,
      "grad_norm": 0.0,
      "learning_rate": 9.768674067072793e-06,
      "loss": 0.0,
      "step": 3360
    },
    {
      "epoch": 0.023249877541446185,
      "grad_norm": 0.0,
      "learning_rate": 9.767984159727349e-06,
      "loss": 0.0064,
      "step": 3370
    },
    {
      "epoch": 0.023318868275990536,
      "grad_norm": 0.0,
      "learning_rate": 9.767294252381906e-06,
      "loss": 0.001,
      "step": 3380
    },
    {
      "epoch": 0.023387859010534886,
      "grad_norm": 0.22072994709014893,
      "learning_rate": 9.766604345036462e-06,
      "loss": 0.0001,
      "step": 3390
    },
    {
      "epoch": 0.023456849745079236,
      "grad_norm": 0.0,
      "learning_rate": 9.765914437691019e-06,
      "loss": 0.0,
      "step": 3400
    },
    {
      "epoch": 0.023525840479623587,
      "grad_norm": 0.001880093477666378,
      "learning_rate": 9.765224530345575e-06,
      "loss": 0.0,
      "step": 3410
    },
    {
      "epoch": 0.023594831214167937,
      "grad_norm": 8.820308372570196e-10,
      "learning_rate": 9.764534623000131e-06,
      "loss": 0.0,
      "step": 3420
    },
    {
      "epoch": 0.023663821948712287,
      "grad_norm": 0.0,
      "learning_rate": 9.763844715654688e-06,
      "loss": 0.0001,
      "step": 3430
    },
    {
      "epoch": 0.023732812683256638,
      "grad_norm": 332.05767822265625,
      "learning_rate": 9.763154808309246e-06,
      "loss": 0.0845,
      "step": 3440
    },
    {
      "epoch": 0.023801803417800988,
      "grad_norm": 0.0,
      "learning_rate": 9.762464900963801e-06,
      "loss": 0.0064,
      "step": 3450
    },
    {
      "epoch": 0.02387079415234534,
      "grad_norm": 0.0,
      "learning_rate": 9.761774993618357e-06,
      "loss": 0.0,
      "step": 3460
    },
    {
      "epoch": 0.023939784886889692,
      "grad_norm": 0.0018555239075794816,
      "learning_rate": 9.761085086272914e-06,
      "loss": 0.0001,
      "step": 3470
    },
    {
      "epoch": 0.024008775621434043,
      "grad_norm": 0.1549466997385025,
      "learning_rate": 9.76039517892747e-06,
      "loss": 0.7553,
      "step": 3480
    },
    {
      "epoch": 0.024077766355978393,
      "grad_norm": 4.988789328308485e-07,
      "learning_rate": 9.759705271582027e-06,
      "loss": 0.0,
      "step": 3490
    },
    {
      "epoch": 0.024146757090522743,
      "grad_norm": 4.5243656043858493e-10,
      "learning_rate": 9.759015364236583e-06,
      "loss": 0.0,
      "step": 3500
    },
    {
      "epoch": 0.024215747825067094,
      "grad_norm": 3.2983130040520336e-06,
      "learning_rate": 9.75832545689114e-06,
      "loss": 0.0001,
      "step": 3510
    },
    {
      "epoch": 0.024284738559611444,
      "grad_norm": 0.9399678707122803,
      "learning_rate": 9.757635549545696e-06,
      "loss": 0.0002,
      "step": 3520
    },
    {
      "epoch": 0.024353729294155795,
      "grad_norm": 2.515891901566647e-05,
      "learning_rate": 9.756945642200253e-06,
      "loss": 0.1279,
      "step": 3530
    },
    {
      "epoch": 0.024422720028700145,
      "grad_norm": 4.815369265998015e-06,
      "learning_rate": 9.75625573485481e-06,
      "loss": 0.0,
      "step": 3540
    },
    {
      "epoch": 0.024491710763244495,
      "grad_norm": 0.0,
      "learning_rate": 9.755565827509368e-06,
      "loss": 0.0015,
      "step": 3550
    },
    {
      "epoch": 0.024560701497788846,
      "grad_norm": 8.866207212854249e-10,
      "learning_rate": 9.754875920163922e-06,
      "loss": 0.0,
      "step": 3560
    },
    {
      "epoch": 0.024629692232333196,
      "grad_norm": 8.132279617711902e-06,
      "learning_rate": 9.754186012818479e-06,
      "loss": 0.001,
      "step": 3570
    },
    {
      "epoch": 0.02469868296687755,
      "grad_norm": 1.1343439837219194e-05,
      "learning_rate": 9.753496105473035e-06,
      "loss": 0.004,
      "step": 3580
    },
    {
      "epoch": 0.0247676737014219,
      "grad_norm": 2.4812104015836667e-07,
      "learning_rate": 9.752806198127592e-06,
      "loss": 0.0,
      "step": 3590
    },
    {
      "epoch": 0.02483666443596625,
      "grad_norm": 0.00022196717327460647,
      "learning_rate": 9.752116290782148e-06,
      "loss": 0.0001,
      "step": 3600
    },
    {
      "epoch": 0.0249056551705106,
      "grad_norm": 2.8421964088920504e-06,
      "learning_rate": 9.751426383436705e-06,
      "loss": 0.0007,
      "step": 3610
    },
    {
      "epoch": 0.02497464590505495,
      "grad_norm": 0.0,
      "learning_rate": 9.750736476091261e-06,
      "loss": 0.0001,
      "step": 3620
    },
    {
      "epoch": 0.0250436366395993,
      "grad_norm": 4.301780336390948e-06,
      "learning_rate": 9.750046568745818e-06,
      "loss": 0.0026,
      "step": 3630
    },
    {
      "epoch": 0.025112627374143652,
      "grad_norm": 6.158128144306829e-06,
      "learning_rate": 9.749356661400374e-06,
      "loss": 0.0003,
      "step": 3640
    },
    {
      "epoch": 0.025181618108688002,
      "grad_norm": 0.0,
      "learning_rate": 9.74866675405493e-06,
      "loss": 0.0,
      "step": 3650
    },
    {
      "epoch": 0.025250608843232353,
      "grad_norm": 748.8187255859375,
      "learning_rate": 9.747976846709489e-06,
      "loss": 0.334,
      "step": 3660
    },
    {
      "epoch": 0.025319599577776703,
      "grad_norm": 0.0,
      "learning_rate": 9.747286939364044e-06,
      "loss": 0.0,
      "step": 3670
    },
    {
      "epoch": 0.025388590312321057,
      "grad_norm": 0.16381755471229553,
      "learning_rate": 9.7465970320186e-06,
      "loss": 0.0078,
      "step": 3680
    },
    {
      "epoch": 0.025457581046865407,
      "grad_norm": 5.540079541788145e-07,
      "learning_rate": 9.745907124673157e-06,
      "loss": 0.0123,
      "step": 3690
    },
    {
      "epoch": 0.025526571781409758,
      "grad_norm": 0.00024552951799705625,
      "learning_rate": 9.745217217327713e-06,
      "loss": 0.0,
      "step": 3700
    },
    {
      "epoch": 0.025595562515954108,
      "grad_norm": 3.2780512810859364e-06,
      "learning_rate": 9.74452730998227e-06,
      "loss": 0.0101,
      "step": 3710
    },
    {
      "epoch": 0.02566455325049846,
      "grad_norm": 0.00011297911987639964,
      "learning_rate": 9.743837402636826e-06,
      "loss": 0.0,
      "step": 3720
    },
    {
      "epoch": 0.02573354398504281,
      "grad_norm": 1.1949259715038352e-05,
      "learning_rate": 9.743147495291383e-06,
      "loss": 0.7962,
      "step": 3730
    },
    {
      "epoch": 0.02580253471958716,
      "grad_norm": 2.413791833077994e-07,
      "learning_rate": 9.74245758794594e-06,
      "loss": 0.0002,
      "step": 3740
    },
    {
      "epoch": 0.02587152545413151,
      "grad_norm": 5.291036497823143e-10,
      "learning_rate": 9.741767680600496e-06,
      "loss": 0.0,
      "step": 3750
    },
    {
      "epoch": 0.02594051618867586,
      "grad_norm": 1.3636709539355252e-08,
      "learning_rate": 9.741077773255052e-06,
      "loss": 0.0002,
      "step": 3760
    },
    {
      "epoch": 0.02600950692322021,
      "grad_norm": 0.0,
      "learning_rate": 9.74038786590961e-06,
      "loss": 0.0,
      "step": 3770
    },
    {
      "epoch": 0.02607849765776456,
      "grad_norm": 0.0,
      "learning_rate": 9.739697958564167e-06,
      "loss": 0.0,
      "step": 3780
    },
    {
      "epoch": 0.026147488392308915,
      "grad_norm": 0.0,
      "learning_rate": 9.739008051218722e-06,
      "loss": 0.0,
      "step": 3790
    },
    {
      "epoch": 0.026216479126853265,
      "grad_norm": 0.0,
      "learning_rate": 9.738318143873278e-06,
      "loss": 0.2594,
      "step": 3800
    },
    {
      "epoch": 0.026285469861397615,
      "grad_norm": 1.0561984620238718e-09,
      "learning_rate": 9.737628236527835e-06,
      "loss": 0.0019,
      "step": 3810
    },
    {
      "epoch": 0.026354460595941966,
      "grad_norm": 0.0,
      "learning_rate": 9.736938329182391e-06,
      "loss": 0.0001,
      "step": 3820
    },
    {
      "epoch": 0.026423451330486316,
      "grad_norm": 0.0034547920804470778,
      "learning_rate": 9.736248421836948e-06,
      "loss": 0.0,
      "step": 3830
    },
    {
      "epoch": 0.026492442065030666,
      "grad_norm": 0.0,
      "learning_rate": 9.735558514491504e-06,
      "loss": 0.0001,
      "step": 3840
    },
    {
      "epoch": 0.026561432799575017,
      "grad_norm": 0.0,
      "learning_rate": 9.73486860714606e-06,
      "loss": 0.0,
      "step": 3850
    },
    {
      "epoch": 0.026630423534119367,
      "grad_norm": 0.0,
      "learning_rate": 9.734178699800617e-06,
      "loss": 0.0,
      "step": 3860
    },
    {
      "epoch": 0.026699414268663717,
      "grad_norm": 0.0,
      "learning_rate": 9.733488792455174e-06,
      "loss": 0.0092,
      "step": 3870
    },
    {
      "epoch": 0.026768405003208068,
      "grad_norm": 0.00023974105715751648,
      "learning_rate": 9.732798885109732e-06,
      "loss": 0.0002,
      "step": 3880
    },
    {
      "epoch": 0.026837395737752418,
      "grad_norm": 0.0,
      "learning_rate": 9.732108977764288e-06,
      "loss": 0.0001,
      "step": 3890
    },
    {
      "epoch": 0.026906386472296772,
      "grad_norm": 0.0,
      "learning_rate": 9.731419070418843e-06,
      "loss": 0.0,
      "step": 3900
    },
    {
      "epoch": 0.026975377206841122,
      "grad_norm": 0.0,
      "learning_rate": 9.7307291630734e-06,
      "loss": 0.0061,
      "step": 3910
    },
    {
      "epoch": 0.027044367941385473,
      "grad_norm": 0.0,
      "learning_rate": 9.730039255727956e-06,
      "loss": 0.0,
      "step": 3920
    },
    {
      "epoch": 0.027113358675929823,
      "grad_norm": 0.0,
      "learning_rate": 9.729349348382512e-06,
      "loss": 0.0221,
      "step": 3930
    },
    {
      "epoch": 0.027182349410474173,
      "grad_norm": 0.007771818898618221,
      "learning_rate": 9.728659441037069e-06,
      "loss": 0.0,
      "step": 3940
    },
    {
      "epoch": 0.027251340145018524,
      "grad_norm": 2.0619160068235942e-07,
      "learning_rate": 9.727969533691625e-06,
      "loss": 0.0007,
      "step": 3950
    },
    {
      "epoch": 0.027320330879562874,
      "grad_norm": 7.617491974087898e-06,
      "learning_rate": 9.727279626346182e-06,
      "loss": 0.0,
      "step": 3960
    },
    {
      "epoch": 0.027389321614107225,
      "grad_norm": 1.6165176930371672e-07,
      "learning_rate": 9.726589719000738e-06,
      "loss": 0.0806,
      "step": 3970
    },
    {
      "epoch": 0.027458312348651575,
      "grad_norm": 0.0,
      "learning_rate": 9.725899811655295e-06,
      "loss": 0.0,
      "step": 3980
    },
    {
      "epoch": 0.027527303083195925,
      "grad_norm": 0.0,
      "learning_rate": 9.725209904309853e-06,
      "loss": 0.0,
      "step": 3990
    },
    {
      "epoch": 0.02759629381774028,
      "grad_norm": 0.00930849090218544,
      "learning_rate": 9.72451999696441e-06,
      "loss": 0.0,
      "step": 4000
    },
    {
      "epoch": 0.02766528455228463,
      "grad_norm": 0.0,
      "learning_rate": 9.723830089618964e-06,
      "loss": 0.0002,
      "step": 4010
    },
    {
      "epoch": 0.02773427528682898,
      "grad_norm": 0.04425029456615448,
      "learning_rate": 9.723140182273521e-06,
      "loss": 0.0001,
      "step": 4020
    },
    {
      "epoch": 0.02780326602137333,
      "grad_norm": 0.0,
      "learning_rate": 9.722450274928077e-06,
      "loss": 0.0,
      "step": 4030
    },
    {
      "epoch": 0.02787225675591768,
      "grad_norm": 8.142186743498314e-06,
      "learning_rate": 9.721760367582634e-06,
      "loss": 0.0001,
      "step": 4040
    },
    {
      "epoch": 0.02794124749046203,
      "grad_norm": 4.0958007048175205e-06,
      "learning_rate": 9.72107046023719e-06,
      "loss": 0.0003,
      "step": 4050
    },
    {
      "epoch": 0.02801023822500638,
      "grad_norm": 0.0,
      "learning_rate": 9.720380552891747e-06,
      "loss": 0.0005,
      "step": 4060
    },
    {
      "epoch": 0.028079228959550732,
      "grad_norm": 0.0,
      "learning_rate": 9.719690645546303e-06,
      "loss": 0.0,
      "step": 4070
    },
    {
      "epoch": 0.028148219694095082,
      "grad_norm": 0.0,
      "learning_rate": 9.71900073820086e-06,
      "loss": 0.0018,
      "step": 4080
    },
    {
      "epoch": 0.028217210428639432,
      "grad_norm": 2.0809545198119395e-09,
      "learning_rate": 9.718310830855416e-06,
      "loss": 0.0,
      "step": 4090
    },
    {
      "epoch": 0.028286201163183783,
      "grad_norm": 0.0,
      "learning_rate": 9.717620923509975e-06,
      "loss": 0.0001,
      "step": 4100
    },
    {
      "epoch": 0.028355191897728137,
      "grad_norm": 0.0001445161906303838,
      "learning_rate": 9.716931016164531e-06,
      "loss": 0.468,
      "step": 4110
    },
    {
      "epoch": 0.028424182632272487,
      "grad_norm": 0.0,
      "learning_rate": 9.716241108819088e-06,
      "loss": 0.0005,
      "step": 4120
    },
    {
      "epoch": 0.028493173366816837,
      "grad_norm": 0.0,
      "learning_rate": 9.715551201473642e-06,
      "loss": 0.0001,
      "step": 4130
    },
    {
      "epoch": 0.028562164101361188,
      "grad_norm": 1.503429842841797e-07,
      "learning_rate": 9.714861294128199e-06,
      "loss": 0.0,
      "step": 4140
    },
    {
      "epoch": 0.028631154835905538,
      "grad_norm": 0.0,
      "learning_rate": 9.714171386782755e-06,
      "loss": 0.0,
      "step": 4150
    },
    {
      "epoch": 0.02870014557044989,
      "grad_norm": 0.0,
      "learning_rate": 9.713481479437312e-06,
      "loss": 0.0,
      "step": 4160
    },
    {
      "epoch": 0.02876913630499424,
      "grad_norm": 4.6401057995915096e-10,
      "learning_rate": 9.712791572091868e-06,
      "loss": 0.017,
      "step": 4170
    },
    {
      "epoch": 0.02883812703953859,
      "grad_norm": 0.001769772032275796,
      "learning_rate": 9.712101664746425e-06,
      "loss": 0.0,
      "step": 4180
    },
    {
      "epoch": 0.02890711777408294,
      "grad_norm": 0.0,
      "learning_rate": 9.711411757400981e-06,
      "loss": 0.0001,
      "step": 4190
    },
    {
      "epoch": 0.02897610850862729,
      "grad_norm": 1.992263378269854e-06,
      "learning_rate": 9.710721850055538e-06,
      "loss": 0.0,
      "step": 4200
    },
    {
      "epoch": 0.02904509924317164,
      "grad_norm": 0.0,
      "learning_rate": 9.710031942710096e-06,
      "loss": 0.0,
      "step": 4210
    },
    {
      "epoch": 0.029114089977715994,
      "grad_norm": 0.0,
      "learning_rate": 9.709342035364652e-06,
      "loss": 0.0339,
      "step": 4220
    },
    {
      "epoch": 0.029183080712260345,
      "grad_norm": 0.0,
      "learning_rate": 9.708652128019209e-06,
      "loss": 0.0,
      "step": 4230
    },
    {
      "epoch": 0.029252071446804695,
      "grad_norm": 0.0,
      "learning_rate": 9.707962220673764e-06,
      "loss": 0.0,
      "step": 4240
    },
    {
      "epoch": 0.029321062181349045,
      "grad_norm": 0.0,
      "learning_rate": 9.70727231332832e-06,
      "loss": 0.0,
      "step": 4250
    },
    {
      "epoch": 0.029390052915893396,
      "grad_norm": 0.0,
      "learning_rate": 9.706582405982877e-06,
      "loss": 0.0006,
      "step": 4260
    },
    {
      "epoch": 0.029459043650437746,
      "grad_norm": 0.0,
      "learning_rate": 9.705892498637433e-06,
      "loss": 0.0,
      "step": 4270
    },
    {
      "epoch": 0.029528034384982096,
      "grad_norm": 0.2454601377248764,
      "learning_rate": 9.70520259129199e-06,
      "loss": 0.0001,
      "step": 4280
    },
    {
      "epoch": 0.029597025119526447,
      "grad_norm": 0.0,
      "learning_rate": 9.704512683946546e-06,
      "loss": 0.0,
      "step": 4290
    },
    {
      "epoch": 0.029666015854070797,
      "grad_norm": 0.0,
      "learning_rate": 9.703822776601103e-06,
      "loss": 0.0,
      "step": 4300
    },
    {
      "epoch": 0.029735006588615148,
      "grad_norm": 0.0,
      "learning_rate": 9.703132869255659e-06,
      "loss": 0.0,
      "step": 4310
    },
    {
      "epoch": 0.0298039973231595,
      "grad_norm": 33.98332595825195,
      "learning_rate": 9.702442961910217e-06,
      "loss": 0.01,
      "step": 4320
    },
    {
      "epoch": 0.029872988057703852,
      "grad_norm": 0.00013368955114856362,
      "learning_rate": 9.701753054564774e-06,
      "loss": 0.0,
      "step": 4330
    },
    {
      "epoch": 0.029941978792248202,
      "grad_norm": 0.0,
      "learning_rate": 9.70106314721933e-06,
      "loss": 0.0,
      "step": 4340
    },
    {
      "epoch": 0.030010969526792552,
      "grad_norm": 0.0,
      "learning_rate": 9.700373239873885e-06,
      "loss": 0.1638,
      "step": 4350
    },
    {
      "epoch": 0.030079960261336903,
      "grad_norm": 0.0,
      "learning_rate": 9.699683332528442e-06,
      "loss": 0.0,
      "step": 4360
    },
    {
      "epoch": 0.030148950995881253,
      "grad_norm": 1.900823320966083e-07,
      "learning_rate": 9.698993425182998e-06,
      "loss": 0.1563,
      "step": 4370
    },
    {
      "epoch": 0.030217941730425604,
      "grad_norm": 13.705475807189941,
      "learning_rate": 9.698303517837555e-06,
      "loss": 0.0025,
      "step": 4380
    },
    {
      "epoch": 0.030286932464969954,
      "grad_norm": 0.0,
      "learning_rate": 9.697613610492111e-06,
      "loss": 0.0,
      "step": 4390
    },
    {
      "epoch": 0.030355923199514304,
      "grad_norm": 0.0,
      "learning_rate": 9.696923703146668e-06,
      "loss": 0.0,
      "step": 4400
    },
    {
      "epoch": 0.030424913934058655,
      "grad_norm": 0.0,
      "learning_rate": 9.696233795801224e-06,
      "loss": 0.0,
      "step": 4410
    },
    {
      "epoch": 0.030493904668603005,
      "grad_norm": 0.0,
      "learning_rate": 9.69554388845578e-06,
      "loss": 0.407,
      "step": 4420
    },
    {
      "epoch": 0.03056289540314736,
      "grad_norm": 0.0,
      "learning_rate": 9.694853981110339e-06,
      "loss": 0.0,
      "step": 4430
    },
    {
      "epoch": 0.03063188613769171,
      "grad_norm": 0.0,
      "learning_rate": 9.694164073764895e-06,
      "loss": 0.0011,
      "step": 4440
    },
    {
      "epoch": 0.03070087687223606,
      "grad_norm": 0.0,
      "learning_rate": 9.693474166419452e-06,
      "loss": 0.2846,
      "step": 4450
    },
    {
      "epoch": 0.03076986760678041,
      "grad_norm": 4.794936669938465e-10,
      "learning_rate": 9.692784259074008e-06,
      "loss": 0.0295,
      "step": 4460
    },
    {
      "epoch": 0.03083885834132476,
      "grad_norm": 0.0,
      "learning_rate": 9.692094351728563e-06,
      "loss": 0.6908,
      "step": 4470
    },
    {
      "epoch": 0.03090784907586911,
      "grad_norm": 0.0,
      "learning_rate": 9.69140444438312e-06,
      "loss": 0.0008,
      "step": 4480
    },
    {
      "epoch": 0.03097683981041346,
      "grad_norm": 0.0,
      "learning_rate": 9.690714537037676e-06,
      "loss": 0.0,
      "step": 4490
    },
    {
      "epoch": 0.03104583054495781,
      "grad_norm": 0.0,
      "learning_rate": 9.690024629692232e-06,
      "loss": 0.0002,
      "step": 4500
    },
    {
      "epoch": 0.031114821279502162,
      "grad_norm": 4.078829363152181e-08,
      "learning_rate": 9.689334722346789e-06,
      "loss": 0.0,
      "step": 4510
    },
    {
      "epoch": 0.031183812014046512,
      "grad_norm": 0.0,
      "learning_rate": 9.688644815001345e-06,
      "loss": 0.0613,
      "step": 4520
    },
    {
      "epoch": 0.031252802748590866,
      "grad_norm": 0.0,
      "learning_rate": 9.687954907655902e-06,
      "loss": 0.0013,
      "step": 4530
    },
    {
      "epoch": 0.03132179348313521,
      "grad_norm": 0.0,
      "learning_rate": 9.68726500031046e-06,
      "loss": 0.0,
      "step": 4540
    },
    {
      "epoch": 0.03139078421767957,
      "grad_norm": 0.002421376761049032,
      "learning_rate": 9.686575092965017e-06,
      "loss": 0.0,
      "step": 4550
    },
    {
      "epoch": 0.031459774952223914,
      "grad_norm": 0.0,
      "learning_rate": 9.685885185619573e-06,
      "loss": 0.0,
      "step": 4560
    },
    {
      "epoch": 0.03152876568676827,
      "grad_norm": 0.022848736494779587,
      "learning_rate": 9.68519527827413e-06,
      "loss": 0.0001,
      "step": 4570
    },
    {
      "epoch": 0.031597756421312614,
      "grad_norm": 0.0,
      "learning_rate": 9.684505370928684e-06,
      "loss": 0.0024,
      "step": 4580
    },
    {
      "epoch": 0.03166674715585697,
      "grad_norm": 0.0,
      "learning_rate": 9.683815463583241e-06,
      "loss": 0.0009,
      "step": 4590
    },
    {
      "epoch": 0.03173573789040132,
      "grad_norm": 1.9884506485823294e-08,
      "learning_rate": 9.683125556237797e-06,
      "loss": 0.0001,
      "step": 4600
    },
    {
      "epoch": 0.03180472862494567,
      "grad_norm": 0.0003966403310187161,
      "learning_rate": 9.682435648892354e-06,
      "loss": 0.0001,
      "step": 4610
    },
    {
      "epoch": 0.03187371935949002,
      "grad_norm": 0.0,
      "learning_rate": 9.68174574154691e-06,
      "loss": 0.0126,
      "step": 4620
    },
    {
      "epoch": 0.03194271009403437,
      "grad_norm": 5.939396032772493e-06,
      "learning_rate": 9.681055834201467e-06,
      "loss": 0.0012,
      "step": 4630
    },
    {
      "epoch": 0.032011700828578724,
      "grad_norm": 1.195463028125232e-06,
      "learning_rate": 9.680365926856023e-06,
      "loss": 0.016,
      "step": 4640
    },
    {
      "epoch": 0.03208069156312307,
      "grad_norm": 0.0,
      "learning_rate": 9.679676019510581e-06,
      "loss": 0.0031,
      "step": 4650
    },
    {
      "epoch": 0.032149682297667424,
      "grad_norm": 0.003359755966812372,
      "learning_rate": 9.678986112165138e-06,
      "loss": 0.0,
      "step": 4660
    },
    {
      "epoch": 0.03221867303221177,
      "grad_norm": 0.0,
      "learning_rate": 9.678296204819694e-06,
      "loss": 0.0018,
      "step": 4670
    },
    {
      "epoch": 0.032287663766756125,
      "grad_norm": 0.0,
      "learning_rate": 9.677606297474251e-06,
      "loss": 0.0,
      "step": 4680
    },
    {
      "epoch": 0.03235665450130047,
      "grad_norm": 1.3364478945732117e-05,
      "learning_rate": 9.676916390128806e-06,
      "loss": 0.0,
      "step": 4690
    },
    {
      "epoch": 0.032425645235844826,
      "grad_norm": 1.3066127735328337e-07,
      "learning_rate": 9.676226482783362e-06,
      "loss": 0.002,
      "step": 4700
    },
    {
      "epoch": 0.03249463597038918,
      "grad_norm": 0.0,
      "learning_rate": 9.675536575437919e-06,
      "loss": 0.0006,
      "step": 4710
    },
    {
      "epoch": 0.032563626704933527,
      "grad_norm": 0.0,
      "learning_rate": 9.674846668092475e-06,
      "loss": 0.0,
      "step": 4720
    },
    {
      "epoch": 0.03263261743947788,
      "grad_norm": 0.0,
      "learning_rate": 9.674156760747032e-06,
      "loss": 0.0,
      "step": 4730
    },
    {
      "epoch": 0.03270160817402223,
      "grad_norm": 0.0,
      "learning_rate": 9.673466853401588e-06,
      "loss": 0.0,
      "step": 4740
    },
    {
      "epoch": 0.03277059890856658,
      "grad_norm": 0.00019785226322710514,
      "learning_rate": 9.672776946056145e-06,
      "loss": 0.001,
      "step": 4750
    },
    {
      "epoch": 0.03283958964311093,
      "grad_norm": 0.0,
      "learning_rate": 9.672087038710703e-06,
      "loss": 0.0,
      "step": 4760
    },
    {
      "epoch": 0.03290858037765528,
      "grad_norm": 0.00433505279943347,
      "learning_rate": 9.67139713136526e-06,
      "loss": 0.0,
      "step": 4770
    },
    {
      "epoch": 0.03297757111219963,
      "grad_norm": 0.0,
      "learning_rate": 9.670707224019816e-06,
      "loss": 0.0,
      "step": 4780
    },
    {
      "epoch": 0.03304656184674398,
      "grad_norm": 0.0,
      "learning_rate": 9.670086307408917e-06,
      "loss": 0.2395,
      "step": 4790
    },
    {
      "epoch": 0.033115552581288336,
      "grad_norm": 0.0004065260582137853,
      "learning_rate": 9.669396400063471e-06,
      "loss": 0.0,
      "step": 4800
    },
    {
      "epoch": 0.03318454331583268,
      "grad_norm": 4.916876350513633e-10,
      "learning_rate": 9.668706492718028e-06,
      "loss": 0.0,
      "step": 4810
    },
    {
      "epoch": 0.03325353405037704,
      "grad_norm": 0.0,
      "learning_rate": 9.668016585372584e-06,
      "loss": 0.0,
      "step": 4820
    },
    {
      "epoch": 0.033322524784921384,
      "grad_norm": 0.015064802020788193,
      "learning_rate": 9.66732667802714e-06,
      "loss": 0.0005,
      "step": 4830
    },
    {
      "epoch": 0.03339151551946574,
      "grad_norm": 0.0,
      "learning_rate": 9.666636770681697e-06,
      "loss": 0.0,
      "step": 4840
    },
    {
      "epoch": 0.033460506254010085,
      "grad_norm": 0.0,
      "learning_rate": 9.665946863336255e-06,
      "loss": 0.0,
      "step": 4850
    },
    {
      "epoch": 0.03352949698855444,
      "grad_norm": 0.054494451731443405,
      "learning_rate": 9.665256955990812e-06,
      "loss": 0.0,
      "step": 4860
    },
    {
      "epoch": 0.033598487723098786,
      "grad_norm": 4.759086458250295e-10,
      "learning_rate": 9.664567048645368e-06,
      "loss": 0.0,
      "step": 4870
    },
    {
      "epoch": 0.03366747845764314,
      "grad_norm": 0.5415263175964355,
      "learning_rate": 9.663877141299925e-06,
      "loss": 0.0028,
      "step": 4880
    },
    {
      "epoch": 0.033736469192187486,
      "grad_norm": 0.0,
      "learning_rate": 9.663187233954481e-06,
      "loss": 0.0,
      "step": 4890
    },
    {
      "epoch": 0.03380545992673184,
      "grad_norm": 0.0,
      "learning_rate": 9.662497326609038e-06,
      "loss": 0.0,
      "step": 4900
    },
    {
      "epoch": 0.033874450661276194,
      "grad_norm": 2.639603614807129,
      "learning_rate": 9.661807419263593e-06,
      "loss": 0.0007,
      "step": 4910
    },
    {
      "epoch": 0.03394344139582054,
      "grad_norm": 0.0,
      "learning_rate": 9.66111751191815e-06,
      "loss": 0.0,
      "step": 4920
    },
    {
      "epoch": 0.034012432130364895,
      "grad_norm": 0.0,
      "learning_rate": 9.660427604572706e-06,
      "loss": 0.0,
      "step": 4930
    },
    {
      "epoch": 0.03408142286490924,
      "grad_norm": 8.912328720092773,
      "learning_rate": 9.659737697227262e-06,
      "loss": 0.0065,
      "step": 4940
    },
    {
      "epoch": 0.034150413599453595,
      "grad_norm": 0.011088664643466473,
      "learning_rate": 9.659047789881819e-06,
      "loss": 0.0,
      "step": 4950
    },
    {
      "epoch": 0.03421940433399794,
      "grad_norm": 0.0,
      "learning_rate": 9.658357882536377e-06,
      "loss": 0.0011,
      "step": 4960
    },
    {
      "epoch": 0.034288395068542296,
      "grad_norm": 0.0,
      "learning_rate": 9.657667975190933e-06,
      "loss": 0.0,
      "step": 4970
    },
    {
      "epoch": 0.03435738580308664,
      "grad_norm": 0.0,
      "learning_rate": 9.65697806784549e-06,
      "loss": 0.0007,
      "step": 4980
    },
    {
      "epoch": 0.034426376537631,
      "grad_norm": 0.0,
      "learning_rate": 9.656288160500046e-06,
      "loss": 0.0,
      "step": 4990
    },
    {
      "epoch": 0.034495367272175344,
      "grad_norm": 0.0,
      "learning_rate": 9.655598253154603e-06,
      "loss": 0.0,
      "step": 5000
    },
    {
      "epoch": 0.0345643580067197,
      "grad_norm": 0.0,
      "learning_rate": 9.65490834580916e-06,
      "loss": 0.0,
      "step": 5010
    },
    {
      "epoch": 0.03463334874126405,
      "grad_norm": 0.0032571053598076105,
      "learning_rate": 9.654218438463716e-06,
      "loss": 0.0,
      "step": 5020
    },
    {
      "epoch": 0.0347023394758084,
      "grad_norm": 0.0,
      "learning_rate": 9.65352853111827e-06,
      "loss": 0.0,
      "step": 5030
    },
    {
      "epoch": 0.03477133021035275,
      "grad_norm": 0.0,
      "learning_rate": 9.652838623772827e-06,
      "loss": 0.0,
      "step": 5040
    },
    {
      "epoch": 0.0348403209448971,
      "grad_norm": 0.0,
      "learning_rate": 9.652148716427384e-06,
      "loss": 0.0001,
      "step": 5050
    },
    {
      "epoch": 0.03490931167944145,
      "grad_norm": 0.0,
      "learning_rate": 9.65145880908194e-06,
      "loss": 0.0,
      "step": 5060
    },
    {
      "epoch": 0.0349783024139858,
      "grad_norm": 8.928826567000669e-10,
      "learning_rate": 9.650768901736498e-06,
      "loss": 0.0,
      "step": 5070
    },
    {
      "epoch": 0.035047293148530154,
      "grad_norm": 0.0,
      "learning_rate": 9.650078994391055e-06,
      "loss": 0.0,
      "step": 5080
    },
    {
      "epoch": 0.0351162838830745,
      "grad_norm": 0.0,
      "learning_rate": 9.649389087045611e-06,
      "loss": 0.0,
      "step": 5090
    },
    {
      "epoch": 0.035185274617618854,
      "grad_norm": 0.0,
      "learning_rate": 9.648699179700168e-06,
      "loss": 0.0,
      "step": 5100
    },
    {
      "epoch": 0.0352542653521632,
      "grad_norm": 7.675723552703857,
      "learning_rate": 9.648009272354724e-06,
      "loss": 0.0656,
      "step": 5110
    },
    {
      "epoch": 0.035323256086707555,
      "grad_norm": 0.0,
      "learning_rate": 9.64731936500928e-06,
      "loss": 0.0,
      "step": 5120
    },
    {
      "epoch": 0.03539224682125191,
      "grad_norm": 1.144585371017456,
      "learning_rate": 9.646629457663837e-06,
      "loss": 0.3927,
      "step": 5130
    },
    {
      "epoch": 0.035461237555796256,
      "grad_norm": 4.366060935012683e-08,
      "learning_rate": 9.645939550318392e-06,
      "loss": 0.0,
      "step": 5140
    },
    {
      "epoch": 0.03553022829034061,
      "grad_norm": 0.0,
      "learning_rate": 9.645249642972948e-06,
      "loss": 0.0,
      "step": 5150
    },
    {
      "epoch": 0.03559921902488496,
      "grad_norm": 0.00010898914479184896,
      "learning_rate": 9.644559735627505e-06,
      "loss": 0.0023,
      "step": 5160
    },
    {
      "epoch": 0.03566820975942931,
      "grad_norm": 7.119273170985707e-09,
      "learning_rate": 9.643869828282061e-06,
      "loss": 0.0,
      "step": 5170
    },
    {
      "epoch": 0.03573720049397366,
      "grad_norm": 0.0,
      "learning_rate": 9.64317992093662e-06,
      "loss": 0.0001,
      "step": 5180
    },
    {
      "epoch": 0.03580619122851801,
      "grad_norm": 0.0,
      "learning_rate": 9.642490013591176e-06,
      "loss": 0.1288,
      "step": 5190
    },
    {
      "epoch": 0.03587518196306236,
      "grad_norm": 0.0,
      "learning_rate": 9.641800106245733e-06,
      "loss": 0.1805,
      "step": 5200
    },
    {
      "epoch": 0.03594417269760671,
      "grad_norm": 8.536252607882489e-06,
      "learning_rate": 9.641110198900289e-06,
      "loss": 0.0,
      "step": 5210
    },
    {
      "epoch": 0.03601316343215106,
      "grad_norm": 0.0,
      "learning_rate": 9.640420291554846e-06,
      "loss": 0.0005,
      "step": 5220
    },
    {
      "epoch": 0.03608215416669541,
      "grad_norm": 0.008400880731642246,
      "learning_rate": 9.639730384209402e-06,
      "loss": 0.0,
      "step": 5230
    },
    {
      "epoch": 0.036151144901239766,
      "grad_norm": 0.0,
      "learning_rate": 9.639040476863959e-06,
      "loss": 0.3464,
      "step": 5240
    },
    {
      "epoch": 0.03622013563578411,
      "grad_norm": 0.0,
      "learning_rate": 9.638350569518513e-06,
      "loss": 0.011,
      "step": 5250
    },
    {
      "epoch": 0.03628912637032847,
      "grad_norm": 6.390629181396434e-08,
      "learning_rate": 9.63766066217307e-06,
      "loss": 0.0,
      "step": 5260
    },
    {
      "epoch": 0.036358117104872814,
      "grad_norm": 0.0,
      "learning_rate": 9.636970754827626e-06,
      "loss": 0.0,
      "step": 5270
    },
    {
      "epoch": 0.03642710783941717,
      "grad_norm": 0.0009792015189304948,
      "learning_rate": 9.636280847482183e-06,
      "loss": 0.0,
      "step": 5280
    },
    {
      "epoch": 0.036496098573961515,
      "grad_norm": 0.0,
      "learning_rate": 9.635590940136741e-06,
      "loss": 0.0,
      "step": 5290
    },
    {
      "epoch": 0.03656508930850587,
      "grad_norm": 0.0,
      "learning_rate": 9.634901032791298e-06,
      "loss": 0.0001,
      "step": 5300
    },
    {
      "epoch": 0.036634080043050216,
      "grad_norm": 2.826805484801298e-06,
      "learning_rate": 9.634211125445854e-06,
      "loss": 0.0,
      "step": 5310
    },
    {
      "epoch": 0.03670307077759457,
      "grad_norm": 14.800854682922363,
      "learning_rate": 9.63352121810041e-06,
      "loss": 0.0052,
      "step": 5320
    },
    {
      "epoch": 0.036772061512138916,
      "grad_norm": 1.2943366527906619e-05,
      "learning_rate": 9.632831310754967e-06,
      "loss": 0.1085,
      "step": 5330
    },
    {
      "epoch": 0.03684105224668327,
      "grad_norm": 0.0,
      "learning_rate": 9.632141403409524e-06,
      "loss": 0.0044,
      "step": 5340
    },
    {
      "epoch": 0.036910042981227624,
      "grad_norm": 0.0,
      "learning_rate": 9.63145149606408e-06,
      "loss": 0.0005,
      "step": 5350
    },
    {
      "epoch": 0.03697903371577197,
      "grad_norm": 0.0,
      "learning_rate": 9.630761588718636e-06,
      "loss": 0.0002,
      "step": 5360
    },
    {
      "epoch": 0.037048024450316325,
      "grad_norm": 0.6750802397727966,
      "learning_rate": 9.630071681373191e-06,
      "loss": 0.0002,
      "step": 5370
    },
    {
      "epoch": 0.03711701518486067,
      "grad_norm": 0.0,
      "learning_rate": 9.629381774027748e-06,
      "loss": 0.0002,
      "step": 5380
    },
    {
      "epoch": 0.037186005919405025,
      "grad_norm": 0.0,
      "learning_rate": 9.628691866682304e-06,
      "loss": 0.0,
      "step": 5390
    },
    {
      "epoch": 0.03725499665394937,
      "grad_norm": 0.0,
      "learning_rate": 9.628001959336862e-06,
      "loss": 0.0,
      "step": 5400
    },
    {
      "epoch": 0.037323987388493726,
      "grad_norm": 0.0,
      "learning_rate": 9.627312051991419e-06,
      "loss": 0.0,
      "step": 5410
    },
    {
      "epoch": 0.03739297812303807,
      "grad_norm": 1.7866976520508615e-07,
      "learning_rate": 9.626622144645975e-06,
      "loss": 0.0,
      "step": 5420
    },
    {
      "epoch": 0.03746196885758243,
      "grad_norm": 0.0,
      "learning_rate": 9.625932237300532e-06,
      "loss": 0.0,
      "step": 5430
    },
    {
      "epoch": 0.03753095959212678,
      "grad_norm": 4.721626423176417e-10,
      "learning_rate": 9.625242329955088e-06,
      "loss": 0.0001,
      "step": 5440
    },
    {
      "epoch": 0.03759995032667113,
      "grad_norm": 0.0,
      "learning_rate": 9.624552422609645e-06,
      "loss": 0.0,
      "step": 5450
    },
    {
      "epoch": 0.03766894106121548,
      "grad_norm": 4.6970615130703663e-07,
      "learning_rate": 9.623862515264201e-06,
      "loss": 0.0,
      "step": 5460
    },
    {
      "epoch": 0.03773793179575983,
      "grad_norm": 2.382010535484369e-07,
      "learning_rate": 9.623172607918758e-06,
      "loss": 0.0,
      "step": 5470
    },
    {
      "epoch": 0.03780692253030418,
      "grad_norm": 5.510453759960399e-10,
      "learning_rate": 9.622482700573313e-06,
      "loss": 0.0002,
      "step": 5480
    },
    {
      "epoch": 0.03787591326484853,
      "grad_norm": 5.554588794708252,
      "learning_rate": 9.62179279322787e-06,
      "loss": 0.0557,
      "step": 5490
    },
    {
      "epoch": 0.03794490399939288,
      "grad_norm": 7.04203102941392e-07,
      "learning_rate": 9.621102885882426e-06,
      "loss": 0.0,
      "step": 5500
    },
    {
      "epoch": 0.03801389473393723,
      "grad_norm": 1.1076947450637817,
      "learning_rate": 9.620412978536984e-06,
      "loss": 0.0001,
      "step": 5510
    },
    {
      "epoch": 0.038082885468481584,
      "grad_norm": 5.152320682100253e-06,
      "learning_rate": 9.61972307119154e-06,
      "loss": 0.0,
      "step": 5520
    },
    {
      "epoch": 0.03815187620302593,
      "grad_norm": 0.05701514706015587,
      "learning_rate": 9.619033163846097e-06,
      "loss": 0.0,
      "step": 5530
    },
    {
      "epoch": 0.038220866937570284,
      "grad_norm": 0.0,
      "learning_rate": 9.618343256500653e-06,
      "loss": 0.0,
      "step": 5540
    },
    {
      "epoch": 0.03828985767211464,
      "grad_norm": 4.2681944090183777e-10,
      "learning_rate": 9.61765334915521e-06,
      "loss": 0.0004,
      "step": 5550
    },
    {
      "epoch": 0.038358848406658985,
      "grad_norm": 0.0,
      "learning_rate": 9.616963441809766e-06,
      "loss": 0.0051,
      "step": 5560
    },
    {
      "epoch": 0.03842783914120334,
      "grad_norm": 0.0,
      "learning_rate": 9.616273534464323e-06,
      "loss": 0.0001,
      "step": 5570
    },
    {
      "epoch": 0.038496829875747686,
      "grad_norm": 0.0,
      "learning_rate": 9.61558362711888e-06,
      "loss": 0.0,
      "step": 5580
    },
    {
      "epoch": 0.03856582061029204,
      "grad_norm": 0.005056244321167469,
      "learning_rate": 9.614893719773434e-06,
      "loss": 0.0,
      "step": 5590
    },
    {
      "epoch": 0.03863481134483639,
      "grad_norm": 0.0,
      "learning_rate": 9.61420381242799e-06,
      "loss": 0.0,
      "step": 5600
    },
    {
      "epoch": 0.03870380207938074,
      "grad_norm": 1.372959650325356e-05,
      "learning_rate": 9.613513905082547e-06,
      "loss": 0.0,
      "step": 5610
    },
    {
      "epoch": 0.03877279281392509,
      "grad_norm": 9.360952013537371e-10,
      "learning_rate": 9.612823997737105e-06,
      "loss": 0.0,
      "step": 5620
    },
    {
      "epoch": 0.03884178354846944,
      "grad_norm": 0.0,
      "learning_rate": 9.612134090391662e-06,
      "loss": 0.0025,
      "step": 5630
    },
    {
      "epoch": 0.03891077428301379,
      "grad_norm": 0.047646623104810715,
      "learning_rate": 9.611444183046218e-06,
      "loss": 0.0,
      "step": 5640
    },
    {
      "epoch": 0.03897976501755814,
      "grad_norm": 0.000264080852502957,
      "learning_rate": 9.610754275700775e-06,
      "loss": 0.0,
      "step": 5650
    },
    {
      "epoch": 0.039048755752102496,
      "grad_norm": 0.07369674742221832,
      "learning_rate": 9.610064368355331e-06,
      "loss": 0.0861,
      "step": 5660
    },
    {
      "epoch": 0.03911774648664684,
      "grad_norm": 114.97257995605469,
      "learning_rate": 9.609374461009888e-06,
      "loss": 0.1089,
      "step": 5670
    },
    {
      "epoch": 0.0391867372211912,
      "grad_norm": 0.0,
      "learning_rate": 9.608684553664444e-06,
      "loss": 0.0,
      "step": 5680
    },
    {
      "epoch": 0.03925572795573554,
      "grad_norm": 7.174975991119936e-08,
      "learning_rate": 9.607994646319e-06,
      "loss": 0.0,
      "step": 5690
    },
    {
      "epoch": 0.0393247186902799,
      "grad_norm": 0.0,
      "learning_rate": 9.607304738973555e-06,
      "loss": 0.0,
      "step": 5700
    },
    {
      "epoch": 0.039393709424824244,
      "grad_norm": 0.0019032327691093087,
      "learning_rate": 9.606614831628112e-06,
      "loss": 0.0,
      "step": 5710
    },
    {
      "epoch": 0.0394627001593686,
      "grad_norm": 9.654377297607653e-10,
      "learning_rate": 9.605924924282668e-06,
      "loss": 0.0,
      "step": 5720
    },
    {
      "epoch": 0.039531690893912945,
      "grad_norm": 0.0,
      "learning_rate": 9.605235016937227e-06,
      "loss": 0.0004,
      "step": 5730
    },
    {
      "epoch": 0.0396006816284573,
      "grad_norm": 1.1168336255806821e-09,
      "learning_rate": 9.604545109591783e-06,
      "loss": 0.0002,
      "step": 5740
    },
    {
      "epoch": 0.039669672363001646,
      "grad_norm": 0.0015918408753350377,
      "learning_rate": 9.60385520224634e-06,
      "loss": 0.0,
      "step": 5750
    },
    {
      "epoch": 0.039738663097546,
      "grad_norm": 8.74627148732543e-05,
      "learning_rate": 9.603165294900896e-06,
      "loss": 0.0143,
      "step": 5760
    },
    {
      "epoch": 0.03980765383209035,
      "grad_norm": 0.0065117026679217815,
      "learning_rate": 9.602475387555453e-06,
      "loss": 0.0,
      "step": 5770
    },
    {
      "epoch": 0.0398766445666347,
      "grad_norm": 0.0,
      "learning_rate": 9.601785480210009e-06,
      "loss": 0.0062,
      "step": 5780
    },
    {
      "epoch": 0.039945635301179054,
      "grad_norm": 0.0,
      "learning_rate": 9.601095572864566e-06,
      "loss": 0.0003,
      "step": 5790
    },
    {
      "epoch": 0.0400146260357234,
      "grad_norm": 0.0,
      "learning_rate": 9.600405665519122e-06,
      "loss": 0.0,
      "step": 5800
    },
    {
      "epoch": 0.040083616770267755,
      "grad_norm": 0.0,
      "learning_rate": 9.599715758173679e-06,
      "loss": 0.0,
      "step": 5810
    },
    {
      "epoch": 0.0401526075048121,
      "grad_norm": 0.00011242792243137956,
      "learning_rate": 9.599025850828233e-06,
      "loss": 0.0,
      "step": 5820
    },
    {
      "epoch": 0.040221598239356456,
      "grad_norm": 0.0,
      "learning_rate": 9.59833594348279e-06,
      "loss": 0.1506,
      "step": 5830
    },
    {
      "epoch": 0.0402905889739008,
      "grad_norm": 0.0,
      "learning_rate": 9.597646036137348e-06,
      "loss": 0.0,
      "step": 5840
    },
    {
      "epoch": 0.040359579708445156,
      "grad_norm": 0.012896775268018246,
      "learning_rate": 9.596956128791905e-06,
      "loss": 0.0,
      "step": 5850
    },
    {
      "epoch": 0.0404285704429895,
      "grad_norm": 9.144940360528153e-10,
      "learning_rate": 9.596266221446461e-06,
      "loss": 0.0,
      "step": 5860
    },
    {
      "epoch": 0.04049756117753386,
      "grad_norm": 0.0,
      "learning_rate": 9.595576314101017e-06,
      "loss": 0.002,
      "step": 5870
    },
    {
      "epoch": 0.04056655191207821,
      "grad_norm": 4.3111356151648295e-10,
      "learning_rate": 9.594886406755574e-06,
      "loss": 0.0277,
      "step": 5880
    },
    {
      "epoch": 0.04063554264662256,
      "grad_norm": 0.0,
      "learning_rate": 9.59419649941013e-06,
      "loss": 0.1753,
      "step": 5890
    },
    {
      "epoch": 0.04070453338116691,
      "grad_norm": 0.0,
      "learning_rate": 9.593506592064687e-06,
      "loss": 0.0001,
      "step": 5900
    },
    {
      "epoch": 0.04077352411571126,
      "grad_norm": 0.0,
      "learning_rate": 9.592816684719243e-06,
      "loss": 0.0,
      "step": 5910
    },
    {
      "epoch": 0.04084251485025561,
      "grad_norm": 1.0126439775604013e-07,
      "learning_rate": 9.5921267773738e-06,
      "loss": 0.0001,
      "step": 5920
    },
    {
      "epoch": 0.04091150558479996,
      "grad_norm": 0.0,
      "learning_rate": 9.591436870028355e-06,
      "loss": 0.0,
      "step": 5930
    },
    {
      "epoch": 0.04098049631934431,
      "grad_norm": 0.0,
      "learning_rate": 9.590746962682911e-06,
      "loss": 0.0001,
      "step": 5940
    },
    {
      "epoch": 0.04104948705388866,
      "grad_norm": 0.13898760080337524,
      "learning_rate": 9.59005705533747e-06,
      "loss": 0.0,
      "step": 5950
    },
    {
      "epoch": 0.041118477788433014,
      "grad_norm": 0.0,
      "learning_rate": 9.589367147992026e-06,
      "loss": 0.0002,
      "step": 5960
    },
    {
      "epoch": 0.04118746852297736,
      "grad_norm": 0.0,
      "learning_rate": 9.588677240646582e-06,
      "loss": 0.0,
      "step": 5970
    },
    {
      "epoch": 0.041256459257521715,
      "grad_norm": 0.0,
      "learning_rate": 9.587987333301139e-06,
      "loss": 0.0005,
      "step": 5980
    },
    {
      "epoch": 0.04132544999206607,
      "grad_norm": 0.0,
      "learning_rate": 9.587297425955695e-06,
      "loss": 0.0,
      "step": 5990
    },
    {
      "epoch": 0.041394440726610415,
      "grad_norm": 0.001535198651254177,
      "learning_rate": 9.586607518610252e-06,
      "loss": 1.9234,
      "step": 6000
    },
    {
      "epoch": 0.04146343146115477,
      "grad_norm": 0.0,
      "learning_rate": 9.585917611264808e-06,
      "loss": 0.0001,
      "step": 6010
    },
    {
      "epoch": 0.041532422195699116,
      "grad_norm": 1.0176873729506042e-05,
      "learning_rate": 9.585227703919365e-06,
      "loss": 0.0,
      "step": 6020
    },
    {
      "epoch": 0.04160141293024347,
      "grad_norm": 0.00411896500736475,
      "learning_rate": 9.584537796573921e-06,
      "loss": 0.0,
      "step": 6030
    },
    {
      "epoch": 0.04167040366478782,
      "grad_norm": 0.0,
      "learning_rate": 9.583847889228476e-06,
      "loss": 0.0,
      "step": 6040
    },
    {
      "epoch": 0.04173939439933217,
      "grad_norm": 1.2623466716377152e-08,
      "learning_rate": 9.583157981883033e-06,
      "loss": 0.0538,
      "step": 6050
    },
    {
      "epoch": 0.04180838513387652,
      "grad_norm": 2.3080203845893266e-06,
      "learning_rate": 9.58246807453759e-06,
      "loss": 0.0,
      "step": 6060
    },
    {
      "epoch": 0.04187737586842087,
      "grad_norm": 3.325064596992888e-07,
      "learning_rate": 9.581778167192147e-06,
      "loss": 0.0008,
      "step": 6070
    },
    {
      "epoch": 0.041946366602965225,
      "grad_norm": 0.0031659924425184727,
      "learning_rate": 9.581088259846704e-06,
      "loss": 0.0003,
      "step": 6080
    },
    {
      "epoch": 0.04201535733750957,
      "grad_norm": 0.0,
      "learning_rate": 9.58039835250126e-06,
      "loss": 0.0,
      "step": 6090
    },
    {
      "epoch": 0.042084348072053926,
      "grad_norm": 4.2391895549442893e-10,
      "learning_rate": 9.579708445155817e-06,
      "loss": 0.0,
      "step": 6100
    },
    {
      "epoch": 0.04215333880659827,
      "grad_norm": 0.0009250965085811913,
      "learning_rate": 9.579018537810373e-06,
      "loss": 0.0001,
      "step": 6110
    },
    {
      "epoch": 0.04222232954114263,
      "grad_norm": 8.939063933510738e-10,
      "learning_rate": 9.57832863046493e-06,
      "loss": 0.0,
      "step": 6120
    },
    {
      "epoch": 0.042291320275686974,
      "grad_norm": 5.28221335116541e-06,
      "learning_rate": 9.577638723119486e-06,
      "loss": 0.0693,
      "step": 6130
    },
    {
      "epoch": 0.04236031101023133,
      "grad_norm": 0.0,
      "learning_rate": 9.576948815774043e-06,
      "loss": 0.0,
      "step": 6140
    },
    {
      "epoch": 0.042429301744775674,
      "grad_norm": 0.0,
      "learning_rate": 9.5762589084286e-06,
      "loss": 0.0009,
      "step": 6150
    },
    {
      "epoch": 0.04249829247932003,
      "grad_norm": 0.0,
      "learning_rate": 9.575569001083154e-06,
      "loss": 0.0,
      "step": 6160
    },
    {
      "epoch": 0.042567283213864375,
      "grad_norm": 0.0,
      "learning_rate": 9.574879093737712e-06,
      "loss": 0.0011,
      "step": 6170
    },
    {
      "epoch": 0.04263627394840873,
      "grad_norm": 669.7105102539062,
      "learning_rate": 9.574189186392269e-06,
      "loss": 0.1937,
      "step": 6180
    },
    {
      "epoch": 0.04270526468295308,
      "grad_norm": 0.0,
      "learning_rate": 9.573499279046825e-06,
      "loss": 0.0,
      "step": 6190
    },
    {
      "epoch": 0.04277425541749743,
      "grad_norm": 9.243245585821569e-05,
      "learning_rate": 9.572809371701382e-06,
      "loss": 0.0,
      "step": 6200
    },
    {
      "epoch": 0.04284324615204178,
      "grad_norm": 159.6192169189453,
      "learning_rate": 9.572119464355938e-06,
      "loss": 0.0274,
      "step": 6210
    },
    {
      "epoch": 0.04291223688658613,
      "grad_norm": 0.0,
      "learning_rate": 9.571429557010495e-06,
      "loss": 0.0,
      "step": 6220
    },
    {
      "epoch": 0.042981227621130484,
      "grad_norm": 0.0,
      "learning_rate": 9.570739649665051e-06,
      "loss": 0.0,
      "step": 6230
    },
    {
      "epoch": 0.04305021835567483,
      "grad_norm": 0.0,
      "learning_rate": 9.570049742319608e-06,
      "loss": 0.0133,
      "step": 6240
    },
    {
      "epoch": 0.043119209090219185,
      "grad_norm": 0.01762031577527523,
      "learning_rate": 9.569359834974164e-06,
      "loss": 0.0,
      "step": 6250
    },
    {
      "epoch": 0.04318819982476353,
      "grad_norm": 0.0,
      "learning_rate": 9.56866992762872e-06,
      "loss": 0.0,
      "step": 6260
    },
    {
      "epoch": 0.043257190559307886,
      "grad_norm": 0.000838885607663542,
      "learning_rate": 9.567980020283275e-06,
      "loss": 0.0004,
      "step": 6270
    },
    {
      "epoch": 0.04332618129385223,
      "grad_norm": 0.0,
      "learning_rate": 9.567290112937834e-06,
      "loss": 0.0,
      "step": 6280
    },
    {
      "epoch": 0.043395172028396586,
      "grad_norm": 0.0,
      "learning_rate": 9.56660020559239e-06,
      "loss": 0.0,
      "step": 6290
    },
    {
      "epoch": 0.04346416276294094,
      "grad_norm": 0.0,
      "learning_rate": 9.565910298246947e-06,
      "loss": 0.0,
      "step": 6300
    },
    {
      "epoch": 0.04353315349748529,
      "grad_norm": 0.0,
      "learning_rate": 9.565220390901503e-06,
      "loss": 0.0297,
      "step": 6310
    },
    {
      "epoch": 0.04360214423202964,
      "grad_norm": 0.0,
      "learning_rate": 9.56453048355606e-06,
      "loss": 0.0,
      "step": 6320
    },
    {
      "epoch": 0.04367113496657399,
      "grad_norm": 0.0,
      "learning_rate": 9.563840576210616e-06,
      "loss": 0.0,
      "step": 6330
    },
    {
      "epoch": 0.04374012570111834,
      "grad_norm": 0.0,
      "learning_rate": 9.563150668865173e-06,
      "loss": 0.0015,
      "step": 6340
    },
    {
      "epoch": 0.04380911643566269,
      "grad_norm": 0.0,
      "learning_rate": 9.562460761519729e-06,
      "loss": 0.0,
      "step": 6350
    },
    {
      "epoch": 0.04387810717020704,
      "grad_norm": 0.0,
      "learning_rate": 9.561770854174286e-06,
      "loss": 0.0127,
      "step": 6360
    },
    {
      "epoch": 0.04394709790475139,
      "grad_norm": 1.602687370905187e-05,
      "learning_rate": 9.561080946828842e-06,
      "loss": 0.0072,
      "step": 6370
    },
    {
      "epoch": 0.04401608863929574,
      "grad_norm": 0.00013067858526483178,
      "learning_rate": 9.560391039483397e-06,
      "loss": 0.0254,
      "step": 6380
    },
    {
      "epoch": 0.04408507937384009,
      "grad_norm": 0.0,
      "learning_rate": 9.559701132137955e-06,
      "loss": 0.0,
      "step": 6390
    },
    {
      "epoch": 0.044154070108384444,
      "grad_norm": 0.0,
      "learning_rate": 9.559011224792511e-06,
      "loss": 0.0,
      "step": 6400
    },
    {
      "epoch": 0.0442230608429288,
      "grad_norm": 0.10909099131822586,
      "learning_rate": 9.558321317447068e-06,
      "loss": 0.0,
      "step": 6410
    },
    {
      "epoch": 0.044292051577473145,
      "grad_norm": 0.0,
      "learning_rate": 9.557631410101624e-06,
      "loss": 0.1475,
      "step": 6420
    },
    {
      "epoch": 0.0443610423120175,
      "grad_norm": 0.0,
      "learning_rate": 9.556941502756181e-06,
      "loss": 0.0,
      "step": 6430
    },
    {
      "epoch": 0.044430033046561845,
      "grad_norm": 0.0,
      "learning_rate": 9.556251595410737e-06,
      "loss": 0.0537,
      "step": 6440
    },
    {
      "epoch": 0.0444990237811062,
      "grad_norm": 0.0,
      "learning_rate": 9.555561688065294e-06,
      "loss": 0.0002,
      "step": 6450
    },
    {
      "epoch": 0.044568014515650546,
      "grad_norm": 0.0,
      "learning_rate": 9.55487178071985e-06,
      "loss": 0.0024,
      "step": 6460
    },
    {
      "epoch": 0.0446370052501949,
      "grad_norm": 0.0,
      "learning_rate": 9.554181873374407e-06,
      "loss": 0.0,
      "step": 6470
    },
    {
      "epoch": 0.04470599598473925,
      "grad_norm": 0.0,
      "learning_rate": 9.553491966028963e-06,
      "loss": 0.0002,
      "step": 6480
    },
    {
      "epoch": 0.0447749867192836,
      "grad_norm": 1.6138839721679688,
      "learning_rate": 9.55280205868352e-06,
      "loss": 0.0025,
      "step": 6490
    },
    {
      "epoch": 0.04484397745382795,
      "grad_norm": 9.541729628637086e-10,
      "learning_rate": 9.552112151338076e-06,
      "loss": 0.012,
      "step": 6500
    },
    {
      "epoch": 0.0449129681883723,
      "grad_norm": 0.0,
      "learning_rate": 9.551422243992633e-06,
      "loss": 0.0023,
      "step": 6510
    },
    {
      "epoch": 0.044981958922916655,
      "grad_norm": 5.9090962167829275e-06,
      "learning_rate": 9.55073233664719e-06,
      "loss": 0.0123,
      "step": 6520
    },
    {
      "epoch": 0.045050949657461,
      "grad_norm": 0.0,
      "learning_rate": 9.550042429301746e-06,
      "loss": 0.0001,
      "step": 6530
    },
    {
      "epoch": 0.045119940392005356,
      "grad_norm": 0.0,
      "learning_rate": 9.549352521956302e-06,
      "loss": 0.0,
      "step": 6540
    },
    {
      "epoch": 0.0451889311265497,
      "grad_norm": 0.0,
      "learning_rate": 9.548662614610859e-06,
      "loss": 0.0,
      "step": 6550
    },
    {
      "epoch": 0.04525792186109406,
      "grad_norm": 0.0,
      "learning_rate": 9.547972707265415e-06,
      "loss": 0.0,
      "step": 6560
    },
    {
      "epoch": 0.045326912595638404,
      "grad_norm": 0.0,
      "learning_rate": 9.547282799919972e-06,
      "loss": 0.0,
      "step": 6570
    },
    {
      "epoch": 0.04539590333018276,
      "grad_norm": 0.0,
      "learning_rate": 9.546592892574528e-06,
      "loss": 0.0008,
      "step": 6580
    },
    {
      "epoch": 0.045464894064727104,
      "grad_norm": 0.029501691460609436,
      "learning_rate": 9.545902985229085e-06,
      "loss": 0.0,
      "step": 6590
    },
    {
      "epoch": 0.04553388479927146,
      "grad_norm": 0.28499457240104675,
      "learning_rate": 9.545213077883641e-06,
      "loss": 0.0001,
      "step": 6600
    },
    {
      "epoch": 0.045602875533815805,
      "grad_norm": 0.0,
      "learning_rate": 9.544523170538198e-06,
      "loss": 0.0,
      "step": 6610
    },
    {
      "epoch": 0.04567186626836016,
      "grad_norm": 0.0,
      "learning_rate": 9.543833263192754e-06,
      "loss": 0.0392,
      "step": 6620
    },
    {
      "epoch": 0.04574085700290451,
      "grad_norm": 4.873459968912641e-10,
      "learning_rate": 9.54314335584731e-06,
      "loss": 0.0,
      "step": 6630
    },
    {
      "epoch": 0.04580984773744886,
      "grad_norm": 0.0,
      "learning_rate": 9.542453448501867e-06,
      "loss": 0.0,
      "step": 6640
    },
    {
      "epoch": 0.045878838471993214,
      "grad_norm": 1.0703723773985985e-06,
      "learning_rate": 9.541763541156424e-06,
      "loss": 0.0,
      "step": 6650
    },
    {
      "epoch": 0.04594782920653756,
      "grad_norm": 1.568539664731361e-05,
      "learning_rate": 9.54107363381098e-06,
      "loss": 0.0,
      "step": 6660
    },
    {
      "epoch": 0.046016819941081914,
      "grad_norm": 1.4101584344672347e-09,
      "learning_rate": 9.540383726465537e-06,
      "loss": 0.0,
      "step": 6670
    },
    {
      "epoch": 0.04608581067562626,
      "grad_norm": 0.0,
      "learning_rate": 9.539693819120093e-06,
      "loss": 0.0,
      "step": 6680
    },
    {
      "epoch": 0.046154801410170615,
      "grad_norm": 0.00038418613257817924,
      "learning_rate": 9.53900391177465e-06,
      "loss": 0.0,
      "step": 6690
    },
    {
      "epoch": 0.04622379214471496,
      "grad_norm": 0.0,
      "learning_rate": 9.538314004429206e-06,
      "loss": 0.1913,
      "step": 6700
    },
    {
      "epoch": 0.046292782879259316,
      "grad_norm": 0.0,
      "learning_rate": 9.537624097083763e-06,
      "loss": 0.0,
      "step": 6710
    },
    {
      "epoch": 0.04636177361380367,
      "grad_norm": 0.0,
      "learning_rate": 9.53693418973832e-06,
      "loss": 0.0,
      "step": 6720
    },
    {
      "epoch": 0.046430764348348016,
      "grad_norm": 0.0,
      "learning_rate": 9.536244282392876e-06,
      "loss": 0.0,
      "step": 6730
    },
    {
      "epoch": 0.04649975508289237,
      "grad_norm": 0.0,
      "learning_rate": 9.535554375047432e-06,
      "loss": 0.0,
      "step": 6740
    },
    {
      "epoch": 0.04656874581743672,
      "grad_norm": 0.0,
      "learning_rate": 9.534864467701989e-06,
      "loss": 0.0,
      "step": 6750
    },
    {
      "epoch": 0.04663773655198107,
      "grad_norm": 0.0,
      "learning_rate": 9.534174560356545e-06,
      "loss": 0.0089,
      "step": 6760
    },
    {
      "epoch": 0.04670672728652542,
      "grad_norm": 4.657961483189865e-07,
      "learning_rate": 9.533484653011102e-06,
      "loss": 0.0133,
      "step": 6770
    },
    {
      "epoch": 0.04677571802106977,
      "grad_norm": 0.0,
      "learning_rate": 9.532794745665658e-06,
      "loss": 0.0,
      "step": 6780
    },
    {
      "epoch": 0.04684470875561412,
      "grad_norm": 0.0,
      "learning_rate": 9.532104838320215e-06,
      "loss": 0.0,
      "step": 6790
    },
    {
      "epoch": 0.04691369949015847,
      "grad_norm": 0.0,
      "learning_rate": 9.531414930974771e-06,
      "loss": 0.0,
      "step": 6800
    },
    {
      "epoch": 0.04698269022470282,
      "grad_norm": 0.001140069798566401,
      "learning_rate": 9.530725023629328e-06,
      "loss": 0.0248,
      "step": 6810
    },
    {
      "epoch": 0.04705168095924717,
      "grad_norm": 2.2088217366267315e-10,
      "learning_rate": 9.530035116283884e-06,
      "loss": 0.0,
      "step": 6820
    },
    {
      "epoch": 0.04712067169379153,
      "grad_norm": 1.510604079157929e-06,
      "learning_rate": 9.52934520893844e-06,
      "loss": 0.0,
      "step": 6830
    },
    {
      "epoch": 0.047189662428335874,
      "grad_norm": 0.0,
      "learning_rate": 9.528655301592997e-06,
      "loss": 0.0,
      "step": 6840
    },
    {
      "epoch": 0.04725865316288023,
      "grad_norm": 0.0,
      "learning_rate": 9.527965394247554e-06,
      "loss": 0.0,
      "step": 6850
    },
    {
      "epoch": 0.047327643897424575,
      "grad_norm": 0.0,
      "learning_rate": 9.52727548690211e-06,
      "loss": 0.0006,
      "step": 6860
    },
    {
      "epoch": 0.04739663463196893,
      "grad_norm": 0.0,
      "learning_rate": 9.526585579556667e-06,
      "loss": 0.0,
      "step": 6870
    },
    {
      "epoch": 0.047465625366513275,
      "grad_norm": 0.0,
      "learning_rate": 9.525895672211223e-06,
      "loss": 0.0,
      "step": 6880
    },
    {
      "epoch": 0.04753461610105763,
      "grad_norm": 0.0,
      "learning_rate": 9.52520576486578e-06,
      "loss": 0.0,
      "step": 6890
    },
    {
      "epoch": 0.047603606835601976,
      "grad_norm": 0.0,
      "learning_rate": 9.524515857520336e-06,
      "loss": 0.0001,
      "step": 6900
    },
    {
      "epoch": 0.04767259757014633,
      "grad_norm": 5.076705988926733e-08,
      "learning_rate": 9.523825950174892e-06,
      "loss": 0.0,
      "step": 6910
    },
    {
      "epoch": 0.04774158830469068,
      "grad_norm": 0.0,
      "learning_rate": 9.523136042829449e-06,
      "loss": 0.0,
      "step": 6920
    },
    {
      "epoch": 0.04781057903923503,
      "grad_norm": 2.04520400615138e-08,
      "learning_rate": 9.522446135484005e-06,
      "loss": 0.0,
      "step": 6930
    },
    {
      "epoch": 0.047879569773779385,
      "grad_norm": 0.0,
      "learning_rate": 9.521756228138562e-06,
      "loss": 0.0,
      "step": 6940
    },
    {
      "epoch": 0.04794856050832373,
      "grad_norm": 0.0002855256025213748,
      "learning_rate": 9.521066320793118e-06,
      "loss": 0.0089,
      "step": 6950
    },
    {
      "epoch": 0.048017551242868085,
      "grad_norm": 0.0,
      "learning_rate": 9.520376413447675e-06,
      "loss": 0.0,
      "step": 6960
    },
    {
      "epoch": 0.04808654197741243,
      "grad_norm": 0.0,
      "learning_rate": 9.519686506102231e-06,
      "loss": 0.0,
      "step": 6970
    },
    {
      "epoch": 0.048155532711956786,
      "grad_norm": 0.003345920704305172,
      "learning_rate": 9.518996598756788e-06,
      "loss": 0.0,
      "step": 6980
    },
    {
      "epoch": 0.04822452344650113,
      "grad_norm": 0.0,
      "learning_rate": 9.518306691411344e-06,
      "loss": 0.0,
      "step": 6990
    },
    {
      "epoch": 0.04829351418104549,
      "grad_norm": 0.0,
      "learning_rate": 9.517616784065901e-06,
      "loss": 0.0,
      "step": 7000
    },
    {
      "epoch": 0.048362504915589834,
      "grad_norm": 8.28308998279681e-07,
      "learning_rate": 9.516926876720457e-06,
      "loss": 0.0,
      "step": 7010
    },
    {
      "epoch": 0.04843149565013419,
      "grad_norm": 0.0,
      "learning_rate": 9.516236969375014e-06,
      "loss": 0.0,
      "step": 7020
    },
    {
      "epoch": 0.048500486384678534,
      "grad_norm": 0.0,
      "learning_rate": 9.51554706202957e-06,
      "loss": 0.0,
      "step": 7030
    },
    {
      "epoch": 0.04856947711922289,
      "grad_norm": 0.0,
      "learning_rate": 9.514857154684127e-06,
      "loss": 0.0,
      "step": 7040
    },
    {
      "epoch": 0.04863846785376724,
      "grad_norm": 1.907149282942555e-08,
      "learning_rate": 9.514167247338683e-06,
      "loss": 0.0235,
      "step": 7050
    },
    {
      "epoch": 0.04870745858831159,
      "grad_norm": 5.629592851619236e-05,
      "learning_rate": 9.51347733999324e-06,
      "loss": 0.0,
      "step": 7060
    },
    {
      "epoch": 0.04877644932285594,
      "grad_norm": 1.383468628546325e-07,
      "learning_rate": 9.51285642338234e-06,
      "loss": 0.2564,
      "step": 7070
    },
    {
      "epoch": 0.04884544005740029,
      "grad_norm": 0.0,
      "learning_rate": 9.512166516036897e-06,
      "loss": 0.0,
      "step": 7080
    },
    {
      "epoch": 0.048914430791944644,
      "grad_norm": 0.0,
      "learning_rate": 9.511476608691454e-06,
      "loss": 0.0131,
      "step": 7090
    },
    {
      "epoch": 0.04898342152648899,
      "grad_norm": 0.0,
      "learning_rate": 9.51078670134601e-06,
      "loss": 0.0,
      "step": 7100
    },
    {
      "epoch": 0.049052412261033344,
      "grad_norm": 0.0,
      "learning_rate": 9.510096794000566e-06,
      "loss": 0.0,
      "step": 7110
    },
    {
      "epoch": 0.04912140299557769,
      "grad_norm": 0.0,
      "learning_rate": 9.509406886655123e-06,
      "loss": 0.0,
      "step": 7120
    },
    {
      "epoch": 0.049190393730122045,
      "grad_norm": 0.0,
      "learning_rate": 9.50871697930968e-06,
      "loss": 0.0021,
      "step": 7130
    },
    {
      "epoch": 0.04925938446466639,
      "grad_norm": 0.0,
      "learning_rate": 9.508027071964236e-06,
      "loss": 0.0,
      "step": 7140
    },
    {
      "epoch": 0.049328375199210746,
      "grad_norm": 0.0,
      "learning_rate": 9.507337164618792e-06,
      "loss": 0.0,
      "step": 7150
    },
    {
      "epoch": 0.0493973659337551,
      "grad_norm": 0.0,
      "learning_rate": 9.506647257273349e-06,
      "loss": 0.0,
      "step": 7160
    },
    {
      "epoch": 0.04946635666829945,
      "grad_norm": 0.0,
      "learning_rate": 9.505957349927905e-06,
      "loss": 0.0,
      "step": 7170
    },
    {
      "epoch": 0.0495353474028438,
      "grad_norm": 0.0,
      "learning_rate": 9.505267442582462e-06,
      "loss": 0.0,
      "step": 7180
    },
    {
      "epoch": 0.04960433813738815,
      "grad_norm": 1.5708725452423096,
      "learning_rate": 9.504577535237018e-06,
      "loss": 0.0003,
      "step": 7190
    },
    {
      "epoch": 0.0496733288719325,
      "grad_norm": 0.0,
      "learning_rate": 9.503887627891575e-06,
      "loss": 0.0,
      "step": 7200
    },
    {
      "epoch": 0.04974231960647685,
      "grad_norm": 0.012020227499306202,
      "learning_rate": 9.503197720546131e-06,
      "loss": 0.0,
      "step": 7210
    },
    {
      "epoch": 0.0498113103410212,
      "grad_norm": 0.0,
      "learning_rate": 9.502507813200688e-06,
      "loss": 0.0001,
      "step": 7220
    },
    {
      "epoch": 0.04988030107556555,
      "grad_norm": 0.0,
      "learning_rate": 9.501817905855244e-06,
      "loss": 0.0,
      "step": 7230
    },
    {
      "epoch": 0.0499492918101099,
      "grad_norm": 8.542450302684301e-08,
      "learning_rate": 9.501127998509801e-06,
      "loss": 0.0003,
      "step": 7240
    },
    {
      "epoch": 0.05001828254465425,
      "grad_norm": 0.0,
      "learning_rate": 9.500438091164357e-06,
      "loss": 0.0,
      "step": 7250
    },
    {
      "epoch": 0.0500872732791986,
      "grad_norm": 5.3104047775268555,
      "learning_rate": 9.499748183818914e-06,
      "loss": 0.0016,
      "step": 7260
    },
    {
      "epoch": 0.05015626401374296,
      "grad_norm": 4.8928811224868696e-08,
      "learning_rate": 9.49905827647347e-06,
      "loss": 0.0107,
      "step": 7270
    },
    {
      "epoch": 0.050225254748287304,
      "grad_norm": 0.0,
      "learning_rate": 9.498368369128027e-06,
      "loss": 0.0,
      "step": 7280
    },
    {
      "epoch": 0.05029424548283166,
      "grad_norm": 0.0,
      "learning_rate": 9.497678461782583e-06,
      "loss": 0.0,
      "step": 7290
    },
    {
      "epoch": 0.050363236217376005,
      "grad_norm": 0.0,
      "learning_rate": 9.49698855443714e-06,
      "loss": 0.0,
      "step": 7300
    },
    {
      "epoch": 0.05043222695192036,
      "grad_norm": 0.0,
      "learning_rate": 9.496298647091696e-06,
      "loss": 0.0,
      "step": 7310
    },
    {
      "epoch": 0.050501217686464706,
      "grad_norm": 0.0,
      "learning_rate": 9.495608739746253e-06,
      "loss": 0.017,
      "step": 7320
    },
    {
      "epoch": 0.05057020842100906,
      "grad_norm": 4.380771940759587e-07,
      "learning_rate": 9.49491883240081e-06,
      "loss": 0.0,
      "step": 7330
    },
    {
      "epoch": 0.050639199155553406,
      "grad_norm": 0.0,
      "learning_rate": 9.494228925055366e-06,
      "loss": 0.0,
      "step": 7340
    },
    {
      "epoch": 0.05070818989009776,
      "grad_norm": 0.001265982398763299,
      "learning_rate": 9.493539017709922e-06,
      "loss": 0.0,
      "step": 7350
    },
    {
      "epoch": 0.050777180624642114,
      "grad_norm": 0.0,
      "learning_rate": 9.492849110364479e-06,
      "loss": 0.0003,
      "step": 7360
    },
    {
      "epoch": 0.05084617135918646,
      "grad_norm": 5.745352268218994,
      "learning_rate": 9.492159203019035e-06,
      "loss": 0.0009,
      "step": 7370
    },
    {
      "epoch": 0.050915162093730815,
      "grad_norm": 0.0,
      "learning_rate": 9.491469295673592e-06,
      "loss": 0.0,
      "step": 7380
    },
    {
      "epoch": 0.05098415282827516,
      "grad_norm": 6.952218245714903e-05,
      "learning_rate": 9.490779388328148e-06,
      "loss": 0.0017,
      "step": 7390
    },
    {
      "epoch": 0.051053143562819515,
      "grad_norm": 0.00034228002186864614,
      "learning_rate": 9.490089480982705e-06,
      "loss": 0.0,
      "step": 7400
    },
    {
      "epoch": 0.05112213429736386,
      "grad_norm": 0.0,
      "learning_rate": 9.489399573637261e-06,
      "loss": 0.0,
      "step": 7410
    },
    {
      "epoch": 0.051191125031908216,
      "grad_norm": 1.0179088860695629e-07,
      "learning_rate": 9.488709666291818e-06,
      "loss": 0.0137,
      "step": 7420
    },
    {
      "epoch": 0.05126011576645256,
      "grad_norm": 0.0,
      "learning_rate": 9.488019758946374e-06,
      "loss": 0.0005,
      "step": 7430
    },
    {
      "epoch": 0.05132910650099692,
      "grad_norm": 5.103669709072278e-10,
      "learning_rate": 9.48732985160093e-06,
      "loss": 0.058,
      "step": 7440
    },
    {
      "epoch": 0.051398097235541264,
      "grad_norm": 0.0,
      "learning_rate": 9.486639944255487e-06,
      "loss": 0.001,
      "step": 7450
    },
    {
      "epoch": 0.05146708797008562,
      "grad_norm": 0.0,
      "learning_rate": 9.485950036910044e-06,
      "loss": 0.0003,
      "step": 7460
    },
    {
      "epoch": 0.05153607870462997,
      "grad_norm": 0.0001053997257258743,
      "learning_rate": 9.4852601295646e-06,
      "loss": 0.0,
      "step": 7470
    },
    {
      "epoch": 0.05160506943917432,
      "grad_norm": 0.007867366075515747,
      "learning_rate": 9.484570222219157e-06,
      "loss": 0.0,
      "step": 7480
    },
    {
      "epoch": 0.05167406017371867,
      "grad_norm": 0.0,
      "learning_rate": 9.483880314873713e-06,
      "loss": 0.0108,
      "step": 7490
    },
    {
      "epoch": 0.05174305090826302,
      "grad_norm": 6.209116065747367e-08,
      "learning_rate": 9.48319040752827e-06,
      "loss": 0.0007,
      "step": 7500
    },
    {
      "epoch": 0.05181204164280737,
      "grad_norm": 20.135934829711914,
      "learning_rate": 9.482500500182826e-06,
      "loss": 0.0038,
      "step": 7510
    },
    {
      "epoch": 0.05188103237735172,
      "grad_norm": 0.0,
      "learning_rate": 9.481810592837383e-06,
      "loss": 0.0172,
      "step": 7520
    },
    {
      "epoch": 0.051950023111896074,
      "grad_norm": 0.0,
      "learning_rate": 9.481120685491939e-06,
      "loss": 0.0,
      "step": 7530
    },
    {
      "epoch": 0.05201901384644042,
      "grad_norm": 21.914880752563477,
      "learning_rate": 9.480430778146496e-06,
      "loss": 0.007,
      "step": 7540
    },
    {
      "epoch": 0.052088004580984774,
      "grad_norm": 0.0,
      "learning_rate": 9.479740870801052e-06,
      "loss": 0.0226,
      "step": 7550
    },
    {
      "epoch": 0.05215699531552912,
      "grad_norm": 0.0,
      "learning_rate": 9.479050963455609e-06,
      "loss": 0.0002,
      "step": 7560
    },
    {
      "epoch": 0.052225986050073475,
      "grad_norm": 0.0,
      "learning_rate": 9.478361056110165e-06,
      "loss": 0.0,
      "step": 7570
    },
    {
      "epoch": 0.05229497678461783,
      "grad_norm": 3.8140576918976876e-08,
      "learning_rate": 9.477671148764722e-06,
      "loss": 0.0,
      "step": 7580
    },
    {
      "epoch": 0.052363967519162176,
      "grad_norm": 7.151640613045629e-09,
      "learning_rate": 9.476981241419278e-06,
      "loss": 0.0368,
      "step": 7590
    },
    {
      "epoch": 0.05243295825370653,
      "grad_norm": 0.0,
      "learning_rate": 9.476291334073835e-06,
      "loss": 0.0,
      "step": 7600
    },
    {
      "epoch": 0.05250194898825088,
      "grad_norm": 0.0,
      "learning_rate": 9.475601426728391e-06,
      "loss": 0.0131,
      "step": 7610
    },
    {
      "epoch": 0.05257093972279523,
      "grad_norm": 0.0,
      "learning_rate": 9.474911519382947e-06,
      "loss": 0.0,
      "step": 7620
    },
    {
      "epoch": 0.05263993045733958,
      "grad_norm": 0.0,
      "learning_rate": 9.474221612037504e-06,
      "loss": 0.0,
      "step": 7630
    },
    {
      "epoch": 0.05270892119188393,
      "grad_norm": 0.0,
      "learning_rate": 9.47353170469206e-06,
      "loss": 0.0013,
      "step": 7640
    },
    {
      "epoch": 0.05277791192642828,
      "grad_norm": 0.0,
      "learning_rate": 9.472841797346617e-06,
      "loss": 0.0,
      "step": 7650
    },
    {
      "epoch": 0.05284690266097263,
      "grad_norm": 0.0,
      "learning_rate": 9.472151890001173e-06,
      "loss": 0.002,
      "step": 7660
    },
    {
      "epoch": 0.05291589339551698,
      "grad_norm": 0.0,
      "learning_rate": 9.47146198265573e-06,
      "loss": 0.0,
      "step": 7670
    },
    {
      "epoch": 0.05298488413006133,
      "grad_norm": 0.0,
      "learning_rate": 9.470772075310286e-06,
      "loss": 0.0,
      "step": 7680
    },
    {
      "epoch": 0.053053874864605687,
      "grad_norm": 4.2308642700383814e-10,
      "learning_rate": 9.470082167964843e-06,
      "loss": 0.0,
      "step": 7690
    },
    {
      "epoch": 0.05312286559915003,
      "grad_norm": 0.0,
      "learning_rate": 9.4693922606194e-06,
      "loss": 0.0,
      "step": 7700
    },
    {
      "epoch": 0.05319185633369439,
      "grad_norm": 0.0,
      "learning_rate": 9.468702353273956e-06,
      "loss": 0.0,
      "step": 7710
    },
    {
      "epoch": 0.053260847068238734,
      "grad_norm": 0.0,
      "learning_rate": 9.468012445928512e-06,
      "loss": 0.0,
      "step": 7720
    },
    {
      "epoch": 0.05332983780278309,
      "grad_norm": 9.698508662836502e-09,
      "learning_rate": 9.467322538583069e-06,
      "loss": 0.0139,
      "step": 7730
    },
    {
      "epoch": 0.053398828537327435,
      "grad_norm": 0.0,
      "learning_rate": 9.466632631237625e-06,
      "loss": 0.0,
      "step": 7740
    },
    {
      "epoch": 0.05346781927187179,
      "grad_norm": 0.0,
      "learning_rate": 9.465942723892182e-06,
      "loss": 0.0,
      "step": 7750
    },
    {
      "epoch": 0.053536810006416136,
      "grad_norm": 0.0,
      "learning_rate": 9.465252816546738e-06,
      "loss": 0.014,
      "step": 7760
    },
    {
      "epoch": 0.05360580074096049,
      "grad_norm": 0.0,
      "learning_rate": 9.464562909201295e-06,
      "loss": 0.0,
      "step": 7770
    },
    {
      "epoch": 0.053674791475504836,
      "grad_norm": 0.0,
      "learning_rate": 9.463873001855851e-06,
      "loss": 0.0,
      "step": 7780
    },
    {
      "epoch": 0.05374378221004919,
      "grad_norm": 0.0,
      "learning_rate": 9.463183094510408e-06,
      "loss": 0.0,
      "step": 7790
    },
    {
      "epoch": 0.053812772944593544,
      "grad_norm": 1.6528139894944616e-05,
      "learning_rate": 9.462493187164964e-06,
      "loss": 0.0002,
      "step": 7800
    },
    {
      "epoch": 0.05388176367913789,
      "grad_norm": 0.0,
      "learning_rate": 9.46180327981952e-06,
      "loss": 0.0011,
      "step": 7810
    },
    {
      "epoch": 0.053950754413682245,
      "grad_norm": 0.0,
      "learning_rate": 9.461113372474077e-06,
      "loss": 0.0,
      "step": 7820
    },
    {
      "epoch": 0.05401974514822659,
      "grad_norm": 0.0,
      "learning_rate": 9.460423465128634e-06,
      "loss": 0.0003,
      "step": 7830
    },
    {
      "epoch": 0.054088735882770946,
      "grad_norm": 1.1157307624816895,
      "learning_rate": 9.45973355778319e-06,
      "loss": 0.0002,
      "step": 7840
    },
    {
      "epoch": 0.05415772661731529,
      "grad_norm": 0.0,
      "learning_rate": 9.459043650437747e-06,
      "loss": 0.0002,
      "step": 7850
    },
    {
      "epoch": 0.054226717351859646,
      "grad_norm": 3.301774853525785e-09,
      "learning_rate": 9.458353743092303e-06,
      "loss": 0.0,
      "step": 7860
    },
    {
      "epoch": 0.05429570808640399,
      "grad_norm": 0.0,
      "learning_rate": 9.45766383574686e-06,
      "loss": 0.0,
      "step": 7870
    },
    {
      "epoch": 0.05436469882094835,
      "grad_norm": 0.0,
      "learning_rate": 9.456973928401416e-06,
      "loss": 0.0,
      "step": 7880
    },
    {
      "epoch": 0.054433689555492694,
      "grad_norm": 0.0,
      "learning_rate": 9.456284021055973e-06,
      "loss": 0.0001,
      "step": 7890
    },
    {
      "epoch": 0.05450268029003705,
      "grad_norm": 0.0,
      "learning_rate": 9.45559411371053e-06,
      "loss": 0.0005,
      "step": 7900
    },
    {
      "epoch": 0.0545716710245814,
      "grad_norm": 0.0,
      "learning_rate": 9.454904206365086e-06,
      "loss": 0.0002,
      "step": 7910
    },
    {
      "epoch": 0.05464066175912575,
      "grad_norm": 2.030442090017459e-07,
      "learning_rate": 9.454214299019642e-06,
      "loss": 0.0,
      "step": 7920
    },
    {
      "epoch": 0.0547096524936701,
      "grad_norm": 4.900098105053985e-10,
      "learning_rate": 9.453524391674199e-06,
      "loss": 0.226,
      "step": 7930
    },
    {
      "epoch": 0.05477864322821445,
      "grad_norm": 0.0,
      "learning_rate": 9.452834484328755e-06,
      "loss": 0.0,
      "step": 7940
    },
    {
      "epoch": 0.0548476339627588,
      "grad_norm": 0.0,
      "learning_rate": 9.452144576983312e-06,
      "loss": 0.0,
      "step": 7950
    },
    {
      "epoch": 0.05491662469730315,
      "grad_norm": 0.0,
      "learning_rate": 9.451454669637868e-06,
      "loss": 0.0,
      "step": 7960
    },
    {
      "epoch": 0.054985615431847504,
      "grad_norm": 0.19920563697814941,
      "learning_rate": 9.450764762292425e-06,
      "loss": 0.0,
      "step": 7970
    },
    {
      "epoch": 0.05505460616639185,
      "grad_norm": 0.00010274054511683062,
      "learning_rate": 9.450074854946981e-06,
      "loss": 0.0,
      "step": 7980
    },
    {
      "epoch": 0.055123596900936205,
      "grad_norm": 2.021112413785886e-05,
      "learning_rate": 9.449384947601538e-06,
      "loss": 0.0002,
      "step": 7990
    },
    {
      "epoch": 0.05519258763548056,
      "grad_norm": 0.00023247135686688125,
      "learning_rate": 9.448695040256094e-06,
      "loss": 0.0058,
      "step": 8000
    },
    {
      "epoch": 0.055261578370024905,
      "grad_norm": 211.25997924804688,
      "learning_rate": 9.44800513291065e-06,
      "loss": 0.0423,
      "step": 8010
    },
    {
      "epoch": 0.05533056910456926,
      "grad_norm": 0.0,
      "learning_rate": 9.447384216299751e-06,
      "loss": 0.7328,
      "step": 8020
    },
    {
      "epoch": 0.055399559839113606,
      "grad_norm": 0.0,
      "learning_rate": 9.446694308954308e-06,
      "loss": 0.0001,
      "step": 8030
    },
    {
      "epoch": 0.05546855057365796,
      "grad_norm": 0.0,
      "learning_rate": 9.446004401608864e-06,
      "loss": 0.0,
      "step": 8040
    },
    {
      "epoch": 0.05553754130820231,
      "grad_norm": 4.76493555723323e-09,
      "learning_rate": 9.44531449426342e-06,
      "loss": 0.0,
      "step": 8050
    },
    {
      "epoch": 0.05560653204274666,
      "grad_norm": 9.497167496874681e-10,
      "learning_rate": 9.444624586917977e-06,
      "loss": 0.0,
      "step": 8060
    },
    {
      "epoch": 0.05567552277729101,
      "grad_norm": 0.0023989698383957148,
      "learning_rate": 9.443934679572534e-06,
      "loss": 0.0,
      "step": 8070
    },
    {
      "epoch": 0.05574451351183536,
      "grad_norm": 0.0,
      "learning_rate": 9.44324477222709e-06,
      "loss": 0.0,
      "step": 8080
    },
    {
      "epoch": 0.05581350424637971,
      "grad_norm": 0.0,
      "learning_rate": 9.442554864881647e-06,
      "loss": 0.0003,
      "step": 8090
    },
    {
      "epoch": 0.05588249498092406,
      "grad_norm": 0.009688925929367542,
      "learning_rate": 9.441864957536203e-06,
      "loss": 0.0,
      "step": 8100
    },
    {
      "epoch": 0.055951485715468416,
      "grad_norm": 0.0,
      "learning_rate": 9.441175050190761e-06,
      "loss": 0.0,
      "step": 8110
    },
    {
      "epoch": 0.05602047645001276,
      "grad_norm": 0.0040071685798466206,
      "learning_rate": 9.440485142845316e-06,
      "loss": 0.0,
      "step": 8120
    },
    {
      "epoch": 0.05608946718455712,
      "grad_norm": 0.0,
      "learning_rate": 9.439795235499873e-06,
      "loss": 0.0009,
      "step": 8130
    },
    {
      "epoch": 0.056158457919101464,
      "grad_norm": 0.0,
      "learning_rate": 9.43910532815443e-06,
      "loss": 0.0,
      "step": 8140
    },
    {
      "epoch": 0.05622744865364582,
      "grad_norm": 0.0,
      "learning_rate": 9.438415420808986e-06,
      "loss": 0.0,
      "step": 8150
    },
    {
      "epoch": 0.056296439388190164,
      "grad_norm": 0.0001427438110113144,
      "learning_rate": 9.437725513463542e-06,
      "loss": 0.1418,
      "step": 8160
    },
    {
      "epoch": 0.05636543012273452,
      "grad_norm": 0.0,
      "learning_rate": 9.437035606118099e-06,
      "loss": 0.0,
      "step": 8170
    },
    {
      "epoch": 0.056434420857278865,
      "grad_norm": 0.0,
      "learning_rate": 9.436345698772655e-06,
      "loss": 0.0,
      "step": 8180
    },
    {
      "epoch": 0.05650341159182322,
      "grad_norm": 9.834275260800496e-06,
      "learning_rate": 9.435655791427212e-06,
      "loss": 0.0,
      "step": 8190
    },
    {
      "epoch": 0.056572402326367566,
      "grad_norm": 0.0,
      "learning_rate": 9.434965884081768e-06,
      "loss": 0.0,
      "step": 8200
    },
    {
      "epoch": 0.05664139306091192,
      "grad_norm": 1.7513029888505116e-05,
      "learning_rate": 9.434275976736325e-06,
      "loss": 0.0,
      "step": 8210
    },
    {
      "epoch": 0.05671038379545627,
      "grad_norm": 0.0,
      "learning_rate": 9.433586069390883e-06,
      "loss": 0.0304,
      "step": 8220
    },
    {
      "epoch": 0.05677937453000062,
      "grad_norm": 32.892425537109375,
      "learning_rate": 9.43289616204544e-06,
      "loss": 0.0072,
      "step": 8230
    },
    {
      "epoch": 0.056848365264544974,
      "grad_norm": 0.0,
      "learning_rate": 9.432206254699994e-06,
      "loss": 0.0,
      "step": 8240
    },
    {
      "epoch": 0.05691735599908932,
      "grad_norm": 0.0,
      "learning_rate": 9.43151634735455e-06,
      "loss": 0.0,
      "step": 8250
    },
    {
      "epoch": 0.056986346733633675,
      "grad_norm": 0.04160303622484207,
      "learning_rate": 9.430826440009107e-06,
      "loss": 0.0,
      "step": 8260
    },
    {
      "epoch": 0.05705533746817802,
      "grad_norm": 0.0,
      "learning_rate": 9.430136532663664e-06,
      "loss": 0.6965,
      "step": 8270
    },
    {
      "epoch": 0.057124328202722376,
      "grad_norm": 0.0,
      "learning_rate": 9.42944662531822e-06,
      "loss": 0.0,
      "step": 8280
    },
    {
      "epoch": 0.05719331893726672,
      "grad_norm": 9.750915808126592e-08,
      "learning_rate": 9.428756717972777e-06,
      "loss": 0.0,
      "step": 8290
    },
    {
      "epoch": 0.057262309671811076,
      "grad_norm": 0.0,
      "learning_rate": 9.428066810627333e-06,
      "loss": 0.0058,
      "step": 8300
    },
    {
      "epoch": 0.05733130040635542,
      "grad_norm": 0.0,
      "learning_rate": 9.42737690328189e-06,
      "loss": 0.0964,
      "step": 8310
    },
    {
      "epoch": 0.05740029114089978,
      "grad_norm": 0.0,
      "learning_rate": 9.426686995936446e-06,
      "loss": 0.0,
      "step": 8320
    },
    {
      "epoch": 0.05746928187544413,
      "grad_norm": 0.12781912088394165,
      "learning_rate": 9.425997088591004e-06,
      "loss": 0.0,
      "step": 8330
    },
    {
      "epoch": 0.05753827260998848,
      "grad_norm": 0.0,
      "learning_rate": 9.42530718124556e-06,
      "loss": 0.0001,
      "step": 8340
    },
    {
      "epoch": 0.05760726334453283,
      "grad_norm": 0.0053254892118275166,
      "learning_rate": 9.424617273900115e-06,
      "loss": 0.0071,
      "step": 8350
    },
    {
      "epoch": 0.05767625407907718,
      "grad_norm": 0.0,
      "learning_rate": 9.423927366554672e-06,
      "loss": 0.0,
      "step": 8360
    },
    {
      "epoch": 0.05774524481362153,
      "grad_norm": 0.002254385966807604,
      "learning_rate": 9.423237459209228e-06,
      "loss": 0.0,
      "step": 8370
    },
    {
      "epoch": 0.05781423554816588,
      "grad_norm": 0.0,
      "learning_rate": 9.422547551863785e-06,
      "loss": 0.0,
      "step": 8380
    },
    {
      "epoch": 0.05788322628271023,
      "grad_norm": 1.6992156588457874e-06,
      "learning_rate": 9.421857644518341e-06,
      "loss": 0.0,
      "step": 8390
    },
    {
      "epoch": 0.05795221701725458,
      "grad_norm": 2.8296074816580585e-08,
      "learning_rate": 9.421167737172898e-06,
      "loss": 0.0,
      "step": 8400
    },
    {
      "epoch": 0.058021207751798934,
      "grad_norm": 0.0,
      "learning_rate": 9.420477829827454e-06,
      "loss": 0.0754,
      "step": 8410
    },
    {
      "epoch": 0.05809019848634328,
      "grad_norm": 0.0,
      "learning_rate": 9.419787922482011e-06,
      "loss": 0.0,
      "step": 8420
    },
    {
      "epoch": 0.058159189220887635,
      "grad_norm": 9.419382877240423e-06,
      "learning_rate": 9.419098015136567e-06,
      "loss": 0.0,
      "step": 8430
    },
    {
      "epoch": 0.05822817995543199,
      "grad_norm": 0.0,
      "learning_rate": 9.418408107791126e-06,
      "loss": 0.4281,
      "step": 8440
    },
    {
      "epoch": 0.058297170689976335,
      "grad_norm": 0.0,
      "learning_rate": 9.417718200445682e-06,
      "loss": 0.0,
      "step": 8450
    },
    {
      "epoch": 0.05836616142452069,
      "grad_norm": 0.0,
      "learning_rate": 9.417028293100237e-06,
      "loss": 0.004,
      "step": 8460
    },
    {
      "epoch": 0.058435152159065036,
      "grad_norm": 0.0,
      "learning_rate": 9.416338385754793e-06,
      "loss": 0.0,
      "step": 8470
    },
    {
      "epoch": 0.05850414289360939,
      "grad_norm": 0.0,
      "learning_rate": 9.41564847840935e-06,
      "loss": 0.0,
      "step": 8480
    },
    {
      "epoch": 0.05857313362815374,
      "grad_norm": 0.0010733334347605705,
      "learning_rate": 9.414958571063906e-06,
      "loss": 0.0,
      "step": 8490
    },
    {
      "epoch": 0.05864212436269809,
      "grad_norm": 0.0,
      "learning_rate": 9.414268663718463e-06,
      "loss": 0.0267,
      "step": 8500
    },
    {
      "epoch": 0.05871111509724244,
      "grad_norm": 0.0,
      "learning_rate": 9.41357875637302e-06,
      "loss": 0.0,
      "step": 8510
    },
    {
      "epoch": 0.05878010583178679,
      "grad_norm": 2.7318827051203698e-05,
      "learning_rate": 9.412888849027576e-06,
      "loss": 0.0,
      "step": 8520
    },
    {
      "epoch": 0.05884909656633114,
      "grad_norm": 0.0,
      "learning_rate": 9.412198941682132e-06,
      "loss": 0.0007,
      "step": 8530
    },
    {
      "epoch": 0.05891808730087549,
      "grad_norm": 0.0,
      "learning_rate": 9.411509034336689e-06,
      "loss": 0.0,
      "step": 8540
    },
    {
      "epoch": 0.058987078035419846,
      "grad_norm": 0.0,
      "learning_rate": 9.410819126991247e-06,
      "loss": 0.0087,
      "step": 8550
    },
    {
      "epoch": 0.05905606876996419,
      "grad_norm": 0.0,
      "learning_rate": 9.410129219645803e-06,
      "loss": 0.0008,
      "step": 8560
    },
    {
      "epoch": 0.05912505950450855,
      "grad_norm": 0.0,
      "learning_rate": 9.40943931230036e-06,
      "loss": 0.0,
      "step": 8570
    },
    {
      "epoch": 0.059194050239052894,
      "grad_norm": 0.0,
      "learning_rate": 9.408749404954915e-06,
      "loss": 0.0003,
      "step": 8580
    },
    {
      "epoch": 0.05926304097359725,
      "grad_norm": 9.69321511945509e-10,
      "learning_rate": 9.408059497609471e-06,
      "loss": 0.0012,
      "step": 8590
    },
    {
      "epoch": 0.059332031708141594,
      "grad_norm": 0.00021176903101149946,
      "learning_rate": 9.407369590264028e-06,
      "loss": 0.0,
      "step": 8600
    },
    {
      "epoch": 0.05940102244268595,
      "grad_norm": 3.326289061078569e-06,
      "learning_rate": 9.406679682918584e-06,
      "loss": 0.0029,
      "step": 8610
    },
    {
      "epoch": 0.059470013177230295,
      "grad_norm": 0.0,
      "learning_rate": 9.40598977557314e-06,
      "loss": 0.0,
      "step": 8620
    },
    {
      "epoch": 0.05953900391177465,
      "grad_norm": 0.0,
      "learning_rate": 9.405299868227697e-06,
      "loss": 0.0,
      "step": 8630
    },
    {
      "epoch": 0.059607994646319,
      "grad_norm": 0.0,
      "learning_rate": 9.404609960882254e-06,
      "loss": 0.0007,
      "step": 8640
    },
    {
      "epoch": 0.05967698538086335,
      "grad_norm": 7.834999996703118e-05,
      "learning_rate": 9.40392005353681e-06,
      "loss": 0.0,
      "step": 8650
    },
    {
      "epoch": 0.059745976115407703,
      "grad_norm": 0.0001593024644535035,
      "learning_rate": 9.403230146191368e-06,
      "loss": 0.0001,
      "step": 8660
    },
    {
      "epoch": 0.05981496684995205,
      "grad_norm": 0.0,
      "learning_rate": 9.402540238845925e-06,
      "loss": 0.0,
      "step": 8670
    },
    {
      "epoch": 0.059883957584496404,
      "grad_norm": 1.7578850020072423e-07,
      "learning_rate": 9.401850331500481e-06,
      "loss": 0.0,
      "step": 8680
    },
    {
      "epoch": 0.05995294831904075,
      "grad_norm": 0.013763878494501114,
      "learning_rate": 9.401160424155036e-06,
      "loss": 0.0,
      "step": 8690
    },
    {
      "epoch": 0.060021939053585105,
      "grad_norm": 0.0,
      "learning_rate": 9.400470516809593e-06,
      "loss": 0.0001,
      "step": 8700
    },
    {
      "epoch": 0.06009092978812945,
      "grad_norm": 0.0,
      "learning_rate": 9.399780609464149e-06,
      "loss": 0.0,
      "step": 8710
    },
    {
      "epoch": 0.060159920522673806,
      "grad_norm": 2.6696572604123503e-05,
      "learning_rate": 9.399090702118706e-06,
      "loss": 0.0003,
      "step": 8720
    },
    {
      "epoch": 0.06022891125721815,
      "grad_norm": 0.0005639257724396884,
      "learning_rate": 9.398400794773262e-06,
      "loss": 0.0,
      "step": 8730
    },
    {
      "epoch": 0.060297901991762506,
      "grad_norm": 0.0,
      "learning_rate": 9.397710887427819e-06,
      "loss": 0.0,
      "step": 8740
    },
    {
      "epoch": 0.06036689272630686,
      "grad_norm": 0.0,
      "learning_rate": 9.397020980082375e-06,
      "loss": 0.0041,
      "step": 8750
    },
    {
      "epoch": 0.06043588346085121,
      "grad_norm": 0.0,
      "learning_rate": 9.396331072736932e-06,
      "loss": 0.0004,
      "step": 8760
    },
    {
      "epoch": 0.06050487419539556,
      "grad_norm": 0.029361022636294365,
      "learning_rate": 9.39564116539149e-06,
      "loss": 0.0,
      "step": 8770
    },
    {
      "epoch": 0.06057386492993991,
      "grad_norm": 0.00011345672828610986,
      "learning_rate": 9.394951258046046e-06,
      "loss": 0.0,
      "step": 8780
    },
    {
      "epoch": 0.06064285566448426,
      "grad_norm": 0.0,
      "learning_rate": 9.394261350700603e-06,
      "loss": 0.0,
      "step": 8790
    },
    {
      "epoch": 0.06071184639902861,
      "grad_norm": 0.0,
      "learning_rate": 9.393571443355158e-06,
      "loss": 0.0,
      "step": 8800
    },
    {
      "epoch": 0.06078083713357296,
      "grad_norm": 0.0,
      "learning_rate": 9.392881536009714e-06,
      "loss": 0.302,
      "step": 8810
    },
    {
      "epoch": 0.06084982786811731,
      "grad_norm": 0.0,
      "learning_rate": 9.39219162866427e-06,
      "loss": 0.0,
      "step": 8820
    },
    {
      "epoch": 0.06091881860266166,
      "grad_norm": 0.0,
      "learning_rate": 9.391501721318827e-06,
      "loss": 0.0,
      "step": 8830
    },
    {
      "epoch": 0.06098780933720601,
      "grad_norm": 0.0,
      "learning_rate": 9.390811813973383e-06,
      "loss": 0.0,
      "step": 8840
    },
    {
      "epoch": 0.061056800071750364,
      "grad_norm": 0.0,
      "learning_rate": 9.39012190662794e-06,
      "loss": 0.0,
      "step": 8850
    },
    {
      "epoch": 0.06112579080629472,
      "grad_norm": 0.0,
      "learning_rate": 9.389431999282496e-06,
      "loss": 0.0,
      "step": 8860
    },
    {
      "epoch": 0.061194781540839065,
      "grad_norm": 1.7742786440066993e-05,
      "learning_rate": 9.388742091937053e-06,
      "loss": 0.0,
      "step": 8870
    },
    {
      "epoch": 0.06126377227538342,
      "grad_norm": 0.001497775549069047,
      "learning_rate": 9.388052184591611e-06,
      "loss": 0.0232,
      "step": 8880
    },
    {
      "epoch": 0.061332763009927765,
      "grad_norm": 0.00032529831514693797,
      "learning_rate": 9.387362277246168e-06,
      "loss": 0.0004,
      "step": 8890
    },
    {
      "epoch": 0.06140175374447212,
      "grad_norm": 0.00010866899538086727,
      "learning_rate": 9.386672369900724e-06,
      "loss": 0.0021,
      "step": 8900
    },
    {
      "epoch": 0.061470744479016466,
      "grad_norm": 1.0541617712078732e-07,
      "learning_rate": 9.38598246255528e-06,
      "loss": 0.0014,
      "step": 8910
    },
    {
      "epoch": 0.06153973521356082,
      "grad_norm": 0.0,
      "learning_rate": 9.385292555209835e-06,
      "loss": 0.0,
      "step": 8920
    },
    {
      "epoch": 0.06160872594810517,
      "grad_norm": 0.0,
      "learning_rate": 9.384602647864392e-06,
      "loss": 0.0017,
      "step": 8930
    },
    {
      "epoch": 0.06167771668264952,
      "grad_norm": 0.0,
      "learning_rate": 9.383912740518948e-06,
      "loss": 0.0,
      "step": 8940
    },
    {
      "epoch": 0.06174670741719387,
      "grad_norm": 0.0,
      "learning_rate": 9.383222833173505e-06,
      "loss": 0.0014,
      "step": 8950
    },
    {
      "epoch": 0.06181569815173822,
      "grad_norm": 0.0,
      "learning_rate": 9.382532925828061e-06,
      "loss": 0.0,
      "step": 8960
    },
    {
      "epoch": 0.061884688886282575,
      "grad_norm": 0.0,
      "learning_rate": 9.381843018482618e-06,
      "loss": 0.0,
      "step": 8970
    },
    {
      "epoch": 0.06195367962082692,
      "grad_norm": 0.0,
      "learning_rate": 9.381153111137174e-06,
      "loss": 0.0,
      "step": 8980
    },
    {
      "epoch": 0.062022670355371276,
      "grad_norm": 0.0,
      "learning_rate": 9.380463203791733e-06,
      "loss": 0.0,
      "step": 8990
    },
    {
      "epoch": 0.06209166108991562,
      "grad_norm": 0.0,
      "learning_rate": 9.379773296446289e-06,
      "loss": 0.0,
      "step": 9000
    },
    {
      "epoch": 0.06216065182445998,
      "grad_norm": 0.0,
      "learning_rate": 9.379083389100846e-06,
      "loss": 0.0017,
      "step": 9010
    },
    {
      "epoch": 0.062229642559004324,
      "grad_norm": 3.527745502651669e-05,
      "learning_rate": 9.378393481755402e-06,
      "loss": 0.0,
      "step": 9020
    },
    {
      "epoch": 0.06229863329354868,
      "grad_norm": 0.0,
      "learning_rate": 9.377703574409957e-06,
      "loss": 0.0001,
      "step": 9030
    },
    {
      "epoch": 0.062367624028093024,
      "grad_norm": 0.0,
      "learning_rate": 9.377013667064513e-06,
      "loss": 0.0011,
      "step": 9040
    },
    {
      "epoch": 0.06243661476263738,
      "grad_norm": 0.011230291798710823,
      "learning_rate": 9.37632375971907e-06,
      "loss": 0.0011,
      "step": 9050
    },
    {
      "epoch": 0.06250560549718173,
      "grad_norm": 0.0,
      "learning_rate": 9.375633852373626e-06,
      "loss": 0.0,
      "step": 9060
    },
    {
      "epoch": 0.06257459623172608,
      "grad_norm": 0.0,
      "learning_rate": 9.374943945028183e-06,
      "loss": 0.0007,
      "step": 9070
    },
    {
      "epoch": 0.06264358696627043,
      "grad_norm": 0.0,
      "learning_rate": 9.37425403768274e-06,
      "loss": 0.0,
      "step": 9080
    },
    {
      "epoch": 0.06271257770081479,
      "grad_norm": 0.0,
      "learning_rate": 9.373564130337296e-06,
      "loss": 0.0,
      "step": 9090
    },
    {
      "epoch": 0.06278156843535913,
      "grad_norm": 0.0,
      "learning_rate": 9.372874222991854e-06,
      "loss": 0.0,
      "step": 9100
    },
    {
      "epoch": 0.06285055916990348,
      "grad_norm": 0.0,
      "learning_rate": 9.37218431564641e-06,
      "loss": 0.1701,
      "step": 9110
    },
    {
      "epoch": 0.06291954990444783,
      "grad_norm": 1.9915714233320614e-07,
      "learning_rate": 9.371494408300967e-06,
      "loss": 0.0,
      "step": 9120
    },
    {
      "epoch": 0.06298854063899219,
      "grad_norm": 184.47238159179688,
      "learning_rate": 9.370804500955523e-06,
      "loss": 0.0305,
      "step": 9130
    },
    {
      "epoch": 0.06305753137353654,
      "grad_norm": 0.0001221702987095341,
      "learning_rate": 9.370114593610078e-06,
      "loss": 0.0,
      "step": 9140
    },
    {
      "epoch": 0.06312652210808088,
      "grad_norm": 0.0,
      "learning_rate": 9.369424686264635e-06,
      "loss": 0.0,
      "step": 9150
    },
    {
      "epoch": 0.06319551284262523,
      "grad_norm": 0.0,
      "learning_rate": 9.368734778919191e-06,
      "loss": 0.5698,
      "step": 9160
    },
    {
      "epoch": 0.06326450357716959,
      "grad_norm": 0.0,
      "learning_rate": 9.368044871573748e-06,
      "loss": 0.0,
      "step": 9170
    },
    {
      "epoch": 0.06333349431171394,
      "grad_norm": 0.0,
      "learning_rate": 9.367354964228304e-06,
      "loss": 0.0001,
      "step": 9180
    },
    {
      "epoch": 0.06340248504625828,
      "grad_norm": 0.0,
      "learning_rate": 9.36666505688286e-06,
      "loss": 0.0,
      "step": 9190
    },
    {
      "epoch": 0.06347147578080264,
      "grad_norm": 1.4486658983514644e-08,
      "learning_rate": 9.365975149537417e-06,
      "loss": 0.0,
      "step": 9200
    },
    {
      "epoch": 0.06354046651534699,
      "grad_norm": 0.0,
      "learning_rate": 9.365285242191975e-06,
      "loss": 0.0027,
      "step": 9210
    },
    {
      "epoch": 0.06360945724989134,
      "grad_norm": 0.0004837385204154998,
      "learning_rate": 9.364595334846532e-06,
      "loss": 0.0001,
      "step": 9220
    },
    {
      "epoch": 0.06367844798443568,
      "grad_norm": 0.0,
      "learning_rate": 9.363905427501088e-06,
      "loss": 0.0,
      "step": 9230
    },
    {
      "epoch": 0.06374743871898005,
      "grad_norm": 0.0,
      "learning_rate": 9.363215520155645e-06,
      "loss": 0.0,
      "step": 9240
    },
    {
      "epoch": 0.06381642945352439,
      "grad_norm": 0.0,
      "learning_rate": 9.362525612810201e-06,
      "loss": 0.0,
      "step": 9250
    },
    {
      "epoch": 0.06388542018806874,
      "grad_norm": 1.93648362159729,
      "learning_rate": 9.361835705464756e-06,
      "loss": 0.0003,
      "step": 9260
    },
    {
      "epoch": 0.06395441092261309,
      "grad_norm": 0.0,
      "learning_rate": 9.361145798119313e-06,
      "loss": 0.0003,
      "step": 9270
    },
    {
      "epoch": 0.06402340165715745,
      "grad_norm": 0.0,
      "learning_rate": 9.360455890773869e-06,
      "loss": 0.0,
      "step": 9280
    },
    {
      "epoch": 0.0640923923917018,
      "grad_norm": 0.00012504179903771728,
      "learning_rate": 9.359765983428426e-06,
      "loss": 0.0,
      "step": 9290
    },
    {
      "epoch": 0.06416138312624614,
      "grad_norm": 0.0,
      "learning_rate": 9.359076076082982e-06,
      "loss": 0.0,
      "step": 9300
    },
    {
      "epoch": 0.0642303738607905,
      "grad_norm": 0.0024786265566945076,
      "learning_rate": 9.358386168737539e-06,
      "loss": 0.0,
      "step": 9310
    },
    {
      "epoch": 0.06429936459533485,
      "grad_norm": 0.0,
      "learning_rate": 9.357696261392097e-06,
      "loss": 0.0,
      "step": 9320
    },
    {
      "epoch": 0.0643683553298792,
      "grad_norm": 0.0005138832493685186,
      "learning_rate": 9.357006354046653e-06,
      "loss": 0.0002,
      "step": 9330
    },
    {
      "epoch": 0.06443734606442354,
      "grad_norm": 0.0,
      "learning_rate": 9.35631644670121e-06,
      "loss": 0.0001,
      "step": 9340
    },
    {
      "epoch": 0.0645063367989679,
      "grad_norm": 0.0,
      "learning_rate": 9.355626539355766e-06,
      "loss": 0.0033,
      "step": 9350
    },
    {
      "epoch": 0.06457532753351225,
      "grad_norm": 0.0,
      "learning_rate": 9.354936632010323e-06,
      "loss": 0.0,
      "step": 9360
    },
    {
      "epoch": 0.0646443182680566,
      "grad_norm": 0.0,
      "learning_rate": 9.354246724664877e-06,
      "loss": 0.0002,
      "step": 9370
    },
    {
      "epoch": 0.06471330900260094,
      "grad_norm": 0.0,
      "learning_rate": 9.353556817319434e-06,
      "loss": 0.0,
      "step": 9380
    },
    {
      "epoch": 0.0647822997371453,
      "grad_norm": 0.0,
      "learning_rate": 9.35286690997399e-06,
      "loss": 0.0,
      "step": 9390
    },
    {
      "epoch": 0.06485129047168965,
      "grad_norm": 0.0,
      "learning_rate": 9.352177002628547e-06,
      "loss": 0.0,
      "step": 9400
    },
    {
      "epoch": 0.064920281206234,
      "grad_norm": 0.0,
      "learning_rate": 9.351487095283103e-06,
      "loss": 0.0017,
      "step": 9410
    },
    {
      "epoch": 0.06498927194077836,
      "grad_norm": 0.0,
      "learning_rate": 9.35079718793766e-06,
      "loss": 0.0,
      "step": 9420
    },
    {
      "epoch": 0.0650582626753227,
      "grad_norm": 0.0,
      "learning_rate": 9.350107280592218e-06,
      "loss": 0.0,
      "step": 9430
    },
    {
      "epoch": 0.06512725340986705,
      "grad_norm": 0.0,
      "learning_rate": 9.349417373246775e-06,
      "loss": 0.011,
      "step": 9440
    },
    {
      "epoch": 0.0651962441444114,
      "grad_norm": 84.49259948730469,
      "learning_rate": 9.348727465901331e-06,
      "loss": 0.0343,
      "step": 9450
    },
    {
      "epoch": 0.06526523487895576,
      "grad_norm": 0.8491132259368896,
      "learning_rate": 9.348037558555888e-06,
      "loss": 0.0012,
      "step": 9460
    },
    {
      "epoch": 0.06533422561350011,
      "grad_norm": 0.0,
      "learning_rate": 9.347347651210444e-06,
      "loss": 0.0,
      "step": 9470
    },
    {
      "epoch": 0.06540321634804445,
      "grad_norm": 2.747396683844272e-05,
      "learning_rate": 9.346657743864999e-06,
      "loss": 0.0,
      "step": 9480
    },
    {
      "epoch": 0.0654722070825888,
      "grad_norm": 0.0,
      "learning_rate": 9.345967836519555e-06,
      "loss": 0.0005,
      "step": 9490
    },
    {
      "epoch": 0.06554119781713316,
      "grad_norm": 0.0,
      "learning_rate": 9.345277929174112e-06,
      "loss": 0.0001,
      "step": 9500
    },
    {
      "epoch": 0.06561018855167751,
      "grad_norm": 0.0,
      "learning_rate": 9.344588021828668e-06,
      "loss": 0.0131,
      "step": 9510
    },
    {
      "epoch": 0.06567917928622186,
      "grad_norm": 0.0,
      "learning_rate": 9.343898114483225e-06,
      "loss": 0.0107,
      "step": 9520
    },
    {
      "epoch": 0.06574817002076622,
      "grad_norm": 0.0,
      "learning_rate": 9.343208207137781e-06,
      "loss": 0.002,
      "step": 9530
    },
    {
      "epoch": 0.06581716075531056,
      "grad_norm": 0.0,
      "learning_rate": 9.34251829979234e-06,
      "loss": 0.0,
      "step": 9540
    },
    {
      "epoch": 0.06588615148985491,
      "grad_norm": 0.07264412939548492,
      "learning_rate": 9.341828392446896e-06,
      "loss": 0.0,
      "step": 9550
    },
    {
      "epoch": 0.06595514222439926,
      "grad_norm": 1.20164349937113e-05,
      "learning_rate": 9.341138485101452e-06,
      "loss": 0.0,
      "step": 9560
    },
    {
      "epoch": 0.06602413295894362,
      "grad_norm": 0.0,
      "learning_rate": 9.340448577756009e-06,
      "loss": 0.0,
      "step": 9570
    },
    {
      "epoch": 0.06609312369348797,
      "grad_norm": 0.0,
      "learning_rate": 9.339758670410565e-06,
      "loss": 0.0,
      "step": 9580
    },
    {
      "epoch": 0.06616211442803231,
      "grad_norm": 0.0,
      "learning_rate": 9.33906876306512e-06,
      "loss": 0.0,
      "step": 9590
    },
    {
      "epoch": 0.06623110516257667,
      "grad_norm": 0.0,
      "learning_rate": 9.338378855719677e-06,
      "loss": 0.0,
      "step": 9600
    },
    {
      "epoch": 0.06630009589712102,
      "grad_norm": 0.0,
      "learning_rate": 9.337688948374233e-06,
      "loss": 0.0,
      "step": 9610
    },
    {
      "epoch": 0.06636908663166537,
      "grad_norm": 0.0,
      "learning_rate": 9.33699904102879e-06,
      "loss": 0.0,
      "step": 9620
    },
    {
      "epoch": 0.06643807736620971,
      "grad_norm": 0.0,
      "learning_rate": 9.336309133683346e-06,
      "loss": 0.0004,
      "step": 9630
    },
    {
      "epoch": 0.06650706810075407,
      "grad_norm": 0.0,
      "learning_rate": 9.335619226337903e-06,
      "loss": 0.0,
      "step": 9640
    },
    {
      "epoch": 0.06657605883529842,
      "grad_norm": 0.0,
      "learning_rate": 9.334929318992461e-06,
      "loss": 0.0,
      "step": 9650
    },
    {
      "epoch": 0.06664504956984277,
      "grad_norm": 0.0,
      "learning_rate": 9.334239411647017e-06,
      "loss": 0.0,
      "step": 9660
    },
    {
      "epoch": 0.06671404030438711,
      "grad_norm": 0.0,
      "learning_rate": 9.333549504301574e-06,
      "loss": 0.0,
      "step": 9670
    },
    {
      "epoch": 0.06678303103893148,
      "grad_norm": 0.0,
      "learning_rate": 9.33285959695613e-06,
      "loss": 0.0,
      "step": 9680
    },
    {
      "epoch": 0.06685202177347582,
      "grad_norm": 9.999718167819083e-05,
      "learning_rate": 9.332169689610687e-06,
      "loss": 0.0007,
      "step": 9690
    },
    {
      "epoch": 0.06692101250802017,
      "grad_norm": 0.0,
      "learning_rate": 9.331479782265243e-06,
      "loss": 0.0001,
      "step": 9700
    },
    {
      "epoch": 0.06699000324256453,
      "grad_norm": 0.0,
      "learning_rate": 9.330789874919798e-06,
      "loss": 0.0,
      "step": 9710
    },
    {
      "epoch": 0.06705899397710888,
      "grad_norm": 0.0,
      "learning_rate": 9.330099967574355e-06,
      "loss": 0.1551,
      "step": 9720
    },
    {
      "epoch": 0.06712798471165322,
      "grad_norm": 3.7993876844666374e-07,
      "learning_rate": 9.329410060228911e-06,
      "loss": 0.2504,
      "step": 9730
    },
    {
      "epoch": 0.06719697544619757,
      "grad_norm": 0.0,
      "learning_rate": 9.328720152883468e-06,
      "loss": 0.0,
      "step": 9740
    },
    {
      "epoch": 0.06726596618074193,
      "grad_norm": 0.0,
      "learning_rate": 9.328030245538024e-06,
      "loss": 0.0002,
      "step": 9750
    },
    {
      "epoch": 0.06733495691528628,
      "grad_norm": 1.4668836811893016e-08,
      "learning_rate": 9.327340338192582e-06,
      "loss": 0.0418,
      "step": 9760
    },
    {
      "epoch": 0.06740394764983063,
      "grad_norm": 0.0013452752027660608,
      "learning_rate": 9.326650430847139e-06,
      "loss": 0.0,
      "step": 9770
    },
    {
      "epoch": 0.06747293838437497,
      "grad_norm": 0.0003280251694377512,
      "learning_rate": 9.325960523501695e-06,
      "loss": 0.0,
      "step": 9780
    },
    {
      "epoch": 0.06754192911891933,
      "grad_norm": 0.0004096842894796282,
      "learning_rate": 9.325270616156252e-06,
      "loss": 0.0,
      "step": 9790
    },
    {
      "epoch": 0.06761091985346368,
      "grad_norm": 0.0,
      "learning_rate": 9.324580708810808e-06,
      "loss": 0.0,
      "step": 9800
    },
    {
      "epoch": 0.06767991058800803,
      "grad_norm": 0.0,
      "learning_rate": 9.323890801465365e-06,
      "loss": 0.044,
      "step": 9810
    },
    {
      "epoch": 0.06774890132255239,
      "grad_norm": 0.00012699179933406413,
      "learning_rate": 9.32320089411992e-06,
      "loss": 0.0,
      "step": 9820
    },
    {
      "epoch": 0.06781789205709673,
      "grad_norm": 0.0,
      "learning_rate": 9.322510986774476e-06,
      "loss": 0.0108,
      "step": 9830
    },
    {
      "epoch": 0.06788688279164108,
      "grad_norm": 0.0,
      "learning_rate": 9.321821079429033e-06,
      "loss": 0.0003,
      "step": 9840
    },
    {
      "epoch": 0.06795587352618543,
      "grad_norm": 0.0,
      "learning_rate": 9.321131172083589e-06,
      "loss": 0.0,
      "step": 9850
    },
    {
      "epoch": 0.06802486426072979,
      "grad_norm": 0.0,
      "learning_rate": 9.320441264738146e-06,
      "loss": 0.0004,
      "step": 9860
    },
    {
      "epoch": 0.06809385499527414,
      "grad_norm": 0.0,
      "learning_rate": 9.319751357392704e-06,
      "loss": 0.0,
      "step": 9870
    },
    {
      "epoch": 0.06816284572981848,
      "grad_norm": 9.521896604525182e-10,
      "learning_rate": 9.31906145004726e-06,
      "loss": 0.0,
      "step": 9880
    },
    {
      "epoch": 0.06823183646436283,
      "grad_norm": 0.0,
      "learning_rate": 9.318371542701817e-06,
      "loss": 0.0,
      "step": 9890
    },
    {
      "epoch": 0.06830082719890719,
      "grad_norm": 0.00018718295905273408,
      "learning_rate": 9.317681635356373e-06,
      "loss": 0.0216,
      "step": 9900
    },
    {
      "epoch": 0.06836981793345154,
      "grad_norm": 0.0,
      "learning_rate": 9.31699172801093e-06,
      "loss": 0.002,
      "step": 9910
    },
    {
      "epoch": 0.06843880866799588,
      "grad_norm": 0.0,
      "learning_rate": 9.316301820665486e-06,
      "loss": 0.0,
      "step": 9920
    },
    {
      "epoch": 0.06850779940254025,
      "grad_norm": 73.73307037353516,
      "learning_rate": 9.315611913320041e-06,
      "loss": 0.0091,
      "step": 9930
    },
    {
      "epoch": 0.06857679013708459,
      "grad_norm": 3.7655553342119674e-08,
      "learning_rate": 9.314922005974597e-06,
      "loss": 0.0,
      "step": 9940
    },
    {
      "epoch": 0.06864578087162894,
      "grad_norm": 0.0,
      "learning_rate": 9.314232098629154e-06,
      "loss": 0.0081,
      "step": 9950
    },
    {
      "epoch": 0.06871477160617329,
      "grad_norm": 9.238464437899552e-10,
      "learning_rate": 9.31354219128371e-06,
      "loss": 0.0,
      "step": 9960
    },
    {
      "epoch": 0.06878376234071765,
      "grad_norm": 0.0003441699664108455,
      "learning_rate": 9.312852283938267e-06,
      "loss": 0.0,
      "step": 9970
    },
    {
      "epoch": 0.068852753075262,
      "grad_norm": 4.399174429181585e-07,
      "learning_rate": 9.312162376592825e-06,
      "loss": 0.0,
      "step": 9980
    },
    {
      "epoch": 0.06892174380980634,
      "grad_norm": 0.0,
      "learning_rate": 9.311472469247382e-06,
      "loss": 0.0,
      "step": 9990
    },
    {
      "epoch": 0.06899073454435069,
      "grad_norm": 0.0,
      "learning_rate": 9.310782561901938e-06,
      "loss": 0.0,
      "step": 10000
    },
    {
      "epoch": 0.06905972527889505,
      "grad_norm": 0.0,
      "learning_rate": 9.310092654556495e-06,
      "loss": 0.0003,
      "step": 10010
    },
    {
      "epoch": 0.0691287160134394,
      "grad_norm": 0.0,
      "learning_rate": 9.309402747211051e-06,
      "loss": 0.0,
      "step": 10020
    },
    {
      "epoch": 0.06919770674798374,
      "grad_norm": 0.0,
      "learning_rate": 9.308712839865608e-06,
      "loss": 0.0009,
      "step": 10030
    },
    {
      "epoch": 0.0692666974825281,
      "grad_norm": 0.0,
      "learning_rate": 9.308022932520164e-06,
      "loss": 0.0,
      "step": 10040
    },
    {
      "epoch": 0.06933568821707245,
      "grad_norm": 0.0,
      "learning_rate": 9.307333025174719e-06,
      "loss": 0.0,
      "step": 10050
    },
    {
      "epoch": 0.0694046789516168,
      "grad_norm": 0.0,
      "learning_rate": 9.306643117829275e-06,
      "loss": 0.0,
      "step": 10060
    },
    {
      "epoch": 0.06947366968616114,
      "grad_norm": 5.556299242925888e-07,
      "learning_rate": 9.305953210483832e-06,
      "loss": 0.0,
      "step": 10070
    },
    {
      "epoch": 0.0695426604207055,
      "grad_norm": 0.0,
      "learning_rate": 9.305263303138388e-06,
      "loss": 0.0,
      "step": 10080
    },
    {
      "epoch": 0.06961165115524985,
      "grad_norm": 0.0004480912466533482,
      "learning_rate": 9.304573395792946e-06,
      "loss": 0.0002,
      "step": 10090
    },
    {
      "epoch": 0.0696806418897942,
      "grad_norm": 0.18314163386821747,
      "learning_rate": 9.303883488447503e-06,
      "loss": 0.2578,
      "step": 10100
    },
    {
      "epoch": 0.06974963262433855,
      "grad_norm": 3.770653123069678e-09,
      "learning_rate": 9.30319358110206e-06,
      "loss": 0.0,
      "step": 10110
    },
    {
      "epoch": 0.0698186233588829,
      "grad_norm": 1.0981832474499242e-06,
      "learning_rate": 9.302503673756616e-06,
      "loss": 0.0004,
      "step": 10120
    },
    {
      "epoch": 0.06988761409342725,
      "grad_norm": 0.0,
      "learning_rate": 9.301813766411172e-06,
      "loss": 0.0,
      "step": 10130
    },
    {
      "epoch": 0.0699566048279716,
      "grad_norm": 0.0,
      "learning_rate": 9.301123859065729e-06,
      "loss": 0.0001,
      "step": 10140
    },
    {
      "epoch": 0.07002559556251596,
      "grad_norm": 0.0,
      "learning_rate": 9.300433951720285e-06,
      "loss": 0.0,
      "step": 10150
    },
    {
      "epoch": 0.07009458629706031,
      "grad_norm": 0.0006030003423802555,
      "learning_rate": 9.29974404437484e-06,
      "loss": 0.0008,
      "step": 10160
    },
    {
      "epoch": 0.07016357703160465,
      "grad_norm": 3.748638377487623e-08,
      "learning_rate": 9.299054137029397e-06,
      "loss": 0.0001,
      "step": 10170
    },
    {
      "epoch": 0.070232567766149,
      "grad_norm": 0.0,
      "learning_rate": 9.298364229683953e-06,
      "loss": 0.0,
      "step": 10180
    },
    {
      "epoch": 0.07030155850069336,
      "grad_norm": 1.876984242699109e-05,
      "learning_rate": 9.29767432233851e-06,
      "loss": 0.0285,
      "step": 10190
    },
    {
      "epoch": 0.07037054923523771,
      "grad_norm": 0.0,
      "learning_rate": 9.296984414993068e-06,
      "loss": 0.0,
      "step": 10200
    },
    {
      "epoch": 0.07043953996978206,
      "grad_norm": 0.0,
      "learning_rate": 9.296294507647624e-06,
      "loss": 0.0,
      "step": 10210
    },
    {
      "epoch": 0.0705085307043264,
      "grad_norm": 1.489681289967848e-05,
      "learning_rate": 9.295604600302181e-06,
      "loss": 0.0067,
      "step": 10220
    },
    {
      "epoch": 0.07057752143887076,
      "grad_norm": 2.6676072462095135e-09,
      "learning_rate": 9.294914692956737e-06,
      "loss": 0.0,
      "step": 10230
    },
    {
      "epoch": 0.07064651217341511,
      "grad_norm": 0.00038919990765862167,
      "learning_rate": 9.294224785611294e-06,
      "loss": 0.0001,
      "step": 10240
    },
    {
      "epoch": 0.07071550290795946,
      "grad_norm": 0.0,
      "learning_rate": 9.29353487826585e-06,
      "loss": 0.0,
      "step": 10250
    },
    {
      "epoch": 0.07078449364250382,
      "grad_norm": 0.0,
      "learning_rate": 9.292844970920407e-06,
      "loss": 0.0,
      "step": 10260
    },
    {
      "epoch": 0.07085348437704816,
      "grad_norm": 0.0,
      "learning_rate": 9.292155063574962e-06,
      "loss": 0.0001,
      "step": 10270
    },
    {
      "epoch": 0.07092247511159251,
      "grad_norm": 0.0,
      "learning_rate": 9.291465156229518e-06,
      "loss": 0.0,
      "step": 10280
    },
    {
      "epoch": 0.07099146584613686,
      "grad_norm": 0.0,
      "learning_rate": 9.290775248884075e-06,
      "loss": 0.0,
      "step": 10290
    },
    {
      "epoch": 0.07106045658068122,
      "grad_norm": 8.037535735638812e-05,
      "learning_rate": 9.290085341538631e-06,
      "loss": 0.0,
      "step": 10300
    },
    {
      "epoch": 0.07112944731522557,
      "grad_norm": 1.4089010619500186e-05,
      "learning_rate": 9.28939543419319e-06,
      "loss": 0.0,
      "step": 10310
    },
    {
      "epoch": 0.07119843804976991,
      "grad_norm": 9.730531047580371e-08,
      "learning_rate": 9.288705526847746e-06,
      "loss": 0.0266,
      "step": 10320
    },
    {
      "epoch": 0.07126742878431426,
      "grad_norm": 0.0,
      "learning_rate": 9.288015619502302e-06,
      "loss": 0.0139,
      "step": 10330
    },
    {
      "epoch": 0.07133641951885862,
      "grad_norm": 0.0,
      "learning_rate": 9.287325712156859e-06,
      "loss": 0.0,
      "step": 10340
    },
    {
      "epoch": 0.07140541025340297,
      "grad_norm": 0.0,
      "learning_rate": 9.286635804811415e-06,
      "loss": 0.0,
      "step": 10350
    },
    {
      "epoch": 0.07147440098794731,
      "grad_norm": 0.0,
      "learning_rate": 9.285945897465972e-06,
      "loss": 0.0,
      "step": 10360
    },
    {
      "epoch": 0.07154339172249168,
      "grad_norm": 0.0,
      "learning_rate": 9.285255990120528e-06,
      "loss": 0.0,
      "step": 10370
    },
    {
      "epoch": 0.07161238245703602,
      "grad_norm": 0.10475505143404007,
      "learning_rate": 9.284566082775085e-06,
      "loss": 0.0,
      "step": 10380
    },
    {
      "epoch": 0.07168137319158037,
      "grad_norm": 0.0,
      "learning_rate": 9.28387617542964e-06,
      "loss": 0.0,
      "step": 10390
    },
    {
      "epoch": 0.07175036392612472,
      "grad_norm": 0.0,
      "learning_rate": 9.283186268084196e-06,
      "loss": 0.0,
      "step": 10400
    },
    {
      "epoch": 0.07181935466066908,
      "grad_norm": 0.0,
      "learning_rate": 9.282496360738752e-06,
      "loss": 0.0,
      "step": 10410
    },
    {
      "epoch": 0.07188834539521342,
      "grad_norm": 0.039523545652627945,
      "learning_rate": 9.28180645339331e-06,
      "loss": 0.0,
      "step": 10420
    },
    {
      "epoch": 0.07195733612975777,
      "grad_norm": 0.000720381794963032,
      "learning_rate": 9.281116546047867e-06,
      "loss": 0.0001,
      "step": 10430
    },
    {
      "epoch": 0.07202632686430212,
      "grad_norm": 2.1007210762036266e-06,
      "learning_rate": 9.280426638702424e-06,
      "loss": 0.0,
      "step": 10440
    },
    {
      "epoch": 0.07209531759884648,
      "grad_norm": 0.8526678681373596,
      "learning_rate": 9.27973673135698e-06,
      "loss": 0.0002,
      "step": 10450
    },
    {
      "epoch": 0.07216430833339083,
      "grad_norm": 53.298274993896484,
      "learning_rate": 9.279046824011537e-06,
      "loss": 0.0058,
      "step": 10460
    },
    {
      "epoch": 0.07223329906793517,
      "grad_norm": 2.7131068502939115e-09,
      "learning_rate": 9.278356916666093e-06,
      "loss": 0.0,
      "step": 10470
    },
    {
      "epoch": 0.07230228980247953,
      "grad_norm": 0.0,
      "learning_rate": 9.27766700932065e-06,
      "loss": 0.0001,
      "step": 10480
    },
    {
      "epoch": 0.07237128053702388,
      "grad_norm": 0.00017692340770736337,
      "learning_rate": 9.276977101975206e-06,
      "loss": 1.0219,
      "step": 10490
    },
    {
      "epoch": 0.07244027127156823,
      "grad_norm": 0.06454253196716309,
      "learning_rate": 9.276287194629761e-06,
      "loss": 0.0,
      "step": 10500
    },
    {
      "epoch": 0.07250926200611257,
      "grad_norm": 0.33074289560317993,
      "learning_rate": 9.275597287284317e-06,
      "loss": 0.0001,
      "step": 10510
    },
    {
      "epoch": 0.07257825274065693,
      "grad_norm": 0.0,
      "learning_rate": 9.274907379938874e-06,
      "loss": 0.0,
      "step": 10520
    },
    {
      "epoch": 0.07264724347520128,
      "grad_norm": 0.0,
      "learning_rate": 9.274217472593432e-06,
      "loss": 0.0,
      "step": 10530
    },
    {
      "epoch": 0.07271623420974563,
      "grad_norm": 5.425266014214003e-09,
      "learning_rate": 9.273527565247989e-06,
      "loss": 0.1383,
      "step": 10540
    },
    {
      "epoch": 0.07278522494428998,
      "grad_norm": 0.0,
      "learning_rate": 9.272837657902545e-06,
      "loss": 0.0,
      "step": 10550
    },
    {
      "epoch": 0.07285421567883434,
      "grad_norm": 0.0010381789179518819,
      "learning_rate": 9.272147750557102e-06,
      "loss": 0.0,
      "step": 10560
    },
    {
      "epoch": 0.07292320641337868,
      "grad_norm": 0.0001351818209514022,
      "learning_rate": 9.271457843211658e-06,
      "loss": 0.0,
      "step": 10570
    },
    {
      "epoch": 0.07299219714792303,
      "grad_norm": 125.50298309326172,
      "learning_rate": 9.270767935866215e-06,
      "loss": 0.0289,
      "step": 10580
    },
    {
      "epoch": 0.07306118788246739,
      "grad_norm": 0.00037610583240166306,
      "learning_rate": 9.270078028520771e-06,
      "loss": 0.0,
      "step": 10590
    },
    {
      "epoch": 0.07313017861701174,
      "grad_norm": 0.0,
      "learning_rate": 9.269388121175327e-06,
      "loss": 0.0,
      "step": 10600
    },
    {
      "epoch": 0.07319916935155608,
      "grad_norm": 0.0,
      "learning_rate": 9.268698213829882e-06,
      "loss": 0.0,
      "step": 10610
    },
    {
      "epoch": 0.07326816008610043,
      "grad_norm": 3.6756453482666984e-05,
      "learning_rate": 9.268008306484439e-06,
      "loss": 0.0003,
      "step": 10620
    },
    {
      "epoch": 0.07333715082064479,
      "grad_norm": 0.8746658563613892,
      "learning_rate": 9.267318399138995e-06,
      "loss": 0.0014,
      "step": 10630
    },
    {
      "epoch": 0.07340614155518914,
      "grad_norm": 1.756311576173175e-05,
      "learning_rate": 9.266628491793553e-06,
      "loss": 0.0,
      "step": 10640
    },
    {
      "epoch": 0.07347513228973349,
      "grad_norm": 0.0003972606500610709,
      "learning_rate": 9.26593858444811e-06,
      "loss": 0.0024,
      "step": 10650
    },
    {
      "epoch": 0.07354412302427783,
      "grad_norm": 1.6024492651922628e-05,
      "learning_rate": 9.265248677102666e-06,
      "loss": 0.0,
      "step": 10660
    },
    {
      "epoch": 0.0736131137588222,
      "grad_norm": 3.27560119330883e-05,
      "learning_rate": 9.264558769757223e-06,
      "loss": 0.0,
      "step": 10670
    },
    {
      "epoch": 0.07368210449336654,
      "grad_norm": 0.0,
      "learning_rate": 9.26386886241178e-06,
      "loss": 0.0007,
      "step": 10680
    },
    {
      "epoch": 0.07375109522791089,
      "grad_norm": 3.7856531143188477,
      "learning_rate": 9.263178955066336e-06,
      "loss": 0.0015,
      "step": 10690
    },
    {
      "epoch": 0.07382008596245525,
      "grad_norm": 0.00032519918750040233,
      "learning_rate": 9.262489047720892e-06,
      "loss": 0.0,
      "step": 10700
    },
    {
      "epoch": 0.0738890766969996,
      "grad_norm": 0.0,
      "learning_rate": 9.261799140375449e-06,
      "loss": 0.0,
      "step": 10710
    },
    {
      "epoch": 0.07395806743154394,
      "grad_norm": 0.0,
      "learning_rate": 9.261109233030005e-06,
      "loss": 0.4078,
      "step": 10720
    },
    {
      "epoch": 0.07402705816608829,
      "grad_norm": 2.287138256562571e-09,
      "learning_rate": 9.26041932568456e-06,
      "loss": 0.0,
      "step": 10730
    },
    {
      "epoch": 0.07409604890063265,
      "grad_norm": 0.0,
      "learning_rate": 9.259729418339117e-06,
      "loss": 0.0,
      "step": 10740
    },
    {
      "epoch": 0.074165039635177,
      "grad_norm": 0.0,
      "learning_rate": 9.259039510993675e-06,
      "loss": 0.0001,
      "step": 10750
    },
    {
      "epoch": 0.07423403036972134,
      "grad_norm": 0.0,
      "learning_rate": 9.258349603648231e-06,
      "loss": 0.0001,
      "step": 10760
    },
    {
      "epoch": 0.0743030211042657,
      "grad_norm": 0.00011291095142951235,
      "learning_rate": 9.257659696302788e-06,
      "loss": 0.0001,
      "step": 10770
    },
    {
      "epoch": 0.07437201183881005,
      "grad_norm": 0.07228134572505951,
      "learning_rate": 9.256969788957344e-06,
      "loss": 0.0001,
      "step": 10780
    },
    {
      "epoch": 0.0744410025733544,
      "grad_norm": 15.591343879699707,
      "learning_rate": 9.2562798816119e-06,
      "loss": 0.0033,
      "step": 10790
    },
    {
      "epoch": 0.07450999330789874,
      "grad_norm": 6.5625176429748535,
      "learning_rate": 9.255589974266457e-06,
      "loss": 0.0017,
      "step": 10800
    },
    {
      "epoch": 0.0745789840424431,
      "grad_norm": 0.0,
      "learning_rate": 9.254900066921014e-06,
      "loss": 0.0,
      "step": 10810
    },
    {
      "epoch": 0.07464797477698745,
      "grad_norm": 0.0,
      "learning_rate": 9.25421015957557e-06,
      "loss": 0.0,
      "step": 10820
    },
    {
      "epoch": 0.0747169655115318,
      "grad_norm": 0.0,
      "learning_rate": 9.253520252230127e-06,
      "loss": 0.0,
      "step": 10830
    },
    {
      "epoch": 0.07478595624607615,
      "grad_norm": 0.0,
      "learning_rate": 9.252830344884682e-06,
      "loss": 0.0005,
      "step": 10840
    },
    {
      "epoch": 0.07485494698062051,
      "grad_norm": 1.6427664206730697e-07,
      "learning_rate": 9.252140437539238e-06,
      "loss": 0.0,
      "step": 10850
    },
    {
      "epoch": 0.07492393771516485,
      "grad_norm": 0.00334505969658494,
      "learning_rate": 9.251450530193796e-06,
      "loss": 0.0009,
      "step": 10860
    },
    {
      "epoch": 0.0749929284497092,
      "grad_norm": 0.0,
      "learning_rate": 9.250760622848353e-06,
      "loss": 0.0,
      "step": 10870
    },
    {
      "epoch": 0.07506191918425356,
      "grad_norm": 0.0012655858881771564,
      "learning_rate": 9.25007071550291e-06,
      "loss": 0.0044,
      "step": 10880
    },
    {
      "epoch": 0.07513090991879791,
      "grad_norm": 1.8479919816627444e-08,
      "learning_rate": 9.249380808157466e-06,
      "loss": 0.0,
      "step": 10890
    },
    {
      "epoch": 0.07519990065334226,
      "grad_norm": 0.0,
      "learning_rate": 9.248690900812022e-06,
      "loss": 0.0,
      "step": 10900
    },
    {
      "epoch": 0.0752688913878866,
      "grad_norm": 4.5717191143879177e-10,
      "learning_rate": 9.248000993466579e-06,
      "loss": 0.0,
      "step": 10910
    },
    {
      "epoch": 0.07533788212243096,
      "grad_norm": 0.0,
      "learning_rate": 9.247311086121135e-06,
      "loss": 0.168,
      "step": 10920
    },
    {
      "epoch": 0.07540687285697531,
      "grad_norm": 0.07586784660816193,
      "learning_rate": 9.246621178775692e-06,
      "loss": 0.0,
      "step": 10930
    },
    {
      "epoch": 0.07547586359151966,
      "grad_norm": 0.0,
      "learning_rate": 9.245931271430248e-06,
      "loss": 0.0,
      "step": 10940
    },
    {
      "epoch": 0.075544854326064,
      "grad_norm": 0.0,
      "learning_rate": 9.245241364084803e-06,
      "loss": 0.0,
      "step": 10950
    },
    {
      "epoch": 0.07561384506060836,
      "grad_norm": 0.0,
      "learning_rate": 9.24455145673936e-06,
      "loss": 0.0001,
      "step": 10960
    },
    {
      "epoch": 0.07568283579515271,
      "grad_norm": 0.0,
      "learning_rate": 9.243861549393918e-06,
      "loss": 0.0,
      "step": 10970
    },
    {
      "epoch": 0.07575182652969706,
      "grad_norm": 0.0,
      "learning_rate": 9.243171642048474e-06,
      "loss": 0.0004,
      "step": 10980
    },
    {
      "epoch": 0.07582081726424142,
      "grad_norm": 0.0,
      "learning_rate": 9.24248173470303e-06,
      "loss": 0.4141,
      "step": 10990
    },
    {
      "epoch": 0.07588980799878577,
      "grad_norm": 0.0,
      "learning_rate": 9.241791827357587e-06,
      "loss": 0.0,
      "step": 11000
    },
    {
      "epoch": 0.07595879873333011,
      "grad_norm": 0.0,
      "learning_rate": 9.241101920012144e-06,
      "loss": 0.0294,
      "step": 11010
    },
    {
      "epoch": 0.07602778946787446,
      "grad_norm": 8.191610589847187e-08,
      "learning_rate": 9.2404120126667e-06,
      "loss": 0.0,
      "step": 11020
    },
    {
      "epoch": 0.07609678020241882,
      "grad_norm": 0.0,
      "learning_rate": 9.239722105321257e-06,
      "loss": 0.0,
      "step": 11030
    },
    {
      "epoch": 0.07616577093696317,
      "grad_norm": 0.0,
      "learning_rate": 9.239032197975813e-06,
      "loss": 0.0,
      "step": 11040
    },
    {
      "epoch": 0.07623476167150751,
      "grad_norm": 3.5035125620197505e-05,
      "learning_rate": 9.23834229063037e-06,
      "loss": 0.0,
      "step": 11050
    },
    {
      "epoch": 0.07630375240605186,
      "grad_norm": 0.0,
      "learning_rate": 9.237652383284924e-06,
      "loss": 0.0015,
      "step": 11060
    },
    {
      "epoch": 0.07637274314059622,
      "grad_norm": 0.0,
      "learning_rate": 9.23696247593948e-06,
      "loss": 0.0,
      "step": 11070
    },
    {
      "epoch": 0.07644173387514057,
      "grad_norm": 0.0,
      "learning_rate": 9.236341559328583e-06,
      "loss": 0.2465,
      "step": 11080
    },
    {
      "epoch": 0.07651072460968492,
      "grad_norm": 0.0,
      "learning_rate": 9.23565165198314e-06,
      "loss": 0.0,
      "step": 11090
    },
    {
      "epoch": 0.07657971534422928,
      "grad_norm": 0.07076693326234818,
      "learning_rate": 9.234961744637696e-06,
      "loss": 0.0001,
      "step": 11100
    },
    {
      "epoch": 0.07664870607877362,
      "grad_norm": 0.0,
      "learning_rate": 9.234271837292253e-06,
      "loss": 0.0,
      "step": 11110
    },
    {
      "epoch": 0.07671769681331797,
      "grad_norm": 0.0,
      "learning_rate": 9.23358192994681e-06,
      "loss": 0.1383,
      "step": 11120
    },
    {
      "epoch": 0.07678668754786232,
      "grad_norm": 1.263082822333672e-06,
      "learning_rate": 9.232892022601366e-06,
      "loss": 0.0,
      "step": 11130
    },
    {
      "epoch": 0.07685567828240668,
      "grad_norm": 0.0,
      "learning_rate": 9.232202115255922e-06,
      "loss": 0.0,
      "step": 11140
    },
    {
      "epoch": 0.07692466901695102,
      "grad_norm": 0.0006859014392830431,
      "learning_rate": 9.231512207910479e-06,
      "loss": 0.0,
      "step": 11150
    },
    {
      "epoch": 0.07699365975149537,
      "grad_norm": 208.89547729492188,
      "learning_rate": 9.230822300565035e-06,
      "loss": 0.0165,
      "step": 11160
    },
    {
      "epoch": 0.07706265048603972,
      "grad_norm": 0.0,
      "learning_rate": 9.230132393219592e-06,
      "loss": 0.0,
      "step": 11170
    },
    {
      "epoch": 0.07713164122058408,
      "grad_norm": 0.00013802248577121645,
      "learning_rate": 9.229442485874148e-06,
      "loss": 0.0,
      "step": 11180
    },
    {
      "epoch": 0.07720063195512843,
      "grad_norm": 0.0,
      "learning_rate": 9.228752578528705e-06,
      "loss": 0.0241,
      "step": 11190
    },
    {
      "epoch": 0.07726962268967277,
      "grad_norm": 0.0,
      "learning_rate": 9.228062671183261e-06,
      "loss": 0.376,
      "step": 11200
    },
    {
      "epoch": 0.07733861342421713,
      "grad_norm": 9.614611329311629e-10,
      "learning_rate": 9.227372763837818e-06,
      "loss": 0.0,
      "step": 11210
    },
    {
      "epoch": 0.07740760415876148,
      "grad_norm": 0.0,
      "learning_rate": 9.226682856492374e-06,
      "loss": 0.002,
      "step": 11220
    },
    {
      "epoch": 0.07747659489330583,
      "grad_norm": 0.0,
      "learning_rate": 9.22599294914693e-06,
      "loss": 0.0,
      "step": 11230
    },
    {
      "epoch": 0.07754558562785017,
      "grad_norm": 0.0,
      "learning_rate": 9.225303041801487e-06,
      "loss": 0.0,
      "step": 11240
    },
    {
      "epoch": 0.07761457636239454,
      "grad_norm": 0.0,
      "learning_rate": 9.224613134456044e-06,
      "loss": 0.0057,
      "step": 11250
    },
    {
      "epoch": 0.07768356709693888,
      "grad_norm": 1.3128318786621094,
      "learning_rate": 9.2239232271106e-06,
      "loss": 0.0003,
      "step": 11260
    },
    {
      "epoch": 0.07775255783148323,
      "grad_norm": 0.0,
      "learning_rate": 9.223233319765157e-06,
      "loss": 0.0,
      "step": 11270
    },
    {
      "epoch": 0.07782154856602758,
      "grad_norm": 0.0,
      "learning_rate": 9.222543412419713e-06,
      "loss": 0.0,
      "step": 11280
    },
    {
      "epoch": 0.07789053930057194,
      "grad_norm": 0.0,
      "learning_rate": 9.22185350507427e-06,
      "loss": 0.0001,
      "step": 11290
    },
    {
      "epoch": 0.07795953003511628,
      "grad_norm": 0.0,
      "learning_rate": 9.221163597728826e-06,
      "loss": 0.0003,
      "step": 11300
    },
    {
      "epoch": 0.07802852076966063,
      "grad_norm": 0.0,
      "learning_rate": 9.220473690383382e-06,
      "loss": 0.0,
      "step": 11310
    },
    {
      "epoch": 0.07809751150420499,
      "grad_norm": 710.934814453125,
      "learning_rate": 9.219783783037939e-06,
      "loss": 0.281,
      "step": 11320
    },
    {
      "epoch": 0.07816650223874934,
      "grad_norm": 0.0,
      "learning_rate": 9.219093875692495e-06,
      "loss": 0.0023,
      "step": 11330
    },
    {
      "epoch": 0.07823549297329369,
      "grad_norm": 9.878846629618465e-10,
      "learning_rate": 9.218403968347052e-06,
      "loss": 0.0002,
      "step": 11340
    },
    {
      "epoch": 0.07830448370783803,
      "grad_norm": 0.0,
      "learning_rate": 9.217714061001608e-06,
      "loss": 0.0,
      "step": 11350
    },
    {
      "epoch": 0.0783734744423824,
      "grad_norm": 2.8398371796356514e-06,
      "learning_rate": 9.217024153656165e-06,
      "loss": 0.0,
      "step": 11360
    },
    {
      "epoch": 0.07844246517692674,
      "grad_norm": 0.0021105895284563303,
      "learning_rate": 9.216334246310721e-06,
      "loss": 0.0,
      "step": 11370
    },
    {
      "epoch": 0.07851145591147109,
      "grad_norm": 0.0,
      "learning_rate": 9.215644338965278e-06,
      "loss": 0.0009,
      "step": 11380
    },
    {
      "epoch": 0.07858044664601543,
      "grad_norm": 0.0,
      "learning_rate": 9.214954431619834e-06,
      "loss": 0.0,
      "step": 11390
    },
    {
      "epoch": 0.0786494373805598,
      "grad_norm": 3.435726370071279e-08,
      "learning_rate": 9.214264524274391e-06,
      "loss": 0.0,
      "step": 11400
    },
    {
      "epoch": 0.07871842811510414,
      "grad_norm": 0.0,
      "learning_rate": 9.213574616928947e-06,
      "loss": 0.0004,
      "step": 11410
    },
    {
      "epoch": 0.07878741884964849,
      "grad_norm": 3.9642894989810884e-05,
      "learning_rate": 9.212884709583504e-06,
      "loss": 0.0666,
      "step": 11420
    },
    {
      "epoch": 0.07885640958419285,
      "grad_norm": 0.0,
      "learning_rate": 9.21219480223806e-06,
      "loss": 0.0001,
      "step": 11430
    },
    {
      "epoch": 0.0789254003187372,
      "grad_norm": 0.0,
      "learning_rate": 9.211504894892617e-06,
      "loss": 0.1877,
      "step": 11440
    },
    {
      "epoch": 0.07899439105328154,
      "grad_norm": 0.0,
      "learning_rate": 9.210814987547173e-06,
      "loss": 0.0,
      "step": 11450
    },
    {
      "epoch": 0.07906338178782589,
      "grad_norm": 0.0,
      "learning_rate": 9.21012508020173e-06,
      "loss": 0.0,
      "step": 11460
    },
    {
      "epoch": 0.07913237252237025,
      "grad_norm": 0.0,
      "learning_rate": 9.209435172856286e-06,
      "loss": 0.0,
      "step": 11470
    },
    {
      "epoch": 0.0792013632569146,
      "grad_norm": 0.0,
      "learning_rate": 9.208745265510843e-06,
      "loss": 0.0,
      "step": 11480
    },
    {
      "epoch": 0.07927035399145894,
      "grad_norm": 0.0,
      "learning_rate": 9.2080553581654e-06,
      "loss": 0.0001,
      "step": 11490
    },
    {
      "epoch": 0.07933934472600329,
      "grad_norm": 0.0,
      "learning_rate": 9.207365450819956e-06,
      "loss": 0.0011,
      "step": 11500
    },
    {
      "epoch": 0.07940833546054765,
      "grad_norm": 0.0009103399352170527,
      "learning_rate": 9.206675543474512e-06,
      "loss": 0.0004,
      "step": 11510
    },
    {
      "epoch": 0.079477326195092,
      "grad_norm": 0.0,
      "learning_rate": 9.205985636129069e-06,
      "loss": 0.0,
      "step": 11520
    },
    {
      "epoch": 0.07954631692963635,
      "grad_norm": 0.0,
      "learning_rate": 9.205295728783625e-06,
      "loss": 0.0,
      "step": 11530
    },
    {
      "epoch": 0.0796153076641807,
      "grad_norm": 0.0,
      "learning_rate": 9.204605821438182e-06,
      "loss": 0.0,
      "step": 11540
    },
    {
      "epoch": 0.07968429839872505,
      "grad_norm": 0.00013076679897494614,
      "learning_rate": 9.203915914092738e-06,
      "loss": 0.0,
      "step": 11550
    },
    {
      "epoch": 0.0797532891332694,
      "grad_norm": 0.0,
      "learning_rate": 9.203226006747295e-06,
      "loss": 0.0,
      "step": 11560
    },
    {
      "epoch": 0.07982227986781375,
      "grad_norm": 0.0,
      "learning_rate": 9.202536099401851e-06,
      "loss": 0.0,
      "step": 11570
    },
    {
      "epoch": 0.07989127060235811,
      "grad_norm": 7.369824743364006e-05,
      "learning_rate": 9.201846192056408e-06,
      "loss": 0.0,
      "step": 11580
    },
    {
      "epoch": 0.07996026133690246,
      "grad_norm": 0.0,
      "learning_rate": 9.201156284710964e-06,
      "loss": 0.0019,
      "step": 11590
    },
    {
      "epoch": 0.0800292520714468,
      "grad_norm": 0.0452035553753376,
      "learning_rate": 9.20046637736552e-06,
      "loss": 0.069,
      "step": 11600
    },
    {
      "epoch": 0.08009824280599115,
      "grad_norm": 4.348968474232606e-08,
      "learning_rate": 9.199776470020077e-06,
      "loss": 0.0,
      "step": 11610
    },
    {
      "epoch": 0.08016723354053551,
      "grad_norm": 0.0,
      "learning_rate": 9.199086562674634e-06,
      "loss": 0.0,
      "step": 11620
    },
    {
      "epoch": 0.08023622427507986,
      "grad_norm": 0.0,
      "learning_rate": 9.19839665532919e-06,
      "loss": 0.0001,
      "step": 11630
    },
    {
      "epoch": 0.0803052150096242,
      "grad_norm": 0.00019220069225411862,
      "learning_rate": 9.197706747983747e-06,
      "loss": 0.0,
      "step": 11640
    },
    {
      "epoch": 0.08037420574416856,
      "grad_norm": 8.480259339194163e-08,
      "learning_rate": 9.197016840638303e-06,
      "loss": 0.0,
      "step": 11650
    },
    {
      "epoch": 0.08044319647871291,
      "grad_norm": 0.0,
      "learning_rate": 9.19632693329286e-06,
      "loss": 0.0005,
      "step": 11660
    },
    {
      "epoch": 0.08051218721325726,
      "grad_norm": 0.0007987026474438608,
      "learning_rate": 9.195637025947416e-06,
      "loss": 0.0003,
      "step": 11670
    },
    {
      "epoch": 0.0805811779478016,
      "grad_norm": 0.0,
      "learning_rate": 9.194947118601973e-06,
      "loss": 0.0,
      "step": 11680
    },
    {
      "epoch": 0.08065016868234597,
      "grad_norm": 1.3973802197142504e-05,
      "learning_rate": 9.194257211256529e-06,
      "loss": 0.0,
      "step": 11690
    },
    {
      "epoch": 0.08071915941689031,
      "grad_norm": 0.0,
      "learning_rate": 9.193567303911086e-06,
      "loss": 0.0,
      "step": 11700
    },
    {
      "epoch": 0.08078815015143466,
      "grad_norm": 0.0,
      "learning_rate": 9.192877396565642e-06,
      "loss": 0.0,
      "step": 11710
    },
    {
      "epoch": 0.080857140885979,
      "grad_norm": 0.0,
      "learning_rate": 9.192187489220199e-06,
      "loss": 0.0023,
      "step": 11720
    },
    {
      "epoch": 0.08092613162052337,
      "grad_norm": 0.0,
      "learning_rate": 9.191497581874755e-06,
      "loss": 0.0,
      "step": 11730
    },
    {
      "epoch": 0.08099512235506771,
      "grad_norm": 0.0022115407045930624,
      "learning_rate": 9.190807674529312e-06,
      "loss": 0.0,
      "step": 11740
    },
    {
      "epoch": 0.08106411308961206,
      "grad_norm": 0.0,
      "learning_rate": 9.190117767183868e-06,
      "loss": 0.0143,
      "step": 11750
    },
    {
      "epoch": 0.08113310382415642,
      "grad_norm": 6.433498583646724e-06,
      "learning_rate": 9.189427859838425e-06,
      "loss": 0.0066,
      "step": 11760
    },
    {
      "epoch": 0.08120209455870077,
      "grad_norm": 0.0,
      "learning_rate": 9.188737952492981e-06,
      "loss": 0.0,
      "step": 11770
    },
    {
      "epoch": 0.08127108529324512,
      "grad_norm": 0.0,
      "learning_rate": 9.188048045147538e-06,
      "loss": 0.0697,
      "step": 11780
    },
    {
      "epoch": 0.08134007602778946,
      "grad_norm": 0.001297208247706294,
      "learning_rate": 9.187358137802094e-06,
      "loss": 0.0449,
      "step": 11790
    },
    {
      "epoch": 0.08140906676233382,
      "grad_norm": 0.0,
      "learning_rate": 9.18666823045665e-06,
      "loss": 0.0257,
      "step": 11800
    },
    {
      "epoch": 0.08147805749687817,
      "grad_norm": 0.0,
      "learning_rate": 9.185978323111207e-06,
      "loss": 0.2309,
      "step": 11810
    },
    {
      "epoch": 0.08154704823142252,
      "grad_norm": 0.024593088775873184,
      "learning_rate": 9.185288415765763e-06,
      "loss": 0.0,
      "step": 11820
    },
    {
      "epoch": 0.08161603896596686,
      "grad_norm": 0.0,
      "learning_rate": 9.18459850842032e-06,
      "loss": 0.8766,
      "step": 11830
    },
    {
      "epoch": 0.08168502970051122,
      "grad_norm": 5.3823820053366944e-05,
      "learning_rate": 9.183908601074876e-06,
      "loss": 0.6586,
      "step": 11840
    },
    {
      "epoch": 0.08175402043505557,
      "grad_norm": 0.0,
      "learning_rate": 9.183218693729433e-06,
      "loss": 0.2819,
      "step": 11850
    },
    {
      "epoch": 0.08182301116959992,
      "grad_norm": 0.0,
      "learning_rate": 9.18252878638399e-06,
      "loss": 0.0006,
      "step": 11860
    },
    {
      "epoch": 0.08189200190414428,
      "grad_norm": 0.0,
      "learning_rate": 9.181838879038546e-06,
      "loss": 0.0024,
      "step": 11870
    },
    {
      "epoch": 0.08196099263868863,
      "grad_norm": 0.0,
      "learning_rate": 9.181148971693102e-06,
      "loss": 0.0001,
      "step": 11880
    },
    {
      "epoch": 0.08202998337323297,
      "grad_norm": 0.0,
      "learning_rate": 9.180459064347659e-06,
      "loss": 0.0003,
      "step": 11890
    },
    {
      "epoch": 0.08209897410777732,
      "grad_norm": 0.0,
      "learning_rate": 9.179769157002215e-06,
      "loss": 0.0,
      "step": 11900
    },
    {
      "epoch": 0.08216796484232168,
      "grad_norm": 0.0,
      "learning_rate": 9.179079249656772e-06,
      "loss": 0.0,
      "step": 11910
    },
    {
      "epoch": 0.08223695557686603,
      "grad_norm": 0.0,
      "learning_rate": 9.178389342311328e-06,
      "loss": 0.0,
      "step": 11920
    },
    {
      "epoch": 0.08230594631141037,
      "grad_norm": 0.0040501696057617664,
      "learning_rate": 9.177699434965885e-06,
      "loss": 0.0,
      "step": 11930
    },
    {
      "epoch": 0.08237493704595472,
      "grad_norm": 0.0,
      "learning_rate": 9.177009527620441e-06,
      "loss": 0.0,
      "step": 11940
    },
    {
      "epoch": 0.08244392778049908,
      "grad_norm": 154.3331756591797,
      "learning_rate": 9.176319620274998e-06,
      "loss": 0.0255,
      "step": 11950
    },
    {
      "epoch": 0.08251291851504343,
      "grad_norm": 0.0,
      "learning_rate": 9.175629712929554e-06,
      "loss": 0.0003,
      "step": 11960
    },
    {
      "epoch": 0.08258190924958778,
      "grad_norm": 8.946063889681e-10,
      "learning_rate": 9.174939805584111e-06,
      "loss": 0.0848,
      "step": 11970
    },
    {
      "epoch": 0.08265089998413214,
      "grad_norm": 17.25187110900879,
      "learning_rate": 9.174249898238667e-06,
      "loss": 0.0018,
      "step": 11980
    },
    {
      "epoch": 0.08271989071867648,
      "grad_norm": 0.0,
      "learning_rate": 9.173559990893224e-06,
      "loss": 0.0002,
      "step": 11990
    },
    {
      "epoch": 0.08278888145322083,
      "grad_norm": 0.0,
      "learning_rate": 9.17287008354778e-06,
      "loss": 0.0096,
      "step": 12000
    },
    {
      "epoch": 0.08285787218776518,
      "grad_norm": 0.0,
      "learning_rate": 9.172180176202337e-06,
      "loss": 0.0,
      "step": 12010
    },
    {
      "epoch": 0.08292686292230954,
      "grad_norm": 0.0,
      "learning_rate": 9.171490268856893e-06,
      "loss": 0.6281,
      "step": 12020
    },
    {
      "epoch": 0.08299585365685389,
      "grad_norm": 0.0,
      "learning_rate": 9.17080036151145e-06,
      "loss": 0.0003,
      "step": 12030
    },
    {
      "epoch": 0.08306484439139823,
      "grad_norm": 0.0,
      "learning_rate": 9.170110454166006e-06,
      "loss": 0.0,
      "step": 12040
    },
    {
      "epoch": 0.08313383512594259,
      "grad_norm": 0.0,
      "learning_rate": 9.169420546820563e-06,
      "loss": 0.0,
      "step": 12050
    },
    {
      "epoch": 0.08320282586048694,
      "grad_norm": 0.0,
      "learning_rate": 9.16873063947512e-06,
      "loss": 0.0,
      "step": 12060
    },
    {
      "epoch": 0.08327181659503129,
      "grad_norm": 0.0,
      "learning_rate": 9.168040732129676e-06,
      "loss": 0.0064,
      "step": 12070
    },
    {
      "epoch": 0.08334080732957563,
      "grad_norm": 0.0,
      "learning_rate": 9.167350824784232e-06,
      "loss": 0.0004,
      "step": 12080
    },
    {
      "epoch": 0.08340979806412,
      "grad_norm": 0.05091177672147751,
      "learning_rate": 9.166660917438789e-06,
      "loss": 0.0,
      "step": 12090
    },
    {
      "epoch": 0.08347878879866434,
      "grad_norm": 0.007914368063211441,
      "learning_rate": 9.165971010093345e-06,
      "loss": 0.0003,
      "step": 12100
    },
    {
      "epoch": 0.08354777953320869,
      "grad_norm": 0.0,
      "learning_rate": 9.165281102747902e-06,
      "loss": 0.2137,
      "step": 12110
    },
    {
      "epoch": 0.08361677026775304,
      "grad_norm": 0.0,
      "learning_rate": 9.164591195402458e-06,
      "loss": 0.0,
      "step": 12120
    },
    {
      "epoch": 0.0836857610022974,
      "grad_norm": 0.0,
      "learning_rate": 9.163901288057015e-06,
      "loss": 0.0131,
      "step": 12130
    },
    {
      "epoch": 0.08375475173684174,
      "grad_norm": 0.0020862892270088196,
      "learning_rate": 9.163211380711571e-06,
      "loss": 0.0009,
      "step": 12140
    },
    {
      "epoch": 0.08382374247138609,
      "grad_norm": 0.0,
      "learning_rate": 9.162521473366128e-06,
      "loss": 0.0,
      "step": 12150
    },
    {
      "epoch": 0.08389273320593045,
      "grad_norm": 1.7016177480400074e-06,
      "learning_rate": 9.161831566020684e-06,
      "loss": 0.0,
      "step": 12160
    },
    {
      "epoch": 0.0839617239404748,
      "grad_norm": 0.0,
      "learning_rate": 9.16114165867524e-06,
      "loss": 0.001,
      "step": 12170
    },
    {
      "epoch": 0.08403071467501914,
      "grad_norm": 0.0,
      "learning_rate": 9.160451751329797e-06,
      "loss": 0.0001,
      "step": 12180
    },
    {
      "epoch": 0.08409970540956349,
      "grad_norm": 0.0,
      "learning_rate": 9.159761843984354e-06,
      "loss": 0.0001,
      "step": 12190
    },
    {
      "epoch": 0.08416869614410785,
      "grad_norm": 0.00013891565322410315,
      "learning_rate": 9.15907193663891e-06,
      "loss": 0.0,
      "step": 12200
    },
    {
      "epoch": 0.0842376868786522,
      "grad_norm": 0.0,
      "learning_rate": 9.158382029293467e-06,
      "loss": 0.0,
      "step": 12210
    },
    {
      "epoch": 0.08430667761319655,
      "grad_norm": 0.0,
      "learning_rate": 9.157692121948023e-06,
      "loss": 0.0,
      "step": 12220
    },
    {
      "epoch": 0.08437566834774089,
      "grad_norm": 0.0,
      "learning_rate": 9.15700221460258e-06,
      "loss": 0.0,
      "step": 12230
    },
    {
      "epoch": 0.08444465908228525,
      "grad_norm": 0.0,
      "learning_rate": 9.156312307257136e-06,
      "loss": 0.0,
      "step": 12240
    },
    {
      "epoch": 0.0845136498168296,
      "grad_norm": 0.0,
      "learning_rate": 9.155622399911693e-06,
      "loss": 0.0,
      "step": 12250
    },
    {
      "epoch": 0.08458264055137395,
      "grad_norm": 4.975483580693663e-09,
      "learning_rate": 9.154932492566249e-06,
      "loss": 0.1714,
      "step": 12260
    },
    {
      "epoch": 0.08465163128591831,
      "grad_norm": 0.0,
      "learning_rate": 9.154242585220806e-06,
      "loss": 0.0,
      "step": 12270
    },
    {
      "epoch": 0.08472062202046265,
      "grad_norm": 0.0,
      "learning_rate": 9.153552677875362e-06,
      "loss": 0.0,
      "step": 12280
    },
    {
      "epoch": 0.084789612755007,
      "grad_norm": 0.0,
      "learning_rate": 9.152862770529919e-06,
      "loss": 0.0,
      "step": 12290
    },
    {
      "epoch": 0.08485860348955135,
      "grad_norm": 0.00012580711336340755,
      "learning_rate": 9.152172863184475e-06,
      "loss": 0.0061,
      "step": 12300
    },
    {
      "epoch": 0.08492759422409571,
      "grad_norm": 0.0,
      "learning_rate": 9.151482955839032e-06,
      "loss": 0.0,
      "step": 12310
    },
    {
      "epoch": 0.08499658495864006,
      "grad_norm": 0.0,
      "learning_rate": 9.150793048493588e-06,
      "loss": 0.0,
      "step": 12320
    },
    {
      "epoch": 0.0850655756931844,
      "grad_norm": 0.0,
      "learning_rate": 9.150103141148145e-06,
      "loss": 0.0,
      "step": 12330
    },
    {
      "epoch": 0.08513456642772875,
      "grad_norm": 0.0,
      "learning_rate": 9.149413233802701e-06,
      "loss": 0.0,
      "step": 12340
    },
    {
      "epoch": 0.08520355716227311,
      "grad_norm": 0.0,
      "learning_rate": 9.148723326457257e-06,
      "loss": 0.0,
      "step": 12350
    },
    {
      "epoch": 0.08527254789681746,
      "grad_norm": 0.0011592888040468097,
      "learning_rate": 9.148033419111814e-06,
      "loss": 0.0106,
      "step": 12360
    },
    {
      "epoch": 0.0853415386313618,
      "grad_norm": 0.0,
      "learning_rate": 9.14734351176637e-06,
      "loss": 0.0,
      "step": 12370
    },
    {
      "epoch": 0.08541052936590617,
      "grad_norm": 0.00942144077271223,
      "learning_rate": 9.146653604420927e-06,
      "loss": 0.0,
      "step": 12380
    },
    {
      "epoch": 0.08547952010045051,
      "grad_norm": 3.95666211261414e-06,
      "learning_rate": 9.145963697075483e-06,
      "loss": 0.0,
      "step": 12390
    },
    {
      "epoch": 0.08554851083499486,
      "grad_norm": 0.0,
      "learning_rate": 9.14527378973004e-06,
      "loss": 0.0,
      "step": 12400
    },
    {
      "epoch": 0.0856175015695392,
      "grad_norm": 0.0,
      "learning_rate": 9.144583882384596e-06,
      "loss": 0.0008,
      "step": 12410
    },
    {
      "epoch": 0.08568649230408357,
      "grad_norm": 0.0026933508925139904,
      "learning_rate": 9.143893975039153e-06,
      "loss": 0.0001,
      "step": 12420
    },
    {
      "epoch": 0.08575548303862791,
      "grad_norm": 1.0207188552158186e-06,
      "learning_rate": 9.14320406769371e-06,
      "loss": 0.1169,
      "step": 12430
    },
    {
      "epoch": 0.08582447377317226,
      "grad_norm": 0.0,
      "learning_rate": 9.142514160348266e-06,
      "loss": 0.0002,
      "step": 12440
    },
    {
      "epoch": 0.08589346450771661,
      "grad_norm": 9.92602777749596e-10,
      "learning_rate": 9.141824253002822e-06,
      "loss": 0.0,
      "step": 12450
    },
    {
      "epoch": 0.08596245524226097,
      "grad_norm": 0.0,
      "learning_rate": 9.141134345657379e-06,
      "loss": 0.0,
      "step": 12460
    },
    {
      "epoch": 0.08603144597680532,
      "grad_norm": 9.948919341695728e-07,
      "learning_rate": 9.140444438311935e-06,
      "loss": 0.0001,
      "step": 12470
    },
    {
      "epoch": 0.08610043671134966,
      "grad_norm": 0.0,
      "learning_rate": 9.139754530966492e-06,
      "loss": 0.0,
      "step": 12480
    },
    {
      "epoch": 0.08616942744589402,
      "grad_norm": 0.0,
      "learning_rate": 9.139064623621048e-06,
      "loss": 0.0,
      "step": 12490
    },
    {
      "epoch": 0.08623841818043837,
      "grad_norm": 0.0,
      "learning_rate": 9.138374716275605e-06,
      "loss": 0.0,
      "step": 12500
    },
    {
      "epoch": 0.08630740891498272,
      "grad_norm": 0.0,
      "learning_rate": 9.137684808930161e-06,
      "loss": 0.0,
      "step": 12510
    },
    {
      "epoch": 0.08637639964952706,
      "grad_norm": 0.0015388409374281764,
      "learning_rate": 9.136994901584718e-06,
      "loss": 0.0,
      "step": 12520
    },
    {
      "epoch": 0.08644539038407142,
      "grad_norm": 7.780840678606182e-05,
      "learning_rate": 9.136304994239274e-06,
      "loss": 0.0,
      "step": 12530
    },
    {
      "epoch": 0.08651438111861577,
      "grad_norm": 0.0,
      "learning_rate": 9.13561508689383e-06,
      "loss": 0.0,
      "step": 12540
    },
    {
      "epoch": 0.08658337185316012,
      "grad_norm": 0.08898631483316422,
      "learning_rate": 9.134925179548387e-06,
      "loss": 0.0,
      "step": 12550
    },
    {
      "epoch": 0.08665236258770447,
      "grad_norm": 0.0,
      "learning_rate": 9.134235272202944e-06,
      "loss": 0.0,
      "step": 12560
    },
    {
      "epoch": 0.08672135332224883,
      "grad_norm": 0.0,
      "learning_rate": 9.1335453648575e-06,
      "loss": 0.0001,
      "step": 12570
    },
    {
      "epoch": 0.08679034405679317,
      "grad_norm": 5.562814209270073e-09,
      "learning_rate": 9.132855457512057e-06,
      "loss": 0.0005,
      "step": 12580
    },
    {
      "epoch": 0.08685933479133752,
      "grad_norm": 0.0,
      "learning_rate": 9.132165550166613e-06,
      "loss": 0.0889,
      "step": 12590
    },
    {
      "epoch": 0.08692832552588188,
      "grad_norm": 0.0,
      "learning_rate": 9.13147564282117e-06,
      "loss": 0.0,
      "step": 12600
    },
    {
      "epoch": 0.08699731626042623,
      "grad_norm": 0.0,
      "learning_rate": 9.130785735475726e-06,
      "loss": 0.0,
      "step": 12610
    },
    {
      "epoch": 0.08706630699497057,
      "grad_norm": 5.688956434823922e-07,
      "learning_rate": 9.130095828130283e-06,
      "loss": 0.0,
      "step": 12620
    },
    {
      "epoch": 0.08713529772951492,
      "grad_norm": 0.0,
      "learning_rate": 9.12940592078484e-06,
      "loss": 0.0,
      "step": 12630
    },
    {
      "epoch": 0.08720428846405928,
      "grad_norm": 0.0,
      "learning_rate": 9.128716013439396e-06,
      "loss": 0.0,
      "step": 12640
    },
    {
      "epoch": 0.08727327919860363,
      "grad_norm": 0.0,
      "learning_rate": 9.128026106093952e-06,
      "loss": 0.0001,
      "step": 12650
    },
    {
      "epoch": 0.08734226993314798,
      "grad_norm": 36.286399841308594,
      "learning_rate": 9.127336198748509e-06,
      "loss": 0.0075,
      "step": 12660
    },
    {
      "epoch": 0.08741126066769232,
      "grad_norm": 0.0,
      "learning_rate": 9.126646291403065e-06,
      "loss": 0.0,
      "step": 12670
    },
    {
      "epoch": 0.08748025140223668,
      "grad_norm": 0.0,
      "learning_rate": 9.125956384057622e-06,
      "loss": 0.0,
      "step": 12680
    },
    {
      "epoch": 0.08754924213678103,
      "grad_norm": 0.0,
      "learning_rate": 9.125266476712178e-06,
      "loss": 0.0,
      "step": 12690
    },
    {
      "epoch": 0.08761823287132538,
      "grad_norm": 1.579142292484903e-07,
      "learning_rate": 9.124576569366735e-06,
      "loss": 0.0,
      "step": 12700
    },
    {
      "epoch": 0.08768722360586974,
      "grad_norm": 0.0,
      "learning_rate": 9.123886662021291e-06,
      "loss": 0.0,
      "step": 12710
    },
    {
      "epoch": 0.08775621434041408,
      "grad_norm": 0.0,
      "learning_rate": 9.123196754675848e-06,
      "loss": 0.0,
      "step": 12720
    },
    {
      "epoch": 0.08782520507495843,
      "grad_norm": 0.0,
      "learning_rate": 9.122506847330404e-06,
      "loss": 0.0,
      "step": 12730
    },
    {
      "epoch": 0.08789419580950278,
      "grad_norm": 0.0,
      "learning_rate": 9.12181693998496e-06,
      "loss": 0.0,
      "step": 12740
    },
    {
      "epoch": 0.08796318654404714,
      "grad_norm": 0.0,
      "learning_rate": 9.121127032639517e-06,
      "loss": 0.0,
      "step": 12750
    },
    {
      "epoch": 0.08803217727859149,
      "grad_norm": 8.586281130540385e-10,
      "learning_rate": 9.120437125294074e-06,
      "loss": 0.0,
      "step": 12760
    },
    {
      "epoch": 0.08810116801313583,
      "grad_norm": 0.0,
      "learning_rate": 9.11974721794863e-06,
      "loss": 0.0,
      "step": 12770
    },
    {
      "epoch": 0.08817015874768018,
      "grad_norm": 0.0,
      "learning_rate": 9.119057310603187e-06,
      "loss": 0.0,
      "step": 12780
    },
    {
      "epoch": 0.08823914948222454,
      "grad_norm": 0.0,
      "learning_rate": 9.118367403257743e-06,
      "loss": 0.0,
      "step": 12790
    },
    {
      "epoch": 0.08830814021676889,
      "grad_norm": 0.0,
      "learning_rate": 9.1176774959123e-06,
      "loss": 0.0,
      "step": 12800
    },
    {
      "epoch": 0.08837713095131323,
      "grad_norm": 0.0,
      "learning_rate": 9.116987588566856e-06,
      "loss": 0.0,
      "step": 12810
    },
    {
      "epoch": 0.0884461216858576,
      "grad_norm": 0.0004610455653164536,
      "learning_rate": 9.116297681221413e-06,
      "loss": 0.0,
      "step": 12820
    },
    {
      "epoch": 0.08851511242040194,
      "grad_norm": 0.0,
      "learning_rate": 9.115607773875969e-06,
      "loss": 0.0006,
      "step": 12830
    },
    {
      "epoch": 0.08858410315494629,
      "grad_norm": 0.0,
      "learning_rate": 9.114917866530526e-06,
      "loss": 0.0001,
      "step": 12840
    },
    {
      "epoch": 0.08865309388949064,
      "grad_norm": 1.3961353033664636e-06,
      "learning_rate": 9.114227959185082e-06,
      "loss": 0.0,
      "step": 12850
    },
    {
      "epoch": 0.088722084624035,
      "grad_norm": 0.0,
      "learning_rate": 9.113538051839638e-06,
      "loss": 0.0,
      "step": 12860
    },
    {
      "epoch": 0.08879107535857934,
      "grad_norm": 0.0,
      "learning_rate": 9.112848144494195e-06,
      "loss": 0.0,
      "step": 12870
    },
    {
      "epoch": 0.08886006609312369,
      "grad_norm": 0.0,
      "learning_rate": 9.112158237148751e-06,
      "loss": 0.0,
      "step": 12880
    },
    {
      "epoch": 0.08892905682766804,
      "grad_norm": 8.310395969601814e-06,
      "learning_rate": 9.111468329803308e-06,
      "loss": 0.0,
      "step": 12890
    },
    {
      "epoch": 0.0889980475622124,
      "grad_norm": 0.0,
      "learning_rate": 9.110778422457864e-06,
      "loss": 0.0,
      "step": 12900
    },
    {
      "epoch": 0.08906703829675675,
      "grad_norm": 0.0,
      "learning_rate": 9.110088515112421e-06,
      "loss": 0.0,
      "step": 12910
    },
    {
      "epoch": 0.08913602903130109,
      "grad_norm": 0.0,
      "learning_rate": 9.109398607766977e-06,
      "loss": 0.0003,
      "step": 12920
    },
    {
      "epoch": 0.08920501976584545,
      "grad_norm": 0.0,
      "learning_rate": 9.108708700421534e-06,
      "loss": 0.0,
      "step": 12930
    },
    {
      "epoch": 0.0892740105003898,
      "grad_norm": 2.1934381422283877e-08,
      "learning_rate": 9.10801879307609e-06,
      "loss": 0.0,
      "step": 12940
    },
    {
      "epoch": 0.08934300123493415,
      "grad_norm": 0.0,
      "learning_rate": 9.107328885730647e-06,
      "loss": 0.0193,
      "step": 12950
    },
    {
      "epoch": 0.0894119919694785,
      "grad_norm": 15.994109153747559,
      "learning_rate": 9.106638978385203e-06,
      "loss": 0.0037,
      "step": 12960
    },
    {
      "epoch": 0.08948098270402285,
      "grad_norm": 0.05119892954826355,
      "learning_rate": 9.10594907103976e-06,
      "loss": 0.0,
      "step": 12970
    },
    {
      "epoch": 0.0895499734385672,
      "grad_norm": 2.489688633033893e-08,
      "learning_rate": 9.105259163694316e-06,
      "loss": 0.0871,
      "step": 12980
    },
    {
      "epoch": 0.08961896417311155,
      "grad_norm": 6.265150318540691e-07,
      "learning_rate": 9.104569256348873e-06,
      "loss": 0.0,
      "step": 12990
    },
    {
      "epoch": 0.0896879549076559,
      "grad_norm": 0.0,
      "learning_rate": 9.10387934900343e-06,
      "loss": 0.0044,
      "step": 13000
    },
    {
      "epoch": 0.08975694564220026,
      "grad_norm": 0.0,
      "learning_rate": 9.103189441657986e-06,
      "loss": 0.0,
      "step": 13010
    },
    {
      "epoch": 0.0898259363767446,
      "grad_norm": 0.0,
      "learning_rate": 9.102499534312542e-06,
      "loss": 0.0,
      "step": 13020
    },
    {
      "epoch": 0.08989492711128895,
      "grad_norm": 3.9424712383606675e-09,
      "learning_rate": 9.101809626967099e-06,
      "loss": 0.0,
      "step": 13030
    },
    {
      "epoch": 0.08996391784583331,
      "grad_norm": 0.0,
      "learning_rate": 9.101119719621655e-06,
      "loss": 0.0,
      "step": 13040
    },
    {
      "epoch": 0.09003290858037766,
      "grad_norm": 0.0,
      "learning_rate": 9.100429812276212e-06,
      "loss": 0.0007,
      "step": 13050
    },
    {
      "epoch": 0.090101899314922,
      "grad_norm": 66.62147521972656,
      "learning_rate": 9.099739904930768e-06,
      "loss": 0.0115,
      "step": 13060
    },
    {
      "epoch": 0.09017089004946635,
      "grad_norm": 1.4568859338760376,
      "learning_rate": 9.099049997585325e-06,
      "loss": 0.0003,
      "step": 13070
    },
    {
      "epoch": 0.09023988078401071,
      "grad_norm": 0.0056158327497541904,
      "learning_rate": 9.098360090239881e-06,
      "loss": 0.0,
      "step": 13080
    },
    {
      "epoch": 0.09030887151855506,
      "grad_norm": 0.0,
      "learning_rate": 9.097670182894438e-06,
      "loss": 0.0,
      "step": 13090
    },
    {
      "epoch": 0.0903778622530994,
      "grad_norm": 0.0,
      "learning_rate": 9.096980275548994e-06,
      "loss": 0.0,
      "step": 13100
    },
    {
      "epoch": 0.09044685298764375,
      "grad_norm": 1.6741969375289045e-05,
      "learning_rate": 9.09629036820355e-06,
      "loss": 0.0,
      "step": 13110
    },
    {
      "epoch": 0.09051584372218811,
      "grad_norm": 0.0,
      "learning_rate": 9.095600460858107e-06,
      "loss": 0.0003,
      "step": 13120
    },
    {
      "epoch": 0.09058483445673246,
      "grad_norm": 2.2316659453736065e-07,
      "learning_rate": 9.094910553512664e-06,
      "loss": 0.0,
      "step": 13130
    },
    {
      "epoch": 0.09065382519127681,
      "grad_norm": 0.0,
      "learning_rate": 9.09422064616722e-06,
      "loss": 0.0326,
      "step": 13140
    },
    {
      "epoch": 0.09072281592582117,
      "grad_norm": 100.9626693725586,
      "learning_rate": 9.093530738821777e-06,
      "loss": 0.0394,
      "step": 13150
    },
    {
      "epoch": 0.09079180666036551,
      "grad_norm": 1.0193910071620849e-07,
      "learning_rate": 9.092840831476333e-06,
      "loss": 0.0225,
      "step": 13160
    },
    {
      "epoch": 0.09086079739490986,
      "grad_norm": 0.0,
      "learning_rate": 9.09215092413089e-06,
      "loss": 0.0,
      "step": 13170
    },
    {
      "epoch": 0.09092978812945421,
      "grad_norm": 0.0,
      "learning_rate": 9.091461016785446e-06,
      "loss": 0.0005,
      "step": 13180
    },
    {
      "epoch": 0.09099877886399857,
      "grad_norm": 3.2125100357660585e-09,
      "learning_rate": 9.090771109440003e-06,
      "loss": 0.0,
      "step": 13190
    },
    {
      "epoch": 0.09106776959854292,
      "grad_norm": 8.771229431658867e-07,
      "learning_rate": 9.09008120209456e-06,
      "loss": 0.5305,
      "step": 13200
    },
    {
      "epoch": 0.09113676033308726,
      "grad_norm": 0.011078632436692715,
      "learning_rate": 9.089391294749116e-06,
      "loss": 0.0,
      "step": 13210
    },
    {
      "epoch": 0.09120575106763161,
      "grad_norm": 0.00012009937927359715,
      "learning_rate": 9.088701387403672e-06,
      "loss": 0.0,
      "step": 13220
    },
    {
      "epoch": 0.09127474180217597,
      "grad_norm": 0.0,
      "learning_rate": 9.088011480058229e-06,
      "loss": 0.0,
      "step": 13230
    },
    {
      "epoch": 0.09134373253672032,
      "grad_norm": 0.023638155311346054,
      "learning_rate": 9.087321572712785e-06,
      "loss": 0.002,
      "step": 13240
    },
    {
      "epoch": 0.09141272327126466,
      "grad_norm": 8.871408607724618e-10,
      "learning_rate": 9.086631665367342e-06,
      "loss": 0.0002,
      "step": 13250
    },
    {
      "epoch": 0.09148171400580903,
      "grad_norm": 0.0,
      "learning_rate": 9.085941758021898e-06,
      "loss": 0.0,
      "step": 13260
    },
    {
      "epoch": 0.09155070474035337,
      "grad_norm": 0.0,
      "learning_rate": 9.085251850676455e-06,
      "loss": 0.0001,
      "step": 13270
    },
    {
      "epoch": 0.09161969547489772,
      "grad_norm": 0.0001510678994236514,
      "learning_rate": 9.084561943331011e-06,
      "loss": 0.0,
      "step": 13280
    },
    {
      "epoch": 0.09168868620944207,
      "grad_norm": 0.0,
      "learning_rate": 9.083872035985568e-06,
      "loss": 0.0,
      "step": 13290
    },
    {
      "epoch": 0.09175767694398643,
      "grad_norm": 0.0,
      "learning_rate": 9.083182128640124e-06,
      "loss": 0.0,
      "step": 13300
    },
    {
      "epoch": 0.09182666767853077,
      "grad_norm": 1.1380863725207746e-05,
      "learning_rate": 9.08249222129468e-06,
      "loss": 0.0,
      "step": 13310
    },
    {
      "epoch": 0.09189565841307512,
      "grad_norm": 2.755751302174758e-05,
      "learning_rate": 9.081802313949237e-06,
      "loss": 0.0097,
      "step": 13320
    },
    {
      "epoch": 0.09196464914761948,
      "grad_norm": 0.0005176571430638433,
      "learning_rate": 9.081112406603794e-06,
      "loss": 0.0006,
      "step": 13330
    },
    {
      "epoch": 0.09203363988216383,
      "grad_norm": 0.00021993771952111274,
      "learning_rate": 9.08042249925835e-06,
      "loss": 0.0,
      "step": 13340
    },
    {
      "epoch": 0.09210263061670818,
      "grad_norm": 5.26480370410809e-08,
      "learning_rate": 9.079732591912907e-06,
      "loss": 0.0,
      "step": 13350
    },
    {
      "epoch": 0.09217162135125252,
      "grad_norm": 0.0,
      "learning_rate": 9.079042684567463e-06,
      "loss": 0.0,
      "step": 13360
    },
    {
      "epoch": 0.09224061208579688,
      "grad_norm": 0.0003403661830816418,
      "learning_rate": 9.078352777222021e-06,
      "loss": 0.0,
      "step": 13370
    },
    {
      "epoch": 0.09230960282034123,
      "grad_norm": 9.105828313593634e-10,
      "learning_rate": 9.077662869876576e-06,
      "loss": 0.019,
      "step": 13380
    },
    {
      "epoch": 0.09237859355488558,
      "grad_norm": 0.0,
      "learning_rate": 9.076972962531132e-06,
      "loss": 0.0,
      "step": 13390
    },
    {
      "epoch": 0.09244758428942992,
      "grad_norm": 0.0,
      "learning_rate": 9.076283055185689e-06,
      "loss": 0.0,
      "step": 13400
    },
    {
      "epoch": 0.09251657502397428,
      "grad_norm": 0.0,
      "learning_rate": 9.075593147840245e-06,
      "loss": 0.0004,
      "step": 13410
    },
    {
      "epoch": 0.09258556575851863,
      "grad_norm": 0.0,
      "learning_rate": 9.074903240494802e-06,
      "loss": 0.0,
      "step": 13420
    },
    {
      "epoch": 0.09265455649306298,
      "grad_norm": 0.0,
      "learning_rate": 9.074213333149358e-06,
      "loss": 0.0,
      "step": 13430
    },
    {
      "epoch": 0.09272354722760734,
      "grad_norm": 0.003086378099396825,
      "learning_rate": 9.073523425803915e-06,
      "loss": 0.0,
      "step": 13440
    },
    {
      "epoch": 0.09279253796215169,
      "grad_norm": 0.0,
      "learning_rate": 9.072833518458471e-06,
      "loss": 0.0003,
      "step": 13450
    },
    {
      "epoch": 0.09286152869669603,
      "grad_norm": 0.0,
      "learning_rate": 9.072143611113028e-06,
      "loss": 0.0,
      "step": 13460
    },
    {
      "epoch": 0.09293051943124038,
      "grad_norm": 0.0,
      "learning_rate": 9.071453703767584e-06,
      "loss": 0.0001,
      "step": 13470
    },
    {
      "epoch": 0.09299951016578474,
      "grad_norm": 0.0,
      "learning_rate": 9.070763796422143e-06,
      "loss": 0.0014,
      "step": 13480
    },
    {
      "epoch": 0.09306850090032909,
      "grad_norm": 0.0,
      "learning_rate": 9.070073889076697e-06,
      "loss": 0.0,
      "step": 13490
    },
    {
      "epoch": 0.09313749163487343,
      "grad_norm": 0.0,
      "learning_rate": 9.069383981731254e-06,
      "loss": 0.0,
      "step": 13500
    },
    {
      "epoch": 0.09320648236941778,
      "grad_norm": 0.0,
      "learning_rate": 9.06869407438581e-06,
      "loss": 0.0,
      "step": 13510
    },
    {
      "epoch": 0.09327547310396214,
      "grad_norm": 0.0,
      "learning_rate": 9.068004167040367e-06,
      "loss": 0.0,
      "step": 13520
    },
    {
      "epoch": 0.09334446383850649,
      "grad_norm": 0.0,
      "learning_rate": 9.067314259694923e-06,
      "loss": 0.0517,
      "step": 13530
    },
    {
      "epoch": 0.09341345457305084,
      "grad_norm": 0.0,
      "learning_rate": 9.06662435234948e-06,
      "loss": 0.0,
      "step": 13540
    },
    {
      "epoch": 0.0934824453075952,
      "grad_norm": 0.0,
      "learning_rate": 9.065934445004036e-06,
      "loss": 0.0,
      "step": 13550
    },
    {
      "epoch": 0.09355143604213954,
      "grad_norm": 45.55268859863281,
      "learning_rate": 9.065244537658593e-06,
      "loss": 0.0076,
      "step": 13560
    },
    {
      "epoch": 0.09362042677668389,
      "grad_norm": 5.459155905107593e-10,
      "learning_rate": 9.06455463031315e-06,
      "loss": 0.0,
      "step": 13570
    },
    {
      "epoch": 0.09368941751122824,
      "grad_norm": 0.006124570034444332,
      "learning_rate": 9.063864722967706e-06,
      "loss": 0.0,
      "step": 13580
    },
    {
      "epoch": 0.0937584082457726,
      "grad_norm": 0.0008759878692217171,
      "learning_rate": 9.063174815622264e-06,
      "loss": 0.0039,
      "step": 13590
    },
    {
      "epoch": 0.09382739898031695,
      "grad_norm": 0.0,
      "learning_rate": 9.062484908276819e-06,
      "loss": 0.199,
      "step": 13600
    },
    {
      "epoch": 0.09389638971486129,
      "grad_norm": 0.0,
      "learning_rate": 9.061795000931375e-06,
      "loss": 0.0,
      "step": 13610
    },
    {
      "epoch": 0.09396538044940564,
      "grad_norm": 0.0,
      "learning_rate": 9.061105093585932e-06,
      "loss": 0.0,
      "step": 13620
    },
    {
      "epoch": 0.09403437118395,
      "grad_norm": 0.0,
      "learning_rate": 9.060415186240488e-06,
      "loss": 0.0,
      "step": 13630
    },
    {
      "epoch": 0.09410336191849435,
      "grad_norm": 0.0,
      "learning_rate": 9.059725278895045e-06,
      "loss": 0.0,
      "step": 13640
    },
    {
      "epoch": 0.0941723526530387,
      "grad_norm": 8.18364143371582,
      "learning_rate": 9.059035371549601e-06,
      "loss": 0.0018,
      "step": 13650
    },
    {
      "epoch": 0.09424134338758305,
      "grad_norm": 109.45488739013672,
      "learning_rate": 9.058345464204158e-06,
      "loss": 0.0498,
      "step": 13660
    },
    {
      "epoch": 0.0943103341221274,
      "grad_norm": 7.464707962867578e-09,
      "learning_rate": 9.057655556858714e-06,
      "loss": 0.004,
      "step": 13670
    },
    {
      "epoch": 0.09437932485667175,
      "grad_norm": 0.01129795704036951,
      "learning_rate": 9.05696564951327e-06,
      "loss": 0.0041,
      "step": 13680
    },
    {
      "epoch": 0.0944483155912161,
      "grad_norm": 0.0,
      "learning_rate": 9.056275742167827e-06,
      "loss": 0.0,
      "step": 13690
    },
    {
      "epoch": 0.09451730632576046,
      "grad_norm": 0.0,
      "learning_rate": 9.055585834822385e-06,
      "loss": 0.0,
      "step": 13700
    },
    {
      "epoch": 0.0945862970603048,
      "grad_norm": 2.364405372645706e-05,
      "learning_rate": 9.05489592747694e-06,
      "loss": 0.0,
      "step": 13710
    },
    {
      "epoch": 0.09465528779484915,
      "grad_norm": 0.0,
      "learning_rate": 9.054206020131497e-06,
      "loss": 0.0001,
      "step": 13720
    },
    {
      "epoch": 0.0947242785293935,
      "grad_norm": 0.0004204988945275545,
      "learning_rate": 9.053516112786053e-06,
      "loss": 0.0149,
      "step": 13730
    },
    {
      "epoch": 0.09479326926393786,
      "grad_norm": 1.9861205657889514e-07,
      "learning_rate": 9.05282620544061e-06,
      "loss": 0.0,
      "step": 13740
    },
    {
      "epoch": 0.0948622599984822,
      "grad_norm": 0.0,
      "learning_rate": 9.052136298095166e-06,
      "loss": 0.0001,
      "step": 13750
    },
    {
      "epoch": 0.09493125073302655,
      "grad_norm": 0.0,
      "learning_rate": 9.051446390749723e-06,
      "loss": 0.0,
      "step": 13760
    },
    {
      "epoch": 0.09500024146757091,
      "grad_norm": 3.38435079960675e-09,
      "learning_rate": 9.050756483404279e-06,
      "loss": 0.0,
      "step": 13770
    },
    {
      "epoch": 0.09506923220211526,
      "grad_norm": 0.0006970739341340959,
      "learning_rate": 9.050066576058836e-06,
      "loss": 0.0,
      "step": 13780
    },
    {
      "epoch": 0.0951382229366596,
      "grad_norm": 1.2562564734253101e-05,
      "learning_rate": 9.049376668713392e-06,
      "loss": 0.006,
      "step": 13790
    },
    {
      "epoch": 0.09520721367120395,
      "grad_norm": 4.783676232911205e-10,
      "learning_rate": 9.048686761367949e-06,
      "loss": 0.0,
      "step": 13800
    },
    {
      "epoch": 0.09527620440574831,
      "grad_norm": 0.0,
      "learning_rate": 9.047996854022507e-06,
      "loss": 0.0,
      "step": 13810
    },
    {
      "epoch": 0.09534519514029266,
      "grad_norm": 0.0,
      "learning_rate": 9.047306946677063e-06,
      "loss": 0.0,
      "step": 13820
    },
    {
      "epoch": 0.09541418587483701,
      "grad_norm": 0.0,
      "learning_rate": 9.046617039331618e-06,
      "loss": 0.363,
      "step": 13830
    },
    {
      "epoch": 0.09548317660938135,
      "grad_norm": 0.0,
      "learning_rate": 9.045927131986175e-06,
      "loss": 0.0,
      "step": 13840
    },
    {
      "epoch": 0.09555216734392571,
      "grad_norm": 0.0002969199267681688,
      "learning_rate": 9.045237224640731e-06,
      "loss": 0.0,
      "step": 13850
    },
    {
      "epoch": 0.09562115807847006,
      "grad_norm": 0.0,
      "learning_rate": 9.044547317295288e-06,
      "loss": 0.0001,
      "step": 13860
    },
    {
      "epoch": 0.09569014881301441,
      "grad_norm": 0.0,
      "learning_rate": 9.043857409949844e-06,
      "loss": 0.0,
      "step": 13870
    },
    {
      "epoch": 0.09575913954755877,
      "grad_norm": 0.0,
      "learning_rate": 9.0431675026044e-06,
      "loss": 0.0033,
      "step": 13880
    },
    {
      "epoch": 0.09582813028210312,
      "grad_norm": 0.007406032178550959,
      "learning_rate": 9.042477595258957e-06,
      "loss": 0.0003,
      "step": 13890
    },
    {
      "epoch": 0.09589712101664746,
      "grad_norm": 0.0,
      "learning_rate": 9.041787687913513e-06,
      "loss": 0.0,
      "step": 13900
    },
    {
      "epoch": 0.09596611175119181,
      "grad_norm": 0.0,
      "learning_rate": 9.04109778056807e-06,
      "loss": 0.0,
      "step": 13910
    },
    {
      "epoch": 0.09603510248573617,
      "grad_norm": 6.853615559521131e-07,
      "learning_rate": 9.040407873222628e-06,
      "loss": 0.0005,
      "step": 13920
    },
    {
      "epoch": 0.09610409322028052,
      "grad_norm": 0.0,
      "learning_rate": 9.039717965877185e-06,
      "loss": 0.0,
      "step": 13930
    },
    {
      "epoch": 0.09617308395482486,
      "grad_norm": 0.0,
      "learning_rate": 9.03902805853174e-06,
      "loss": 0.0,
      "step": 13940
    },
    {
      "epoch": 0.09624207468936921,
      "grad_norm": 0.0,
      "learning_rate": 9.038338151186296e-06,
      "loss": 0.0002,
      "step": 13950
    },
    {
      "epoch": 0.09631106542391357,
      "grad_norm": 0.0,
      "learning_rate": 9.037648243840852e-06,
      "loss": 0.0,
      "step": 13960
    },
    {
      "epoch": 0.09638005615845792,
      "grad_norm": 0.0,
      "learning_rate": 9.036958336495409e-06,
      "loss": 0.0,
      "step": 13970
    },
    {
      "epoch": 0.09644904689300227,
      "grad_norm": 0.0,
      "learning_rate": 9.036268429149965e-06,
      "loss": 0.0,
      "step": 13980
    },
    {
      "epoch": 0.09651803762754663,
      "grad_norm": 0.0,
      "learning_rate": 9.035578521804522e-06,
      "loss": 0.0,
      "step": 13990
    },
    {
      "epoch": 0.09658702836209097,
      "grad_norm": 5.598350227842275e-08,
      "learning_rate": 9.034888614459078e-06,
      "loss": 0.0018,
      "step": 14000
    },
    {
      "epoch": 0.09665601909663532,
      "grad_norm": 0.0,
      "learning_rate": 9.034198707113635e-06,
      "loss": 0.0,
      "step": 14010
    },
    {
      "epoch": 0.09672500983117967,
      "grad_norm": 0.00027740251971408725,
      "learning_rate": 9.033508799768191e-06,
      "loss": 0.0032,
      "step": 14020
    },
    {
      "epoch": 0.09679400056572403,
      "grad_norm": 0.0,
      "learning_rate": 9.03281889242275e-06,
      "loss": 0.0,
      "step": 14030
    },
    {
      "epoch": 0.09686299130026838,
      "grad_norm": 0.0,
      "learning_rate": 9.032128985077306e-06,
      "loss": 0.0,
      "step": 14040
    },
    {
      "epoch": 0.09693198203481272,
      "grad_norm": 0.0,
      "learning_rate": 9.03143907773186e-06,
      "loss": 0.0,
      "step": 14050
    },
    {
      "epoch": 0.09700097276935707,
      "grad_norm": 0.002469313330948353,
      "learning_rate": 9.030749170386417e-06,
      "loss": 0.0,
      "step": 14060
    },
    {
      "epoch": 0.09706996350390143,
      "grad_norm": 256.684814453125,
      "learning_rate": 9.030059263040974e-06,
      "loss": 0.0274,
      "step": 14070
    },
    {
      "epoch": 0.09713895423844578,
      "grad_norm": 0.0,
      "learning_rate": 9.02936935569553e-06,
      "loss": 0.005,
      "step": 14080
    },
    {
      "epoch": 0.09720794497299012,
      "grad_norm": 0.0,
      "learning_rate": 9.028679448350087e-06,
      "loss": 0.0,
      "step": 14090
    },
    {
      "epoch": 0.09727693570753448,
      "grad_norm": 0.0,
      "learning_rate": 9.027989541004643e-06,
      "loss": 0.0,
      "step": 14100
    },
    {
      "epoch": 0.09734592644207883,
      "grad_norm": 0.0,
      "learning_rate": 9.0272996336592e-06,
      "loss": 0.0,
      "step": 14110
    },
    {
      "epoch": 0.09741491717662318,
      "grad_norm": 0.0,
      "learning_rate": 9.026609726313756e-06,
      "loss": 0.0,
      "step": 14120
    },
    {
      "epoch": 0.09748390791116752,
      "grad_norm": 3.649547352324589e-06,
      "learning_rate": 9.025919818968313e-06,
      "loss": 0.5051,
      "step": 14130
    },
    {
      "epoch": 0.09755289864571189,
      "grad_norm": 2.96873192695557e-09,
      "learning_rate": 9.025229911622871e-06,
      "loss": 0.0102,
      "step": 14140
    },
    {
      "epoch": 0.09762188938025623,
      "grad_norm": 0.017666010186076164,
      "learning_rate": 9.024540004277427e-06,
      "loss": 0.0,
      "step": 14150
    },
    {
      "epoch": 0.09769088011480058,
      "grad_norm": 0.0,
      "learning_rate": 9.023850096931984e-06,
      "loss": 0.0105,
      "step": 14160
    },
    {
      "epoch": 0.09775987084934493,
      "grad_norm": 0.0,
      "learning_rate": 9.023160189586539e-06,
      "loss": 0.0,
      "step": 14170
    },
    {
      "epoch": 0.09782886158388929,
      "grad_norm": 0.0,
      "learning_rate": 9.022470282241095e-06,
      "loss": 0.0003,
      "step": 14180
    },
    {
      "epoch": 0.09789785231843363,
      "grad_norm": 0.0,
      "learning_rate": 9.021780374895652e-06,
      "loss": 0.0007,
      "step": 14190
    },
    {
      "epoch": 0.09796684305297798,
      "grad_norm": 4.652412066707967e-10,
      "learning_rate": 9.021090467550208e-06,
      "loss": 0.0,
      "step": 14200
    },
    {
      "epoch": 0.09803583378752234,
      "grad_norm": 0.0,
      "learning_rate": 9.020400560204765e-06,
      "loss": 0.0,
      "step": 14210
    },
    {
      "epoch": 0.09810482452206669,
      "grad_norm": 0.0,
      "learning_rate": 9.019710652859321e-06,
      "loss": 0.0,
      "step": 14220
    },
    {
      "epoch": 0.09817381525661104,
      "grad_norm": 0.0,
      "learning_rate": 9.019020745513878e-06,
      "loss": 0.0,
      "step": 14230
    },
    {
      "epoch": 0.09824280599115538,
      "grad_norm": 0.6789950728416443,
      "learning_rate": 9.018330838168434e-06,
      "loss": 0.0002,
      "step": 14240
    },
    {
      "epoch": 0.09831179672569974,
      "grad_norm": 1.3683045096968272e-07,
      "learning_rate": 9.017640930822992e-06,
      "loss": 0.0,
      "step": 14250
    },
    {
      "epoch": 0.09838078746024409,
      "grad_norm": 3.689347849444857e-08,
      "learning_rate": 9.016951023477549e-06,
      "loss": 0.0001,
      "step": 14260
    },
    {
      "epoch": 0.09844977819478844,
      "grad_norm": 0.0,
      "learning_rate": 9.016261116132105e-06,
      "loss": 0.0,
      "step": 14270
    },
    {
      "epoch": 0.09851876892933278,
      "grad_norm": 0.0,
      "learning_rate": 9.01557120878666e-06,
      "loss": 0.0,
      "step": 14280
    },
    {
      "epoch": 0.09858775966387714,
      "grad_norm": 0.17829005420207977,
      "learning_rate": 9.014881301441217e-06,
      "loss": 0.011,
      "step": 14290
    },
    {
      "epoch": 0.09865675039842149,
      "grad_norm": 0.9580913186073303,
      "learning_rate": 9.014191394095773e-06,
      "loss": 0.0001,
      "step": 14300
    },
    {
      "epoch": 0.09872574113296584,
      "grad_norm": 0.0,
      "learning_rate": 9.01350148675033e-06,
      "loss": 0.0,
      "step": 14310
    },
    {
      "epoch": 0.0987947318675102,
      "grad_norm": 0.0,
      "learning_rate": 9.01288057013943e-06,
      "loss": 0.9633,
      "step": 14320
    },
    {
      "epoch": 0.09886372260205455,
      "grad_norm": 0.0,
      "learning_rate": 9.012190662793987e-06,
      "loss": 0.0,
      "step": 14330
    },
    {
      "epoch": 0.0989327133365989,
      "grad_norm": 0.0,
      "learning_rate": 9.011500755448545e-06,
      "loss": 0.0,
      "step": 14340
    },
    {
      "epoch": 0.09900170407114324,
      "grad_norm": 0.0,
      "learning_rate": 9.010810848103101e-06,
      "loss": 0.0,
      "step": 14350
    },
    {
      "epoch": 0.0990706948056876,
      "grad_norm": 0.0,
      "learning_rate": 9.010120940757658e-06,
      "loss": 0.0062,
      "step": 14360
    },
    {
      "epoch": 0.09913968554023195,
      "grad_norm": 0.0,
      "learning_rate": 9.009431033412214e-06,
      "loss": 0.0,
      "step": 14370
    },
    {
      "epoch": 0.0992086762747763,
      "grad_norm": 0.0013788517098873854,
      "learning_rate": 9.008741126066771e-06,
      "loss": 0.0,
      "step": 14380
    },
    {
      "epoch": 0.09927766700932064,
      "grad_norm": 0.009718854911625385,
      "learning_rate": 9.008051218721326e-06,
      "loss": 0.0,
      "step": 14390
    },
    {
      "epoch": 0.099346657743865,
      "grad_norm": 2.6873924525716575e-06,
      "learning_rate": 9.007361311375882e-06,
      "loss": 0.0,
      "step": 14400
    },
    {
      "epoch": 0.09941564847840935,
      "grad_norm": 0.0,
      "learning_rate": 9.006671404030439e-06,
      "loss": 0.0,
      "step": 14410
    },
    {
      "epoch": 0.0994846392129537,
      "grad_norm": 0.0,
      "learning_rate": 9.005981496684995e-06,
      "loss": 0.0092,
      "step": 14420
    },
    {
      "epoch": 0.09955362994749806,
      "grad_norm": 0.0,
      "learning_rate": 9.005291589339552e-06,
      "loss": 0.0,
      "step": 14430
    },
    {
      "epoch": 0.0996226206820424,
      "grad_norm": 0.17684686183929443,
      "learning_rate": 9.004601681994108e-06,
      "loss": 0.0,
      "step": 14440
    },
    {
      "epoch": 0.09969161141658675,
      "grad_norm": 0.0,
      "learning_rate": 9.003911774648666e-06,
      "loss": 0.3758,
      "step": 14450
    },
    {
      "epoch": 0.0997606021511311,
      "grad_norm": 0.0,
      "learning_rate": 9.003221867303223e-06,
      "loss": 0.0004,
      "step": 14460
    },
    {
      "epoch": 0.09982959288567546,
      "grad_norm": 0.0,
      "learning_rate": 9.00253195995778e-06,
      "loss": 0.0,
      "step": 14470
    },
    {
      "epoch": 0.0998985836202198,
      "grad_norm": 7.986645221710205,
      "learning_rate": 9.001842052612336e-06,
      "loss": 0.0057,
      "step": 14480
    },
    {
      "epoch": 0.09996757435476415,
      "grad_norm": 8.702115883352235e-05,
      "learning_rate": 9.001152145266892e-06,
      "loss": 0.0,
      "step": 14490
    },
    {
      "epoch": 0.1000365650893085,
      "grad_norm": 0.0,
      "learning_rate": 9.000462237921447e-06,
      "loss": 0.0032,
      "step": 14500
    },
    {
      "epoch": 0.10010555582385286,
      "grad_norm": 0.0,
      "learning_rate": 8.999772330576004e-06,
      "loss": 0.0008,
      "step": 14510
    },
    {
      "epoch": 0.1001745465583972,
      "grad_norm": 0.0,
      "learning_rate": 8.99908242323056e-06,
      "loss": 0.0,
      "step": 14520
    },
    {
      "epoch": 0.10024353729294155,
      "grad_norm": 0.0,
      "learning_rate": 8.998392515885117e-06,
      "loss": 0.0,
      "step": 14530
    },
    {
      "epoch": 0.10031252802748591,
      "grad_norm": 3.892130848726083e-07,
      "learning_rate": 8.997702608539673e-06,
      "loss": 0.0025,
      "step": 14540
    },
    {
      "epoch": 0.10038151876203026,
      "grad_norm": 0.0,
      "learning_rate": 8.99701270119423e-06,
      "loss": 0.1964,
      "step": 14550
    },
    {
      "epoch": 0.10045050949657461,
      "grad_norm": 0.0,
      "learning_rate": 8.996322793848788e-06,
      "loss": 0.0038,
      "step": 14560
    },
    {
      "epoch": 0.10051950023111896,
      "grad_norm": 0.0,
      "learning_rate": 8.995632886503344e-06,
      "loss": 0.0,
      "step": 14570
    },
    {
      "epoch": 0.10058849096566332,
      "grad_norm": 334.4906311035156,
      "learning_rate": 8.9949429791579e-06,
      "loss": 0.1375,
      "step": 14580
    },
    {
      "epoch": 0.10065748170020766,
      "grad_norm": 1.3659666776657104,
      "learning_rate": 8.994253071812457e-06,
      "loss": 0.0003,
      "step": 14590
    },
    {
      "epoch": 0.10072647243475201,
      "grad_norm": 0.004187623504549265,
      "learning_rate": 8.993563164467014e-06,
      "loss": 2.1156,
      "step": 14600
    },
    {
      "epoch": 0.10079546316929637,
      "grad_norm": 0.0,
      "learning_rate": 8.99287325712157e-06,
      "loss": 0.0,
      "step": 14610
    },
    {
      "epoch": 0.10086445390384072,
      "grad_norm": 0.00265240459702909,
      "learning_rate": 8.992183349776125e-06,
      "loss": 0.0,
      "step": 14620
    },
    {
      "epoch": 0.10093344463838506,
      "grad_norm": 1.7600315871391103e-08,
      "learning_rate": 8.991493442430681e-06,
      "loss": 0.0156,
      "step": 14630
    },
    {
      "epoch": 0.10100243537292941,
      "grad_norm": 0.0,
      "learning_rate": 8.990803535085238e-06,
      "loss": 0.6057,
      "step": 14640
    },
    {
      "epoch": 0.10107142610747377,
      "grad_norm": 1.6776948541519232e-07,
      "learning_rate": 8.990113627739794e-06,
      "loss": 0.1937,
      "step": 14650
    },
    {
      "epoch": 0.10114041684201812,
      "grad_norm": 0.0,
      "learning_rate": 8.989423720394351e-06,
      "loss": 0.0,
      "step": 14660
    },
    {
      "epoch": 0.10120940757656247,
      "grad_norm": 1.028349757194519,
      "learning_rate": 8.988733813048909e-06,
      "loss": 0.0169,
      "step": 14670
    },
    {
      "epoch": 0.10127839831110681,
      "grad_norm": 0.0,
      "learning_rate": 8.988043905703466e-06,
      "loss": 0.0,
      "step": 14680
    },
    {
      "epoch": 0.10134738904565117,
      "grad_norm": 0.0011982250725850463,
      "learning_rate": 8.987353998358022e-06,
      "loss": 0.0,
      "step": 14690
    },
    {
      "epoch": 0.10141637978019552,
      "grad_norm": 0.0,
      "learning_rate": 8.986664091012579e-06,
      "loss": 0.0,
      "step": 14700
    },
    {
      "epoch": 0.10148537051473987,
      "grad_norm": 0.0,
      "learning_rate": 8.985974183667135e-06,
      "loss": 0.359,
      "step": 14710
    },
    {
      "epoch": 0.10155436124928423,
      "grad_norm": 0.0,
      "learning_rate": 8.985284276321692e-06,
      "loss": 0.0,
      "step": 14720
    },
    {
      "epoch": 0.10162335198382857,
      "grad_norm": 0.0,
      "learning_rate": 8.984594368976246e-06,
      "loss": 0.0016,
      "step": 14730
    },
    {
      "epoch": 0.10169234271837292,
      "grad_norm": 0.0,
      "learning_rate": 8.983904461630803e-06,
      "loss": 0.0008,
      "step": 14740
    },
    {
      "epoch": 0.10176133345291727,
      "grad_norm": 0.0,
      "learning_rate": 8.98321455428536e-06,
      "loss": 0.0015,
      "step": 14750
    },
    {
      "epoch": 0.10183032418746163,
      "grad_norm": 53.06954574584961,
      "learning_rate": 8.982524646939916e-06,
      "loss": 0.0094,
      "step": 14760
    },
    {
      "epoch": 0.10189931492200598,
      "grad_norm": 0.0,
      "learning_rate": 8.981834739594472e-06,
      "loss": 0.0002,
      "step": 14770
    },
    {
      "epoch": 0.10196830565655032,
      "grad_norm": 0.0,
      "learning_rate": 8.98114483224903e-06,
      "loss": 0.0,
      "step": 14780
    },
    {
      "epoch": 0.10203729639109467,
      "grad_norm": 0.0,
      "learning_rate": 8.980454924903587e-06,
      "loss": 0.2904,
      "step": 14790
    },
    {
      "epoch": 0.10210628712563903,
      "grad_norm": 0.0007439965265803039,
      "learning_rate": 8.979765017558143e-06,
      "loss": 0.0,
      "step": 14800
    },
    {
      "epoch": 0.10217527786018338,
      "grad_norm": 0.0,
      "learning_rate": 8.9790751102127e-06,
      "loss": 0.0,
      "step": 14810
    },
    {
      "epoch": 0.10224426859472772,
      "grad_norm": 0.6486353278160095,
      "learning_rate": 8.978385202867256e-06,
      "loss": 0.0001,
      "step": 14820
    },
    {
      "epoch": 0.10231325932927209,
      "grad_norm": 0.0,
      "learning_rate": 8.977695295521813e-06,
      "loss": 0.0001,
      "step": 14830
    },
    {
      "epoch": 0.10238225006381643,
      "grad_norm": 0.0,
      "learning_rate": 8.977005388176368e-06,
      "loss": 0.0,
      "step": 14840
    },
    {
      "epoch": 0.10245124079836078,
      "grad_norm": 20.890771865844727,
      "learning_rate": 8.976315480830924e-06,
      "loss": 0.8796,
      "step": 14850
    },
    {
      "epoch": 0.10252023153290513,
      "grad_norm": 0.0,
      "learning_rate": 8.97562557348548e-06,
      "loss": 0.0,
      "step": 14860
    },
    {
      "epoch": 0.10258922226744949,
      "grad_norm": 0.0,
      "learning_rate": 8.974935666140037e-06,
      "loss": 0.0,
      "step": 14870
    },
    {
      "epoch": 0.10265821300199383,
      "grad_norm": 21.972917556762695,
      "learning_rate": 8.974245758794594e-06,
      "loss": 0.0038,
      "step": 14880
    },
    {
      "epoch": 0.10272720373653818,
      "grad_norm": 0.0,
      "learning_rate": 8.973555851449152e-06,
      "loss": 0.7,
      "step": 14890
    },
    {
      "epoch": 0.10279619447108253,
      "grad_norm": 2.8864340251288922e-08,
      "learning_rate": 8.972865944103708e-06,
      "loss": 0.0005,
      "step": 14900
    },
    {
      "epoch": 0.10286518520562689,
      "grad_norm": 0.0,
      "learning_rate": 8.972176036758265e-06,
      "loss": 0.0006,
      "step": 14910
    },
    {
      "epoch": 0.10293417594017124,
      "grad_norm": 0.0,
      "learning_rate": 8.971486129412821e-06,
      "loss": 0.0,
      "step": 14920
    },
    {
      "epoch": 0.10300316667471558,
      "grad_norm": 0.0,
      "learning_rate": 8.970796222067378e-06,
      "loss": 0.0,
      "step": 14930
    },
    {
      "epoch": 0.10307215740925994,
      "grad_norm": 0.0,
      "learning_rate": 8.970106314721934e-06,
      "loss": 0.0044,
      "step": 14940
    },
    {
      "epoch": 0.10314114814380429,
      "grad_norm": 0.0,
      "learning_rate": 8.96941640737649e-06,
      "loss": 0.0,
      "step": 14950
    },
    {
      "epoch": 0.10321013887834864,
      "grad_norm": 0.0,
      "learning_rate": 8.968726500031046e-06,
      "loss": 0.0015,
      "step": 14960
    },
    {
      "epoch": 0.10327912961289298,
      "grad_norm": 0.0,
      "learning_rate": 8.968036592685602e-06,
      "loss": 0.0,
      "step": 14970
    },
    {
      "epoch": 0.10334812034743734,
      "grad_norm": 0.11133863776922226,
      "learning_rate": 8.967346685340159e-06,
      "loss": 0.0005,
      "step": 14980
    },
    {
      "epoch": 0.10341711108198169,
      "grad_norm": 0.0,
      "learning_rate": 8.966656777994715e-06,
      "loss": 0.0002,
      "step": 14990
    },
    {
      "epoch": 0.10348610181652604,
      "grad_norm": 0.0,
      "learning_rate": 8.965966870649273e-06,
      "loss": 0.0001,
      "step": 15000
    },
    {
      "epoch": 0.10355509255107039,
      "grad_norm": 0.0,
      "learning_rate": 8.96527696330383e-06,
      "loss": 0.0,
      "step": 15010
    },
    {
      "epoch": 0.10362408328561475,
      "grad_norm": 0.0,
      "learning_rate": 8.964587055958386e-06,
      "loss": 0.0,
      "step": 15020
    },
    {
      "epoch": 0.10369307402015909,
      "grad_norm": 0.0,
      "learning_rate": 8.963897148612943e-06,
      "loss": 0.0004,
      "step": 15030
    },
    {
      "epoch": 0.10376206475470344,
      "grad_norm": 0.0,
      "learning_rate": 8.9632072412675e-06,
      "loss": 0.0,
      "step": 15040
    },
    {
      "epoch": 0.1038310554892478,
      "grad_norm": 0.0,
      "learning_rate": 8.962517333922056e-06,
      "loss": 0.0002,
      "step": 15050
    },
    {
      "epoch": 0.10390004622379215,
      "grad_norm": 0.0,
      "learning_rate": 8.961827426576612e-06,
      "loss": 0.0,
      "step": 15060
    },
    {
      "epoch": 0.1039690369583365,
      "grad_norm": 1.1922156772925518e-07,
      "learning_rate": 8.961137519231167e-06,
      "loss": 0.0,
      "step": 15070
    },
    {
      "epoch": 0.10403802769288084,
      "grad_norm": 0.007567184045910835,
      "learning_rate": 8.960447611885724e-06,
      "loss": 0.0115,
      "step": 15080
    },
    {
      "epoch": 0.1041070184274252,
      "grad_norm": 0.0,
      "learning_rate": 8.95975770454028e-06,
      "loss": 0.0031,
      "step": 15090
    },
    {
      "epoch": 0.10417600916196955,
      "grad_norm": 0.0,
      "learning_rate": 8.959067797194837e-06,
      "loss": 0.0,
      "step": 15100
    },
    {
      "epoch": 0.1042449998965139,
      "grad_norm": 0.0,
      "learning_rate": 8.958377889849395e-06,
      "loss": 0.0,
      "step": 15110
    },
    {
      "epoch": 0.10431399063105824,
      "grad_norm": 0.0,
      "learning_rate": 8.957687982503951e-06,
      "loss": 0.0,
      "step": 15120
    },
    {
      "epoch": 0.1043829813656026,
      "grad_norm": 0.00018382474081590772,
      "learning_rate": 8.956998075158508e-06,
      "loss": 0.075,
      "step": 15130
    },
    {
      "epoch": 0.10445197210014695,
      "grad_norm": 0.0,
      "learning_rate": 8.956308167813064e-06,
      "loss": 0.107,
      "step": 15140
    },
    {
      "epoch": 0.1045209628346913,
      "grad_norm": 6.728784001097665e-07,
      "learning_rate": 8.95561826046762e-06,
      "loss": 0.0,
      "step": 15150
    },
    {
      "epoch": 0.10458995356923566,
      "grad_norm": 0.0,
      "learning_rate": 8.954928353122177e-06,
      "loss": 0.0,
      "step": 15160
    },
    {
      "epoch": 0.10465894430378,
      "grad_norm": 1.3022390604019165,
      "learning_rate": 8.954238445776734e-06,
      "loss": 0.0124,
      "step": 15170
    },
    {
      "epoch": 0.10472793503832435,
      "grad_norm": 0.0,
      "learning_rate": 8.953548538431288e-06,
      "loss": 0.0014,
      "step": 15180
    },
    {
      "epoch": 0.1047969257728687,
      "grad_norm": 0.0,
      "learning_rate": 8.952858631085845e-06,
      "loss": 0.4665,
      "step": 15190
    },
    {
      "epoch": 0.10486591650741306,
      "grad_norm": 0.000387391890399158,
      "learning_rate": 8.952168723740401e-06,
      "loss": 0.0002,
      "step": 15200
    },
    {
      "epoch": 0.1049349072419574,
      "grad_norm": 0.0,
      "learning_rate": 8.951478816394958e-06,
      "loss": 0.0,
      "step": 15210
    },
    {
      "epoch": 0.10500389797650175,
      "grad_norm": 0.0,
      "learning_rate": 8.950788909049516e-06,
      "loss": 0.2579,
      "step": 15220
    },
    {
      "epoch": 0.1050728887110461,
      "grad_norm": 6.467067237281299e-08,
      "learning_rate": 8.950099001704073e-06,
      "loss": 0.0,
      "step": 15230
    },
    {
      "epoch": 0.10514187944559046,
      "grad_norm": 0.0,
      "learning_rate": 8.949409094358629e-06,
      "loss": 0.0,
      "step": 15240
    },
    {
      "epoch": 0.10521087018013481,
      "grad_norm": 95.11103057861328,
      "learning_rate": 8.948719187013186e-06,
      "loss": 0.0141,
      "step": 15250
    },
    {
      "epoch": 0.10527986091467915,
      "grad_norm": 0.0,
      "learning_rate": 8.948029279667742e-06,
      "loss": 0.0012,
      "step": 15260
    },
    {
      "epoch": 0.10534885164922352,
      "grad_norm": 0.0,
      "learning_rate": 8.947339372322299e-06,
      "loss": 0.0219,
      "step": 15270
    },
    {
      "epoch": 0.10541784238376786,
      "grad_norm": 7.990665471879765e-05,
      "learning_rate": 8.946649464976855e-06,
      "loss": 0.0353,
      "step": 15280
    },
    {
      "epoch": 0.10548683311831221,
      "grad_norm": 0.0,
      "learning_rate": 8.94595955763141e-06,
      "loss": 0.0,
      "step": 15290
    },
    {
      "epoch": 0.10555582385285656,
      "grad_norm": 0.01425836980342865,
      "learning_rate": 8.945269650285966e-06,
      "loss": 0.0,
      "step": 15300
    },
    {
      "epoch": 0.10562481458740092,
      "grad_norm": 0.0,
      "learning_rate": 8.944579742940523e-06,
      "loss": 0.0001,
      "step": 15310
    },
    {
      "epoch": 0.10569380532194526,
      "grad_norm": 0.0,
      "learning_rate": 8.94388983559508e-06,
      "loss": 0.0011,
      "step": 15320
    },
    {
      "epoch": 0.10576279605648961,
      "grad_norm": 0.0,
      "learning_rate": 8.943199928249637e-06,
      "loss": 0.0001,
      "step": 15330
    },
    {
      "epoch": 0.10583178679103396,
      "grad_norm": 0.0,
      "learning_rate": 8.942510020904194e-06,
      "loss": 0.0,
      "step": 15340
    },
    {
      "epoch": 0.10590077752557832,
      "grad_norm": 0.0,
      "learning_rate": 8.94182011355875e-06,
      "loss": 0.0,
      "step": 15350
    },
    {
      "epoch": 0.10596976826012267,
      "grad_norm": 0.0,
      "learning_rate": 8.941130206213307e-06,
      "loss": 0.0,
      "step": 15360
    },
    {
      "epoch": 0.10603875899466701,
      "grad_norm": 0.0,
      "learning_rate": 8.940440298867863e-06,
      "loss": 0.0,
      "step": 15370
    },
    {
      "epoch": 0.10610774972921137,
      "grad_norm": 0.0,
      "learning_rate": 8.93975039152242e-06,
      "loss": 0.0003,
      "step": 15380
    },
    {
      "epoch": 0.10617674046375572,
      "grad_norm": 0.0014280946925282478,
      "learning_rate": 8.939060484176976e-06,
      "loss": 0.0,
      "step": 15390
    },
    {
      "epoch": 0.10624573119830007,
      "grad_norm": 0.0,
      "learning_rate": 8.938370576831533e-06,
      "loss": 0.0,
      "step": 15400
    },
    {
      "epoch": 0.10631472193284441,
      "grad_norm": 0.0,
      "learning_rate": 8.937680669486088e-06,
      "loss": 0.0,
      "step": 15410
    },
    {
      "epoch": 0.10638371266738877,
      "grad_norm": 0.0,
      "learning_rate": 8.936990762140644e-06,
      "loss": 0.0003,
      "step": 15420
    },
    {
      "epoch": 0.10645270340193312,
      "grad_norm": 0.48397955298423767,
      "learning_rate": 8.9363008547952e-06,
      "loss": 0.0014,
      "step": 15430
    },
    {
      "epoch": 0.10652169413647747,
      "grad_norm": 4.2676206248870585e-06,
      "learning_rate": 8.935610947449759e-06,
      "loss": 0.0,
      "step": 15440
    },
    {
      "epoch": 0.10659068487102182,
      "grad_norm": 0.0,
      "learning_rate": 8.934921040104315e-06,
      "loss": 0.0718,
      "step": 15450
    },
    {
      "epoch": 0.10665967560556618,
      "grad_norm": 0.0,
      "learning_rate": 8.934231132758872e-06,
      "loss": 0.0001,
      "step": 15460
    },
    {
      "epoch": 0.10672866634011052,
      "grad_norm": 0.03356366232037544,
      "learning_rate": 8.933541225413428e-06,
      "loss": 0.0,
      "step": 15470
    },
    {
      "epoch": 0.10679765707465487,
      "grad_norm": 0.004412492737174034,
      "learning_rate": 8.932851318067985e-06,
      "loss": 0.0,
      "step": 15480
    },
    {
      "epoch": 0.10686664780919923,
      "grad_norm": 0.0,
      "learning_rate": 8.932161410722541e-06,
      "loss": 0.0,
      "step": 15490
    },
    {
      "epoch": 0.10693563854374358,
      "grad_norm": 0.0,
      "learning_rate": 8.931471503377098e-06,
      "loss": 0.0,
      "step": 15500
    },
    {
      "epoch": 0.10700462927828792,
      "grad_norm": 0.0,
      "learning_rate": 8.930781596031654e-06,
      "loss": 0.0,
      "step": 15510
    },
    {
      "epoch": 0.10707362001283227,
      "grad_norm": 0.0,
      "learning_rate": 8.930091688686209e-06,
      "loss": 0.0001,
      "step": 15520
    },
    {
      "epoch": 0.10714261074737663,
      "grad_norm": 1.216089273725629e-08,
      "learning_rate": 8.929401781340766e-06,
      "loss": 0.0,
      "step": 15530
    },
    {
      "epoch": 0.10721160148192098,
      "grad_norm": 0.0,
      "learning_rate": 8.928711873995322e-06,
      "loss": 0.0,
      "step": 15540
    },
    {
      "epoch": 0.10728059221646533,
      "grad_norm": 0.005477714352309704,
      "learning_rate": 8.92802196664988e-06,
      "loss": 0.0,
      "step": 15550
    },
    {
      "epoch": 0.10734958295100967,
      "grad_norm": 0.0,
      "learning_rate": 8.927332059304437e-06,
      "loss": 0.0,
      "step": 15560
    },
    {
      "epoch": 0.10741857368555403,
      "grad_norm": 0.002477702684700489,
      "learning_rate": 8.926642151958993e-06,
      "loss": 0.0,
      "step": 15570
    },
    {
      "epoch": 0.10748756442009838,
      "grad_norm": 5.71883845168486e-07,
      "learning_rate": 8.92595224461355e-06,
      "loss": 0.0,
      "step": 15580
    },
    {
      "epoch": 0.10755655515464273,
      "grad_norm": 0.0,
      "learning_rate": 8.925262337268106e-06,
      "loss": 0.1838,
      "step": 15590
    },
    {
      "epoch": 0.10762554588918709,
      "grad_norm": 0.005755654536187649,
      "learning_rate": 8.924572429922663e-06,
      "loss": 0.0001,
      "step": 15600
    },
    {
      "epoch": 0.10769453662373143,
      "grad_norm": 0.0,
      "learning_rate": 8.92388252257722e-06,
      "loss": 0.0005,
      "step": 15610
    },
    {
      "epoch": 0.10776352735827578,
      "grad_norm": 0.0,
      "learning_rate": 8.923192615231776e-06,
      "loss": 0.0,
      "step": 15620
    },
    {
      "epoch": 0.10783251809282013,
      "grad_norm": 0.0015298360958695412,
      "learning_rate": 8.92250270788633e-06,
      "loss": 0.0,
      "step": 15630
    },
    {
      "epoch": 0.10790150882736449,
      "grad_norm": 0.0,
      "learning_rate": 8.921812800540887e-06,
      "loss": 0.0,
      "step": 15640
    },
    {
      "epoch": 0.10797049956190884,
      "grad_norm": 0.0,
      "learning_rate": 8.921122893195443e-06,
      "loss": 0.0,
      "step": 15650
    },
    {
      "epoch": 0.10803949029645318,
      "grad_norm": 0.0,
      "learning_rate": 8.920432985850002e-06,
      "loss": 0.0,
      "step": 15660
    },
    {
      "epoch": 0.10810848103099753,
      "grad_norm": 2.5580108697909054e-08,
      "learning_rate": 8.919743078504558e-06,
      "loss": 0.0001,
      "step": 15670
    },
    {
      "epoch": 0.10817747176554189,
      "grad_norm": 0.0,
      "learning_rate": 8.919053171159115e-06,
      "loss": 0.5555,
      "step": 15680
    },
    {
      "epoch": 0.10824646250008624,
      "grad_norm": 0.0,
      "learning_rate": 8.918363263813671e-06,
      "loss": 0.0,
      "step": 15690
    },
    {
      "epoch": 0.10831545323463058,
      "grad_norm": 0.45190346240997314,
      "learning_rate": 8.917673356468228e-06,
      "loss": 0.0001,
      "step": 15700
    },
    {
      "epoch": 0.10838444396917495,
      "grad_norm": 0.0,
      "learning_rate": 8.916983449122784e-06,
      "loss": 0.0,
      "step": 15710
    },
    {
      "epoch": 0.10845343470371929,
      "grad_norm": 0.0002167319762520492,
      "learning_rate": 8.91629354177734e-06,
      "loss": 0.0083,
      "step": 15720
    },
    {
      "epoch": 0.10852242543826364,
      "grad_norm": 0.0,
      "learning_rate": 8.915603634431897e-06,
      "loss": 0.0121,
      "step": 15730
    },
    {
      "epoch": 0.10859141617280799,
      "grad_norm": 0.0,
      "learning_rate": 8.914913727086454e-06,
      "loss": 0.0,
      "step": 15740
    },
    {
      "epoch": 0.10866040690735235,
      "grad_norm": 0.0015758657827973366,
      "learning_rate": 8.914223819741008e-06,
      "loss": 0.0094,
      "step": 15750
    },
    {
      "epoch": 0.1087293976418967,
      "grad_norm": 0.0,
      "learning_rate": 8.913533912395565e-06,
      "loss": 0.0001,
      "step": 15760
    },
    {
      "epoch": 0.10879838837644104,
      "grad_norm": 0.0,
      "learning_rate": 8.912844005050123e-06,
      "loss": 0.0,
      "step": 15770
    },
    {
      "epoch": 0.10886737911098539,
      "grad_norm": 0.0,
      "learning_rate": 8.91215409770468e-06,
      "loss": 0.0004,
      "step": 15780
    },
    {
      "epoch": 0.10893636984552975,
      "grad_norm": 0.0,
      "learning_rate": 8.911464190359236e-06,
      "loss": 0.0,
      "step": 15790
    },
    {
      "epoch": 0.1090053605800741,
      "grad_norm": 0.0,
      "learning_rate": 8.910774283013793e-06,
      "loss": 0.0,
      "step": 15800
    },
    {
      "epoch": 0.10907435131461844,
      "grad_norm": 1.0306755939382128e-05,
      "learning_rate": 8.910084375668349e-06,
      "loss": 0.0,
      "step": 15810
    },
    {
      "epoch": 0.1091433420491628,
      "grad_norm": 0.0,
      "learning_rate": 8.909394468322906e-06,
      "loss": 0.0,
      "step": 15820
    },
    {
      "epoch": 0.10921233278370715,
      "grad_norm": 0.0,
      "learning_rate": 8.908704560977462e-06,
      "loss": 0.0,
      "step": 15830
    },
    {
      "epoch": 0.1092813235182515,
      "grad_norm": 3.588132244658482e-07,
      "learning_rate": 8.908014653632018e-06,
      "loss": 0.0,
      "step": 15840
    },
    {
      "epoch": 0.10935031425279584,
      "grad_norm": 141.24696350097656,
      "learning_rate": 8.907324746286575e-06,
      "loss": 0.0244,
      "step": 15850
    },
    {
      "epoch": 0.1094193049873402,
      "grad_norm": 0.03340895101428032,
      "learning_rate": 8.90663483894113e-06,
      "loss": 0.0,
      "step": 15860
    },
    {
      "epoch": 0.10948829572188455,
      "grad_norm": 0.0,
      "learning_rate": 8.905944931595686e-06,
      "loss": 0.2949,
      "step": 15870
    },
    {
      "epoch": 0.1095572864564289,
      "grad_norm": 8.070172043517232e-05,
      "learning_rate": 8.905255024250244e-06,
      "loss": 0.0889,
      "step": 15880
    },
    {
      "epoch": 0.10962627719097326,
      "grad_norm": 0.0,
      "learning_rate": 8.904565116904801e-06,
      "loss": 0.0,
      "step": 15890
    },
    {
      "epoch": 0.1096952679255176,
      "grad_norm": 4.5419182777404785,
      "learning_rate": 8.903875209559357e-06,
      "loss": 0.0013,
      "step": 15900
    },
    {
      "epoch": 0.10976425866006195,
      "grad_norm": 0.0,
      "learning_rate": 8.903185302213914e-06,
      "loss": 0.0,
      "step": 15910
    },
    {
      "epoch": 0.1098332493946063,
      "grad_norm": 0.0,
      "learning_rate": 8.90249539486847e-06,
      "loss": 0.0,
      "step": 15920
    },
    {
      "epoch": 0.10990224012915066,
      "grad_norm": 5.2723315491221e-08,
      "learning_rate": 8.901805487523027e-06,
      "loss": 0.0,
      "step": 15930
    },
    {
      "epoch": 0.10997123086369501,
      "grad_norm": 9.635378006578321e-08,
      "learning_rate": 8.901115580177583e-06,
      "loss": 0.107,
      "step": 15940
    },
    {
      "epoch": 0.11004022159823935,
      "grad_norm": 0.0,
      "learning_rate": 8.90042567283214e-06,
      "loss": 0.0001,
      "step": 15950
    },
    {
      "epoch": 0.1101092123327837,
      "grad_norm": 0.0,
      "learning_rate": 8.899735765486696e-06,
      "loss": 0.0,
      "step": 15960
    },
    {
      "epoch": 0.11017820306732806,
      "grad_norm": 0.0,
      "learning_rate": 8.899045858141251e-06,
      "loss": 0.0008,
      "step": 15970
    },
    {
      "epoch": 0.11024719380187241,
      "grad_norm": 0.0,
      "learning_rate": 8.898355950795808e-06,
      "loss": 0.0,
      "step": 15980
    },
    {
      "epoch": 0.11031618453641676,
      "grad_norm": 0.0,
      "learning_rate": 8.897666043450366e-06,
      "loss": 0.0,
      "step": 15990
    },
    {
      "epoch": 0.11038517527096112,
      "grad_norm": 0.0,
      "learning_rate": 8.896976136104922e-06,
      "loss": 0.0,
      "step": 16000
    },
    {
      "epoch": 0.11045416600550546,
      "grad_norm": 1.895869493484497,
      "learning_rate": 8.896286228759479e-06,
      "loss": 0.0004,
      "step": 16010
    },
    {
      "epoch": 0.11052315674004981,
      "grad_norm": 0.0,
      "learning_rate": 8.895596321414035e-06,
      "loss": 0.0,
      "step": 16020
    },
    {
      "epoch": 0.11059214747459416,
      "grad_norm": 0.0,
      "learning_rate": 8.894906414068592e-06,
      "loss": 0.0,
      "step": 16030
    },
    {
      "epoch": 0.11066113820913852,
      "grad_norm": 0.0001389685639878735,
      "learning_rate": 8.894216506723148e-06,
      "loss": 0.0001,
      "step": 16040
    },
    {
      "epoch": 0.11073012894368287,
      "grad_norm": 0.0,
      "learning_rate": 8.893526599377705e-06,
      "loss": 0.0,
      "step": 16050
    },
    {
      "epoch": 0.11079911967822721,
      "grad_norm": 0.0,
      "learning_rate": 8.892836692032261e-06,
      "loss": 0.0,
      "step": 16060
    },
    {
      "epoch": 0.11086811041277156,
      "grad_norm": 0.0,
      "learning_rate": 8.892146784686818e-06,
      "loss": 0.0,
      "step": 16070
    },
    {
      "epoch": 0.11093710114731592,
      "grad_norm": 0.0,
      "learning_rate": 8.891456877341374e-06,
      "loss": 0.0,
      "step": 16080
    },
    {
      "epoch": 0.11100609188186027,
      "grad_norm": 0.2615267038345337,
      "learning_rate": 8.890766969995929e-06,
      "loss": 0.0,
      "step": 16090
    },
    {
      "epoch": 0.11107508261640461,
      "grad_norm": 0.0004013115249108523,
      "learning_rate": 8.890077062650487e-06,
      "loss": 0.0,
      "step": 16100
    },
    {
      "epoch": 0.11114407335094897,
      "grad_norm": 0.0,
      "learning_rate": 8.889387155305044e-06,
      "loss": 0.0036,
      "step": 16110
    },
    {
      "epoch": 0.11121306408549332,
      "grad_norm": 0.0,
      "learning_rate": 8.8886972479596e-06,
      "loss": 0.0,
      "step": 16120
    },
    {
      "epoch": 0.11128205482003767,
      "grad_norm": 0.0,
      "learning_rate": 8.888007340614157e-06,
      "loss": 0.0,
      "step": 16130
    },
    {
      "epoch": 0.11135104555458201,
      "grad_norm": 3.3078337846603745e-09,
      "learning_rate": 8.887317433268713e-06,
      "loss": 0.0,
      "step": 16140
    },
    {
      "epoch": 0.11142003628912638,
      "grad_norm": 0.0035774235147982836,
      "learning_rate": 8.88662752592327e-06,
      "loss": 0.0,
      "step": 16150
    },
    {
      "epoch": 0.11148902702367072,
      "grad_norm": 0.0,
      "learning_rate": 8.885937618577826e-06,
      "loss": 0.0,
      "step": 16160
    },
    {
      "epoch": 0.11155801775821507,
      "grad_norm": 0.0,
      "learning_rate": 8.885247711232383e-06,
      "loss": 0.0,
      "step": 16170
    },
    {
      "epoch": 0.11162700849275942,
      "grad_norm": 0.0,
      "learning_rate": 8.88455780388694e-06,
      "loss": 0.0,
      "step": 16180
    },
    {
      "epoch": 0.11169599922730378,
      "grad_norm": 0.0,
      "learning_rate": 8.883867896541496e-06,
      "loss": 0.0,
      "step": 16190
    },
    {
      "epoch": 0.11176498996184812,
      "grad_norm": 0.0,
      "learning_rate": 8.88317798919605e-06,
      "loss": 0.0001,
      "step": 16200
    },
    {
      "epoch": 0.11183398069639247,
      "grad_norm": 0.0,
      "learning_rate": 8.882488081850609e-06,
      "loss": 0.0,
      "step": 16210
    },
    {
      "epoch": 0.11190297143093683,
      "grad_norm": 0.0002957073738798499,
      "learning_rate": 8.881798174505165e-06,
      "loss": 0.0,
      "step": 16220
    },
    {
      "epoch": 0.11197196216548118,
      "grad_norm": 0.0,
      "learning_rate": 8.881108267159722e-06,
      "loss": 0.0,
      "step": 16230
    },
    {
      "epoch": 0.11204095290002553,
      "grad_norm": 9.604125079931691e-05,
      "learning_rate": 8.880418359814278e-06,
      "loss": 0.0,
      "step": 16240
    },
    {
      "epoch": 0.11210994363456987,
      "grad_norm": 9.580921611629378e-10,
      "learning_rate": 8.879728452468835e-06,
      "loss": 0.0,
      "step": 16250
    },
    {
      "epoch": 0.11217893436911423,
      "grad_norm": 0.0,
      "learning_rate": 8.879038545123391e-06,
      "loss": 0.002,
      "step": 16260
    },
    {
      "epoch": 0.11224792510365858,
      "grad_norm": 8.946687279909327e-10,
      "learning_rate": 8.878348637777948e-06,
      "loss": 0.0,
      "step": 16270
    },
    {
      "epoch": 0.11231691583820293,
      "grad_norm": 0.0,
      "learning_rate": 8.877658730432504e-06,
      "loss": 0.0,
      "step": 16280
    },
    {
      "epoch": 0.11238590657274727,
      "grad_norm": 0.00015398212417494506,
      "learning_rate": 8.87696882308706e-06,
      "loss": 0.0005,
      "step": 16290
    },
    {
      "epoch": 0.11245489730729163,
      "grad_norm": 0.0,
      "learning_rate": 8.876278915741617e-06,
      "loss": 0.0,
      "step": 16300
    },
    {
      "epoch": 0.11252388804183598,
      "grad_norm": 0.0,
      "learning_rate": 8.875589008396172e-06,
      "loss": 0.0,
      "step": 16310
    },
    {
      "epoch": 0.11259287877638033,
      "grad_norm": 2.7241250677434437e-07,
      "learning_rate": 8.87489910105073e-06,
      "loss": 0.0011,
      "step": 16320
    },
    {
      "epoch": 0.11266186951092469,
      "grad_norm": 0.0,
      "learning_rate": 8.874209193705287e-06,
      "loss": 0.0,
      "step": 16330
    },
    {
      "epoch": 0.11273086024546904,
      "grad_norm": 0.0,
      "learning_rate": 8.873519286359843e-06,
      "loss": 0.0,
      "step": 16340
    },
    {
      "epoch": 0.11279985098001338,
      "grad_norm": 0.0,
      "learning_rate": 8.8728293790144e-06,
      "loss": 0.0344,
      "step": 16350
    },
    {
      "epoch": 0.11286884171455773,
      "grad_norm": 0.0,
      "learning_rate": 8.872139471668956e-06,
      "loss": 0.0,
      "step": 16360
    },
    {
      "epoch": 0.11293783244910209,
      "grad_norm": 1.253390382771613e-05,
      "learning_rate": 8.871449564323512e-06,
      "loss": 0.0,
      "step": 16370
    },
    {
      "epoch": 0.11300682318364644,
      "grad_norm": 0.0,
      "learning_rate": 8.870759656978069e-06,
      "loss": 0.0,
      "step": 16380
    },
    {
      "epoch": 0.11307581391819078,
      "grad_norm": 1.9625547793111764e-05,
      "learning_rate": 8.870069749632625e-06,
      "loss": 0.0026,
      "step": 16390
    },
    {
      "epoch": 0.11314480465273513,
      "grad_norm": 1.0650897941388848e-09,
      "learning_rate": 8.869379842287182e-06,
      "loss": 0.0185,
      "step": 16400
    },
    {
      "epoch": 0.11321379538727949,
      "grad_norm": 0.0,
      "learning_rate": 8.868689934941738e-06,
      "loss": 0.0,
      "step": 16410
    },
    {
      "epoch": 0.11328278612182384,
      "grad_norm": 0.0,
      "learning_rate": 8.868000027596293e-06,
      "loss": 0.0084,
      "step": 16420
    },
    {
      "epoch": 0.11335177685636819,
      "grad_norm": 0.0,
      "learning_rate": 8.867310120250851e-06,
      "loss": 0.0,
      "step": 16430
    },
    {
      "epoch": 0.11342076759091255,
      "grad_norm": 1.967303433048073e-06,
      "learning_rate": 8.866620212905408e-06,
      "loss": 0.0,
      "step": 16440
    },
    {
      "epoch": 0.1134897583254569,
      "grad_norm": 0.0,
      "learning_rate": 8.865930305559964e-06,
      "loss": 0.0,
      "step": 16450
    },
    {
      "epoch": 0.11355874906000124,
      "grad_norm": 1.1420366519132585e-07,
      "learning_rate": 8.865240398214521e-06,
      "loss": 0.0,
      "step": 16460
    },
    {
      "epoch": 0.11362773979454559,
      "grad_norm": 0.0,
      "learning_rate": 8.864550490869077e-06,
      "loss": 0.0,
      "step": 16470
    },
    {
      "epoch": 0.11369673052908995,
      "grad_norm": 8.06945317890495e-05,
      "learning_rate": 8.863860583523634e-06,
      "loss": 0.0,
      "step": 16480
    },
    {
      "epoch": 0.1137657212636343,
      "grad_norm": 0.0,
      "learning_rate": 8.86317067617819e-06,
      "loss": 0.0,
      "step": 16490
    },
    {
      "epoch": 0.11383471199817864,
      "grad_norm": 0.013719487935304642,
      "learning_rate": 8.862480768832747e-06,
      "loss": 0.0,
      "step": 16500
    },
    {
      "epoch": 0.11390370273272299,
      "grad_norm": 0.03387153893709183,
      "learning_rate": 8.861790861487303e-06,
      "loss": 0.0003,
      "step": 16510
    },
    {
      "epoch": 0.11397269346726735,
      "grad_norm": 0.0,
      "learning_rate": 8.86110095414186e-06,
      "loss": 0.0,
      "step": 16520
    },
    {
      "epoch": 0.1140416842018117,
      "grad_norm": 0.0,
      "learning_rate": 8.860411046796416e-06,
      "loss": 0.0001,
      "step": 16530
    },
    {
      "epoch": 0.11411067493635604,
      "grad_norm": 0.0,
      "learning_rate": 8.859721139450973e-06,
      "loss": 0.0,
      "step": 16540
    },
    {
      "epoch": 0.1141796656709004,
      "grad_norm": 0.0,
      "learning_rate": 8.85903123210553e-06,
      "loss": 0.0,
      "step": 16550
    },
    {
      "epoch": 0.11424865640544475,
      "grad_norm": 0.0,
      "learning_rate": 8.858341324760086e-06,
      "loss": 0.0,
      "step": 16560
    },
    {
      "epoch": 0.1143176471399891,
      "grad_norm": 0.0,
      "learning_rate": 8.857651417414642e-06,
      "loss": 0.0,
      "step": 16570
    },
    {
      "epoch": 0.11438663787453345,
      "grad_norm": 0.0,
      "learning_rate": 8.856961510069199e-06,
      "loss": 0.0,
      "step": 16580
    },
    {
      "epoch": 0.1144556286090778,
      "grad_norm": 0.0367586687207222,
      "learning_rate": 8.856271602723755e-06,
      "loss": 0.0,
      "step": 16590
    },
    {
      "epoch": 0.11452461934362215,
      "grad_norm": 0.0,
      "learning_rate": 8.855581695378312e-06,
      "loss": 0.0,
      "step": 16600
    },
    {
      "epoch": 0.1145936100781665,
      "grad_norm": 0.0,
      "learning_rate": 8.854891788032868e-06,
      "loss": 0.0001,
      "step": 16610
    },
    {
      "epoch": 0.11466260081271085,
      "grad_norm": 9.291002411870863e-10,
      "learning_rate": 8.854201880687425e-06,
      "loss": 0.0,
      "step": 16620
    },
    {
      "epoch": 0.11473159154725521,
      "grad_norm": 0.0,
      "learning_rate": 8.853511973341981e-06,
      "loss": 0.0,
      "step": 16630
    },
    {
      "epoch": 0.11480058228179955,
      "grad_norm": 0.0,
      "learning_rate": 8.852822065996538e-06,
      "loss": 0.0,
      "step": 16640
    },
    {
      "epoch": 0.1148695730163439,
      "grad_norm": 0.0,
      "learning_rate": 8.852132158651094e-06,
      "loss": 0.0,
      "step": 16650
    },
    {
      "epoch": 0.11493856375088826,
      "grad_norm": 0.0,
      "learning_rate": 8.85144225130565e-06,
      "loss": 0.0,
      "step": 16660
    },
    {
      "epoch": 0.11500755448543261,
      "grad_norm": 0.0,
      "learning_rate": 8.850752343960207e-06,
      "loss": 0.0192,
      "step": 16670
    },
    {
      "epoch": 0.11507654521997696,
      "grad_norm": 0.0,
      "learning_rate": 8.850062436614764e-06,
      "loss": 0.0,
      "step": 16680
    },
    {
      "epoch": 0.1151455359545213,
      "grad_norm": 1.0388605531375106e-09,
      "learning_rate": 8.84937252926932e-06,
      "loss": 0.0,
      "step": 16690
    },
    {
      "epoch": 0.11521452668906566,
      "grad_norm": 0.0,
      "learning_rate": 8.848682621923877e-06,
      "loss": 0.0,
      "step": 16700
    },
    {
      "epoch": 0.11528351742361001,
      "grad_norm": 0.0,
      "learning_rate": 8.847992714578433e-06,
      "loss": 0.0,
      "step": 16710
    },
    {
      "epoch": 0.11535250815815436,
      "grad_norm": 0.0,
      "learning_rate": 8.84730280723299e-06,
      "loss": 0.0018,
      "step": 16720
    },
    {
      "epoch": 0.1154214988926987,
      "grad_norm": 0.0,
      "learning_rate": 8.846612899887546e-06,
      "loss": 0.0003,
      "step": 16730
    },
    {
      "epoch": 0.11549048962724306,
      "grad_norm": 0.0,
      "learning_rate": 8.845922992542103e-06,
      "loss": 0.0,
      "step": 16740
    },
    {
      "epoch": 0.11555948036178741,
      "grad_norm": 0.0,
      "learning_rate": 8.845233085196659e-06,
      "loss": 0.0,
      "step": 16750
    },
    {
      "epoch": 0.11562847109633176,
      "grad_norm": 0.0,
      "learning_rate": 8.844543177851216e-06,
      "loss": 0.0,
      "step": 16760
    },
    {
      "epoch": 0.11569746183087612,
      "grad_norm": 0.023466231301426888,
      "learning_rate": 8.843922261240316e-06,
      "loss": 0.6711,
      "step": 16770
    },
    {
      "epoch": 0.11576645256542047,
      "grad_norm": 9.900904115056619e-05,
      "learning_rate": 8.843232353894873e-06,
      "loss": 0.0,
      "step": 16780
    },
    {
      "epoch": 0.11583544329996481,
      "grad_norm": 0.0,
      "learning_rate": 8.84254244654943e-06,
      "loss": 0.0,
      "step": 16790
    },
    {
      "epoch": 0.11590443403450916,
      "grad_norm": 0.0,
      "learning_rate": 8.841852539203986e-06,
      "loss": 0.0,
      "step": 16800
    },
    {
      "epoch": 0.11597342476905352,
      "grad_norm": 0.0,
      "learning_rate": 8.841162631858542e-06,
      "loss": 0.0012,
      "step": 16810
    },
    {
      "epoch": 0.11604241550359787,
      "grad_norm": 0.0,
      "learning_rate": 8.840472724513099e-06,
      "loss": 0.0,
      "step": 16820
    },
    {
      "epoch": 0.11611140623814221,
      "grad_norm": 0.0,
      "learning_rate": 8.839782817167655e-06,
      "loss": 0.0011,
      "step": 16830
    },
    {
      "epoch": 0.11618039697268656,
      "grad_norm": 6.143480959508452e-07,
      "learning_rate": 8.839092909822212e-06,
      "loss": 0.0,
      "step": 16840
    },
    {
      "epoch": 0.11624938770723092,
      "grad_norm": 0.000599917839281261,
      "learning_rate": 8.838403002476768e-06,
      "loss": 0.011,
      "step": 16850
    },
    {
      "epoch": 0.11631837844177527,
      "grad_norm": 0.0,
      "learning_rate": 8.837713095131325e-06,
      "loss": 0.0,
      "step": 16860
    },
    {
      "epoch": 0.11638736917631962,
      "grad_norm": 0.0,
      "learning_rate": 8.837023187785881e-06,
      "loss": 0.0,
      "step": 16870
    },
    {
      "epoch": 0.11645635991086398,
      "grad_norm": 0.0,
      "learning_rate": 8.836333280440438e-06,
      "loss": 0.0002,
      "step": 16880
    },
    {
      "epoch": 0.11652535064540832,
      "grad_norm": 0.00526492390781641,
      "learning_rate": 8.835643373094994e-06,
      "loss": 0.0,
      "step": 16890
    },
    {
      "epoch": 0.11659434137995267,
      "grad_norm": 0.0,
      "learning_rate": 8.83495346574955e-06,
      "loss": 0.0001,
      "step": 16900
    },
    {
      "epoch": 0.11666333211449702,
      "grad_norm": 1.077324673914859e-09,
      "learning_rate": 8.834263558404107e-06,
      "loss": 0.0898,
      "step": 16910
    },
    {
      "epoch": 0.11673232284904138,
      "grad_norm": 0.0,
      "learning_rate": 8.833573651058664e-06,
      "loss": 0.0,
      "step": 16920
    },
    {
      "epoch": 0.11680131358358573,
      "grad_norm": 0.0,
      "learning_rate": 8.83288374371322e-06,
      "loss": 0.0008,
      "step": 16930
    },
    {
      "epoch": 0.11687030431813007,
      "grad_norm": 0.0,
      "learning_rate": 8.832193836367777e-06,
      "loss": 0.0108,
      "step": 16940
    },
    {
      "epoch": 0.11693929505267442,
      "grad_norm": 0.752956211566925,
      "learning_rate": 8.831503929022333e-06,
      "loss": 0.0001,
      "step": 16950
    },
    {
      "epoch": 0.11700828578721878,
      "grad_norm": 0.0,
      "learning_rate": 8.83081402167689e-06,
      "loss": 0.0,
      "step": 16960
    },
    {
      "epoch": 0.11707727652176313,
      "grad_norm": 2.4848066004778957e-06,
      "learning_rate": 8.830124114331446e-06,
      "loss": 0.0,
      "step": 16970
    },
    {
      "epoch": 0.11714626725630747,
      "grad_norm": 0.0,
      "learning_rate": 8.829434206986003e-06,
      "loss": 0.0,
      "step": 16980
    },
    {
      "epoch": 0.11721525799085183,
      "grad_norm": 0.0,
      "learning_rate": 8.828744299640559e-06,
      "loss": 0.0,
      "step": 16990
    },
    {
      "epoch": 0.11728424872539618,
      "grad_norm": 0.0,
      "learning_rate": 8.828054392295116e-06,
      "loss": 0.0,
      "step": 17000
    },
    {
      "epoch": 0.11735323945994053,
      "grad_norm": 0.0,
      "learning_rate": 8.827364484949672e-06,
      "loss": 0.0001,
      "step": 17010
    },
    {
      "epoch": 0.11742223019448488,
      "grad_norm": 0.0,
      "learning_rate": 8.826674577604229e-06,
      "loss": 0.3004,
      "step": 17020
    },
    {
      "epoch": 0.11749122092902924,
      "grad_norm": 0.0,
      "learning_rate": 8.825984670258785e-06,
      "loss": 0.0,
      "step": 17030
    },
    {
      "epoch": 0.11756021166357358,
      "grad_norm": 0.0,
      "learning_rate": 8.825294762913342e-06,
      "loss": 0.0,
      "step": 17040
    },
    {
      "epoch": 0.11762920239811793,
      "grad_norm": 3.680776834487915,
      "learning_rate": 8.824604855567898e-06,
      "loss": 0.0008,
      "step": 17050
    },
    {
      "epoch": 0.11769819313266228,
      "grad_norm": 3.346624311006963e-08,
      "learning_rate": 8.823914948222454e-06,
      "loss": 0.0,
      "step": 17060
    },
    {
      "epoch": 0.11776718386720664,
      "grad_norm": 0.0,
      "learning_rate": 8.823225040877011e-06,
      "loss": 0.0,
      "step": 17070
    },
    {
      "epoch": 0.11783617460175098,
      "grad_norm": 0.0,
      "learning_rate": 8.822535133531567e-06,
      "loss": 0.0,
      "step": 17080
    },
    {
      "epoch": 0.11790516533629533,
      "grad_norm": 0.0,
      "learning_rate": 8.821845226186124e-06,
      "loss": 0.0,
      "step": 17090
    },
    {
      "epoch": 0.11797415607083969,
      "grad_norm": 224.3075408935547,
      "learning_rate": 8.82115531884068e-06,
      "loss": 0.0576,
      "step": 17100
    },
    {
      "epoch": 0.11804314680538404,
      "grad_norm": 0.0,
      "learning_rate": 8.820465411495237e-06,
      "loss": 0.0011,
      "step": 17110
    },
    {
      "epoch": 0.11811213753992839,
      "grad_norm": 0.0,
      "learning_rate": 8.819775504149793e-06,
      "loss": 0.0051,
      "step": 17120
    },
    {
      "epoch": 0.11818112827447273,
      "grad_norm": 0.0003071391547564417,
      "learning_rate": 8.81908559680435e-06,
      "loss": 0.0,
      "step": 17130
    },
    {
      "epoch": 0.1182501190090171,
      "grad_norm": 0.05422855168581009,
      "learning_rate": 8.818395689458906e-06,
      "loss": 0.0,
      "step": 17140
    },
    {
      "epoch": 0.11831910974356144,
      "grad_norm": 0.00013419974129647017,
      "learning_rate": 8.817705782113463e-06,
      "loss": 0.0,
      "step": 17150
    },
    {
      "epoch": 0.11838810047810579,
      "grad_norm": 0.0,
      "learning_rate": 8.81701587476802e-06,
      "loss": 0.0,
      "step": 17160
    },
    {
      "epoch": 0.11845709121265015,
      "grad_norm": 7.59888716856949e-05,
      "learning_rate": 8.816325967422576e-06,
      "loss": 0.0001,
      "step": 17170
    },
    {
      "epoch": 0.1185260819471945,
      "grad_norm": 1.506629843106566e-08,
      "learning_rate": 8.815636060077132e-06,
      "loss": 0.1693,
      "step": 17180
    },
    {
      "epoch": 0.11859507268173884,
      "grad_norm": 0.0,
      "learning_rate": 8.814946152731689e-06,
      "loss": 0.001,
      "step": 17190
    },
    {
      "epoch": 0.11866406341628319,
      "grad_norm": 0.0009273168398067355,
      "learning_rate": 8.814256245386245e-06,
      "loss": 0.0,
      "step": 17200
    },
    {
      "epoch": 0.11873305415082755,
      "grad_norm": 0.0,
      "learning_rate": 8.813566338040802e-06,
      "loss": 0.0001,
      "step": 17210
    },
    {
      "epoch": 0.1188020448853719,
      "grad_norm": 0.0,
      "learning_rate": 8.812876430695358e-06,
      "loss": 0.0,
      "step": 17220
    },
    {
      "epoch": 0.11887103561991624,
      "grad_norm": 0.0,
      "learning_rate": 8.812186523349915e-06,
      "loss": 0.0,
      "step": 17230
    },
    {
      "epoch": 0.11894002635446059,
      "grad_norm": 0.0,
      "learning_rate": 8.811496616004471e-06,
      "loss": 0.0001,
      "step": 17240
    },
    {
      "epoch": 0.11900901708900495,
      "grad_norm": 0.0,
      "learning_rate": 8.810806708659028e-06,
      "loss": 0.0,
      "step": 17250
    },
    {
      "epoch": 0.1190780078235493,
      "grad_norm": 0.0,
      "learning_rate": 8.810116801313584e-06,
      "loss": 0.0,
      "step": 17260
    },
    {
      "epoch": 0.11914699855809364,
      "grad_norm": 0.0,
      "learning_rate": 8.80942689396814e-06,
      "loss": 0.0,
      "step": 17270
    },
    {
      "epoch": 0.119215989292638,
      "grad_norm": 0.0,
      "learning_rate": 8.808736986622697e-06,
      "loss": 0.0289,
      "step": 17280
    },
    {
      "epoch": 0.11928498002718235,
      "grad_norm": 0.0,
      "learning_rate": 8.808047079277254e-06,
      "loss": 0.0,
      "step": 17290
    },
    {
      "epoch": 0.1193539707617267,
      "grad_norm": 0.0,
      "learning_rate": 8.80735717193181e-06,
      "loss": 0.0,
      "step": 17300
    },
    {
      "epoch": 0.11942296149627105,
      "grad_norm": 0.0,
      "learning_rate": 8.806667264586367e-06,
      "loss": 0.0,
      "step": 17310
    },
    {
      "epoch": 0.11949195223081541,
      "grad_norm": 0.0,
      "learning_rate": 8.805977357240923e-06,
      "loss": 0.0001,
      "step": 17320
    },
    {
      "epoch": 0.11956094296535975,
      "grad_norm": 0.0004927848931401968,
      "learning_rate": 8.80528744989548e-06,
      "loss": 0.0,
      "step": 17330
    },
    {
      "epoch": 0.1196299336999041,
      "grad_norm": 0.0,
      "learning_rate": 8.804597542550036e-06,
      "loss": 0.0,
      "step": 17340
    },
    {
      "epoch": 0.11969892443444845,
      "grad_norm": 0.0,
      "learning_rate": 8.803907635204593e-06,
      "loss": 0.0055,
      "step": 17350
    },
    {
      "epoch": 0.11976791516899281,
      "grad_norm": 0.0,
      "learning_rate": 8.80321772785915e-06,
      "loss": 0.0048,
      "step": 17360
    },
    {
      "epoch": 0.11983690590353716,
      "grad_norm": 0.0,
      "learning_rate": 8.802527820513706e-06,
      "loss": 0.0,
      "step": 17370
    },
    {
      "epoch": 0.1199058966380815,
      "grad_norm": 0.0,
      "learning_rate": 8.801837913168262e-06,
      "loss": 0.0,
      "step": 17380
    },
    {
      "epoch": 0.11997488737262586,
      "grad_norm": 9.931895306181104e-10,
      "learning_rate": 8.801148005822819e-06,
      "loss": 0.0,
      "step": 17390
    },
    {
      "epoch": 0.12004387810717021,
      "grad_norm": 0.00020726821094285697,
      "learning_rate": 8.800458098477375e-06,
      "loss": 0.2492,
      "step": 17400
    },
    {
      "epoch": 0.12011286884171456,
      "grad_norm": 0.00019490814884193242,
      "learning_rate": 8.799768191131932e-06,
      "loss": 0.0,
      "step": 17410
    },
    {
      "epoch": 0.1201818595762589,
      "grad_norm": 0.0,
      "learning_rate": 8.799078283786488e-06,
      "loss": 0.0007,
      "step": 17420
    },
    {
      "epoch": 0.12025085031080326,
      "grad_norm": 0.0,
      "learning_rate": 8.798388376441045e-06,
      "loss": 0.0,
      "step": 17430
    },
    {
      "epoch": 0.12031984104534761,
      "grad_norm": 9.201019945948019e-10,
      "learning_rate": 8.797698469095601e-06,
      "loss": 0.0,
      "step": 17440
    },
    {
      "epoch": 0.12038883177989196,
      "grad_norm": 0.0,
      "learning_rate": 8.797008561750158e-06,
      "loss": 0.2594,
      "step": 17450
    },
    {
      "epoch": 0.1204578225144363,
      "grad_norm": 0.0,
      "learning_rate": 8.796318654404714e-06,
      "loss": 0.0,
      "step": 17460
    },
    {
      "epoch": 0.12052681324898067,
      "grad_norm": 0.0,
      "learning_rate": 8.79562874705927e-06,
      "loss": 0.0804,
      "step": 17470
    },
    {
      "epoch": 0.12059580398352501,
      "grad_norm": 1.2530210646843898e-08,
      "learning_rate": 8.794938839713827e-06,
      "loss": 0.0001,
      "step": 17480
    },
    {
      "epoch": 0.12066479471806936,
      "grad_norm": 0.0,
      "learning_rate": 8.794248932368384e-06,
      "loss": 0.0,
      "step": 17490
    },
    {
      "epoch": 0.12073378545261372,
      "grad_norm": 1.2081340551376343,
      "learning_rate": 8.79355902502294e-06,
      "loss": 0.0003,
      "step": 17500
    },
    {
      "epoch": 0.12080277618715807,
      "grad_norm": 0.0,
      "learning_rate": 8.792869117677497e-06,
      "loss": 0.0002,
      "step": 17510
    },
    {
      "epoch": 0.12087176692170241,
      "grad_norm": 0.0,
      "learning_rate": 8.792179210332053e-06,
      "loss": 0.0003,
      "step": 17520
    },
    {
      "epoch": 0.12094075765624676,
      "grad_norm": 2.628364370593772e-07,
      "learning_rate": 8.79148930298661e-06,
      "loss": 0.0,
      "step": 17530
    },
    {
      "epoch": 0.12100974839079112,
      "grad_norm": 0.0,
      "learning_rate": 8.790799395641166e-06,
      "loss": 0.0008,
      "step": 17540
    },
    {
      "epoch": 0.12107873912533547,
      "grad_norm": 9.124003554461524e-05,
      "learning_rate": 8.790109488295723e-06,
      "loss": 0.0,
      "step": 17550
    },
    {
      "epoch": 0.12114772985987982,
      "grad_norm": 0.0013275255914777517,
      "learning_rate": 8.789419580950279e-06,
      "loss": 0.0028,
      "step": 17560
    },
    {
      "epoch": 0.12121672059442416,
      "grad_norm": 0.18724779784679413,
      "learning_rate": 8.788729673604836e-06,
      "loss": 0.0,
      "step": 17570
    },
    {
      "epoch": 0.12128571132896852,
      "grad_norm": 0.0,
      "learning_rate": 8.788039766259392e-06,
      "loss": 0.0,
      "step": 17580
    },
    {
      "epoch": 0.12135470206351287,
      "grad_norm": 4.172948138148058e-07,
      "learning_rate": 8.787349858913948e-06,
      "loss": 0.0,
      "step": 17590
    },
    {
      "epoch": 0.12142369279805722,
      "grad_norm": 0.0,
      "learning_rate": 8.786659951568505e-06,
      "loss": 0.0,
      "step": 17600
    },
    {
      "epoch": 0.12149268353260158,
      "grad_norm": 0.0,
      "learning_rate": 8.785970044223061e-06,
      "loss": 0.0,
      "step": 17610
    },
    {
      "epoch": 0.12156167426714592,
      "grad_norm": 0.0,
      "learning_rate": 8.785280136877618e-06,
      "loss": 0.0,
      "step": 17620
    },
    {
      "epoch": 0.12163066500169027,
      "grad_norm": 0.0,
      "learning_rate": 8.784590229532174e-06,
      "loss": 0.0,
      "step": 17630
    },
    {
      "epoch": 0.12169965573623462,
      "grad_norm": 0.0,
      "learning_rate": 8.783900322186731e-06,
      "loss": 0.0,
      "step": 17640
    },
    {
      "epoch": 0.12176864647077898,
      "grad_norm": 0.004924447275698185,
      "learning_rate": 8.783210414841287e-06,
      "loss": 0.0,
      "step": 17650
    },
    {
      "epoch": 0.12183763720532333,
      "grad_norm": 8.936698603356774e-10,
      "learning_rate": 8.782520507495844e-06,
      "loss": 0.0248,
      "step": 17660
    },
    {
      "epoch": 0.12190662793986767,
      "grad_norm": 2.3284343114937656e-05,
      "learning_rate": 8.7818306001504e-06,
      "loss": 0.0022,
      "step": 17670
    },
    {
      "epoch": 0.12197561867441202,
      "grad_norm": 1.5944105768994632e-07,
      "learning_rate": 8.781140692804957e-06,
      "loss": 0.0,
      "step": 17680
    },
    {
      "epoch": 0.12204460940895638,
      "grad_norm": 5.706326192012057e-05,
      "learning_rate": 8.780450785459513e-06,
      "loss": 0.0,
      "step": 17690
    },
    {
      "epoch": 0.12211360014350073,
      "grad_norm": 0.0,
      "learning_rate": 8.77976087811407e-06,
      "loss": 0.3295,
      "step": 17700
    },
    {
      "epoch": 0.12218259087804507,
      "grad_norm": 0.0,
      "learning_rate": 8.779070970768626e-06,
      "loss": 0.0,
      "step": 17710
    },
    {
      "epoch": 0.12225158161258944,
      "grad_norm": 0.0,
      "learning_rate": 8.778381063423183e-06,
      "loss": 0.04,
      "step": 17720
    },
    {
      "epoch": 0.12232057234713378,
      "grad_norm": 0.011400078423321247,
      "learning_rate": 8.77769115607774e-06,
      "loss": 0.0002,
      "step": 17730
    },
    {
      "epoch": 0.12238956308167813,
      "grad_norm": 0.0,
      "learning_rate": 8.777001248732296e-06,
      "loss": 0.0,
      "step": 17740
    },
    {
      "epoch": 0.12245855381622248,
      "grad_norm": 0.0,
      "learning_rate": 8.776311341386852e-06,
      "loss": 0.0006,
      "step": 17750
    },
    {
      "epoch": 0.12252754455076684,
      "grad_norm": 0.0,
      "learning_rate": 8.775621434041409e-06,
      "loss": 0.0,
      "step": 17760
    },
    {
      "epoch": 0.12259653528531118,
      "grad_norm": 0.0,
      "learning_rate": 8.774931526695965e-06,
      "loss": 0.0,
      "step": 17770
    },
    {
      "epoch": 0.12266552601985553,
      "grad_norm": 0.30097076296806335,
      "learning_rate": 8.774241619350522e-06,
      "loss": 0.002,
      "step": 17780
    },
    {
      "epoch": 0.12273451675439988,
      "grad_norm": 0.018053485080599785,
      "learning_rate": 8.773551712005078e-06,
      "loss": 0.0001,
      "step": 17790
    },
    {
      "epoch": 0.12280350748894424,
      "grad_norm": 0.0,
      "learning_rate": 8.772861804659635e-06,
      "loss": 0.1314,
      "step": 17800
    },
    {
      "epoch": 0.12287249822348859,
      "grad_norm": 0.0,
      "learning_rate": 8.772171897314191e-06,
      "loss": 0.0,
      "step": 17810
    },
    {
      "epoch": 0.12294148895803293,
      "grad_norm": 0.0,
      "learning_rate": 8.771481989968748e-06,
      "loss": 0.0,
      "step": 17820
    },
    {
      "epoch": 0.1230104796925773,
      "grad_norm": 0.0,
      "learning_rate": 8.770792082623304e-06,
      "loss": 0.0,
      "step": 17830
    },
    {
      "epoch": 0.12307947042712164,
      "grad_norm": 0.0,
      "learning_rate": 8.77010217527786e-06,
      "loss": 0.0283,
      "step": 17840
    },
    {
      "epoch": 0.12314846116166599,
      "grad_norm": 0.0,
      "learning_rate": 8.769412267932417e-06,
      "loss": 0.0,
      "step": 17850
    },
    {
      "epoch": 0.12321745189621033,
      "grad_norm": 0.002129849512130022,
      "learning_rate": 8.768722360586974e-06,
      "loss": 0.0,
      "step": 17860
    },
    {
      "epoch": 0.1232864426307547,
      "grad_norm": 0.0,
      "learning_rate": 8.76803245324153e-06,
      "loss": 0.0,
      "step": 17870
    },
    {
      "epoch": 0.12335543336529904,
      "grad_norm": 8.501918613568193e-10,
      "learning_rate": 8.767342545896087e-06,
      "loss": 0.0,
      "step": 17880
    },
    {
      "epoch": 0.12342442409984339,
      "grad_norm": 0.0,
      "learning_rate": 8.766652638550643e-06,
      "loss": 0.0,
      "step": 17890
    },
    {
      "epoch": 0.12349341483438774,
      "grad_norm": 0.00872646551579237,
      "learning_rate": 8.7659627312052e-06,
      "loss": 0.0,
      "step": 17900
    },
    {
      "epoch": 0.1235624055689321,
      "grad_norm": 0.0,
      "learning_rate": 8.765272823859756e-06,
      "loss": 0.0,
      "step": 17910
    },
    {
      "epoch": 0.12363139630347644,
      "grad_norm": 0.5586554408073425,
      "learning_rate": 8.764582916514313e-06,
      "loss": 0.0001,
      "step": 17920
    },
    {
      "epoch": 0.12370038703802079,
      "grad_norm": 0.7736665606498718,
      "learning_rate": 8.76389300916887e-06,
      "loss": 0.0003,
      "step": 17930
    },
    {
      "epoch": 0.12376937777256515,
      "grad_norm": 0.0,
      "learning_rate": 8.763203101823426e-06,
      "loss": 0.0,
      "step": 17940
    },
    {
      "epoch": 0.1238383685071095,
      "grad_norm": 0.0,
      "learning_rate": 8.762513194477982e-06,
      "loss": 0.0088,
      "step": 17950
    },
    {
      "epoch": 0.12390735924165384,
      "grad_norm": 4.878197845670229e-09,
      "learning_rate": 8.761823287132539e-06,
      "loss": 0.0,
      "step": 17960
    },
    {
      "epoch": 0.12397634997619819,
      "grad_norm": 0.0,
      "learning_rate": 8.761133379787095e-06,
      "loss": 0.0,
      "step": 17970
    },
    {
      "epoch": 0.12404534071074255,
      "grad_norm": 0.0,
      "learning_rate": 8.760443472441652e-06,
      "loss": 0.0,
      "step": 17980
    },
    {
      "epoch": 0.1241143314452869,
      "grad_norm": 0.0,
      "learning_rate": 8.759753565096208e-06,
      "loss": 0.0,
      "step": 17990
    },
    {
      "epoch": 0.12418332217983125,
      "grad_norm": 0.0,
      "learning_rate": 8.759063657750765e-06,
      "loss": 0.0099,
      "step": 18000
    },
    {
      "epoch": 0.12425231291437559,
      "grad_norm": 0.0,
      "learning_rate": 8.758373750405321e-06,
      "loss": 0.0,
      "step": 18010
    },
    {
      "epoch": 0.12432130364891995,
      "grad_norm": 0.0,
      "learning_rate": 8.757683843059878e-06,
      "loss": 0.0,
      "step": 18020
    },
    {
      "epoch": 0.1243902943834643,
      "grad_norm": 0.0,
      "learning_rate": 8.756993935714434e-06,
      "loss": 0.0,
      "step": 18030
    },
    {
      "epoch": 0.12445928511800865,
      "grad_norm": 0.0,
      "learning_rate": 8.75630402836899e-06,
      "loss": 0.0,
      "step": 18040
    },
    {
      "epoch": 0.12452827585255301,
      "grad_norm": 0.0,
      "learning_rate": 8.755614121023547e-06,
      "loss": 0.0,
      "step": 18050
    },
    {
      "epoch": 0.12459726658709736,
      "grad_norm": 5.171454517949314e-07,
      "learning_rate": 8.754924213678104e-06,
      "loss": 0.0,
      "step": 18060
    },
    {
      "epoch": 0.1246662573216417,
      "grad_norm": 0.0,
      "learning_rate": 8.75423430633266e-06,
      "loss": 0.0008,
      "step": 18070
    },
    {
      "epoch": 0.12473524805618605,
      "grad_norm": 7.356138667091727e-05,
      "learning_rate": 8.753544398987217e-06,
      "loss": 0.0007,
      "step": 18080
    },
    {
      "epoch": 0.12480423879073041,
      "grad_norm": 415.2155456542969,
      "learning_rate": 8.752854491641773e-06,
      "loss": 0.2725,
      "step": 18090
    },
    {
      "epoch": 0.12487322952527476,
      "grad_norm": 0.0,
      "learning_rate": 8.75216458429633e-06,
      "loss": 0.0,
      "step": 18100
    },
    {
      "epoch": 0.1249422202598191,
      "grad_norm": 8.613947888314044e-10,
      "learning_rate": 8.751474676950886e-06,
      "loss": 1.3031,
      "step": 18110
    },
    {
      "epoch": 0.12501121099436346,
      "grad_norm": 5.406222314263687e-08,
      "learning_rate": 8.750784769605442e-06,
      "loss": 0.0573,
      "step": 18120
    },
    {
      "epoch": 0.1250802017289078,
      "grad_norm": 0.0011144365416839719,
      "learning_rate": 8.750094862259999e-06,
      "loss": 0.0,
      "step": 18130
    },
    {
      "epoch": 0.12514919246345216,
      "grad_norm": 0.0,
      "learning_rate": 8.749404954914555e-06,
      "loss": 0.1248,
      "step": 18140
    },
    {
      "epoch": 0.12521818319799652,
      "grad_norm": 0.0,
      "learning_rate": 8.748715047569112e-06,
      "loss": 0.0005,
      "step": 18150
    },
    {
      "epoch": 0.12528717393254085,
      "grad_norm": 0.0,
      "learning_rate": 8.748025140223668e-06,
      "loss": 0.0,
      "step": 18160
    },
    {
      "epoch": 0.1253561646670852,
      "grad_norm": 0.0,
      "learning_rate": 8.747335232878225e-06,
      "loss": 0.0,
      "step": 18170
    },
    {
      "epoch": 0.12542515540162957,
      "grad_norm": 0.00011594170791795477,
      "learning_rate": 8.746645325532781e-06,
      "loss": 0.0,
      "step": 18180
    },
    {
      "epoch": 0.1254941461361739,
      "grad_norm": 0.0,
      "learning_rate": 8.745955418187338e-06,
      "loss": 0.0,
      "step": 18190
    },
    {
      "epoch": 0.12556313687071827,
      "grad_norm": 0.0,
      "learning_rate": 8.745265510841894e-06,
      "loss": 0.0088,
      "step": 18200
    },
    {
      "epoch": 0.1256321276052626,
      "grad_norm": 9.818076351919558e-10,
      "learning_rate": 8.744575603496451e-06,
      "loss": 0.0003,
      "step": 18210
    },
    {
      "epoch": 0.12570111833980696,
      "grad_norm": 0.0,
      "learning_rate": 8.743885696151007e-06,
      "loss": 0.0001,
      "step": 18220
    },
    {
      "epoch": 0.12577010907435132,
      "grad_norm": 0.0,
      "learning_rate": 8.743195788805564e-06,
      "loss": 0.0,
      "step": 18230
    },
    {
      "epoch": 0.12583909980889565,
      "grad_norm": 0.0,
      "learning_rate": 8.74250588146012e-06,
      "loss": 0.0,
      "step": 18240
    },
    {
      "epoch": 0.12590809054344002,
      "grad_norm": 0.0,
      "learning_rate": 8.741815974114677e-06,
      "loss": 0.0,
      "step": 18250
    },
    {
      "epoch": 0.12597708127798438,
      "grad_norm": 0.0,
      "learning_rate": 8.741126066769233e-06,
      "loss": 0.0,
      "step": 18260
    },
    {
      "epoch": 0.1260460720125287,
      "grad_norm": 0.0,
      "learning_rate": 8.74043615942379e-06,
      "loss": 0.0,
      "step": 18270
    },
    {
      "epoch": 0.12611506274707307,
      "grad_norm": 0.0,
      "learning_rate": 8.739746252078346e-06,
      "loss": 0.0,
      "step": 18280
    },
    {
      "epoch": 0.12618405348161743,
      "grad_norm": 0.0,
      "learning_rate": 8.739056344732903e-06,
      "loss": 0.0,
      "step": 18290
    },
    {
      "epoch": 0.12625304421616176,
      "grad_norm": 0.0,
      "learning_rate": 8.73836643738746e-06,
      "loss": 0.0,
      "step": 18300
    },
    {
      "epoch": 0.12632203495070612,
      "grad_norm": 0.0,
      "learning_rate": 8.737676530042016e-06,
      "loss": 0.0092,
      "step": 18310
    },
    {
      "epoch": 0.12639102568525046,
      "grad_norm": 0.0,
      "learning_rate": 8.736986622696572e-06,
      "loss": 0.0,
      "step": 18320
    },
    {
      "epoch": 0.12646001641979482,
      "grad_norm": 0.0,
      "learning_rate": 8.736296715351129e-06,
      "loss": 0.0,
      "step": 18330
    },
    {
      "epoch": 0.12652900715433918,
      "grad_norm": 2.6131767299375497e-05,
      "learning_rate": 8.735606808005685e-06,
      "loss": 0.0,
      "step": 18340
    },
    {
      "epoch": 0.1265979978888835,
      "grad_norm": 9.021263394970447e-06,
      "learning_rate": 8.734916900660242e-06,
      "loss": 0.0,
      "step": 18350
    },
    {
      "epoch": 0.12666698862342787,
      "grad_norm": 0.0,
      "learning_rate": 8.734226993314798e-06,
      "loss": 0.0,
      "step": 18360
    },
    {
      "epoch": 0.12673597935797223,
      "grad_norm": 2.1192390704527497e-06,
      "learning_rate": 8.733537085969355e-06,
      "loss": 0.0,
      "step": 18370
    },
    {
      "epoch": 0.12680497009251657,
      "grad_norm": 0.0,
      "learning_rate": 8.732847178623911e-06,
      "loss": 0.0,
      "step": 18380
    },
    {
      "epoch": 0.12687396082706093,
      "grad_norm": 0.0,
      "learning_rate": 8.732157271278468e-06,
      "loss": 0.0001,
      "step": 18390
    },
    {
      "epoch": 0.1269429515616053,
      "grad_norm": 0.0,
      "learning_rate": 8.731467363933024e-06,
      "loss": 0.0006,
      "step": 18400
    },
    {
      "epoch": 0.12701194229614962,
      "grad_norm": 0.0,
      "learning_rate": 8.73077745658758e-06,
      "loss": 0.0,
      "step": 18410
    },
    {
      "epoch": 0.12708093303069398,
      "grad_norm": 0.0,
      "learning_rate": 8.730087549242137e-06,
      "loss": 0.0,
      "step": 18420
    },
    {
      "epoch": 0.12714992376523832,
      "grad_norm": 0.0,
      "learning_rate": 8.729397641896694e-06,
      "loss": 0.0917,
      "step": 18430
    },
    {
      "epoch": 0.12721891449978268,
      "grad_norm": 0.0,
      "learning_rate": 8.72870773455125e-06,
      "loss": 0.0,
      "step": 18440
    },
    {
      "epoch": 0.12728790523432704,
      "grad_norm": 0.0,
      "learning_rate": 8.728017827205807e-06,
      "loss": 0.0,
      "step": 18450
    },
    {
      "epoch": 0.12735689596887137,
      "grad_norm": 0.0,
      "learning_rate": 8.727327919860363e-06,
      "loss": 0.0014,
      "step": 18460
    },
    {
      "epoch": 0.12742588670341573,
      "grad_norm": 0.0,
      "learning_rate": 8.72663801251492e-06,
      "loss": 0.0,
      "step": 18470
    },
    {
      "epoch": 0.1274948774379601,
      "grad_norm": 8.226572390412912e-05,
      "learning_rate": 8.725948105169476e-06,
      "loss": 0.0,
      "step": 18480
    },
    {
      "epoch": 0.12756386817250442,
      "grad_norm": 0.0,
      "learning_rate": 8.725258197824033e-06,
      "loss": 0.0,
      "step": 18490
    },
    {
      "epoch": 0.12763285890704879,
      "grad_norm": 0.0,
      "learning_rate": 8.724568290478589e-06,
      "loss": 0.0,
      "step": 18500
    },
    {
      "epoch": 0.12770184964159315,
      "grad_norm": 0.0,
      "learning_rate": 8.723878383133146e-06,
      "loss": 0.0,
      "step": 18510
    },
    {
      "epoch": 0.12777084037613748,
      "grad_norm": 0.0,
      "learning_rate": 8.723188475787702e-06,
      "loss": 0.0,
      "step": 18520
    },
    {
      "epoch": 0.12783983111068184,
      "grad_norm": 0.0,
      "learning_rate": 8.722498568442259e-06,
      "loss": 0.0,
      "step": 18530
    },
    {
      "epoch": 0.12790882184522617,
      "grad_norm": 0.0,
      "learning_rate": 8.721808661096815e-06,
      "loss": 0.0,
      "step": 18540
    },
    {
      "epoch": 0.12797781257977053,
      "grad_norm": 0.0,
      "learning_rate": 8.721118753751372e-06,
      "loss": 0.0,
      "step": 18550
    },
    {
      "epoch": 0.1280468033143149,
      "grad_norm": 0.0,
      "learning_rate": 8.720428846405928e-06,
      "loss": 0.0001,
      "step": 18560
    },
    {
      "epoch": 0.12811579404885923,
      "grad_norm": 0.0,
      "learning_rate": 8.719738939060485e-06,
      "loss": 0.0,
      "step": 18570
    },
    {
      "epoch": 0.1281847847834036,
      "grad_norm": 1.5707319889202154e-09,
      "learning_rate": 8.719049031715041e-06,
      "loss": 0.0145,
      "step": 18580
    },
    {
      "epoch": 0.12825377551794795,
      "grad_norm": 0.0,
      "learning_rate": 8.718359124369598e-06,
      "loss": 0.0,
      "step": 18590
    },
    {
      "epoch": 0.12832276625249228,
      "grad_norm": 0.0,
      "learning_rate": 8.717669217024154e-06,
      "loss": 0.0,
      "step": 18600
    },
    {
      "epoch": 0.12839175698703664,
      "grad_norm": 0.0,
      "learning_rate": 8.71697930967871e-06,
      "loss": 0.3123,
      "step": 18610
    },
    {
      "epoch": 0.128460747721581,
      "grad_norm": 0.0,
      "learning_rate": 8.716289402333267e-06,
      "loss": 0.0,
      "step": 18620
    },
    {
      "epoch": 0.12852973845612534,
      "grad_norm": 8.372144089108247e-10,
      "learning_rate": 8.715599494987823e-06,
      "loss": 0.0,
      "step": 18630
    },
    {
      "epoch": 0.1285987291906697,
      "grad_norm": 0.0,
      "learning_rate": 8.71490958764238e-06,
      "loss": 0.0,
      "step": 18640
    },
    {
      "epoch": 0.12866771992521403,
      "grad_norm": 0.5077363848686218,
      "learning_rate": 8.714219680296936e-06,
      "loss": 0.0002,
      "step": 18650
    },
    {
      "epoch": 0.1287367106597584,
      "grad_norm": 0.0,
      "learning_rate": 8.713529772951493e-06,
      "loss": 0.0001,
      "step": 18660
    },
    {
      "epoch": 0.12880570139430275,
      "grad_norm": 0.0,
      "learning_rate": 8.71283986560605e-06,
      "loss": 0.0201,
      "step": 18670
    },
    {
      "epoch": 0.12887469212884708,
      "grad_norm": 0.0,
      "learning_rate": 8.712149958260606e-06,
      "loss": 0.0,
      "step": 18680
    },
    {
      "epoch": 0.12894368286339145,
      "grad_norm": 0.060998305678367615,
      "learning_rate": 8.711460050915162e-06,
      "loss": 0.0,
      "step": 18690
    },
    {
      "epoch": 0.1290126735979358,
      "grad_norm": 0.0,
      "learning_rate": 8.710770143569719e-06,
      "loss": 0.0002,
      "step": 18700
    },
    {
      "epoch": 0.12908166433248014,
      "grad_norm": 1.6868391036987305,
      "learning_rate": 8.710080236224275e-06,
      "loss": 0.3038,
      "step": 18710
    },
    {
      "epoch": 0.1291506550670245,
      "grad_norm": 0.0,
      "learning_rate": 8.709390328878832e-06,
      "loss": 0.0,
      "step": 18720
    },
    {
      "epoch": 0.12921964580156886,
      "grad_norm": 0.0,
      "learning_rate": 8.70870042153339e-06,
      "loss": 0.0,
      "step": 18730
    },
    {
      "epoch": 0.1292886365361132,
      "grad_norm": 0.18403001129627228,
      "learning_rate": 8.708010514187945e-06,
      "loss": 0.0,
      "step": 18740
    },
    {
      "epoch": 0.12935762727065755,
      "grad_norm": 0.0,
      "learning_rate": 8.707320606842501e-06,
      "loss": 0.0015,
      "step": 18750
    },
    {
      "epoch": 0.1294266180052019,
      "grad_norm": 0.0006686258711852133,
      "learning_rate": 8.706630699497058e-06,
      "loss": 0.0,
      "step": 18760
    },
    {
      "epoch": 0.12949560873974625,
      "grad_norm": 0.0,
      "learning_rate": 8.705940792151614e-06,
      "loss": 0.0,
      "step": 18770
    },
    {
      "epoch": 0.1295645994742906,
      "grad_norm": 0.0,
      "learning_rate": 8.70525088480617e-06,
      "loss": 0.0,
      "step": 18780
    },
    {
      "epoch": 0.12963359020883494,
      "grad_norm": 0.0,
      "learning_rate": 8.704560977460727e-06,
      "loss": 0.0142,
      "step": 18790
    },
    {
      "epoch": 0.1297025809433793,
      "grad_norm": 4.698492706012303e-10,
      "learning_rate": 8.703871070115284e-06,
      "loss": 0.0,
      "step": 18800
    },
    {
      "epoch": 0.12977157167792366,
      "grad_norm": 0.0,
      "learning_rate": 8.70318116276984e-06,
      "loss": 0.0001,
      "step": 18810
    },
    {
      "epoch": 0.129840562412468,
      "grad_norm": 0.00023342363419942558,
      "learning_rate": 8.702491255424397e-06,
      "loss": 0.0,
      "step": 18820
    },
    {
      "epoch": 0.12990955314701236,
      "grad_norm": 0.0,
      "learning_rate": 8.701801348078953e-06,
      "loss": 0.0,
      "step": 18830
    },
    {
      "epoch": 0.12997854388155672,
      "grad_norm": 0.0,
      "learning_rate": 8.701111440733511e-06,
      "loss": 0.0,
      "step": 18840
    },
    {
      "epoch": 0.13004753461610105,
      "grad_norm": 0.0,
      "learning_rate": 8.700421533388066e-06,
      "loss": 0.001,
      "step": 18850
    },
    {
      "epoch": 0.1301165253506454,
      "grad_norm": 0.4395293891429901,
      "learning_rate": 8.699731626042623e-06,
      "loss": 0.0001,
      "step": 18860
    },
    {
      "epoch": 0.13018551608518975,
      "grad_norm": 0.0008274639840237796,
      "learning_rate": 8.69904171869718e-06,
      "loss": 0.0002,
      "step": 18870
    },
    {
      "epoch": 0.1302545068197341,
      "grad_norm": 7.702266202613828e-07,
      "learning_rate": 8.698351811351736e-06,
      "loss": 0.0,
      "step": 18880
    },
    {
      "epoch": 0.13032349755427847,
      "grad_norm": 2.2250200345297344e-05,
      "learning_rate": 8.697661904006292e-06,
      "loss": 0.0,
      "step": 18890
    },
    {
      "epoch": 0.1303924882888228,
      "grad_norm": 0.0,
      "learning_rate": 8.696971996660849e-06,
      "loss": 0.0,
      "step": 18900
    },
    {
      "epoch": 0.13046147902336716,
      "grad_norm": 1.3947058505436871e-05,
      "learning_rate": 8.696282089315405e-06,
      "loss": 0.0,
      "step": 18910
    },
    {
      "epoch": 0.13053046975791152,
      "grad_norm": 0.0,
      "learning_rate": 8.695592181969962e-06,
      "loss": 0.0,
      "step": 18920
    },
    {
      "epoch": 0.13059946049245585,
      "grad_norm": 6.807110912632197e-05,
      "learning_rate": 8.694902274624518e-06,
      "loss": 0.0,
      "step": 18930
    },
    {
      "epoch": 0.13066845122700022,
      "grad_norm": 0.0,
      "learning_rate": 8.694212367279075e-06,
      "loss": 0.0,
      "step": 18940
    },
    {
      "epoch": 0.13073744196154458,
      "grad_norm": 0.0,
      "learning_rate": 8.693522459933633e-06,
      "loss": 0.0047,
      "step": 18950
    },
    {
      "epoch": 0.1308064326960889,
      "grad_norm": 0.0002660874743014574,
      "learning_rate": 8.692832552588188e-06,
      "loss": 0.0,
      "step": 18960
    },
    {
      "epoch": 0.13087542343063327,
      "grad_norm": 6.707267061756284e-07,
      "learning_rate": 8.692142645242744e-06,
      "loss": 0.0418,
      "step": 18970
    },
    {
      "epoch": 0.1309444141651776,
      "grad_norm": 0.0,
      "learning_rate": 8.6914527378973e-06,
      "loss": 0.0,
      "step": 18980
    },
    {
      "epoch": 0.13101340489972196,
      "grad_norm": 0.0022617836948484182,
      "learning_rate": 8.690762830551857e-06,
      "loss": 0.0072,
      "step": 18990
    },
    {
      "epoch": 0.13108239563426632,
      "grad_norm": 0.0,
      "learning_rate": 8.690072923206414e-06,
      "loss": 0.0095,
      "step": 19000
    },
    {
      "epoch": 0.13115138636881066,
      "grad_norm": 0.001342968549579382,
      "learning_rate": 8.68938301586097e-06,
      "loss": 0.0,
      "step": 19010
    },
    {
      "epoch": 0.13122037710335502,
      "grad_norm": 0.0,
      "learning_rate": 8.688693108515527e-06,
      "loss": 0.0001,
      "step": 19020
    },
    {
      "epoch": 0.13128936783789938,
      "grad_norm": 0.0,
      "learning_rate": 8.688003201170083e-06,
      "loss": 0.007,
      "step": 19030
    },
    {
      "epoch": 0.1313583585724437,
      "grad_norm": 0.0,
      "learning_rate": 8.68731329382464e-06,
      "loss": 0.0,
      "step": 19040
    },
    {
      "epoch": 0.13142734930698807,
      "grad_norm": 0.0,
      "learning_rate": 8.686623386479196e-06,
      "loss": 0.0,
      "step": 19050
    },
    {
      "epoch": 0.13149634004153243,
      "grad_norm": 5.441776465886505e-06,
      "learning_rate": 8.685933479133754e-06,
      "loss": 0.0,
      "step": 19060
    },
    {
      "epoch": 0.13156533077607677,
      "grad_norm": 0.0,
      "learning_rate": 8.685243571788309e-06,
      "loss": 0.0,
      "step": 19070
    },
    {
      "epoch": 0.13163432151062113,
      "grad_norm": 0.0,
      "learning_rate": 8.684553664442866e-06,
      "loss": 0.0003,
      "step": 19080
    },
    {
      "epoch": 0.1317033122451655,
      "grad_norm": 2.34255912801018e-05,
      "learning_rate": 8.683863757097422e-06,
      "loss": 0.0,
      "step": 19090
    },
    {
      "epoch": 0.13177230297970982,
      "grad_norm": 0.0007809360395185649,
      "learning_rate": 8.683173849751979e-06,
      "loss": 0.0,
      "step": 19100
    },
    {
      "epoch": 0.13184129371425418,
      "grad_norm": 1.6401661468989914e-06,
      "learning_rate": 8.682483942406535e-06,
      "loss": 0.0,
      "step": 19110
    },
    {
      "epoch": 0.13191028444879851,
      "grad_norm": 0.0,
      "learning_rate": 8.681794035061092e-06,
      "loss": 0.0,
      "step": 19120
    },
    {
      "epoch": 0.13197927518334288,
      "grad_norm": 0.0,
      "learning_rate": 8.681104127715648e-06,
      "loss": 0.0014,
      "step": 19130
    },
    {
      "epoch": 0.13204826591788724,
      "grad_norm": 0.0,
      "learning_rate": 8.680414220370204e-06,
      "loss": 0.0054,
      "step": 19140
    },
    {
      "epoch": 0.13211725665243157,
      "grad_norm": 0.0,
      "learning_rate": 8.679724313024761e-06,
      "loss": 0.0,
      "step": 19150
    },
    {
      "epoch": 0.13218624738697593,
      "grad_norm": 1.6095175681130058e-07,
      "learning_rate": 8.679034405679317e-06,
      "loss": 0.0,
      "step": 19160
    },
    {
      "epoch": 0.1322552381215203,
      "grad_norm": 0.0,
      "learning_rate": 8.678344498333876e-06,
      "loss": 0.0001,
      "step": 19170
    },
    {
      "epoch": 0.13232422885606462,
      "grad_norm": 0.0,
      "learning_rate": 8.677654590988432e-06,
      "loss": 0.0001,
      "step": 19180
    },
    {
      "epoch": 0.13239321959060898,
      "grad_norm": 0.0,
      "learning_rate": 8.676964683642987e-06,
      "loss": 0.0,
      "step": 19190
    },
    {
      "epoch": 0.13246221032515335,
      "grad_norm": 0.0,
      "learning_rate": 8.676274776297543e-06,
      "loss": 0.0,
      "step": 19200
    },
    {
      "epoch": 0.13253120105969768,
      "grad_norm": 0.0,
      "learning_rate": 8.6755848689521e-06,
      "loss": 0.0,
      "step": 19210
    },
    {
      "epoch": 0.13260019179424204,
      "grad_norm": 0.0,
      "learning_rate": 8.674894961606656e-06,
      "loss": 0.0,
      "step": 19220
    },
    {
      "epoch": 0.13266918252878637,
      "grad_norm": 0.0,
      "learning_rate": 8.674205054261213e-06,
      "loss": 0.0,
      "step": 19230
    },
    {
      "epoch": 0.13273817326333073,
      "grad_norm": 0.0,
      "learning_rate": 8.67351514691577e-06,
      "loss": 0.0,
      "step": 19240
    },
    {
      "epoch": 0.1328071639978751,
      "grad_norm": 0.0,
      "learning_rate": 8.672825239570326e-06,
      "loss": 0.0,
      "step": 19250
    },
    {
      "epoch": 0.13287615473241943,
      "grad_norm": 0.0,
      "learning_rate": 8.672135332224882e-06,
      "loss": 0.0,
      "step": 19260
    },
    {
      "epoch": 0.1329451454669638,
      "grad_norm": 0.0,
      "learning_rate": 8.671445424879439e-06,
      "loss": 0.0,
      "step": 19270
    },
    {
      "epoch": 0.13301413620150815,
      "grad_norm": 0.0,
      "learning_rate": 8.670755517533997e-06,
      "loss": 0.0,
      "step": 19280
    },
    {
      "epoch": 0.13308312693605248,
      "grad_norm": 0.0,
      "learning_rate": 8.670065610188554e-06,
      "loss": 0.0,
      "step": 19290
    },
    {
      "epoch": 0.13315211767059684,
      "grad_norm": 0.00677242036908865,
      "learning_rate": 8.669375702843108e-06,
      "loss": 0.0,
      "step": 19300
    },
    {
      "epoch": 0.1332211084051412,
      "grad_norm": 0.0,
      "learning_rate": 8.668685795497665e-06,
      "loss": 0.0,
      "step": 19310
    },
    {
      "epoch": 0.13329009913968554,
      "grad_norm": 0.0,
      "learning_rate": 8.667995888152221e-06,
      "loss": 0.0,
      "step": 19320
    },
    {
      "epoch": 0.1333590898742299,
      "grad_norm": 0.0006715624476782978,
      "learning_rate": 8.667305980806778e-06,
      "loss": 0.0,
      "step": 19330
    },
    {
      "epoch": 0.13342808060877423,
      "grad_norm": 0.0,
      "learning_rate": 8.666616073461334e-06,
      "loss": 0.0,
      "step": 19340
    },
    {
      "epoch": 0.1334970713433186,
      "grad_norm": 1.0780977222069055e-09,
      "learning_rate": 8.66592616611589e-06,
      "loss": 0.0,
      "step": 19350
    },
    {
      "epoch": 0.13356606207786295,
      "grad_norm": 3.8380069732666016,
      "learning_rate": 8.665236258770447e-06,
      "loss": 0.0006,
      "step": 19360
    },
    {
      "epoch": 0.13363505281240728,
      "grad_norm": 0.0,
      "learning_rate": 8.664546351425004e-06,
      "loss": 0.0,
      "step": 19370
    },
    {
      "epoch": 0.13370404354695165,
      "grad_norm": 3.7773427266074577e-06,
      "learning_rate": 8.66385644407956e-06,
      "loss": 0.0,
      "step": 19380
    },
    {
      "epoch": 0.133773034281496,
      "grad_norm": 0.0,
      "learning_rate": 8.663166536734118e-06,
      "loss": 0.0,
      "step": 19390
    },
    {
      "epoch": 0.13384202501604034,
      "grad_norm": 0.0,
      "learning_rate": 8.662476629388675e-06,
      "loss": 0.0,
      "step": 19400
    },
    {
      "epoch": 0.1339110157505847,
      "grad_norm": 0.005529284942895174,
      "learning_rate": 8.66178672204323e-06,
      "loss": 0.0,
      "step": 19410
    },
    {
      "epoch": 0.13398000648512906,
      "grad_norm": 2.355862125114072e-05,
      "learning_rate": 8.661096814697786e-06,
      "loss": 0.0,
      "step": 19420
    },
    {
      "epoch": 0.1340489972196734,
      "grad_norm": 0.0,
      "learning_rate": 8.660406907352343e-06,
      "loss": 0.0,
      "step": 19430
    },
    {
      "epoch": 0.13411798795421775,
      "grad_norm": 0.0,
      "learning_rate": 8.6597170000069e-06,
      "loss": 0.0001,
      "step": 19440
    },
    {
      "epoch": 0.1341869786887621,
      "grad_norm": 0.0,
      "learning_rate": 8.659027092661456e-06,
      "loss": 0.0015,
      "step": 19450
    },
    {
      "epoch": 0.13425596942330645,
      "grad_norm": 9.999886970035732e-05,
      "learning_rate": 8.658337185316012e-06,
      "loss": 0.0,
      "step": 19460
    },
    {
      "epoch": 0.1343249601578508,
      "grad_norm": 0.0,
      "learning_rate": 8.657647277970569e-06,
      "loss": 0.0,
      "step": 19470
    },
    {
      "epoch": 0.13439395089239514,
      "grad_norm": 0.0,
      "learning_rate": 8.656957370625125e-06,
      "loss": 0.0,
      "step": 19480
    },
    {
      "epoch": 0.1344629416269395,
      "grad_norm": 0.0002451766049489379,
      "learning_rate": 8.656267463279682e-06,
      "loss": 0.0024,
      "step": 19490
    },
    {
      "epoch": 0.13453193236148386,
      "grad_norm": 0.0,
      "learning_rate": 8.65557755593424e-06,
      "loss": 0.0,
      "step": 19500
    },
    {
      "epoch": 0.1346009230960282,
      "grad_norm": 0.0,
      "learning_rate": 8.654887648588796e-06,
      "loss": 0.0,
      "step": 19510
    },
    {
      "epoch": 0.13466991383057256,
      "grad_norm": 0.0,
      "learning_rate": 8.654197741243353e-06,
      "loss": 0.0,
      "step": 19520
    },
    {
      "epoch": 0.13473890456511692,
      "grad_norm": 0.0,
      "learning_rate": 8.653507833897908e-06,
      "loss": 0.0005,
      "step": 19530
    },
    {
      "epoch": 0.13480789529966125,
      "grad_norm": 0.0,
      "learning_rate": 8.652817926552464e-06,
      "loss": 0.0001,
      "step": 19540
    },
    {
      "epoch": 0.1348768860342056,
      "grad_norm": 428.6827697753906,
      "learning_rate": 8.65212801920702e-06,
      "loss": 1.448,
      "step": 19550
    },
    {
      "epoch": 0.13494587676874994,
      "grad_norm": 0.0,
      "learning_rate": 8.651438111861577e-06,
      "loss": 0.0,
      "step": 19560
    },
    {
      "epoch": 0.1350148675032943,
      "grad_norm": 0.0,
      "learning_rate": 8.650748204516134e-06,
      "loss": 0.0,
      "step": 19570
    },
    {
      "epoch": 0.13508385823783867,
      "grad_norm": 0.0,
      "learning_rate": 8.65005829717069e-06,
      "loss": 0.0002,
      "step": 19580
    },
    {
      "epoch": 0.135152848972383,
      "grad_norm": 0.0,
      "learning_rate": 8.649368389825247e-06,
      "loss": 0.0001,
      "step": 19590
    },
    {
      "epoch": 0.13522183970692736,
      "grad_norm": 0.3019979000091553,
      "learning_rate": 8.648678482479803e-06,
      "loss": 0.0001,
      "step": 19600
    },
    {
      "epoch": 0.13529083044147172,
      "grad_norm": 8.601876470493153e-05,
      "learning_rate": 8.647988575134361e-06,
      "loss": 0.2959,
      "step": 19610
    },
    {
      "epoch": 0.13535982117601605,
      "grad_norm": 0.0,
      "learning_rate": 8.647298667788918e-06,
      "loss": 0.0,
      "step": 19620
    },
    {
      "epoch": 0.13542881191056041,
      "grad_norm": 6.097987004238803e-09,
      "learning_rate": 8.646608760443474e-06,
      "loss": 0.0002,
      "step": 19630
    },
    {
      "epoch": 0.13549780264510478,
      "grad_norm": 0.0003549469984136522,
      "learning_rate": 8.645918853098029e-06,
      "loss": 0.0115,
      "step": 19640
    },
    {
      "epoch": 0.1355667933796491,
      "grad_norm": 0.0,
      "learning_rate": 8.645228945752585e-06,
      "loss": 0.0,
      "step": 19650
    },
    {
      "epoch": 0.13563578411419347,
      "grad_norm": 0.0,
      "learning_rate": 8.644539038407142e-06,
      "loss": 0.0058,
      "step": 19660
    },
    {
      "epoch": 0.1357047748487378,
      "grad_norm": 0.0,
      "learning_rate": 8.643849131061698e-06,
      "loss": 0.0,
      "step": 19670
    },
    {
      "epoch": 0.13577376558328216,
      "grad_norm": 0.0,
      "learning_rate": 8.643159223716255e-06,
      "loss": 0.0004,
      "step": 19680
    },
    {
      "epoch": 0.13584275631782652,
      "grad_norm": 3.587937116622925,
      "learning_rate": 8.642469316370811e-06,
      "loss": 0.0006,
      "step": 19690
    },
    {
      "epoch": 0.13591174705237086,
      "grad_norm": 6.730941581878369e-09,
      "learning_rate": 8.641779409025368e-06,
      "loss": 0.0009,
      "step": 19700
    },
    {
      "epoch": 0.13598073778691522,
      "grad_norm": 0.0,
      "learning_rate": 8.641089501679924e-06,
      "loss": 0.0013,
      "step": 19710
    },
    {
      "epoch": 0.13604972852145958,
      "grad_norm": 0.0,
      "learning_rate": 8.640399594334483e-06,
      "loss": 0.0,
      "step": 19720
    },
    {
      "epoch": 0.1361187192560039,
      "grad_norm": 0.0,
      "learning_rate": 8.639709686989039e-06,
      "loss": 0.0,
      "step": 19730
    },
    {
      "epoch": 0.13618770999054827,
      "grad_norm": 4.829764366220957e-10,
      "learning_rate": 8.639019779643596e-06,
      "loss": 0.0077,
      "step": 19740
    },
    {
      "epoch": 0.13625670072509263,
      "grad_norm": 0.0,
      "learning_rate": 8.63832987229815e-06,
      "loss": 0.0001,
      "step": 19750
    },
    {
      "epoch": 0.13632569145963697,
      "grad_norm": 0.0,
      "learning_rate": 8.637639964952707e-06,
      "loss": 0.6105,
      "step": 19760
    },
    {
      "epoch": 0.13639468219418133,
      "grad_norm": 0.0,
      "learning_rate": 8.636950057607263e-06,
      "loss": 0.0,
      "step": 19770
    },
    {
      "epoch": 0.13646367292872566,
      "grad_norm": 0.0,
      "learning_rate": 8.63626015026182e-06,
      "loss": 0.0,
      "step": 19780
    },
    {
      "epoch": 0.13653266366327002,
      "grad_norm": 0.0,
      "learning_rate": 8.635570242916376e-06,
      "loss": 0.0067,
      "step": 19790
    },
    {
      "epoch": 0.13660165439781438,
      "grad_norm": 0.0,
      "learning_rate": 8.634880335570933e-06,
      "loss": 0.0,
      "step": 19800
    },
    {
      "epoch": 0.13667064513235871,
      "grad_norm": 0.0,
      "learning_rate": 8.63419042822549e-06,
      "loss": 0.0,
      "step": 19810
    },
    {
      "epoch": 0.13673963586690308,
      "grad_norm": 0.0,
      "learning_rate": 8.633500520880046e-06,
      "loss": 0.0331,
      "step": 19820
    },
    {
      "epoch": 0.13680862660144744,
      "grad_norm": 0.0,
      "learning_rate": 8.632810613534604e-06,
      "loss": 0.0,
      "step": 19830
    },
    {
      "epoch": 0.13687761733599177,
      "grad_norm": 0.0,
      "learning_rate": 8.63212070618916e-06,
      "loss": 0.0,
      "step": 19840
    },
    {
      "epoch": 0.13694660807053613,
      "grad_norm": 377.914306640625,
      "learning_rate": 8.631430798843717e-06,
      "loss": 0.0543,
      "step": 19850
    },
    {
      "epoch": 0.1370155988050805,
      "grad_norm": 0.0,
      "learning_rate": 8.630740891498273e-06,
      "loss": 0.0,
      "step": 19860
    },
    {
      "epoch": 0.13708458953962482,
      "grad_norm": 0.28941604495048523,
      "learning_rate": 8.630050984152828e-06,
      "loss": 0.0,
      "step": 19870
    },
    {
      "epoch": 0.13715358027416918,
      "grad_norm": 0.0,
      "learning_rate": 8.629361076807385e-06,
      "loss": 0.0,
      "step": 19880
    },
    {
      "epoch": 0.13722257100871352,
      "grad_norm": 0.0,
      "learning_rate": 8.628671169461941e-06,
      "loss": 0.0,
      "step": 19890
    },
    {
      "epoch": 0.13729156174325788,
      "grad_norm": 3.1171734917734284e-06,
      "learning_rate": 8.627981262116498e-06,
      "loss": 0.0,
      "step": 19900
    },
    {
      "epoch": 0.13736055247780224,
      "grad_norm": 2.208297100736445e-09,
      "learning_rate": 8.627291354771054e-06,
      "loss": 0.0,
      "step": 19910
    },
    {
      "epoch": 0.13742954321234657,
      "grad_norm": 0.0,
      "learning_rate": 8.62660144742561e-06,
      "loss": 0.0,
      "step": 19920
    },
    {
      "epoch": 0.13749853394689093,
      "grad_norm": 0.000125538426800631,
      "learning_rate": 8.625911540080167e-06,
      "loss": 0.0,
      "step": 19930
    },
    {
      "epoch": 0.1375675246814353,
      "grad_norm": 0.0,
      "learning_rate": 8.625221632734725e-06,
      "loss": 0.365,
      "step": 19940
    },
    {
      "epoch": 0.13763651541597963,
      "grad_norm": 0.0,
      "learning_rate": 8.624531725389282e-06,
      "loss": 0.0,
      "step": 19950
    },
    {
      "epoch": 0.137705506150524,
      "grad_norm": 0.0,
      "learning_rate": 8.623841818043838e-06,
      "loss": 0.0,
      "step": 19960
    },
    {
      "epoch": 0.13777449688506835,
      "grad_norm": 5.0932342787746165e-08,
      "learning_rate": 8.623151910698395e-06,
      "loss": 0.0,
      "step": 19970
    },
    {
      "epoch": 0.13784348761961268,
      "grad_norm": 0.0,
      "learning_rate": 8.62246200335295e-06,
      "loss": 0.0,
      "step": 19980
    },
    {
      "epoch": 0.13791247835415704,
      "grad_norm": 0.0,
      "learning_rate": 8.621772096007506e-06,
      "loss": 0.0005,
      "step": 19990
    },
    {
      "epoch": 0.13798146908870138,
      "grad_norm": 0.0,
      "learning_rate": 8.621082188662063e-06,
      "loss": 0.0009,
      "step": 20000
    },
    {
      "epoch": 0.13805045982324574,
      "grad_norm": 4.4332235120236874e-05,
      "learning_rate": 8.620392281316619e-06,
      "loss": 0.0,
      "step": 20010
    },
    {
      "epoch": 0.1381194505577901,
      "grad_norm": 0.03322438895702362,
      "learning_rate": 8.619702373971176e-06,
      "loss": 0.0,
      "step": 20020
    },
    {
      "epoch": 0.13818844129233443,
      "grad_norm": 0.0,
      "learning_rate": 8.619012466625732e-06,
      "loss": 0.0019,
      "step": 20030
    },
    {
      "epoch": 0.1382574320268788,
      "grad_norm": 1.6360729205189273e-05,
      "learning_rate": 8.618322559280289e-06,
      "loss": 0.0002,
      "step": 20040
    },
    {
      "epoch": 0.13832642276142315,
      "grad_norm": 0.0,
      "learning_rate": 8.617632651934847e-06,
      "loss": 0.0,
      "step": 20050
    },
    {
      "epoch": 0.13839541349596748,
      "grad_norm": 2.6920284312836884e-07,
      "learning_rate": 8.616942744589403e-06,
      "loss": 0.875,
      "step": 20060
    },
    {
      "epoch": 0.13846440423051184,
      "grad_norm": 0.06351414322853088,
      "learning_rate": 8.61625283724396e-06,
      "loss": 0.0,
      "step": 20070
    },
    {
      "epoch": 0.1385333949650562,
      "grad_norm": 0.0,
      "learning_rate": 8.615562929898516e-06,
      "loss": 0.0,
      "step": 20080
    },
    {
      "epoch": 0.13860238569960054,
      "grad_norm": 4.050568591651427e-08,
      "learning_rate": 8.614873022553071e-06,
      "loss": 0.0,
      "step": 20090
    },
    {
      "epoch": 0.1386713764341449,
      "grad_norm": 0.04281700402498245,
      "learning_rate": 8.614183115207628e-06,
      "loss": 0.0001,
      "step": 20100
    },
    {
      "epoch": 0.13874036716868923,
      "grad_norm": 2.1143573576409835e-06,
      "learning_rate": 8.613493207862184e-06,
      "loss": 0.0125,
      "step": 20110
    },
    {
      "epoch": 0.1388093579032336,
      "grad_norm": 0.0,
      "learning_rate": 8.61280330051674e-06,
      "loss": 0.0,
      "step": 20120
    },
    {
      "epoch": 0.13887834863777795,
      "grad_norm": 0.0,
      "learning_rate": 8.612113393171297e-06,
      "loss": 0.0,
      "step": 20130
    },
    {
      "epoch": 0.1389473393723223,
      "grad_norm": 0.0,
      "learning_rate": 8.611423485825854e-06,
      "loss": 0.0,
      "step": 20140
    },
    {
      "epoch": 0.13901633010686665,
      "grad_norm": 0.0,
      "learning_rate": 8.61073357848041e-06,
      "loss": 0.0,
      "step": 20150
    },
    {
      "epoch": 0.139085320841411,
      "grad_norm": 0.0,
      "learning_rate": 8.610043671134968e-06,
      "loss": 0.0,
      "step": 20160
    },
    {
      "epoch": 0.13915431157595534,
      "grad_norm": 0.0,
      "learning_rate": 8.609353763789525e-06,
      "loss": 0.0,
      "step": 20170
    },
    {
      "epoch": 0.1392233023104997,
      "grad_norm": 0.0,
      "learning_rate": 8.608663856444081e-06,
      "loss": 0.0002,
      "step": 20180
    },
    {
      "epoch": 0.13929229304504406,
      "grad_norm": 4.659440056009601e-10,
      "learning_rate": 8.607973949098638e-06,
      "loss": 0.0,
      "step": 20190
    },
    {
      "epoch": 0.1393612837795884,
      "grad_norm": 4.7986690333345905e-05,
      "learning_rate": 8.607284041753194e-06,
      "loss": 0.0,
      "step": 20200
    },
    {
      "epoch": 0.13943027451413276,
      "grad_norm": 2.532990484027664e-09,
      "learning_rate": 8.606594134407749e-06,
      "loss": 0.2168,
      "step": 20210
    },
    {
      "epoch": 0.1394992652486771,
      "grad_norm": 0.0,
      "learning_rate": 8.605904227062305e-06,
      "loss": 0.0,
      "step": 20220
    },
    {
      "epoch": 0.13956825598322145,
      "grad_norm": 0.0,
      "learning_rate": 8.605214319716862e-06,
      "loss": 0.3484,
      "step": 20230
    },
    {
      "epoch": 0.1396372467177658,
      "grad_norm": 0.0,
      "learning_rate": 8.604524412371418e-06,
      "loss": 0.0003,
      "step": 20240
    },
    {
      "epoch": 0.13970623745231014,
      "grad_norm": 0.0,
      "learning_rate": 8.603834505025975e-06,
      "loss": 0.0,
      "step": 20250
    },
    {
      "epoch": 0.1397752281868545,
      "grad_norm": 4.6949327980883027e-08,
      "learning_rate": 8.603144597680531e-06,
      "loss": 0.0,
      "step": 20260
    },
    {
      "epoch": 0.13984421892139887,
      "grad_norm": 0.0,
      "learning_rate": 8.60245469033509e-06,
      "loss": 0.0,
      "step": 20270
    },
    {
      "epoch": 0.1399132096559432,
      "grad_norm": 0.0,
      "learning_rate": 8.601764782989646e-06,
      "loss": 0.0,
      "step": 20280
    },
    {
      "epoch": 0.13998220039048756,
      "grad_norm": 0.0,
      "learning_rate": 8.601074875644203e-06,
      "loss": 0.0,
      "step": 20290
    },
    {
      "epoch": 0.14005119112503192,
      "grad_norm": 0.0011996901594102383,
      "learning_rate": 8.600384968298759e-06,
      "loss": 0.0003,
      "step": 20300
    },
    {
      "epoch": 0.14012018185957625,
      "grad_norm": 7.967141613107742e-08,
      "learning_rate": 8.599695060953316e-06,
      "loss": 0.0911,
      "step": 20310
    },
    {
      "epoch": 0.14018917259412061,
      "grad_norm": 0.0,
      "learning_rate": 8.59900515360787e-06,
      "loss": 0.0,
      "step": 20320
    },
    {
      "epoch": 0.14025816332866495,
      "grad_norm": 0.0,
      "learning_rate": 8.598315246262427e-06,
      "loss": 0.0005,
      "step": 20330
    },
    {
      "epoch": 0.1403271540632093,
      "grad_norm": 0.0,
      "learning_rate": 8.597625338916983e-06,
      "loss": 0.0001,
      "step": 20340
    },
    {
      "epoch": 0.14039614479775367,
      "grad_norm": 0.0,
      "learning_rate": 8.59693543157154e-06,
      "loss": 0.0,
      "step": 20350
    },
    {
      "epoch": 0.140465135532298,
      "grad_norm": 0.0,
      "learning_rate": 8.596245524226096e-06,
      "loss": 0.0002,
      "step": 20360
    },
    {
      "epoch": 0.14053412626684236,
      "grad_norm": 0.0,
      "learning_rate": 8.595555616880653e-06,
      "loss": 0.0,
      "step": 20370
    },
    {
      "epoch": 0.14060311700138672,
      "grad_norm": 0.0,
      "learning_rate": 8.594865709535211e-06,
      "loss": 0.5828,
      "step": 20380
    },
    {
      "epoch": 0.14067210773593106,
      "grad_norm": 145.4329071044922,
      "learning_rate": 8.594175802189767e-06,
      "loss": 0.0276,
      "step": 20390
    },
    {
      "epoch": 0.14074109847047542,
      "grad_norm": 0.0,
      "learning_rate": 8.593485894844324e-06,
      "loss": 0.0,
      "step": 20400
    },
    {
      "epoch": 0.14081008920501978,
      "grad_norm": 0.0,
      "learning_rate": 8.59279598749888e-06,
      "loss": 0.0006,
      "step": 20410
    },
    {
      "epoch": 0.1408790799395641,
      "grad_norm": 0.0,
      "learning_rate": 8.592106080153437e-06,
      "loss": 0.0011,
      "step": 20420
    },
    {
      "epoch": 0.14094807067410847,
      "grad_norm": 0.0,
      "learning_rate": 8.591416172807992e-06,
      "loss": 0.0,
      "step": 20430
    },
    {
      "epoch": 0.1410170614086528,
      "grad_norm": 0.0,
      "learning_rate": 8.590726265462548e-06,
      "loss": 0.0197,
      "step": 20440
    },
    {
      "epoch": 0.14108605214319717,
      "grad_norm": 9.441091970074922e-05,
      "learning_rate": 8.590036358117105e-06,
      "loss": 0.0,
      "step": 20450
    },
    {
      "epoch": 0.14115504287774153,
      "grad_norm": 0.31755271553993225,
      "learning_rate": 8.589346450771661e-06,
      "loss": 0.0001,
      "step": 20460
    },
    {
      "epoch": 0.14122403361228586,
      "grad_norm": 0.0,
      "learning_rate": 8.588656543426218e-06,
      "loss": 0.0,
      "step": 20470
    },
    {
      "epoch": 0.14129302434683022,
      "grad_norm": 0.0,
      "learning_rate": 8.587966636080774e-06,
      "loss": 0.0,
      "step": 20480
    },
    {
      "epoch": 0.14136201508137458,
      "grad_norm": 0.0,
      "learning_rate": 8.587276728735332e-06,
      "loss": 0.0,
      "step": 20490
    },
    {
      "epoch": 0.14143100581591891,
      "grad_norm": 0.0,
      "learning_rate": 8.586586821389889e-06,
      "loss": 0.0,
      "step": 20500
    },
    {
      "epoch": 0.14149999655046328,
      "grad_norm": 0.0,
      "learning_rate": 8.585896914044445e-06,
      "loss": 0.0,
      "step": 20510
    },
    {
      "epoch": 0.14156898728500764,
      "grad_norm": 0.0,
      "learning_rate": 8.585207006699002e-06,
      "loss": 0.0,
      "step": 20520
    },
    {
      "epoch": 0.14163797801955197,
      "grad_norm": 0.0,
      "learning_rate": 8.584517099353558e-06,
      "loss": 0.0,
      "step": 20530
    },
    {
      "epoch": 0.14170696875409633,
      "grad_norm": 0.0026347588282078505,
      "learning_rate": 8.583827192008113e-06,
      "loss": 0.6192,
      "step": 20540
    },
    {
      "epoch": 0.14177595948864066,
      "grad_norm": 3.6158878803253174,
      "learning_rate": 8.58313728466267e-06,
      "loss": 0.0008,
      "step": 20550
    },
    {
      "epoch": 0.14184495022318502,
      "grad_norm": 0.0014746921369805932,
      "learning_rate": 8.582447377317226e-06,
      "loss": 0.5042,
      "step": 20560
    },
    {
      "epoch": 0.14191394095772938,
      "grad_norm": 0.0,
      "learning_rate": 8.581757469971783e-06,
      "loss": 0.0,
      "step": 20570
    },
    {
      "epoch": 0.14198293169227372,
      "grad_norm": 0.0,
      "learning_rate": 8.581067562626339e-06,
      "loss": 0.0,
      "step": 20580
    },
    {
      "epoch": 0.14205192242681808,
      "grad_norm": 0.0,
      "learning_rate": 8.580377655280896e-06,
      "loss": 0.0,
      "step": 20590
    },
    {
      "epoch": 0.14212091316136244,
      "grad_norm": 0.0,
      "learning_rate": 8.579687747935454e-06,
      "loss": 0.0,
      "step": 20600
    },
    {
      "epoch": 0.14218990389590677,
      "grad_norm": 0.22815673053264618,
      "learning_rate": 8.57899784059001e-06,
      "loss": 0.0,
      "step": 20610
    },
    {
      "epoch": 0.14225889463045113,
      "grad_norm": 4.310106160865246e-10,
      "learning_rate": 8.578307933244567e-06,
      "loss": 0.0,
      "step": 20620
    },
    {
      "epoch": 0.1423278853649955,
      "grad_norm": 0.0,
      "learning_rate": 8.577618025899123e-06,
      "loss": 0.0,
      "step": 20630
    },
    {
      "epoch": 0.14239687609953983,
      "grad_norm": 0.0,
      "learning_rate": 8.57692811855368e-06,
      "loss": 0.0,
      "step": 20640
    },
    {
      "epoch": 0.1424658668340842,
      "grad_norm": 0.0,
      "learning_rate": 8.576238211208236e-06,
      "loss": 0.0002,
      "step": 20650
    },
    {
      "epoch": 0.14253485756862852,
      "grad_norm": 0.0,
      "learning_rate": 8.575548303862791e-06,
      "loss": 0.0,
      "step": 20660
    },
    {
      "epoch": 0.14260384830317288,
      "grad_norm": 0.0,
      "learning_rate": 8.574858396517348e-06,
      "loss": 0.0001,
      "step": 20670
    },
    {
      "epoch": 0.14267283903771724,
      "grad_norm": 0.0,
      "learning_rate": 8.574168489171904e-06,
      "loss": 0.0,
      "step": 20680
    },
    {
      "epoch": 0.14274182977226157,
      "grad_norm": 0.0023758942261338234,
      "learning_rate": 8.57347858182646e-06,
      "loss": 0.0002,
      "step": 20690
    },
    {
      "epoch": 0.14281082050680594,
      "grad_norm": 0.0,
      "learning_rate": 8.572788674481017e-06,
      "loss": 0.1063,
      "step": 20700
    },
    {
      "epoch": 0.1428798112413503,
      "grad_norm": 0.0,
      "learning_rate": 8.572098767135575e-06,
      "loss": 0.0,
      "step": 20710
    },
    {
      "epoch": 0.14294880197589463,
      "grad_norm": 1.6173250116935378e-07,
      "learning_rate": 8.571408859790132e-06,
      "loss": 0.0,
      "step": 20720
    },
    {
      "epoch": 0.143017792710439,
      "grad_norm": 0.0,
      "learning_rate": 8.570718952444688e-06,
      "loss": 0.0,
      "step": 20730
    },
    {
      "epoch": 0.14308678344498335,
      "grad_norm": 0.0,
      "learning_rate": 8.570029045099245e-06,
      "loss": 0.0,
      "step": 20740
    },
    {
      "epoch": 0.14315577417952768,
      "grad_norm": 0.0,
      "learning_rate": 8.569339137753801e-06,
      "loss": 0.0003,
      "step": 20750
    },
    {
      "epoch": 0.14322476491407204,
      "grad_norm": 67.3569564819336,
      "learning_rate": 8.568649230408358e-06,
      "loss": 0.0189,
      "step": 20760
    },
    {
      "epoch": 0.14329375564861638,
      "grad_norm": 2.327387865763697e-10,
      "learning_rate": 8.567959323062912e-06,
      "loss": 0.0,
      "step": 20770
    },
    {
      "epoch": 0.14336274638316074,
      "grad_norm": 0.0,
      "learning_rate": 8.567269415717469e-06,
      "loss": 0.0,
      "step": 20780
    },
    {
      "epoch": 0.1434317371177051,
      "grad_norm": 4.710614575742511e-06,
      "learning_rate": 8.566579508372025e-06,
      "loss": 0.0,
      "step": 20790
    },
    {
      "epoch": 0.14350072785224943,
      "grad_norm": 0.0,
      "learning_rate": 8.565889601026582e-06,
      "loss": 0.0,
      "step": 20800
    },
    {
      "epoch": 0.1435697185867938,
      "grad_norm": 0.0010542693780735135,
      "learning_rate": 8.565199693681138e-06,
      "loss": 0.0,
      "step": 20810
    },
    {
      "epoch": 0.14363870932133815,
      "grad_norm": 0.0,
      "learning_rate": 8.564509786335697e-06,
      "loss": 0.0,
      "step": 20820
    },
    {
      "epoch": 0.1437077000558825,
      "grad_norm": 0.0,
      "learning_rate": 8.563819878990253e-06,
      "loss": 0.0,
      "step": 20830
    },
    {
      "epoch": 0.14377669079042685,
      "grad_norm": 4.367059773358051e-06,
      "learning_rate": 8.56312997164481e-06,
      "loss": 0.0006,
      "step": 20840
    },
    {
      "epoch": 0.1438456815249712,
      "grad_norm": 0.0,
      "learning_rate": 8.562440064299366e-06,
      "loss": 0.0,
      "step": 20850
    },
    {
      "epoch": 0.14391467225951554,
      "grad_norm": 0.0,
      "learning_rate": 8.561750156953923e-06,
      "loss": 0.0143,
      "step": 20860
    },
    {
      "epoch": 0.1439836629940599,
      "grad_norm": 0.0,
      "learning_rate": 8.561129240343023e-06,
      "loss": 0.1563,
      "step": 20870
    },
    {
      "epoch": 0.14405265372860424,
      "grad_norm": 0.000269328651484102,
      "learning_rate": 8.560439332997578e-06,
      "loss": 0.0,
      "step": 20880
    },
    {
      "epoch": 0.1441216444631486,
      "grad_norm": 0.0,
      "learning_rate": 8.559749425652134e-06,
      "loss": 0.0,
      "step": 20890
    },
    {
      "epoch": 0.14419063519769296,
      "grad_norm": 0.0,
      "learning_rate": 8.559059518306691e-06,
      "loss": 0.0,
      "step": 20900
    },
    {
      "epoch": 0.1442596259322373,
      "grad_norm": 0.0,
      "learning_rate": 8.55836961096125e-06,
      "loss": 0.0,
      "step": 20910
    },
    {
      "epoch": 0.14432861666678165,
      "grad_norm": 0.0,
      "learning_rate": 8.557679703615806e-06,
      "loss": 0.0,
      "step": 20920
    },
    {
      "epoch": 0.144397607401326,
      "grad_norm": 0.0,
      "learning_rate": 8.556989796270362e-06,
      "loss": 0.0,
      "step": 20930
    },
    {
      "epoch": 0.14446659813587034,
      "grad_norm": 0.0,
      "learning_rate": 8.556299888924919e-06,
      "loss": 0.0,
      "step": 20940
    },
    {
      "epoch": 0.1445355888704147,
      "grad_norm": 0.00013134221080690622,
      "learning_rate": 8.555609981579475e-06,
      "loss": 0.0029,
      "step": 20950
    },
    {
      "epoch": 0.14460457960495907,
      "grad_norm": 0.0,
      "learning_rate": 8.554920074234032e-06,
      "loss": 0.0001,
      "step": 20960
    },
    {
      "epoch": 0.1446735703395034,
      "grad_norm": 0.1842164695262909,
      "learning_rate": 8.554230166888588e-06,
      "loss": 0.7594,
      "step": 20970
    },
    {
      "epoch": 0.14474256107404776,
      "grad_norm": 0.01287443470209837,
      "learning_rate": 8.553540259543145e-06,
      "loss": 0.0,
      "step": 20980
    },
    {
      "epoch": 0.1448115518085921,
      "grad_norm": 0.0,
      "learning_rate": 8.5528503521977e-06,
      "loss": 0.0,
      "step": 20990
    },
    {
      "epoch": 0.14488054254313645,
      "grad_norm": 0.0,
      "learning_rate": 8.552160444852256e-06,
      "loss": 0.0005,
      "step": 21000
    },
    {
      "epoch": 0.14494953327768081,
      "grad_norm": 0.0,
      "learning_rate": 8.551470537506812e-06,
      "loss": 0.0,
      "step": 21010
    },
    {
      "epoch": 0.14501852401222515,
      "grad_norm": 7.392631687253015e-06,
      "learning_rate": 8.55078063016137e-06,
      "loss": 0.0022,
      "step": 21020
    },
    {
      "epoch": 0.1450875147467695,
      "grad_norm": 0.0,
      "learning_rate": 8.550090722815927e-06,
      "loss": 0.0064,
      "step": 21030
    },
    {
      "epoch": 0.14515650548131387,
      "grad_norm": 4.6237408346527786e-10,
      "learning_rate": 8.549400815470484e-06,
      "loss": 0.0,
      "step": 21040
    },
    {
      "epoch": 0.1452254962158582,
      "grad_norm": 0.3659661114215851,
      "learning_rate": 8.54871090812504e-06,
      "loss": 0.0001,
      "step": 21050
    },
    {
      "epoch": 0.14529448695040256,
      "grad_norm": 9.910163800697092e-10,
      "learning_rate": 8.548021000779597e-06,
      "loss": 0.0001,
      "step": 21060
    },
    {
      "epoch": 0.14536347768494692,
      "grad_norm": 0.0,
      "learning_rate": 8.547331093434153e-06,
      "loss": 0.0,
      "step": 21070
    },
    {
      "epoch": 0.14543246841949126,
      "grad_norm": 3.528532943164464e-06,
      "learning_rate": 8.54664118608871e-06,
      "loss": 0.0,
      "step": 21080
    },
    {
      "epoch": 0.14550145915403562,
      "grad_norm": 0.00026312697445973754,
      "learning_rate": 8.545951278743266e-06,
      "loss": 0.0001,
      "step": 21090
    },
    {
      "epoch": 0.14557044988857995,
      "grad_norm": 6.645448991093872e-08,
      "learning_rate": 8.545261371397822e-06,
      "loss": 0.0048,
      "step": 21100
    },
    {
      "epoch": 0.1456394406231243,
      "grad_norm": 0.0,
      "learning_rate": 8.544571464052377e-06,
      "loss": 0.0,
      "step": 21110
    },
    {
      "epoch": 0.14570843135766867,
      "grad_norm": 0.0008390983566641808,
      "learning_rate": 8.543881556706934e-06,
      "loss": 0.0,
      "step": 21120
    },
    {
      "epoch": 0.145777422092213,
      "grad_norm": 0.0,
      "learning_rate": 8.543191649361492e-06,
      "loss": 0.007,
      "step": 21130
    },
    {
      "epoch": 0.14584641282675737,
      "grad_norm": 0.0,
      "learning_rate": 8.542501742016048e-06,
      "loss": 0.0,
      "step": 21140
    },
    {
      "epoch": 0.14591540356130173,
      "grad_norm": 4.408148015500046e-06,
      "learning_rate": 8.541811834670605e-06,
      "loss": 0.0147,
      "step": 21150
    },
    {
      "epoch": 0.14598439429584606,
      "grad_norm": 4.568366795965062e-10,
      "learning_rate": 8.541121927325161e-06,
      "loss": 0.0003,
      "step": 21160
    },
    {
      "epoch": 0.14605338503039042,
      "grad_norm": 0.01356740202754736,
      "learning_rate": 8.540432019979718e-06,
      "loss": 0.0,
      "step": 21170
    },
    {
      "epoch": 0.14612237576493478,
      "grad_norm": 0.0,
      "learning_rate": 8.539742112634274e-06,
      "loss": 0.0,
      "step": 21180
    },
    {
      "epoch": 0.14619136649947911,
      "grad_norm": 0.0,
      "learning_rate": 8.539052205288831e-06,
      "loss": 0.0,
      "step": 21190
    },
    {
      "epoch": 0.14626035723402347,
      "grad_norm": 0.0,
      "learning_rate": 8.538362297943387e-06,
      "loss": 0.0,
      "step": 21200
    },
    {
      "epoch": 0.1463293479685678,
      "grad_norm": 6.831362406956032e-05,
      "learning_rate": 8.537672390597944e-06,
      "loss": 0.0,
      "step": 21210
    },
    {
      "epoch": 0.14639833870311217,
      "grad_norm": 4.554460701911012e-06,
      "learning_rate": 8.536982483252499e-06,
      "loss": 0.1257,
      "step": 21220
    },
    {
      "epoch": 0.14646732943765653,
      "grad_norm": 0.024184178560972214,
      "learning_rate": 8.536292575907055e-06,
      "loss": 0.0,
      "step": 21230
    },
    {
      "epoch": 0.14653632017220086,
      "grad_norm": 1.8706403352553025e-05,
      "learning_rate": 8.535602668561613e-06,
      "loss": 0.0001,
      "step": 21240
    },
    {
      "epoch": 0.14660531090674522,
      "grad_norm": 10.950136184692383,
      "learning_rate": 8.53491276121617e-06,
      "loss": 0.0029,
      "step": 21250
    },
    {
      "epoch": 0.14667430164128958,
      "grad_norm": 0.0,
      "learning_rate": 8.534222853870726e-06,
      "loss": 0.0,
      "step": 21260
    },
    {
      "epoch": 0.14674329237583392,
      "grad_norm": 0.0,
      "learning_rate": 8.533532946525283e-06,
      "loss": 0.0,
      "step": 21270
    },
    {
      "epoch": 0.14681228311037828,
      "grad_norm": 4.391468022557632e-10,
      "learning_rate": 8.53284303917984e-06,
      "loss": 0.0,
      "step": 21280
    },
    {
      "epoch": 0.14688127384492264,
      "grad_norm": 0.03587803989648819,
      "learning_rate": 8.532153131834396e-06,
      "loss": 0.0,
      "step": 21290
    },
    {
      "epoch": 0.14695026457946697,
      "grad_norm": 0.0002556056424509734,
      "learning_rate": 8.531463224488952e-06,
      "loss": 0.0,
      "step": 21300
    },
    {
      "epoch": 0.14701925531401133,
      "grad_norm": 0.0011786416871473193,
      "learning_rate": 8.530773317143509e-06,
      "loss": 0.0,
      "step": 21310
    },
    {
      "epoch": 0.14708824604855567,
      "grad_norm": 0.0024628208484500647,
      "learning_rate": 8.530083409798065e-06,
      "loss": 0.0,
      "step": 21320
    },
    {
      "epoch": 0.14715723678310003,
      "grad_norm": 4.728093472294859e-10,
      "learning_rate": 8.52939350245262e-06,
      "loss": 0.0001,
      "step": 21330
    },
    {
      "epoch": 0.1472262275176444,
      "grad_norm": 0.0,
      "learning_rate": 8.528703595107177e-06,
      "loss": 0.0,
      "step": 21340
    },
    {
      "epoch": 0.14729521825218872,
      "grad_norm": 0.0,
      "learning_rate": 8.528013687761735e-06,
      "loss": 0.0,
      "step": 21350
    },
    {
      "epoch": 0.14736420898673308,
      "grad_norm": 0.0,
      "learning_rate": 8.527323780416291e-06,
      "loss": 0.0839,
      "step": 21360
    },
    {
      "epoch": 0.14743319972127744,
      "grad_norm": 0.0,
      "learning_rate": 8.526633873070848e-06,
      "loss": 0.0,
      "step": 21370
    },
    {
      "epoch": 0.14750219045582177,
      "grad_norm": 5.044664130871013e-10,
      "learning_rate": 8.525943965725404e-06,
      "loss": 0.0,
      "step": 21380
    },
    {
      "epoch": 0.14757118119036614,
      "grad_norm": 9.083923941943794e-07,
      "learning_rate": 8.52525405837996e-06,
      "loss": 0.0,
      "step": 21390
    },
    {
      "epoch": 0.1476401719249105,
      "grad_norm": 9.69434913713485e-05,
      "learning_rate": 8.524564151034517e-06,
      "loss": 0.0,
      "step": 21400
    },
    {
      "epoch": 0.14770916265945483,
      "grad_norm": 0.0,
      "learning_rate": 8.523874243689074e-06,
      "loss": 0.0,
      "step": 21410
    },
    {
      "epoch": 0.1477781533939992,
      "grad_norm": 1.9816625118255615,
      "learning_rate": 8.52318433634363e-06,
      "loss": 0.0003,
      "step": 21420
    },
    {
      "epoch": 0.14784714412854352,
      "grad_norm": 0.0006800815463066101,
      "learning_rate": 8.522494428998187e-06,
      "loss": 0.0,
      "step": 21430
    },
    {
      "epoch": 0.14791613486308788,
      "grad_norm": 2.3139521090342896e-06,
      "learning_rate": 8.521804521652743e-06,
      "loss": 0.0,
      "step": 21440
    },
    {
      "epoch": 0.14798512559763224,
      "grad_norm": 1.0483397261396021e-08,
      "learning_rate": 8.521114614307298e-06,
      "loss": 0.0,
      "step": 21450
    },
    {
      "epoch": 0.14805411633217658,
      "grad_norm": 0.0,
      "learning_rate": 8.520424706961856e-06,
      "loss": 0.0,
      "step": 21460
    },
    {
      "epoch": 0.14812310706672094,
      "grad_norm": 1.1231684027279698e-08,
      "learning_rate": 8.519734799616413e-06,
      "loss": 0.0006,
      "step": 21470
    },
    {
      "epoch": 0.1481920978012653,
      "grad_norm": 3.9493808223767246e-10,
      "learning_rate": 8.519044892270969e-06,
      "loss": 0.0,
      "step": 21480
    },
    {
      "epoch": 0.14826108853580963,
      "grad_norm": 0.0,
      "learning_rate": 8.518354984925526e-06,
      "loss": 0.0,
      "step": 21490
    },
    {
      "epoch": 0.148330079270354,
      "grad_norm": 0.0,
      "learning_rate": 8.517665077580082e-06,
      "loss": 0.0,
      "step": 21500
    },
    {
      "epoch": 0.14839907000489835,
      "grad_norm": 4.80211981290779e-10,
      "learning_rate": 8.516975170234639e-06,
      "loss": 0.0003,
      "step": 21510
    },
    {
      "epoch": 0.1484680607394427,
      "grad_norm": 1.0378970216606831e-07,
      "learning_rate": 8.516285262889195e-06,
      "loss": 0.0,
      "step": 21520
    },
    {
      "epoch": 0.14853705147398705,
      "grad_norm": 0.0,
      "learning_rate": 8.515595355543752e-06,
      "loss": 0.0009,
      "step": 21530
    },
    {
      "epoch": 0.1486060422085314,
      "grad_norm": 0.0,
      "learning_rate": 8.514905448198308e-06,
      "loss": 0.0,
      "step": 21540
    },
    {
      "epoch": 0.14867503294307574,
      "grad_norm": 0.0,
      "learning_rate": 8.514215540852865e-06,
      "loss": 0.0005,
      "step": 21550
    },
    {
      "epoch": 0.1487440236776201,
      "grad_norm": 0.0,
      "learning_rate": 8.51352563350742e-06,
      "loss": 0.0424,
      "step": 21560
    },
    {
      "epoch": 0.14881301441216443,
      "grad_norm": 0.0,
      "learning_rate": 8.512835726161978e-06,
      "loss": 0.2916,
      "step": 21570
    },
    {
      "epoch": 0.1488820051467088,
      "grad_norm": 0.0,
      "learning_rate": 8.512145818816534e-06,
      "loss": 0.0,
      "step": 21580
    },
    {
      "epoch": 0.14895099588125316,
      "grad_norm": 0.0,
      "learning_rate": 8.51145591147109e-06,
      "loss": 0.0,
      "step": 21590
    },
    {
      "epoch": 0.1490199866157975,
      "grad_norm": 0.0,
      "learning_rate": 8.510766004125647e-06,
      "loss": 0.0,
      "step": 21600
    },
    {
      "epoch": 0.14908897735034185,
      "grad_norm": 0.0,
      "learning_rate": 8.510076096780203e-06,
      "loss": 0.0,
      "step": 21610
    },
    {
      "epoch": 0.1491579680848862,
      "grad_norm": 2.675690302567091e-05,
      "learning_rate": 8.50938618943476e-06,
      "loss": 0.0,
      "step": 21620
    },
    {
      "epoch": 0.14922695881943054,
      "grad_norm": 0.0,
      "learning_rate": 8.508696282089316e-06,
      "loss": 0.0,
      "step": 21630
    },
    {
      "epoch": 0.1492959495539749,
      "grad_norm": 0.0017321071354672313,
      "learning_rate": 8.508006374743873e-06,
      "loss": 0.0,
      "step": 21640
    },
    {
      "epoch": 0.14936494028851927,
      "grad_norm": 0.0,
      "learning_rate": 8.50731646739843e-06,
      "loss": 0.0,
      "step": 21650
    },
    {
      "epoch": 0.1494339310230636,
      "grad_norm": 0.00024256906181108207,
      "learning_rate": 8.506626560052986e-06,
      "loss": 0.0,
      "step": 21660
    },
    {
      "epoch": 0.14950292175760796,
      "grad_norm": 0.0,
      "learning_rate": 8.50593665270754e-06,
      "loss": 0.0,
      "step": 21670
    },
    {
      "epoch": 0.1495719124921523,
      "grad_norm": 0.19439753890037537,
      "learning_rate": 8.505246745362099e-06,
      "loss": 0.0,
      "step": 21680
    },
    {
      "epoch": 0.14964090322669665,
      "grad_norm": 0.0,
      "learning_rate": 8.504556838016655e-06,
      "loss": 0.003,
      "step": 21690
    },
    {
      "epoch": 0.14970989396124101,
      "grad_norm": 0.0,
      "learning_rate": 8.503866930671212e-06,
      "loss": 0.0,
      "step": 21700
    },
    {
      "epoch": 0.14977888469578535,
      "grad_norm": 0.0,
      "learning_rate": 8.503177023325768e-06,
      "loss": 0.0,
      "step": 21710
    },
    {
      "epoch": 0.1498478754303297,
      "grad_norm": 0.0,
      "learning_rate": 8.502487115980325e-06,
      "loss": 0.0,
      "step": 21720
    },
    {
      "epoch": 0.14991686616487407,
      "grad_norm": 0.0,
      "learning_rate": 8.501797208634881e-06,
      "loss": 0.0,
      "step": 21730
    },
    {
      "epoch": 0.1499858568994184,
      "grad_norm": 0.0,
      "learning_rate": 8.501107301289438e-06,
      "loss": 0.0,
      "step": 21740
    },
    {
      "epoch": 0.15005484763396276,
      "grad_norm": 2.5094188458751887e-05,
      "learning_rate": 8.500417393943994e-06,
      "loss": 0.0011,
      "step": 21750
    },
    {
      "epoch": 0.15012383836850712,
      "grad_norm": 2.246944745820656e-07,
      "learning_rate": 8.49972748659855e-06,
      "loss": 0.0,
      "step": 21760
    },
    {
      "epoch": 0.15019282910305146,
      "grad_norm": 0.0,
      "learning_rate": 8.499037579253107e-06,
      "loss": 0.0003,
      "step": 21770
    },
    {
      "epoch": 0.15026181983759582,
      "grad_norm": 0.0,
      "learning_rate": 8.498347671907662e-06,
      "loss": 0.0,
      "step": 21780
    },
    {
      "epoch": 0.15033081057214015,
      "grad_norm": 0.1592770665884018,
      "learning_rate": 8.49765776456222e-06,
      "loss": 0.0,
      "step": 21790
    },
    {
      "epoch": 0.1503998013066845,
      "grad_norm": 0.0,
      "learning_rate": 8.496967857216777e-06,
      "loss": 0.0661,
      "step": 21800
    },
    {
      "epoch": 0.15046879204122887,
      "grad_norm": 2.2893316745758057,
      "learning_rate": 8.496277949871333e-06,
      "loss": 0.0045,
      "step": 21810
    },
    {
      "epoch": 0.1505377827757732,
      "grad_norm": 8.9533195932745e-06,
      "learning_rate": 8.49558804252589e-06,
      "loss": 0.0015,
      "step": 21820
    },
    {
      "epoch": 0.15060677351031757,
      "grad_norm": 0.00012073614198016003,
      "learning_rate": 8.494898135180446e-06,
      "loss": 0.0,
      "step": 21830
    },
    {
      "epoch": 0.15067576424486193,
      "grad_norm": 0.0,
      "learning_rate": 8.494208227835003e-06,
      "loss": 0.0007,
      "step": 21840
    },
    {
      "epoch": 0.15074475497940626,
      "grad_norm": 0.0,
      "learning_rate": 8.49351832048956e-06,
      "loss": 0.0,
      "step": 21850
    },
    {
      "epoch": 0.15081374571395062,
      "grad_norm": 9.611086170480121e-06,
      "learning_rate": 8.492828413144116e-06,
      "loss": 0.0007,
      "step": 21860
    },
    {
      "epoch": 0.15088273644849498,
      "grad_norm": 0.0,
      "learning_rate": 8.492138505798672e-06,
      "loss": 0.0917,
      "step": 21870
    },
    {
      "epoch": 0.1509517271830393,
      "grad_norm": 1.0724800176831195e-06,
      "learning_rate": 8.491448598453229e-06,
      "loss": 0.0019,
      "step": 21880
    },
    {
      "epoch": 0.15102071791758367,
      "grad_norm": 0.0,
      "learning_rate": 8.490758691107785e-06,
      "loss": 0.0,
      "step": 21890
    },
    {
      "epoch": 0.151089708652128,
      "grad_norm": 2.9375782105489634e-05,
      "learning_rate": 8.490068783762342e-06,
      "loss": 0.0,
      "step": 21900
    },
    {
      "epoch": 0.15115869938667237,
      "grad_norm": 4.700917988209596e-10,
      "learning_rate": 8.489378876416898e-06,
      "loss": 0.0,
      "step": 21910
    },
    {
      "epoch": 0.15122769012121673,
      "grad_norm": 9.998324390636526e-10,
      "learning_rate": 8.488688969071455e-06,
      "loss": 0.0,
      "step": 21920
    },
    {
      "epoch": 0.15129668085576106,
      "grad_norm": 0.0,
      "learning_rate": 8.487999061726011e-06,
      "loss": 0.1684,
      "step": 21930
    },
    {
      "epoch": 0.15136567159030542,
      "grad_norm": 0.0,
      "learning_rate": 8.487309154380568e-06,
      "loss": 0.0,
      "step": 21940
    },
    {
      "epoch": 0.15143466232484978,
      "grad_norm": 8.427773479091627e-10,
      "learning_rate": 8.486619247035124e-06,
      "loss": 0.0,
      "step": 21950
    },
    {
      "epoch": 0.15150365305939412,
      "grad_norm": 5.873551845550537,
      "learning_rate": 8.48592933968968e-06,
      "loss": 0.0007,
      "step": 21960
    },
    {
      "epoch": 0.15157264379393848,
      "grad_norm": 0.0,
      "learning_rate": 8.485239432344237e-06,
      "loss": 0.0,
      "step": 21970
    },
    {
      "epoch": 0.15164163452848284,
      "grad_norm": 0.0,
      "learning_rate": 8.484549524998794e-06,
      "loss": 0.0,
      "step": 21980
    },
    {
      "epoch": 0.15171062526302717,
      "grad_norm": 0.0,
      "learning_rate": 8.48385961765335e-06,
      "loss": 0.0,
      "step": 21990
    },
    {
      "epoch": 0.15177961599757153,
      "grad_norm": 0.0,
      "learning_rate": 8.483169710307907e-06,
      "loss": 0.0,
      "step": 22000
    },
    {
      "epoch": 0.15184860673211587,
      "grad_norm": 0.0,
      "learning_rate": 8.482479802962463e-06,
      "loss": 0.0,
      "step": 22010
    },
    {
      "epoch": 0.15191759746666023,
      "grad_norm": 0.0,
      "learning_rate": 8.48178989561702e-06,
      "loss": 0.0,
      "step": 22020
    },
    {
      "epoch": 0.1519865882012046,
      "grad_norm": 4.358558177947998,
      "learning_rate": 8.481099988271576e-06,
      "loss": 0.0009,
      "step": 22030
    },
    {
      "epoch": 0.15205557893574892,
      "grad_norm": 2.074382791761309e-06,
      "learning_rate": 8.480410080926133e-06,
      "loss": 0.0,
      "step": 22040
    },
    {
      "epoch": 0.15212456967029328,
      "grad_norm": 0.0,
      "learning_rate": 8.479720173580689e-06,
      "loss": 0.0,
      "step": 22050
    },
    {
      "epoch": 0.15219356040483764,
      "grad_norm": 0.0,
      "learning_rate": 8.479030266235246e-06,
      "loss": 0.0,
      "step": 22060
    },
    {
      "epoch": 0.15226255113938197,
      "grad_norm": 0.0,
      "learning_rate": 8.478340358889802e-06,
      "loss": 0.0103,
      "step": 22070
    },
    {
      "epoch": 0.15233154187392633,
      "grad_norm": 0.0,
      "learning_rate": 8.477650451544359e-06,
      "loss": 0.0906,
      "step": 22080
    },
    {
      "epoch": 0.1524005326084707,
      "grad_norm": 8.175796210707631e-06,
      "learning_rate": 8.476960544198915e-06,
      "loss": 0.0007,
      "step": 22090
    },
    {
      "epoch": 0.15246952334301503,
      "grad_norm": 0.0,
      "learning_rate": 8.476270636853472e-06,
      "loss": 0.2031,
      "step": 22100
    },
    {
      "epoch": 0.1525385140775594,
      "grad_norm": 0.0,
      "learning_rate": 8.475580729508028e-06,
      "loss": 0.0,
      "step": 22110
    },
    {
      "epoch": 0.15260750481210372,
      "grad_norm": 0.0,
      "learning_rate": 8.474890822162584e-06,
      "loss": 0.1562,
      "step": 22120
    },
    {
      "epoch": 0.15267649554664808,
      "grad_norm": 0.4623240828514099,
      "learning_rate": 8.474200914817141e-06,
      "loss": 0.0001,
      "step": 22130
    },
    {
      "epoch": 0.15274548628119244,
      "grad_norm": 0.0,
      "learning_rate": 8.473511007471697e-06,
      "loss": 0.0447,
      "step": 22140
    },
    {
      "epoch": 0.15281447701573678,
      "grad_norm": 0.0,
      "learning_rate": 8.472821100126254e-06,
      "loss": 0.0,
      "step": 22150
    },
    {
      "epoch": 0.15288346775028114,
      "grad_norm": 29.423892974853516,
      "learning_rate": 8.47213119278081e-06,
      "loss": 0.0059,
      "step": 22160
    },
    {
      "epoch": 0.1529524584848255,
      "grad_norm": 0.0,
      "learning_rate": 8.471441285435367e-06,
      "loss": 0.0,
      "step": 22170
    },
    {
      "epoch": 0.15302144921936983,
      "grad_norm": 0.0,
      "learning_rate": 8.470751378089923e-06,
      "loss": 0.0637,
      "step": 22180
    },
    {
      "epoch": 0.1530904399539142,
      "grad_norm": 0.0,
      "learning_rate": 8.47006147074448e-06,
      "loss": 0.0,
      "step": 22190
    },
    {
      "epoch": 0.15315943068845855,
      "grad_norm": 0.0,
      "learning_rate": 8.469371563399036e-06,
      "loss": 0.002,
      "step": 22200
    },
    {
      "epoch": 0.1532284214230029,
      "grad_norm": 0.06562628597021103,
      "learning_rate": 8.468750646788137e-06,
      "loss": 0.8662,
      "step": 22210
    },
    {
      "epoch": 0.15329741215754725,
      "grad_norm": 4.369831458461704e-06,
      "learning_rate": 8.468060739442694e-06,
      "loss": 0.0,
      "step": 22220
    },
    {
      "epoch": 0.15336640289209158,
      "grad_norm": 0.0,
      "learning_rate": 8.46737083209725e-06,
      "loss": 0.0013,
      "step": 22230
    },
    {
      "epoch": 0.15343539362663594,
      "grad_norm": 1.7202839899255196e-06,
      "learning_rate": 8.466680924751807e-06,
      "loss": 0.021,
      "step": 22240
    },
    {
      "epoch": 0.1535043843611803,
      "grad_norm": 0.0,
      "learning_rate": 8.465991017406363e-06,
      "loss": 0.0,
      "step": 22250
    },
    {
      "epoch": 0.15357337509572463,
      "grad_norm": 1.1882375478744507,
      "learning_rate": 8.46530111006092e-06,
      "loss": 0.0002,
      "step": 22260
    },
    {
      "epoch": 0.153642365830269,
      "grad_norm": 4.670365427728029e-09,
      "learning_rate": 8.464611202715476e-06,
      "loss": 0.0045,
      "step": 22270
    },
    {
      "epoch": 0.15371135656481336,
      "grad_norm": 0.0,
      "learning_rate": 8.463921295370033e-06,
      "loss": 0.0,
      "step": 22280
    },
    {
      "epoch": 0.1537803472993577,
      "grad_norm": 0.0,
      "learning_rate": 8.463231388024589e-06,
      "loss": 0.0,
      "step": 22290
    },
    {
      "epoch": 0.15384933803390205,
      "grad_norm": 0.0,
      "learning_rate": 8.462541480679146e-06,
      "loss": 0.0289,
      "step": 22300
    },
    {
      "epoch": 0.1539183287684464,
      "grad_norm": 0.0,
      "learning_rate": 8.461851573333702e-06,
      "loss": 0.1851,
      "step": 22310
    },
    {
      "epoch": 0.15398731950299074,
      "grad_norm": 0.0,
      "learning_rate": 8.461161665988258e-06,
      "loss": 0.0,
      "step": 22320
    },
    {
      "epoch": 0.1540563102375351,
      "grad_norm": 0.0,
      "learning_rate": 8.460471758642815e-06,
      "loss": 0.0,
      "step": 22330
    },
    {
      "epoch": 0.15412530097207944,
      "grad_norm": 1.896715366456192e-05,
      "learning_rate": 8.459781851297371e-06,
      "loss": 0.0,
      "step": 22340
    },
    {
      "epoch": 0.1541942917066238,
      "grad_norm": 0.0,
      "learning_rate": 8.459091943951928e-06,
      "loss": 0.0,
      "step": 22350
    },
    {
      "epoch": 0.15426328244116816,
      "grad_norm": 4.8361735343933105,
      "learning_rate": 8.458402036606484e-06,
      "loss": 0.0007,
      "step": 22360
    },
    {
      "epoch": 0.1543322731757125,
      "grad_norm": 0.0,
      "learning_rate": 8.457712129261041e-06,
      "loss": 0.0,
      "step": 22370
    },
    {
      "epoch": 0.15440126391025685,
      "grad_norm": 0.0,
      "learning_rate": 8.457022221915597e-06,
      "loss": 0.0001,
      "step": 22380
    },
    {
      "epoch": 0.1544702546448012,
      "grad_norm": 0.0,
      "learning_rate": 8.456332314570154e-06,
      "loss": 0.0004,
      "step": 22390
    },
    {
      "epoch": 0.15453924537934555,
      "grad_norm": 0.0,
      "learning_rate": 8.45564240722471e-06,
      "loss": 0.0001,
      "step": 22400
    },
    {
      "epoch": 0.1546082361138899,
      "grad_norm": 0.0,
      "learning_rate": 8.454952499879267e-06,
      "loss": 0.0,
      "step": 22410
    },
    {
      "epoch": 0.15467722684843427,
      "grad_norm": 0.8087576627731323,
      "learning_rate": 8.454262592533823e-06,
      "loss": 0.0001,
      "step": 22420
    },
    {
      "epoch": 0.1547462175829786,
      "grad_norm": 0.0,
      "learning_rate": 8.45357268518838e-06,
      "loss": 0.0,
      "step": 22430
    },
    {
      "epoch": 0.15481520831752296,
      "grad_norm": 0.0,
      "learning_rate": 8.452882777842936e-06,
      "loss": 0.0002,
      "step": 22440
    },
    {
      "epoch": 0.1548841990520673,
      "grad_norm": 0.0,
      "learning_rate": 8.452192870497493e-06,
      "loss": 0.0,
      "step": 22450
    },
    {
      "epoch": 0.15495318978661166,
      "grad_norm": 0.0,
      "learning_rate": 8.45150296315205e-06,
      "loss": 0.0332,
      "step": 22460
    },
    {
      "epoch": 0.15502218052115602,
      "grad_norm": 0.0,
      "learning_rate": 8.450813055806606e-06,
      "loss": 0.0,
      "step": 22470
    },
    {
      "epoch": 0.15509117125570035,
      "grad_norm": 0.0,
      "learning_rate": 8.450123148461162e-06,
      "loss": 0.0,
      "step": 22480
    },
    {
      "epoch": 0.1551601619902447,
      "grad_norm": 0.00048648554366081953,
      "learning_rate": 8.449433241115719e-06,
      "loss": 0.0,
      "step": 22490
    },
    {
      "epoch": 0.15522915272478907,
      "grad_norm": 0.0,
      "learning_rate": 8.448743333770275e-06,
      "loss": 0.0,
      "step": 22500
    },
    {
      "epoch": 0.1552981434593334,
      "grad_norm": 0.0,
      "learning_rate": 8.448053426424832e-06,
      "loss": 0.0001,
      "step": 22510
    },
    {
      "epoch": 0.15536713419387777,
      "grad_norm": 8.714773116480501e-07,
      "learning_rate": 8.447363519079388e-06,
      "loss": 0.0,
      "step": 22520
    },
    {
      "epoch": 0.15543612492842213,
      "grad_norm": 0.0,
      "learning_rate": 8.446673611733945e-06,
      "loss": 0.0,
      "step": 22530
    },
    {
      "epoch": 0.15550511566296646,
      "grad_norm": 0.0,
      "learning_rate": 8.445983704388501e-06,
      "loss": 0.1804,
      "step": 22540
    },
    {
      "epoch": 0.15557410639751082,
      "grad_norm": 0.0,
      "learning_rate": 8.445293797043058e-06,
      "loss": 0.0001,
      "step": 22550
    },
    {
      "epoch": 0.15564309713205515,
      "grad_norm": 0.0,
      "learning_rate": 8.444603889697614e-06,
      "loss": 0.0,
      "step": 22560
    },
    {
      "epoch": 0.1557120878665995,
      "grad_norm": 0.0,
      "learning_rate": 8.44391398235217e-06,
      "loss": 0.0,
      "step": 22570
    },
    {
      "epoch": 0.15578107860114387,
      "grad_norm": 0.0,
      "learning_rate": 8.443224075006727e-06,
      "loss": 0.0,
      "step": 22580
    },
    {
      "epoch": 0.1558500693356882,
      "grad_norm": 0.0,
      "learning_rate": 8.442534167661284e-06,
      "loss": 0.0094,
      "step": 22590
    },
    {
      "epoch": 0.15591906007023257,
      "grad_norm": 0.0,
      "learning_rate": 8.44184426031584e-06,
      "loss": 0.0,
      "step": 22600
    },
    {
      "epoch": 0.15598805080477693,
      "grad_norm": 0.0,
      "learning_rate": 8.441154352970397e-06,
      "loss": 0.0,
      "step": 22610
    },
    {
      "epoch": 0.15605704153932126,
      "grad_norm": 0.0,
      "learning_rate": 8.440464445624953e-06,
      "loss": 0.0,
      "step": 22620
    },
    {
      "epoch": 0.15612603227386562,
      "grad_norm": 0.37789469957351685,
      "learning_rate": 8.43977453827951e-06,
      "loss": 0.0013,
      "step": 22630
    },
    {
      "epoch": 0.15619502300840998,
      "grad_norm": 2.2981808456279396e-07,
      "learning_rate": 8.439084630934066e-06,
      "loss": 0.0,
      "step": 22640
    },
    {
      "epoch": 0.15626401374295432,
      "grad_norm": 0.0,
      "learning_rate": 8.438394723588623e-06,
      "loss": 0.0,
      "step": 22650
    },
    {
      "epoch": 0.15633300447749868,
      "grad_norm": 0.0,
      "learning_rate": 8.437704816243179e-06,
      "loss": 0.0003,
      "step": 22660
    },
    {
      "epoch": 0.156401995212043,
      "grad_norm": 6.841717549832538e-05,
      "learning_rate": 8.437014908897736e-06,
      "loss": 0.0,
      "step": 22670
    },
    {
      "epoch": 0.15647098594658737,
      "grad_norm": 0.0,
      "learning_rate": 8.436325001552292e-06,
      "loss": 0.3749,
      "step": 22680
    },
    {
      "epoch": 0.15653997668113173,
      "grad_norm": 0.0,
      "learning_rate": 8.435635094206849e-06,
      "loss": 0.0003,
      "step": 22690
    },
    {
      "epoch": 0.15660896741567606,
      "grad_norm": 0.0,
      "learning_rate": 8.434945186861405e-06,
      "loss": 0.0001,
      "step": 22700
    },
    {
      "epoch": 0.15667795815022043,
      "grad_norm": 0.0,
      "learning_rate": 8.434255279515962e-06,
      "loss": 0.0,
      "step": 22710
    },
    {
      "epoch": 0.1567469488847648,
      "grad_norm": 0.0,
      "learning_rate": 8.433565372170518e-06,
      "loss": 0.0,
      "step": 22720
    },
    {
      "epoch": 0.15681593961930912,
      "grad_norm": 0.0,
      "learning_rate": 8.432875464825075e-06,
      "loss": 0.0008,
      "step": 22730
    },
    {
      "epoch": 0.15688493035385348,
      "grad_norm": 0.0,
      "learning_rate": 8.432185557479631e-06,
      "loss": 0.0,
      "step": 22740
    },
    {
      "epoch": 0.15695392108839784,
      "grad_norm": 0.0,
      "learning_rate": 8.431495650134188e-06,
      "loss": 0.0,
      "step": 22750
    },
    {
      "epoch": 0.15702291182294217,
      "grad_norm": 0.0,
      "learning_rate": 8.430805742788744e-06,
      "loss": 0.0001,
      "step": 22760
    },
    {
      "epoch": 0.15709190255748653,
      "grad_norm": 1.8915983446277096e-06,
      "learning_rate": 8.4301158354433e-06,
      "loss": 0.0,
      "step": 22770
    },
    {
      "epoch": 0.15716089329203087,
      "grad_norm": 0.0,
      "learning_rate": 8.429425928097857e-06,
      "loss": 0.0186,
      "step": 22780
    },
    {
      "epoch": 0.15722988402657523,
      "grad_norm": 3.625578756327741e-06,
      "learning_rate": 8.428736020752414e-06,
      "loss": 0.0002,
      "step": 22790
    },
    {
      "epoch": 0.1572988747611196,
      "grad_norm": 1.0459139110352567e-09,
      "learning_rate": 8.42804611340697e-06,
      "loss": 0.0,
      "step": 22800
    },
    {
      "epoch": 0.15736786549566392,
      "grad_norm": 2.289341449737549,
      "learning_rate": 8.427356206061527e-06,
      "loss": 0.0007,
      "step": 22810
    },
    {
      "epoch": 0.15743685623020828,
      "grad_norm": 0.0,
      "learning_rate": 8.426666298716083e-06,
      "loss": 0.0003,
      "step": 22820
    },
    {
      "epoch": 0.15750584696475264,
      "grad_norm": 0.0,
      "learning_rate": 8.42597639137064e-06,
      "loss": 0.0,
      "step": 22830
    },
    {
      "epoch": 0.15757483769929698,
      "grad_norm": 1.2251314407407676e-09,
      "learning_rate": 8.425286484025196e-06,
      "loss": 0.0,
      "step": 22840
    },
    {
      "epoch": 0.15764382843384134,
      "grad_norm": 0.0,
      "learning_rate": 8.424596576679752e-06,
      "loss": 0.0,
      "step": 22850
    },
    {
      "epoch": 0.1577128191683857,
      "grad_norm": 0.0,
      "learning_rate": 8.423906669334309e-06,
      "loss": 0.0,
      "step": 22860
    },
    {
      "epoch": 0.15778180990293003,
      "grad_norm": 0.0,
      "learning_rate": 8.423216761988865e-06,
      "loss": 0.0,
      "step": 22870
    },
    {
      "epoch": 0.1578508006374744,
      "grad_norm": 1.9375049760128604e-06,
      "learning_rate": 8.422526854643422e-06,
      "loss": 0.001,
      "step": 22880
    },
    {
      "epoch": 0.15791979137201873,
      "grad_norm": 0.0,
      "learning_rate": 8.421836947297978e-06,
      "loss": 0.1014,
      "step": 22890
    },
    {
      "epoch": 0.15798878210656309,
      "grad_norm": 0.0,
      "learning_rate": 8.421147039952535e-06,
      "loss": 0.0,
      "step": 22900
    },
    {
      "epoch": 0.15805777284110745,
      "grad_norm": 0.17923089861869812,
      "learning_rate": 8.420457132607091e-06,
      "loss": 0.0,
      "step": 22910
    },
    {
      "epoch": 0.15812676357565178,
      "grad_norm": 0.013974872417747974,
      "learning_rate": 8.419767225261648e-06,
      "loss": 0.0,
      "step": 22920
    },
    {
      "epoch": 0.15819575431019614,
      "grad_norm": 0.0,
      "learning_rate": 8.419077317916204e-06,
      "loss": 0.0,
      "step": 22930
    },
    {
      "epoch": 0.1582647450447405,
      "grad_norm": 0.0,
      "learning_rate": 8.418387410570761e-06,
      "loss": 0.0207,
      "step": 22940
    },
    {
      "epoch": 0.15833373577928483,
      "grad_norm": 0.0,
      "learning_rate": 8.417697503225317e-06,
      "loss": 0.0,
      "step": 22950
    },
    {
      "epoch": 0.1584027265138292,
      "grad_norm": 9.797831435065518e-10,
      "learning_rate": 8.417007595879874e-06,
      "loss": 0.0,
      "step": 22960
    },
    {
      "epoch": 0.15847171724837356,
      "grad_norm": 0.0,
      "learning_rate": 8.41631768853443e-06,
      "loss": 0.0001,
      "step": 22970
    },
    {
      "epoch": 0.1585407079829179,
      "grad_norm": 0.0,
      "learning_rate": 8.415627781188987e-06,
      "loss": 0.0,
      "step": 22980
    },
    {
      "epoch": 0.15860969871746225,
      "grad_norm": 5.27865813637618e-06,
      "learning_rate": 8.414937873843543e-06,
      "loss": 0.0,
      "step": 22990
    },
    {
      "epoch": 0.15867868945200658,
      "grad_norm": 0.0015955582493916154,
      "learning_rate": 8.4142479664981e-06,
      "loss": 0.0,
      "step": 23000
    },
    {
      "epoch": 0.15874768018655094,
      "grad_norm": 2.0678694312437074e-09,
      "learning_rate": 8.413558059152656e-06,
      "loss": 0.0001,
      "step": 23010
    },
    {
      "epoch": 0.1588166709210953,
      "grad_norm": 0.0,
      "learning_rate": 8.412868151807213e-06,
      "loss": 0.0,
      "step": 23020
    },
    {
      "epoch": 0.15888566165563964,
      "grad_norm": 7.874262337281834e-06,
      "learning_rate": 8.41217824446177e-06,
      "loss": 0.0,
      "step": 23030
    },
    {
      "epoch": 0.158954652390184,
      "grad_norm": 0.0,
      "learning_rate": 8.411488337116326e-06,
      "loss": 0.0,
      "step": 23040
    },
    {
      "epoch": 0.15902364312472836,
      "grad_norm": 0.0,
      "learning_rate": 8.410798429770882e-06,
      "loss": 0.0,
      "step": 23050
    },
    {
      "epoch": 0.1590926338592727,
      "grad_norm": 0.0,
      "learning_rate": 8.410108522425439e-06,
      "loss": 0.0,
      "step": 23060
    },
    {
      "epoch": 0.15916162459381705,
      "grad_norm": 0.00026652630185708404,
      "learning_rate": 8.409418615079995e-06,
      "loss": 0.0058,
      "step": 23070
    },
    {
      "epoch": 0.1592306153283614,
      "grad_norm": 0.0,
      "learning_rate": 8.408728707734552e-06,
      "loss": 0.0,
      "step": 23080
    },
    {
      "epoch": 0.15929960606290575,
      "grad_norm": 0.0004675178788602352,
      "learning_rate": 8.408038800389108e-06,
      "loss": 0.0001,
      "step": 23090
    },
    {
      "epoch": 0.1593685967974501,
      "grad_norm": 0.0,
      "learning_rate": 8.407348893043665e-06,
      "loss": 0.0,
      "step": 23100
    },
    {
      "epoch": 0.15943758753199444,
      "grad_norm": 0.0,
      "learning_rate": 8.406658985698221e-06,
      "loss": 0.0065,
      "step": 23110
    },
    {
      "epoch": 0.1595065782665388,
      "grad_norm": 0.0,
      "learning_rate": 8.406038069087322e-06,
      "loss": 0.8562,
      "step": 23120
    },
    {
      "epoch": 0.15957556900108316,
      "grad_norm": 0.0008137800032272935,
      "learning_rate": 8.405348161741878e-06,
      "loss": 0.0,
      "step": 23130
    },
    {
      "epoch": 0.1596445597356275,
      "grad_norm": 0.0,
      "learning_rate": 8.404658254396435e-06,
      "loss": 0.0005,
      "step": 23140
    },
    {
      "epoch": 0.15971355047017186,
      "grad_norm": 0.0,
      "learning_rate": 8.403968347050991e-06,
      "loss": 0.0,
      "step": 23150
    },
    {
      "epoch": 0.15978254120471622,
      "grad_norm": 586.30029296875,
      "learning_rate": 8.403278439705548e-06,
      "loss": 0.1831,
      "step": 23160
    },
    {
      "epoch": 0.15985153193926055,
      "grad_norm": 0.0,
      "learning_rate": 8.402588532360104e-06,
      "loss": 0.001,
      "step": 23170
    },
    {
      "epoch": 0.1599205226738049,
      "grad_norm": 0.0002095679083140567,
      "learning_rate": 8.401898625014663e-06,
      "loss": 0.0,
      "step": 23180
    },
    {
      "epoch": 0.15998951340834927,
      "grad_norm": 0.0,
      "learning_rate": 8.401208717669217e-06,
      "loss": 0.0,
      "step": 23190
    },
    {
      "epoch": 0.1600585041428936,
      "grad_norm": 7.866688974900171e-05,
      "learning_rate": 8.400518810323774e-06,
      "loss": 0.0007,
      "step": 23200
    },
    {
      "epoch": 0.16012749487743796,
      "grad_norm": 0.00010515480971662328,
      "learning_rate": 8.39982890297833e-06,
      "loss": 0.0,
      "step": 23210
    },
    {
      "epoch": 0.1601964856119823,
      "grad_norm": 0.0,
      "learning_rate": 8.399138995632887e-06,
      "loss": 0.0,
      "step": 23220
    },
    {
      "epoch": 0.16026547634652666,
      "grad_norm": 0.148875892162323,
      "learning_rate": 8.398449088287443e-06,
      "loss": 0.0,
      "step": 23230
    },
    {
      "epoch": 0.16033446708107102,
      "grad_norm": 0.0,
      "learning_rate": 8.397759180942e-06,
      "loss": 0.1063,
      "step": 23240
    },
    {
      "epoch": 0.16040345781561535,
      "grad_norm": 0.0,
      "learning_rate": 8.397069273596556e-06,
      "loss": 0.0,
      "step": 23250
    },
    {
      "epoch": 0.1604724485501597,
      "grad_norm": 0.0,
      "learning_rate": 8.396379366251113e-06,
      "loss": 0.0,
      "step": 23260
    },
    {
      "epoch": 0.16054143928470407,
      "grad_norm": 0.0,
      "learning_rate": 8.39568945890567e-06,
      "loss": 0.0,
      "step": 23270
    },
    {
      "epoch": 0.1606104300192484,
      "grad_norm": 0.0,
      "learning_rate": 8.394999551560226e-06,
      "loss": 0.0019,
      "step": 23280
    },
    {
      "epoch": 0.16067942075379277,
      "grad_norm": 0.0,
      "learning_rate": 8.394309644214784e-06,
      "loss": 0.0,
      "step": 23290
    },
    {
      "epoch": 0.16074841148833713,
      "grad_norm": 0.0,
      "learning_rate": 8.393619736869339e-06,
      "loss": 0.0,
      "step": 23300
    },
    {
      "epoch": 0.16081740222288146,
      "grad_norm": 3.7193925717105003e-09,
      "learning_rate": 8.392929829523895e-06,
      "loss": 0.0,
      "step": 23310
    },
    {
      "epoch": 0.16088639295742582,
      "grad_norm": 0.0,
      "learning_rate": 8.392239922178452e-06,
      "loss": 0.0,
      "step": 23320
    },
    {
      "epoch": 0.16095538369197016,
      "grad_norm": 0.0,
      "learning_rate": 8.391550014833008e-06,
      "loss": 0.0,
      "step": 23330
    },
    {
      "epoch": 0.16102437442651452,
      "grad_norm": 0.0,
      "learning_rate": 8.390860107487565e-06,
      "loss": 0.0,
      "step": 23340
    },
    {
      "epoch": 0.16109336516105888,
      "grad_norm": 0.0,
      "learning_rate": 8.390170200142121e-06,
      "loss": 0.0,
      "step": 23350
    },
    {
      "epoch": 0.1611623558956032,
      "grad_norm": 0.0,
      "learning_rate": 8.389480292796678e-06,
      "loss": 0.1134,
      "step": 23360
    },
    {
      "epoch": 0.16123134663014757,
      "grad_norm": 0.0,
      "learning_rate": 8.388790385451234e-06,
      "loss": 0.0134,
      "step": 23370
    },
    {
      "epoch": 0.16130033736469193,
      "grad_norm": 0.0,
      "learning_rate": 8.38810047810579e-06,
      "loss": 0.0152,
      "step": 23380
    },
    {
      "epoch": 0.16136932809923626,
      "grad_norm": 0.03064868599176407,
      "learning_rate": 8.387410570760347e-06,
      "loss": 0.007,
      "step": 23390
    },
    {
      "epoch": 0.16143831883378063,
      "grad_norm": 0.0,
      "learning_rate": 8.386720663414905e-06,
      "loss": 0.0,
      "step": 23400
    },
    {
      "epoch": 0.16150730956832499,
      "grad_norm": 0.0,
      "learning_rate": 8.38603075606946e-06,
      "loss": 0.0,
      "step": 23410
    },
    {
      "epoch": 0.16157630030286932,
      "grad_norm": 0.0,
      "learning_rate": 8.385340848724017e-06,
      "loss": 0.0,
      "step": 23420
    },
    {
      "epoch": 0.16164529103741368,
      "grad_norm": 0.0,
      "learning_rate": 8.384650941378573e-06,
      "loss": 0.0002,
      "step": 23430
    },
    {
      "epoch": 0.161714281771958,
      "grad_norm": 9.744911949383095e-05,
      "learning_rate": 8.38396103403313e-06,
      "loss": 0.0001,
      "step": 23440
    },
    {
      "epoch": 0.16178327250650237,
      "grad_norm": 0.0,
      "learning_rate": 8.383271126687686e-06,
      "loss": 0.0038,
      "step": 23450
    },
    {
      "epoch": 0.16185226324104673,
      "grad_norm": 0.0,
      "learning_rate": 8.382581219342243e-06,
      "loss": 0.0,
      "step": 23460
    },
    {
      "epoch": 0.16192125397559107,
      "grad_norm": 0.0,
      "learning_rate": 8.381891311996799e-06,
      "loss": 0.0006,
      "step": 23470
    },
    {
      "epoch": 0.16199024471013543,
      "grad_norm": 0.0,
      "learning_rate": 8.381201404651356e-06,
      "loss": 0.0001,
      "step": 23480
    },
    {
      "epoch": 0.1620592354446798,
      "grad_norm": 53.06589889526367,
      "learning_rate": 8.380511497305912e-06,
      "loss": 0.0054,
      "step": 23490
    },
    {
      "epoch": 0.16212822617922412,
      "grad_norm": 7.495381851185812e-06,
      "learning_rate": 8.379821589960469e-06,
      "loss": 0.0027,
      "step": 23500
    },
    {
      "epoch": 0.16219721691376848,
      "grad_norm": 0.0,
      "learning_rate": 8.379131682615027e-06,
      "loss": 0.0009,
      "step": 23510
    },
    {
      "epoch": 0.16226620764831284,
      "grad_norm": 0.0,
      "learning_rate": 8.378441775269583e-06,
      "loss": 1.0734,
      "step": 23520
    },
    {
      "epoch": 0.16233519838285718,
      "grad_norm": 0.0,
      "learning_rate": 8.377751867924138e-06,
      "loss": 0.0,
      "step": 23530
    },
    {
      "epoch": 0.16240418911740154,
      "grad_norm": 0.0,
      "learning_rate": 8.377061960578694e-06,
      "loss": 0.0,
      "step": 23540
    },
    {
      "epoch": 0.16247317985194587,
      "grad_norm": 0.0004498373600654304,
      "learning_rate": 8.376372053233251e-06,
      "loss": 0.0,
      "step": 23550
    },
    {
      "epoch": 0.16254217058649023,
      "grad_norm": 0.0,
      "learning_rate": 8.375682145887807e-06,
      "loss": 0.0,
      "step": 23560
    },
    {
      "epoch": 0.1626111613210346,
      "grad_norm": 0.0,
      "learning_rate": 8.374992238542364e-06,
      "loss": 0.0004,
      "step": 23570
    },
    {
      "epoch": 0.16268015205557892,
      "grad_norm": 0.0,
      "learning_rate": 8.37430233119692e-06,
      "loss": 0.425,
      "step": 23580
    },
    {
      "epoch": 0.16274914279012329,
      "grad_norm": 0.0,
      "learning_rate": 8.373612423851477e-06,
      "loss": 0.0,
      "step": 23590
    },
    {
      "epoch": 0.16281813352466765,
      "grad_norm": 0.0,
      "learning_rate": 8.372922516506033e-06,
      "loss": 0.0,
      "step": 23600
    },
    {
      "epoch": 0.16288712425921198,
      "grad_norm": 0.0,
      "learning_rate": 8.37223260916059e-06,
      "loss": 0.0,
      "step": 23610
    },
    {
      "epoch": 0.16295611499375634,
      "grad_norm": 0.0,
      "learning_rate": 8.371542701815148e-06,
      "loss": 0.0033,
      "step": 23620
    },
    {
      "epoch": 0.1630251057283007,
      "grad_norm": 7.632367487531155e-05,
      "learning_rate": 8.370852794469705e-06,
      "loss": 0.0,
      "step": 23630
    },
    {
      "epoch": 0.16309409646284503,
      "grad_norm": 0.0,
      "learning_rate": 8.37016288712426e-06,
      "loss": 0.0004,
      "step": 23640
    },
    {
      "epoch": 0.1631630871973894,
      "grad_norm": 0.0,
      "learning_rate": 8.369472979778816e-06,
      "loss": 0.0,
      "step": 23650
    },
    {
      "epoch": 0.16323207793193373,
      "grad_norm": 0.0,
      "learning_rate": 8.368783072433372e-06,
      "loss": 0.0,
      "step": 23660
    },
    {
      "epoch": 0.1633010686664781,
      "grad_norm": 0.0,
      "learning_rate": 8.368093165087929e-06,
      "loss": 0.0024,
      "step": 23670
    },
    {
      "epoch": 0.16337005940102245,
      "grad_norm": 0.0,
      "learning_rate": 8.367403257742485e-06,
      "loss": 0.0,
      "step": 23680
    },
    {
      "epoch": 0.16343905013556678,
      "grad_norm": 0.0,
      "learning_rate": 8.366713350397042e-06,
      "loss": 0.0,
      "step": 23690
    },
    {
      "epoch": 0.16350804087011114,
      "grad_norm": 0.024429699406027794,
      "learning_rate": 8.366023443051598e-06,
      "loss": 0.1677,
      "step": 23700
    },
    {
      "epoch": 0.1635770316046555,
      "grad_norm": 1.4550627724929655e-07,
      "learning_rate": 8.365333535706155e-06,
      "loss": 0.0002,
      "step": 23710
    },
    {
      "epoch": 0.16364602233919984,
      "grad_norm": 0.0,
      "learning_rate": 8.364643628360711e-06,
      "loss": 0.0281,
      "step": 23720
    },
    {
      "epoch": 0.1637150130737442,
      "grad_norm": 0.003637194400653243,
      "learning_rate": 8.36395372101527e-06,
      "loss": 0.0,
      "step": 23730
    },
    {
      "epoch": 0.16378400380828856,
      "grad_norm": 0.0,
      "learning_rate": 8.363263813669826e-06,
      "loss": 0.0,
      "step": 23740
    },
    {
      "epoch": 0.1638529945428329,
      "grad_norm": 1.1732846161294219e-07,
      "learning_rate": 8.36257390632438e-06,
      "loss": 0.0,
      "step": 23750
    },
    {
      "epoch": 0.16392198527737725,
      "grad_norm": 0.12413500994443893,
      "learning_rate": 8.361883998978937e-06,
      "loss": 0.0,
      "step": 23760
    },
    {
      "epoch": 0.16399097601192159,
      "grad_norm": 0.0,
      "learning_rate": 8.361194091633494e-06,
      "loss": 0.003,
      "step": 23770
    },
    {
      "epoch": 0.16405996674646595,
      "grad_norm": 3.282538414001465,
      "learning_rate": 8.36050418428805e-06,
      "loss": 0.0004,
      "step": 23780
    },
    {
      "epoch": 0.1641289574810103,
      "grad_norm": 0.0,
      "learning_rate": 8.359814276942607e-06,
      "loss": 0.0,
      "step": 23790
    },
    {
      "epoch": 0.16419794821555464,
      "grad_norm": 0.0,
      "learning_rate": 8.359124369597163e-06,
      "loss": 0.0006,
      "step": 23800
    },
    {
      "epoch": 0.164266938950099,
      "grad_norm": 0.0,
      "learning_rate": 8.358503452986264e-06,
      "loss": 0.399,
      "step": 23810
    },
    {
      "epoch": 0.16433592968464336,
      "grad_norm": 0.0,
      "learning_rate": 8.357813545640822e-06,
      "loss": 0.0001,
      "step": 23820
    },
    {
      "epoch": 0.1644049204191877,
      "grad_norm": 0.0,
      "learning_rate": 8.357123638295379e-06,
      "loss": 0.0,
      "step": 23830
    },
    {
      "epoch": 0.16447391115373206,
      "grad_norm": 0.0,
      "learning_rate": 8.356433730949935e-06,
      "loss": 0.0006,
      "step": 23840
    },
    {
      "epoch": 0.16454290188827642,
      "grad_norm": 0.0,
      "learning_rate": 8.355743823604492e-06,
      "loss": 0.0001,
      "step": 23850
    },
    {
      "epoch": 0.16461189262282075,
      "grad_norm": 0.0,
      "learning_rate": 8.355053916259046e-06,
      "loss": 0.0002,
      "step": 23860
    },
    {
      "epoch": 0.1646808833573651,
      "grad_norm": 0.0,
      "learning_rate": 8.354364008913603e-06,
      "loss": 0.0,
      "step": 23870
    },
    {
      "epoch": 0.16474987409190944,
      "grad_norm": 0.0,
      "learning_rate": 8.35367410156816e-06,
      "loss": 0.021,
      "step": 23880
    },
    {
      "epoch": 0.1648188648264538,
      "grad_norm": 0.0,
      "learning_rate": 8.352984194222716e-06,
      "loss": 0.0,
      "step": 23890
    },
    {
      "epoch": 0.16488785556099816,
      "grad_norm": 0.019565938040614128,
      "learning_rate": 8.352294286877272e-06,
      "loss": 0.0,
      "step": 23900
    },
    {
      "epoch": 0.1649568462955425,
      "grad_norm": 0.0,
      "learning_rate": 8.351604379531829e-06,
      "loss": 0.0,
      "step": 23910
    },
    {
      "epoch": 0.16502583703008686,
      "grad_norm": 0.0,
      "learning_rate": 8.350914472186385e-06,
      "loss": 0.0,
      "step": 23920
    },
    {
      "epoch": 0.16509482776463122,
      "grad_norm": 0.0,
      "learning_rate": 8.350224564840943e-06,
      "loss": 0.0,
      "step": 23930
    },
    {
      "epoch": 0.16516381849917555,
      "grad_norm": 0.1406061202287674,
      "learning_rate": 8.3495346574955e-06,
      "loss": 0.0026,
      "step": 23940
    },
    {
      "epoch": 0.1652328092337199,
      "grad_norm": 6.7976855255835744e-09,
      "learning_rate": 8.348844750150056e-06,
      "loss": 0.0118,
      "step": 23950
    },
    {
      "epoch": 0.16530179996826427,
      "grad_norm": 0.0,
      "learning_rate": 8.348154842804613e-06,
      "loss": 0.0,
      "step": 23960
    },
    {
      "epoch": 0.1653707907028086,
      "grad_norm": 0.0,
      "learning_rate": 8.347464935459168e-06,
      "loss": 0.0001,
      "step": 23970
    },
    {
      "epoch": 0.16543978143735297,
      "grad_norm": 0.0,
      "learning_rate": 8.346775028113724e-06,
      "loss": 0.0,
      "step": 23980
    },
    {
      "epoch": 0.1655087721718973,
      "grad_norm": 10.209993362426758,
      "learning_rate": 8.34608512076828e-06,
      "loss": 0.0019,
      "step": 23990
    },
    {
      "epoch": 0.16557776290644166,
      "grad_norm": 0.0,
      "learning_rate": 8.345395213422837e-06,
      "loss": 0.0,
      "step": 24000
    },
    {
      "epoch": 0.16564675364098602,
      "grad_norm": 0.11810226738452911,
      "learning_rate": 8.344705306077394e-06,
      "loss": 0.0024,
      "step": 24010
    },
    {
      "epoch": 0.16571574437553035,
      "grad_norm": 0.0,
      "learning_rate": 8.34401539873195e-06,
      "loss": 0.0,
      "step": 24020
    },
    {
      "epoch": 0.16578473511007472,
      "grad_norm": 11.925437927246094,
      "learning_rate": 8.343325491386507e-06,
      "loss": 0.0029,
      "step": 24030
    },
    {
      "epoch": 0.16585372584461908,
      "grad_norm": 0.0,
      "learning_rate": 8.342635584041065e-06,
      "loss": 0.0,
      "step": 24040
    },
    {
      "epoch": 0.1659227165791634,
      "grad_norm": 0.0,
      "learning_rate": 8.341945676695621e-06,
      "loss": 0.0,
      "step": 24050
    },
    {
      "epoch": 0.16599170731370777,
      "grad_norm": 0.0,
      "learning_rate": 8.341255769350178e-06,
      "loss": 0.0,
      "step": 24060
    },
    {
      "epoch": 0.16606069804825213,
      "grad_norm": 0.00028354235109873116,
      "learning_rate": 8.340565862004734e-06,
      "loss": 0.0051,
      "step": 24070
    },
    {
      "epoch": 0.16612968878279646,
      "grad_norm": 0.0,
      "learning_rate": 8.33987595465929e-06,
      "loss": 0.0,
      "step": 24080
    },
    {
      "epoch": 0.16619867951734082,
      "grad_norm": 0.0,
      "learning_rate": 8.339186047313846e-06,
      "loss": 0.0,
      "step": 24090
    },
    {
      "epoch": 0.16626767025188519,
      "grad_norm": 1.2160699952801224e-05,
      "learning_rate": 8.338496139968402e-06,
      "loss": 0.0,
      "step": 24100
    },
    {
      "epoch": 0.16633666098642952,
      "grad_norm": 0.5098678469657898,
      "learning_rate": 8.337806232622959e-06,
      "loss": 0.0001,
      "step": 24110
    },
    {
      "epoch": 0.16640565172097388,
      "grad_norm": 1.0296205488202759e-07,
      "learning_rate": 8.337116325277515e-06,
      "loss": 0.0,
      "step": 24120
    },
    {
      "epoch": 0.1664746424555182,
      "grad_norm": 0.0,
      "learning_rate": 8.336426417932072e-06,
      "loss": 0.0,
      "step": 24130
    },
    {
      "epoch": 0.16654363319006257,
      "grad_norm": 0.0,
      "learning_rate": 8.335736510586628e-06,
      "loss": 0.0,
      "step": 24140
    },
    {
      "epoch": 0.16661262392460693,
      "grad_norm": 0.0,
      "learning_rate": 8.335046603241186e-06,
      "loss": 0.0039,
      "step": 24150
    },
    {
      "epoch": 0.16668161465915127,
      "grad_norm": 0.0,
      "learning_rate": 8.334356695895743e-06,
      "loss": 0.0,
      "step": 24160
    },
    {
      "epoch": 0.16675060539369563,
      "grad_norm": 0.0,
      "learning_rate": 8.3336667885503e-06,
      "loss": 0.0,
      "step": 24170
    },
    {
      "epoch": 0.16681959612824,
      "grad_norm": 0.0,
      "learning_rate": 8.332976881204856e-06,
      "loss": 0.0,
      "step": 24180
    },
    {
      "epoch": 0.16688858686278432,
      "grad_norm": 0.001259670709259808,
      "learning_rate": 8.332286973859412e-06,
      "loss": 0.0,
      "step": 24190
    },
    {
      "epoch": 0.16695757759732868,
      "grad_norm": 0.0,
      "learning_rate": 8.331597066513967e-06,
      "loss": 0.0,
      "step": 24200
    },
    {
      "epoch": 0.16702656833187304,
      "grad_norm": 0.0,
      "learning_rate": 8.330907159168524e-06,
      "loss": 0.1286,
      "step": 24210
    },
    {
      "epoch": 0.16709555906641738,
      "grad_norm": 0.0,
      "learning_rate": 8.33021725182308e-06,
      "loss": 0.0008,
      "step": 24220
    },
    {
      "epoch": 0.16716454980096174,
      "grad_norm": 0.0,
      "learning_rate": 8.329527344477637e-06,
      "loss": 0.0,
      "step": 24230
    },
    {
      "epoch": 0.16723354053550607,
      "grad_norm": 0.0,
      "learning_rate": 8.328837437132193e-06,
      "loss": 0.0,
      "step": 24240
    },
    {
      "epoch": 0.16730253127005043,
      "grad_norm": 0.0,
      "learning_rate": 8.32814752978675e-06,
      "loss": 0.0,
      "step": 24250
    },
    {
      "epoch": 0.1673715220045948,
      "grad_norm": 0.0,
      "learning_rate": 8.327457622441308e-06,
      "loss": 0.0,
      "step": 24260
    },
    {
      "epoch": 0.16744051273913912,
      "grad_norm": 0.0,
      "learning_rate": 8.326767715095864e-06,
      "loss": 0.0,
      "step": 24270
    },
    {
      "epoch": 0.16750950347368349,
      "grad_norm": 0.0,
      "learning_rate": 8.32607780775042e-06,
      "loss": 0.0451,
      "step": 24280
    },
    {
      "epoch": 0.16757849420822785,
      "grad_norm": 0.0,
      "learning_rate": 8.325387900404977e-06,
      "loss": 0.0,
      "step": 24290
    },
    {
      "epoch": 0.16764748494277218,
      "grad_norm": 0.0005077415844425559,
      "learning_rate": 8.324697993059534e-06,
      "loss": 0.0,
      "step": 24300
    },
    {
      "epoch": 0.16771647567731654,
      "grad_norm": 0.0,
      "learning_rate": 8.324008085714088e-06,
      "loss": 0.0002,
      "step": 24310
    },
    {
      "epoch": 0.1677854664118609,
      "grad_norm": 5.61315346203628e-07,
      "learning_rate": 8.323318178368645e-06,
      "loss": 0.0003,
      "step": 24320
    },
    {
      "epoch": 0.16785445714640523,
      "grad_norm": 127.95923614501953,
      "learning_rate": 8.322628271023201e-06,
      "loss": 0.1063,
      "step": 24330
    },
    {
      "epoch": 0.1679234478809496,
      "grad_norm": 0.0,
      "learning_rate": 8.321938363677758e-06,
      "loss": 0.0,
      "step": 24340
    },
    {
      "epoch": 0.16799243861549393,
      "grad_norm": 0.0,
      "learning_rate": 8.321248456332314e-06,
      "loss": 0.0,
      "step": 24350
    },
    {
      "epoch": 0.1680614293500383,
      "grad_norm": 0.09342078864574432,
      "learning_rate": 8.320558548986871e-06,
      "loss": 0.0,
      "step": 24360
    },
    {
      "epoch": 0.16813042008458265,
      "grad_norm": 0.0,
      "learning_rate": 8.319868641641429e-06,
      "loss": 1.486,
      "step": 24370
    },
    {
      "epoch": 0.16819941081912698,
      "grad_norm": 0.0,
      "learning_rate": 8.319178734295986e-06,
      "loss": 0.0,
      "step": 24380
    },
    {
      "epoch": 0.16826840155367134,
      "grad_norm": 0.0,
      "learning_rate": 8.318488826950542e-06,
      "loss": 0.0,
      "step": 24390
    },
    {
      "epoch": 0.1683373922882157,
      "grad_norm": 0.0,
      "learning_rate": 8.317798919605099e-06,
      "loss": 0.0,
      "step": 24400
    },
    {
      "epoch": 0.16840638302276004,
      "grad_norm": 0.0,
      "learning_rate": 8.317109012259655e-06,
      "loss": 0.8625,
      "step": 24410
    },
    {
      "epoch": 0.1684753737573044,
      "grad_norm": 0.0,
      "learning_rate": 8.316419104914212e-06,
      "loss": 0.0529,
      "step": 24420
    },
    {
      "epoch": 0.16854436449184876,
      "grad_norm": 0.0,
      "learning_rate": 8.315729197568766e-06,
      "loss": 0.0,
      "step": 24430
    },
    {
      "epoch": 0.1686133552263931,
      "grad_norm": 0.0,
      "learning_rate": 8.315039290223323e-06,
      "loss": 0.0,
      "step": 24440
    },
    {
      "epoch": 0.16868234596093745,
      "grad_norm": 0.0,
      "learning_rate": 8.31434938287788e-06,
      "loss": 0.0032,
      "step": 24450
    },
    {
      "epoch": 0.16875133669548179,
      "grad_norm": 5.762419732491253e-06,
      "learning_rate": 8.313659475532436e-06,
      "loss": 0.0,
      "step": 24460
    },
    {
      "epoch": 0.16882032743002615,
      "grad_norm": 0.0,
      "learning_rate": 8.312969568186992e-06,
      "loss": 0.0,
      "step": 24470
    },
    {
      "epoch": 0.1688893181645705,
      "grad_norm": 0.0,
      "learning_rate": 8.31227966084155e-06,
      "loss": 0.0002,
      "step": 24480
    },
    {
      "epoch": 0.16895830889911484,
      "grad_norm": 0.0,
      "learning_rate": 8.311589753496107e-06,
      "loss": 0.0,
      "step": 24490
    },
    {
      "epoch": 0.1690272996336592,
      "grad_norm": 0.0,
      "learning_rate": 8.310899846150663e-06,
      "loss": 0.0001,
      "step": 24500
    },
    {
      "epoch": 0.16909629036820356,
      "grad_norm": 0.0,
      "learning_rate": 8.31020993880522e-06,
      "loss": 0.0,
      "step": 24510
    },
    {
      "epoch": 0.1691652811027479,
      "grad_norm": 0.0,
      "learning_rate": 8.309520031459776e-06,
      "loss": 0.0,
      "step": 24520
    },
    {
      "epoch": 0.16923427183729225,
      "grad_norm": 0.0,
      "learning_rate": 8.308830124114333e-06,
      "loss": 0.0,
      "step": 24530
    },
    {
      "epoch": 0.16930326257183662,
      "grad_norm": 0.0,
      "learning_rate": 8.308140216768888e-06,
      "loss": 0.0,
      "step": 24540
    },
    {
      "epoch": 0.16937225330638095,
      "grad_norm": 0.0,
      "learning_rate": 8.307450309423444e-06,
      "loss": 0.6671,
      "step": 24550
    },
    {
      "epoch": 0.1694412440409253,
      "grad_norm": 0.0,
      "learning_rate": 8.306760402078e-06,
      "loss": 0.0,
      "step": 24560
    },
    {
      "epoch": 0.16951023477546964,
      "grad_norm": 0.00019171614258084446,
      "learning_rate": 8.306070494732557e-06,
      "loss": 0.0006,
      "step": 24570
    },
    {
      "epoch": 0.169579225510014,
      "grad_norm": 0.0,
      "learning_rate": 8.305380587387114e-06,
      "loss": 0.0001,
      "step": 24580
    },
    {
      "epoch": 0.16964821624455836,
      "grad_norm": 0.0,
      "learning_rate": 8.304690680041672e-06,
      "loss": 0.0,
      "step": 24590
    },
    {
      "epoch": 0.1697172069791027,
      "grad_norm": 0.0,
      "learning_rate": 8.304000772696228e-06,
      "loss": 0.0001,
      "step": 24600
    },
    {
      "epoch": 0.16978619771364706,
      "grad_norm": 0.0,
      "learning_rate": 8.303310865350785e-06,
      "loss": 0.0,
      "step": 24610
    },
    {
      "epoch": 0.16985518844819142,
      "grad_norm": 0.0,
      "learning_rate": 8.302620958005341e-06,
      "loss": 0.0,
      "step": 24620
    },
    {
      "epoch": 0.16992417918273575,
      "grad_norm": 0.0,
      "learning_rate": 8.301931050659898e-06,
      "loss": 0.0,
      "step": 24630
    },
    {
      "epoch": 0.1699931699172801,
      "grad_norm": 212.1542205810547,
      "learning_rate": 8.301241143314454e-06,
      "loss": 0.0169,
      "step": 24640
    },
    {
      "epoch": 0.17006216065182447,
      "grad_norm": 0.0,
      "learning_rate": 8.300551235969009e-06,
      "loss": 0.0,
      "step": 24650
    },
    {
      "epoch": 0.1701311513863688,
      "grad_norm": 0.0,
      "learning_rate": 8.299861328623566e-06,
      "loss": 0.0021,
      "step": 24660
    },
    {
      "epoch": 0.17020014212091317,
      "grad_norm": 0.0,
      "learning_rate": 8.299171421278122e-06,
      "loss": 0.0,
      "step": 24670
    },
    {
      "epoch": 0.1702691328554575,
      "grad_norm": 0.0,
      "learning_rate": 8.298481513932679e-06,
      "loss": 0.7152,
      "step": 24680
    },
    {
      "epoch": 0.17033812359000186,
      "grad_norm": 0.0,
      "learning_rate": 8.297791606587235e-06,
      "loss": 0.0001,
      "step": 24690
    },
    {
      "epoch": 0.17040711432454622,
      "grad_norm": 1.1909532560139269e-07,
      "learning_rate": 8.297101699241793e-06,
      "loss": 0.0008,
      "step": 24700
    },
    {
      "epoch": 0.17047610505909055,
      "grad_norm": 1.1512995115481317e-05,
      "learning_rate": 8.29641179189635e-06,
      "loss": 0.0,
      "step": 24710
    },
    {
      "epoch": 0.17054509579363492,
      "grad_norm": 0.03892756626009941,
      "learning_rate": 8.295721884550906e-06,
      "loss": 0.0,
      "step": 24720
    },
    {
      "epoch": 0.17061408652817928,
      "grad_norm": 0.0,
      "learning_rate": 8.295031977205463e-06,
      "loss": 0.0001,
      "step": 24730
    },
    {
      "epoch": 0.1706830772627236,
      "grad_norm": 0.0,
      "learning_rate": 8.29434206986002e-06,
      "loss": 0.0,
      "step": 24740
    },
    {
      "epoch": 0.17075206799726797,
      "grad_norm": 0.0,
      "learning_rate": 8.293652162514576e-06,
      "loss": 0.0,
      "step": 24750
    },
    {
      "epoch": 0.17082105873181233,
      "grad_norm": 0.0,
      "learning_rate": 8.29296225516913e-06,
      "loss": 0.0,
      "step": 24760
    },
    {
      "epoch": 0.17089004946635666,
      "grad_norm": 0.0,
      "learning_rate": 8.292272347823687e-06,
      "loss": 0.0,
      "step": 24770
    },
    {
      "epoch": 0.17095904020090102,
      "grad_norm": 7.119715883163735e-05,
      "learning_rate": 8.291582440478243e-06,
      "loss": 0.0,
      "step": 24780
    },
    {
      "epoch": 0.17102803093544536,
      "grad_norm": 0.0,
      "learning_rate": 8.2908925331328e-06,
      "loss": 0.0005,
      "step": 24790
    },
    {
      "epoch": 0.17109702166998972,
      "grad_norm": 0.0,
      "learning_rate": 8.290202625787356e-06,
      "loss": 0.0868,
      "step": 24800
    },
    {
      "epoch": 0.17116601240453408,
      "grad_norm": 0.0,
      "learning_rate": 8.289512718441915e-06,
      "loss": 0.0085,
      "step": 24810
    },
    {
      "epoch": 0.1712350031390784,
      "grad_norm": 0.0,
      "learning_rate": 8.288822811096471e-06,
      "loss": 0.0,
      "step": 24820
    },
    {
      "epoch": 0.17130399387362277,
      "grad_norm": 0.0,
      "learning_rate": 8.288132903751028e-06,
      "loss": 0.0,
      "step": 24830
    },
    {
      "epoch": 0.17137298460816713,
      "grad_norm": 0.0,
      "learning_rate": 8.287442996405584e-06,
      "loss": 0.0002,
      "step": 24840
    },
    {
      "epoch": 0.17144197534271147,
      "grad_norm": 0.0029003354720771313,
      "learning_rate": 8.28675308906014e-06,
      "loss": 0.0001,
      "step": 24850
    },
    {
      "epoch": 0.17151096607725583,
      "grad_norm": 0.0,
      "learning_rate": 8.286063181714697e-06,
      "loss": 0.0002,
      "step": 24860
    },
    {
      "epoch": 0.1715799568118002,
      "grad_norm": 0.0,
      "learning_rate": 8.285373274369254e-06,
      "loss": 0.0,
      "step": 24870
    },
    {
      "epoch": 0.17164894754634452,
      "grad_norm": 0.6882894039154053,
      "learning_rate": 8.284683367023808e-06,
      "loss": 0.0279,
      "step": 24880
    },
    {
      "epoch": 0.17171793828088888,
      "grad_norm": 0.0,
      "learning_rate": 8.283993459678365e-06,
      "loss": 0.0002,
      "step": 24890
    },
    {
      "epoch": 0.17178692901543322,
      "grad_norm": 5.751724785341139e-08,
      "learning_rate": 8.283303552332921e-06,
      "loss": 0.0,
      "step": 24900
    },
    {
      "epoch": 0.17185591974997758,
      "grad_norm": 0.0,
      "learning_rate": 8.282613644987478e-06,
      "loss": 0.0,
      "step": 24910
    },
    {
      "epoch": 0.17192491048452194,
      "grad_norm": 0.0,
      "learning_rate": 8.281923737642036e-06,
      "loss": 0.0032,
      "step": 24920
    },
    {
      "epoch": 0.17199390121906627,
      "grad_norm": 0.006061329040676355,
      "learning_rate": 8.281233830296593e-06,
      "loss": 0.0,
      "step": 24930
    },
    {
      "epoch": 0.17206289195361063,
      "grad_norm": 0.0,
      "learning_rate": 8.280543922951149e-06,
      "loss": 0.0008,
      "step": 24940
    },
    {
      "epoch": 0.172131882688155,
      "grad_norm": 0.0,
      "learning_rate": 8.279854015605706e-06,
      "loss": 0.0,
      "step": 24950
    },
    {
      "epoch": 0.17220087342269932,
      "grad_norm": 0.0,
      "learning_rate": 8.279164108260262e-06,
      "loss": 0.0,
      "step": 24960
    },
    {
      "epoch": 0.17226986415724369,
      "grad_norm": 0.0,
      "learning_rate": 8.278474200914818e-06,
      "loss": 0.0045,
      "step": 24970
    },
    {
      "epoch": 0.17233885489178805,
      "grad_norm": 0.0,
      "learning_rate": 8.277784293569375e-06,
      "loss": 0.0,
      "step": 24980
    },
    {
      "epoch": 0.17240784562633238,
      "grad_norm": 0.0,
      "learning_rate": 8.27709438622393e-06,
      "loss": 0.0,
      "step": 24990
    },
    {
      "epoch": 0.17247683636087674,
      "grad_norm": 0.0,
      "learning_rate": 8.276404478878486e-06,
      "loss": 0.0,
      "step": 25000
    },
    {
      "epoch": 0.17254582709542107,
      "grad_norm": 0.0016783939208835363,
      "learning_rate": 8.275714571533043e-06,
      "loss": 0.0,
      "step": 25010
    },
    {
      "epoch": 0.17261481782996543,
      "grad_norm": 0.0,
      "learning_rate": 8.2750246641876e-06,
      "loss": 0.0,
      "step": 25020
    },
    {
      "epoch": 0.1726838085645098,
      "grad_norm": 0.0,
      "learning_rate": 8.274334756842157e-06,
      "loss": 0.0,
      "step": 25030
    },
    {
      "epoch": 0.17275279929905413,
      "grad_norm": 0.0,
      "learning_rate": 8.273644849496714e-06,
      "loss": 0.0,
      "step": 25040
    },
    {
      "epoch": 0.1728217900335985,
      "grad_norm": 0.0,
      "learning_rate": 8.27295494215127e-06,
      "loss": 0.0002,
      "step": 25050
    },
    {
      "epoch": 0.17289078076814285,
      "grad_norm": 0.0,
      "learning_rate": 8.272265034805827e-06,
      "loss": 0.0,
      "step": 25060
    },
    {
      "epoch": 0.17295977150268718,
      "grad_norm": 0.0,
      "learning_rate": 8.271575127460383e-06,
      "loss": 0.0,
      "step": 25070
    },
    {
      "epoch": 0.17302876223723154,
      "grad_norm": 0.0,
      "learning_rate": 8.27088522011494e-06,
      "loss": 0.0001,
      "step": 25080
    },
    {
      "epoch": 0.1730977529717759,
      "grad_norm": 0.0,
      "learning_rate": 8.270195312769496e-06,
      "loss": 0.0014,
      "step": 25090
    },
    {
      "epoch": 0.17316674370632024,
      "grad_norm": 0.0,
      "learning_rate": 8.269505405424051e-06,
      "loss": 0.0,
      "step": 25100
    },
    {
      "epoch": 0.1732357344408646,
      "grad_norm": 0.0,
      "learning_rate": 8.268815498078608e-06,
      "loss": 0.0,
      "step": 25110
    },
    {
      "epoch": 0.17330472517540893,
      "grad_norm": 0.0,
      "learning_rate": 8.268125590733164e-06,
      "loss": 0.0,
      "step": 25120
    },
    {
      "epoch": 0.1733737159099533,
      "grad_norm": 9.67517371464055e-06,
      "learning_rate": 8.26743568338772e-06,
      "loss": 0.0121,
      "step": 25130
    },
    {
      "epoch": 0.17344270664449765,
      "grad_norm": 4.8759222437411154e-08,
      "learning_rate": 8.266745776042279e-06,
      "loss": 0.0,
      "step": 25140
    },
    {
      "epoch": 0.17351169737904198,
      "grad_norm": 2.282831337652169e-05,
      "learning_rate": 8.266055868696835e-06,
      "loss": 0.0013,
      "step": 25150
    },
    {
      "epoch": 0.17358068811358635,
      "grad_norm": 0.0008630091324448586,
      "learning_rate": 8.265365961351392e-06,
      "loss": 0.0003,
      "step": 25160
    },
    {
      "epoch": 0.1736496788481307,
      "grad_norm": 1.5896519656166674e-08,
      "learning_rate": 8.264676054005948e-06,
      "loss": 0.0,
      "step": 25170
    },
    {
      "epoch": 0.17371866958267504,
      "grad_norm": 0.0,
      "learning_rate": 8.263986146660505e-06,
      "loss": 0.0,
      "step": 25180
    },
    {
      "epoch": 0.1737876603172194,
      "grad_norm": 0.0,
      "learning_rate": 8.263296239315061e-06,
      "loss": 0.0,
      "step": 25190
    },
    {
      "epoch": 0.17385665105176376,
      "grad_norm": 0.0,
      "learning_rate": 8.262606331969618e-06,
      "loss": 0.0,
      "step": 25200
    },
    {
      "epoch": 0.1739256417863081,
      "grad_norm": 483.7286682128906,
      "learning_rate": 8.261916424624174e-06,
      "loss": 0.2293,
      "step": 25210
    },
    {
      "epoch": 0.17399463252085245,
      "grad_norm": 0.0,
      "learning_rate": 8.261226517278729e-06,
      "loss": 0.0,
      "step": 25220
    },
    {
      "epoch": 0.1740636232553968,
      "grad_norm": 0.0,
      "learning_rate": 8.260536609933286e-06,
      "loss": 0.0,
      "step": 25230
    },
    {
      "epoch": 0.17413261398994115,
      "grad_norm": 8.166391580743948e-07,
      "learning_rate": 8.259846702587842e-06,
      "loss": 0.0,
      "step": 25240
    },
    {
      "epoch": 0.1742016047244855,
      "grad_norm": 1.974263312831681e-07,
      "learning_rate": 8.2591567952424e-06,
      "loss": 0.0002,
      "step": 25250
    },
    {
      "epoch": 0.17427059545902984,
      "grad_norm": 0.0,
      "learning_rate": 8.258466887896957e-06,
      "loss": 0.0,
      "step": 25260
    },
    {
      "epoch": 0.1743395861935742,
      "grad_norm": 0.0,
      "learning_rate": 8.257776980551513e-06,
      "loss": 0.0,
      "step": 25270
    },
    {
      "epoch": 0.17440857692811856,
      "grad_norm": 0.0,
      "learning_rate": 8.25708707320607e-06,
      "loss": 0.0,
      "step": 25280
    },
    {
      "epoch": 0.1744775676626629,
      "grad_norm": 0.0,
      "learning_rate": 8.256397165860626e-06,
      "loss": 0.0,
      "step": 25290
    },
    {
      "epoch": 0.17454655839720726,
      "grad_norm": 0.0,
      "learning_rate": 8.255707258515183e-06,
      "loss": 0.001,
      "step": 25300
    },
    {
      "epoch": 0.17461554913175162,
      "grad_norm": 0.0,
      "learning_rate": 8.25501735116974e-06,
      "loss": 0.0,
      "step": 25310
    },
    {
      "epoch": 0.17468453986629595,
      "grad_norm": 0.0,
      "learning_rate": 8.254327443824296e-06,
      "loss": 0.0,
      "step": 25320
    },
    {
      "epoch": 0.1747535306008403,
      "grad_norm": 0.0,
      "learning_rate": 8.25363753647885e-06,
      "loss": 0.0,
      "step": 25330
    },
    {
      "epoch": 0.17482252133538465,
      "grad_norm": 0.0,
      "learning_rate": 8.252947629133407e-06,
      "loss": 0.0,
      "step": 25340
    },
    {
      "epoch": 0.174891512069929,
      "grad_norm": 0.0,
      "learning_rate": 8.252257721787963e-06,
      "loss": 0.0001,
      "step": 25350
    },
    {
      "epoch": 0.17496050280447337,
      "grad_norm": 0.0,
      "learning_rate": 8.251567814442522e-06,
      "loss": 0.0,
      "step": 25360
    },
    {
      "epoch": 0.1750294935390177,
      "grad_norm": 0.0,
      "learning_rate": 8.250877907097078e-06,
      "loss": 0.0,
      "step": 25370
    },
    {
      "epoch": 0.17509848427356206,
      "grad_norm": 7.346125698859396e-07,
      "learning_rate": 8.250187999751635e-06,
      "loss": 0.0,
      "step": 25380
    },
    {
      "epoch": 0.17516747500810642,
      "grad_norm": 0.0,
      "learning_rate": 8.249498092406191e-06,
      "loss": 0.0,
      "step": 25390
    },
    {
      "epoch": 0.17523646574265075,
      "grad_norm": 807.4644165039062,
      "learning_rate": 8.248808185060748e-06,
      "loss": 0.2996,
      "step": 25400
    },
    {
      "epoch": 0.17530545647719512,
      "grad_norm": 0.0,
      "learning_rate": 8.248118277715304e-06,
      "loss": 0.0,
      "step": 25410
    },
    {
      "epoch": 0.17537444721173948,
      "grad_norm": 9.436121035832912e-05,
      "learning_rate": 8.24742837036986e-06,
      "loss": 0.0196,
      "step": 25420
    },
    {
      "epoch": 0.1754434379462838,
      "grad_norm": 1.77166275534546e-05,
      "learning_rate": 8.246738463024417e-06,
      "loss": 0.0,
      "step": 25430
    },
    {
      "epoch": 0.17551242868082817,
      "grad_norm": 0.0,
      "learning_rate": 8.246048555678972e-06,
      "loss": 1.0488,
      "step": 25440
    },
    {
      "epoch": 0.1755814194153725,
      "grad_norm": 0.0,
      "learning_rate": 8.245358648333528e-06,
      "loss": 0.4918,
      "step": 25450
    },
    {
      "epoch": 0.17565041014991686,
      "grad_norm": 1.7942100384971127e-05,
      "learning_rate": 8.244668740988085e-06,
      "loss": 0.0,
      "step": 25460
    },
    {
      "epoch": 0.17571940088446122,
      "grad_norm": 0.0,
      "learning_rate": 8.243978833642643e-06,
      "loss": 0.0,
      "step": 25470
    },
    {
      "epoch": 0.17578839161900556,
      "grad_norm": 0.0007773078978061676,
      "learning_rate": 8.2432889262972e-06,
      "loss": 0.0,
      "step": 25480
    },
    {
      "epoch": 0.17585738235354992,
      "grad_norm": 0.0,
      "learning_rate": 8.242599018951756e-06,
      "loss": 0.0004,
      "step": 25490
    },
    {
      "epoch": 0.17592637308809428,
      "grad_norm": 0.0,
      "learning_rate": 8.241909111606312e-06,
      "loss": 0.0,
      "step": 25500
    },
    {
      "epoch": 0.1759953638226386,
      "grad_norm": 0.0,
      "learning_rate": 8.241219204260869e-06,
      "loss": 0.0,
      "step": 25510
    },
    {
      "epoch": 0.17606435455718297,
      "grad_norm": 0.0,
      "learning_rate": 8.240529296915425e-06,
      "loss": 0.0015,
      "step": 25520
    },
    {
      "epoch": 0.17613334529172733,
      "grad_norm": 0.4363812804222107,
      "learning_rate": 8.239839389569982e-06,
      "loss": 0.0001,
      "step": 25530
    },
    {
      "epoch": 0.17620233602627167,
      "grad_norm": 3.2861378063131497e-09,
      "learning_rate": 8.239149482224538e-06,
      "loss": 0.0002,
      "step": 25540
    },
    {
      "epoch": 0.17627132676081603,
      "grad_norm": 0.0,
      "learning_rate": 8.238459574879095e-06,
      "loss": 0.0,
      "step": 25550
    },
    {
      "epoch": 0.17634031749536036,
      "grad_norm": 0.0,
      "learning_rate": 8.23776966753365e-06,
      "loss": 0.0,
      "step": 25560
    },
    {
      "epoch": 0.17640930822990472,
      "grad_norm": 0.0,
      "learning_rate": 8.237079760188206e-06,
      "loss": 0.0,
      "step": 25570
    },
    {
      "epoch": 0.17647829896444908,
      "grad_norm": 0.0,
      "learning_rate": 8.236389852842764e-06,
      "loss": 0.0,
      "step": 25580
    },
    {
      "epoch": 0.17654728969899341,
      "grad_norm": 3.523582563502714e-05,
      "learning_rate": 8.235699945497321e-06,
      "loss": 0.4141,
      "step": 25590
    },
    {
      "epoch": 0.17661628043353778,
      "grad_norm": 0.0,
      "learning_rate": 8.235010038151877e-06,
      "loss": 0.0,
      "step": 25600
    },
    {
      "epoch": 0.17668527116808214,
      "grad_norm": 0.0,
      "learning_rate": 8.234320130806434e-06,
      "loss": 0.0001,
      "step": 25610
    },
    {
      "epoch": 0.17675426190262647,
      "grad_norm": 0.0,
      "learning_rate": 8.23363022346099e-06,
      "loss": 0.0,
      "step": 25620
    },
    {
      "epoch": 0.17682325263717083,
      "grad_norm": 0.0,
      "learning_rate": 8.232940316115547e-06,
      "loss": 0.0,
      "step": 25630
    },
    {
      "epoch": 0.1768922433717152,
      "grad_norm": 0.0004252506187185645,
      "learning_rate": 8.232250408770103e-06,
      "loss": 0.008,
      "step": 25640
    },
    {
      "epoch": 0.17696123410625952,
      "grad_norm": 0.0,
      "learning_rate": 8.23156050142466e-06,
      "loss": 0.0,
      "step": 25650
    },
    {
      "epoch": 0.17703022484080388,
      "grad_norm": 0.0005287486128509045,
      "learning_rate": 8.230870594079216e-06,
      "loss": 0.0,
      "step": 25660
    },
    {
      "epoch": 0.17709921557534822,
      "grad_norm": 0.0,
      "learning_rate": 8.230180686733771e-06,
      "loss": 0.0377,
      "step": 25670
    },
    {
      "epoch": 0.17716820630989258,
      "grad_norm": 0.0,
      "learning_rate": 8.229490779388328e-06,
      "loss": 0.0,
      "step": 25680
    },
    {
      "epoch": 0.17723719704443694,
      "grad_norm": 0.0,
      "learning_rate": 8.228800872042886e-06,
      "loss": 0.0,
      "step": 25690
    },
    {
      "epoch": 0.17730618777898127,
      "grad_norm": 0.0,
      "learning_rate": 8.228110964697442e-06,
      "loss": 0.0,
      "step": 25700
    },
    {
      "epoch": 0.17737517851352563,
      "grad_norm": 0.0002452524204272777,
      "learning_rate": 8.227421057351999e-06,
      "loss": 0.1014,
      "step": 25710
    },
    {
      "epoch": 0.17744416924807,
      "grad_norm": 0.0,
      "learning_rate": 8.226731150006555e-06,
      "loss": 0.0058,
      "step": 25720
    },
    {
      "epoch": 0.17751315998261433,
      "grad_norm": 7.203065251815133e-06,
      "learning_rate": 8.226041242661112e-06,
      "loss": 0.0001,
      "step": 25730
    },
    {
      "epoch": 0.1775821507171587,
      "grad_norm": 0.0,
      "learning_rate": 8.225351335315668e-06,
      "loss": 0.0,
      "step": 25740
    },
    {
      "epoch": 0.17765114145170305,
      "grad_norm": 0.0,
      "learning_rate": 8.224661427970225e-06,
      "loss": 0.0,
      "step": 25750
    },
    {
      "epoch": 0.17772013218624738,
      "grad_norm": 34.71978759765625,
      "learning_rate": 8.223971520624781e-06,
      "loss": 0.0087,
      "step": 25760
    },
    {
      "epoch": 0.17778912292079174,
      "grad_norm": 1.0024551784226787e-06,
      "learning_rate": 8.223281613279338e-06,
      "loss": 0.0,
      "step": 25770
    },
    {
      "epoch": 0.17785811365533608,
      "grad_norm": 0.0,
      "learning_rate": 8.222591705933893e-06,
      "loss": 0.0,
      "step": 25780
    },
    {
      "epoch": 0.17792710438988044,
      "grad_norm": 0.0,
      "learning_rate": 8.221901798588449e-06,
      "loss": 0.0,
      "step": 25790
    },
    {
      "epoch": 0.1779960951244248,
      "grad_norm": 0.0,
      "learning_rate": 8.221211891243007e-06,
      "loss": 0.0,
      "step": 25800
    },
    {
      "epoch": 0.17806508585896913,
      "grad_norm": 0.0,
      "learning_rate": 8.220521983897564e-06,
      "loss": 0.0006,
      "step": 25810
    },
    {
      "epoch": 0.1781340765935135,
      "grad_norm": 4.79025885624651e-08,
      "learning_rate": 8.21983207655212e-06,
      "loss": 0.1689,
      "step": 25820
    },
    {
      "epoch": 0.17820306732805785,
      "grad_norm": 0.0,
      "learning_rate": 8.219142169206677e-06,
      "loss": 0.0,
      "step": 25830
    },
    {
      "epoch": 0.17827205806260218,
      "grad_norm": 0.0,
      "learning_rate": 8.218452261861233e-06,
      "loss": 0.0,
      "step": 25840
    },
    {
      "epoch": 0.17834104879714655,
      "grad_norm": 0.0,
      "learning_rate": 8.21776235451579e-06,
      "loss": 0.0027,
      "step": 25850
    },
    {
      "epoch": 0.1784100395316909,
      "grad_norm": 1.7561869736226754e-09,
      "learning_rate": 8.217072447170346e-06,
      "loss": 0.0026,
      "step": 25860
    },
    {
      "epoch": 0.17847903026623524,
      "grad_norm": 0.0,
      "learning_rate": 8.216382539824903e-06,
      "loss": 0.0,
      "step": 25870
    },
    {
      "epoch": 0.1785480210007796,
      "grad_norm": 2.0229863650911284e-07,
      "learning_rate": 8.215692632479459e-06,
      "loss": 0.0,
      "step": 25880
    },
    {
      "epoch": 0.17861701173532393,
      "grad_norm": 0.0,
      "learning_rate": 8.215002725134016e-06,
      "loss": 0.0,
      "step": 25890
    },
    {
      "epoch": 0.1786860024698683,
      "grad_norm": 0.0,
      "learning_rate": 8.21431281778857e-06,
      "loss": 0.0,
      "step": 25900
    },
    {
      "epoch": 0.17875499320441265,
      "grad_norm": 0.0,
      "learning_rate": 8.213622910443129e-06,
      "loss": 0.0001,
      "step": 25910
    },
    {
      "epoch": 0.178823983938957,
      "grad_norm": 0.0,
      "learning_rate": 8.212933003097685e-06,
      "loss": 0.0,
      "step": 25920
    },
    {
      "epoch": 0.17889297467350135,
      "grad_norm": 3.810760151878867e-09,
      "learning_rate": 8.212243095752242e-06,
      "loss": 0.0,
      "step": 25930
    },
    {
      "epoch": 0.1789619654080457,
      "grad_norm": 0.0,
      "learning_rate": 8.211553188406798e-06,
      "loss": 0.0,
      "step": 25940
    },
    {
      "epoch": 0.17903095614259004,
      "grad_norm": 0.0,
      "learning_rate": 8.210863281061355e-06,
      "loss": 0.0,
      "step": 25950
    },
    {
      "epoch": 0.1790999468771344,
      "grad_norm": 0.0,
      "learning_rate": 8.210173373715911e-06,
      "loss": 0.0,
      "step": 25960
    },
    {
      "epoch": 0.17916893761167876,
      "grad_norm": 0.007863602600991726,
      "learning_rate": 8.209483466370468e-06,
      "loss": 0.0,
      "step": 25970
    },
    {
      "epoch": 0.1792379283462231,
      "grad_norm": 0.0,
      "learning_rate": 8.208793559025024e-06,
      "loss": 0.0,
      "step": 25980
    },
    {
      "epoch": 0.17930691908076746,
      "grad_norm": 3.773438450593858e-08,
      "learning_rate": 8.20810365167958e-06,
      "loss": 0.0016,
      "step": 25990
    },
    {
      "epoch": 0.1793759098153118,
      "grad_norm": 0.0,
      "learning_rate": 8.207413744334137e-06,
      "loss": 0.0274,
      "step": 26000
    },
    {
      "epoch": 0.17944490054985615,
      "grad_norm": 1.6173011374576163e-07,
      "learning_rate": 8.206723836988692e-06,
      "loss": 0.9105,
      "step": 26010
    },
    {
      "epoch": 0.1795138912844005,
      "grad_norm": 0.0,
      "learning_rate": 8.20603392964325e-06,
      "loss": 0.048,
      "step": 26020
    },
    {
      "epoch": 0.17958288201894484,
      "grad_norm": 0.0,
      "learning_rate": 8.205344022297806e-06,
      "loss": 0.0056,
      "step": 26030
    },
    {
      "epoch": 0.1796518727534892,
      "grad_norm": 0.0,
      "learning_rate": 8.204654114952363e-06,
      "loss": 0.0,
      "step": 26040
    },
    {
      "epoch": 0.17972086348803357,
      "grad_norm": 0.0,
      "learning_rate": 8.20396420760692e-06,
      "loss": 0.0014,
      "step": 26050
    },
    {
      "epoch": 0.1797898542225779,
      "grad_norm": 0.0,
      "learning_rate": 8.203274300261476e-06,
      "loss": 0.0006,
      "step": 26060
    },
    {
      "epoch": 0.17985884495712226,
      "grad_norm": 0.0,
      "learning_rate": 8.202584392916032e-06,
      "loss": 0.0,
      "step": 26070
    },
    {
      "epoch": 0.17992783569166662,
      "grad_norm": 1.7319604639354225e-09,
      "learning_rate": 8.201894485570589e-06,
      "loss": 0.0005,
      "step": 26080
    },
    {
      "epoch": 0.17999682642621095,
      "grad_norm": 0.0,
      "learning_rate": 8.201204578225145e-06,
      "loss": 0.0,
      "step": 26090
    },
    {
      "epoch": 0.18006581716075531,
      "grad_norm": 1.5374783668420378e-09,
      "learning_rate": 8.200514670879702e-06,
      "loss": 0.4062,
      "step": 26100
    },
    {
      "epoch": 0.18013480789529965,
      "grad_norm": 0.0,
      "learning_rate": 8.199824763534258e-06,
      "loss": 0.0,
      "step": 26110
    },
    {
      "epoch": 0.180203798629844,
      "grad_norm": 0.0,
      "learning_rate": 8.199134856188813e-06,
      "loss": 0.0,
      "step": 26120
    },
    {
      "epoch": 0.18027278936438837,
      "grad_norm": 0.0,
      "learning_rate": 8.198444948843371e-06,
      "loss": 0.0,
      "step": 26130
    },
    {
      "epoch": 0.1803417800989327,
      "grad_norm": 0.0,
      "learning_rate": 8.197755041497928e-06,
      "loss": 0.0,
      "step": 26140
    },
    {
      "epoch": 0.18041077083347706,
      "grad_norm": 0.0,
      "learning_rate": 8.197065134152484e-06,
      "loss": 0.0,
      "step": 26150
    },
    {
      "epoch": 0.18047976156802142,
      "grad_norm": 0.0,
      "learning_rate": 8.19637522680704e-06,
      "loss": 0.01,
      "step": 26160
    },
    {
      "epoch": 0.18054875230256576,
      "grad_norm": 0.00018162050400860608,
      "learning_rate": 8.195685319461597e-06,
      "loss": 0.0,
      "step": 26170
    },
    {
      "epoch": 0.18061774303711012,
      "grad_norm": 1.9468604506300835e-09,
      "learning_rate": 8.194995412116154e-06,
      "loss": 0.0,
      "step": 26180
    },
    {
      "epoch": 0.18068673377165448,
      "grad_norm": 0.0,
      "learning_rate": 8.19430550477071e-06,
      "loss": 0.0,
      "step": 26190
    },
    {
      "epoch": 0.1807557245061988,
      "grad_norm": 0.36995938420295715,
      "learning_rate": 8.193615597425267e-06,
      "loss": 0.0,
      "step": 26200
    },
    {
      "epoch": 0.18082471524074317,
      "grad_norm": 0.0,
      "learning_rate": 8.192925690079823e-06,
      "loss": 0.276,
      "step": 26210
    },
    {
      "epoch": 0.1808937059752875,
      "grad_norm": 0.0,
      "learning_rate": 8.19223578273438e-06,
      "loss": 0.0,
      "step": 26220
    },
    {
      "epoch": 0.18096269670983187,
      "grad_norm": 0.0,
      "learning_rate": 8.191545875388936e-06,
      "loss": 0.0,
      "step": 26230
    },
    {
      "epoch": 0.18103168744437623,
      "grad_norm": 0.0,
      "learning_rate": 8.190855968043493e-06,
      "loss": 0.0,
      "step": 26240
    },
    {
      "epoch": 0.18110067817892056,
      "grad_norm": 0.0,
      "learning_rate": 8.19016606069805e-06,
      "loss": 0.0,
      "step": 26250
    },
    {
      "epoch": 0.18116966891346492,
      "grad_norm": 0.0,
      "learning_rate": 8.189476153352606e-06,
      "loss": 0.0,
      "step": 26260
    },
    {
      "epoch": 0.18123865964800928,
      "grad_norm": 0.0,
      "learning_rate": 8.188786246007162e-06,
      "loss": 0.2093,
      "step": 26270
    },
    {
      "epoch": 0.18130765038255361,
      "grad_norm": 0.0,
      "learning_rate": 8.188096338661719e-06,
      "loss": 0.0,
      "step": 26280
    },
    {
      "epoch": 0.18137664111709798,
      "grad_norm": 0.0,
      "learning_rate": 8.187406431316275e-06,
      "loss": 0.0,
      "step": 26290
    },
    {
      "epoch": 0.18144563185164234,
      "grad_norm": 0.0,
      "learning_rate": 8.186716523970832e-06,
      "loss": 0.0,
      "step": 26300
    },
    {
      "epoch": 0.18151462258618667,
      "grad_norm": 0.0,
      "learning_rate": 8.186026616625388e-06,
      "loss": 0.0,
      "step": 26310
    },
    {
      "epoch": 0.18158361332073103,
      "grad_norm": 0.0,
      "learning_rate": 8.185336709279945e-06,
      "loss": 0.1025,
      "step": 26320
    },
    {
      "epoch": 0.18165260405527536,
      "grad_norm": 0.0,
      "learning_rate": 8.184646801934501e-06,
      "loss": 0.0,
      "step": 26330
    },
    {
      "epoch": 0.18172159478981972,
      "grad_norm": 0.0,
      "learning_rate": 8.183956894589058e-06,
      "loss": 0.0,
      "step": 26340
    },
    {
      "epoch": 0.18179058552436408,
      "grad_norm": 0.0,
      "learning_rate": 8.183266987243614e-06,
      "loss": 0.0,
      "step": 26350
    },
    {
      "epoch": 0.18185957625890842,
      "grad_norm": 0.0,
      "learning_rate": 8.18257707989817e-06,
      "loss": 0.0,
      "step": 26360
    },
    {
      "epoch": 0.18192856699345278,
      "grad_norm": 3.433952677767138e-09,
      "learning_rate": 8.181887172552727e-06,
      "loss": 0.0,
      "step": 26370
    },
    {
      "epoch": 0.18199755772799714,
      "grad_norm": 0.0,
      "learning_rate": 8.181197265207284e-06,
      "loss": 0.0,
      "step": 26380
    },
    {
      "epoch": 0.18206654846254147,
      "grad_norm": 0.00016772629169281572,
      "learning_rate": 8.18050735786184e-06,
      "loss": 0.0,
      "step": 26390
    },
    {
      "epoch": 0.18213553919708583,
      "grad_norm": 4.427766597814298e-08,
      "learning_rate": 8.179817450516397e-06,
      "loss": 0.1638,
      "step": 26400
    },
    {
      "epoch": 0.1822045299316302,
      "grad_norm": 7.976954918831325e-08,
      "learning_rate": 8.179127543170953e-06,
      "loss": 0.0004,
      "step": 26410
    },
    {
      "epoch": 0.18227352066617453,
      "grad_norm": 0.0,
      "learning_rate": 8.17843763582551e-06,
      "loss": 0.0,
      "step": 26420
    },
    {
      "epoch": 0.1823425114007189,
      "grad_norm": 0.0,
      "learning_rate": 8.177747728480066e-06,
      "loss": 0.0,
      "step": 26430
    },
    {
      "epoch": 0.18241150213526322,
      "grad_norm": 0.0,
      "learning_rate": 8.177057821134623e-06,
      "loss": 0.0003,
      "step": 26440
    },
    {
      "epoch": 0.18248049286980758,
      "grad_norm": 1.7809059619903564,
      "learning_rate": 8.176367913789179e-06,
      "loss": 0.0169,
      "step": 26450
    },
    {
      "epoch": 0.18254948360435194,
      "grad_norm": 0.0,
      "learning_rate": 8.175678006443736e-06,
      "loss": 0.0,
      "step": 26460
    },
    {
      "epoch": 0.18261847433889628,
      "grad_norm": 0.0,
      "learning_rate": 8.174988099098292e-06,
      "loss": 0.0,
      "step": 26470
    },
    {
      "epoch": 0.18268746507344064,
      "grad_norm": 0.0,
      "learning_rate": 8.174298191752849e-06,
      "loss": 0.0001,
      "step": 26480
    },
    {
      "epoch": 0.182756455807985,
      "grad_norm": 0.0,
      "learning_rate": 8.173608284407405e-06,
      "loss": 0.0014,
      "step": 26490
    },
    {
      "epoch": 0.18282544654252933,
      "grad_norm": 0.5432473421096802,
      "learning_rate": 8.172918377061962e-06,
      "loss": 0.0001,
      "step": 26500
    },
    {
      "epoch": 0.1828944372770737,
      "grad_norm": 0.0,
      "learning_rate": 8.172228469716518e-06,
      "loss": 0.0001,
      "step": 26510
    },
    {
      "epoch": 0.18296342801161805,
      "grad_norm": 0.0,
      "learning_rate": 8.171538562371074e-06,
      "loss": 0.0025,
      "step": 26520
    },
    {
      "epoch": 0.18303241874616238,
      "grad_norm": 0.0,
      "learning_rate": 8.170848655025631e-06,
      "loss": 0.0035,
      "step": 26530
    },
    {
      "epoch": 0.18310140948070674,
      "grad_norm": 0.007199286948889494,
      "learning_rate": 8.170158747680187e-06,
      "loss": 0.3424,
      "step": 26540
    },
    {
      "epoch": 0.18317040021525108,
      "grad_norm": 0.0,
      "learning_rate": 8.169468840334744e-06,
      "loss": 0.0,
      "step": 26550
    },
    {
      "epoch": 0.18323939094979544,
      "grad_norm": 0.0,
      "learning_rate": 8.1687789329893e-06,
      "loss": 0.0,
      "step": 26560
    },
    {
      "epoch": 0.1833083816843398,
      "grad_norm": 9.50700996327214e-06,
      "learning_rate": 8.168089025643857e-06,
      "loss": 0.001,
      "step": 26570
    },
    {
      "epoch": 0.18337737241888413,
      "grad_norm": 0.0,
      "learning_rate": 8.167399118298413e-06,
      "loss": 0.0,
      "step": 26580
    },
    {
      "epoch": 0.1834463631534285,
      "grad_norm": 0.0,
      "learning_rate": 8.16670921095297e-06,
      "loss": 0.0001,
      "step": 26590
    },
    {
      "epoch": 0.18351535388797285,
      "grad_norm": 0.0,
      "learning_rate": 8.166019303607526e-06,
      "loss": 0.0,
      "step": 26600
    },
    {
      "epoch": 0.1835843446225172,
      "grad_norm": 0.0,
      "learning_rate": 8.165329396262083e-06,
      "loss": 0.0,
      "step": 26610
    },
    {
      "epoch": 0.18365333535706155,
      "grad_norm": 0.0,
      "learning_rate": 8.16463948891664e-06,
      "loss": 0.0006,
      "step": 26620
    },
    {
      "epoch": 0.1837223260916059,
      "grad_norm": 0.0,
      "learning_rate": 8.163949581571196e-06,
      "loss": 0.0,
      "step": 26630
    },
    {
      "epoch": 0.18379131682615024,
      "grad_norm": 0.0,
      "learning_rate": 8.163259674225752e-06,
      "loss": 0.0,
      "step": 26640
    },
    {
      "epoch": 0.1838603075606946,
      "grad_norm": 0.00027806765865534544,
      "learning_rate": 8.162569766880309e-06,
      "loss": 0.0,
      "step": 26650
    },
    {
      "epoch": 0.18392929829523896,
      "grad_norm": 0.0001901968935271725,
      "learning_rate": 8.161879859534865e-06,
      "loss": 0.0,
      "step": 26660
    },
    {
      "epoch": 0.1839982890297833,
      "grad_norm": 2.6538904762674065e-07,
      "learning_rate": 8.161189952189422e-06,
      "loss": 0.0,
      "step": 26670
    },
    {
      "epoch": 0.18406727976432766,
      "grad_norm": 0.0,
      "learning_rate": 8.160500044843978e-06,
      "loss": 0.0,
      "step": 26680
    },
    {
      "epoch": 0.184136270498872,
      "grad_norm": 1.838854291058567e-09,
      "learning_rate": 8.159810137498535e-06,
      "loss": 0.0,
      "step": 26690
    },
    {
      "epoch": 0.18420526123341635,
      "grad_norm": 0.0,
      "learning_rate": 8.159120230153091e-06,
      "loss": 0.0006,
      "step": 26700
    },
    {
      "epoch": 0.1842742519679607,
      "grad_norm": 0.0,
      "learning_rate": 8.158430322807648e-06,
      "loss": 0.0,
      "step": 26710
    },
    {
      "epoch": 0.18434324270250504,
      "grad_norm": 4.077456026152504e-07,
      "learning_rate": 8.157740415462204e-06,
      "loss": 0.0,
      "step": 26720
    },
    {
      "epoch": 0.1844122334370494,
      "grad_norm": 0.0,
      "learning_rate": 8.15705050811676e-06,
      "loss": 0.0809,
      "step": 26730
    },
    {
      "epoch": 0.18448122417159377,
      "grad_norm": 0.0,
      "learning_rate": 8.156360600771317e-06,
      "loss": 0.0,
      "step": 26740
    },
    {
      "epoch": 0.1845502149061381,
      "grad_norm": 0.0,
      "learning_rate": 8.155670693425874e-06,
      "loss": 0.0013,
      "step": 26750
    },
    {
      "epoch": 0.18461920564068246,
      "grad_norm": 0.03360108286142349,
      "learning_rate": 8.15498078608043e-06,
      "loss": 0.0,
      "step": 26760
    },
    {
      "epoch": 0.18468819637522682,
      "grad_norm": 0.0,
      "learning_rate": 8.154290878734987e-06,
      "loss": 0.0,
      "step": 26770
    },
    {
      "epoch": 0.18475718710977115,
      "grad_norm": 3.782386848172337e-09,
      "learning_rate": 8.153600971389543e-06,
      "loss": 0.0,
      "step": 26780
    },
    {
      "epoch": 0.18482617784431551,
      "grad_norm": 0.0,
      "learning_rate": 8.1529110640441e-06,
      "loss": 0.0012,
      "step": 26790
    },
    {
      "epoch": 0.18489516857885985,
      "grad_norm": 0.0,
      "learning_rate": 8.152221156698656e-06,
      "loss": 0.0033,
      "step": 26800
    },
    {
      "epoch": 0.1849641593134042,
      "grad_norm": 3.249361384405347e-07,
      "learning_rate": 8.151531249353213e-06,
      "loss": 0.0004,
      "step": 26810
    },
    {
      "epoch": 0.18503315004794857,
      "grad_norm": 0.0,
      "learning_rate": 8.15084134200777e-06,
      "loss": 0.0,
      "step": 26820
    },
    {
      "epoch": 0.1851021407824929,
      "grad_norm": 0.0,
      "learning_rate": 8.150151434662326e-06,
      "loss": 0.0,
      "step": 26830
    },
    {
      "epoch": 0.18517113151703726,
      "grad_norm": 0.0,
      "learning_rate": 8.149461527316882e-06,
      "loss": 0.0,
      "step": 26840
    },
    {
      "epoch": 0.18524012225158162,
      "grad_norm": 0.0,
      "learning_rate": 8.148771619971439e-06,
      "loss": 0.0,
      "step": 26850
    },
    {
      "epoch": 0.18530911298612596,
      "grad_norm": 0.0,
      "learning_rate": 8.148081712625995e-06,
      "loss": 0.0,
      "step": 26860
    },
    {
      "epoch": 0.18537810372067032,
      "grad_norm": 0.0,
      "learning_rate": 8.147391805280552e-06,
      "loss": 0.0019,
      "step": 26870
    },
    {
      "epoch": 0.18544709445521468,
      "grad_norm": 0.0,
      "learning_rate": 8.146701897935108e-06,
      "loss": 0.0,
      "step": 26880
    },
    {
      "epoch": 0.185516085189759,
      "grad_norm": 0.0,
      "learning_rate": 8.146011990589665e-06,
      "loss": 0.0001,
      "step": 26890
    },
    {
      "epoch": 0.18558507592430337,
      "grad_norm": 0.0,
      "learning_rate": 8.145322083244221e-06,
      "loss": 0.0,
      "step": 26900
    },
    {
      "epoch": 0.1856540666588477,
      "grad_norm": 0.0,
      "learning_rate": 8.144632175898778e-06,
      "loss": 0.0,
      "step": 26910
    },
    {
      "epoch": 0.18572305739339207,
      "grad_norm": 0.0,
      "learning_rate": 8.143942268553334e-06,
      "loss": 0.1884,
      "step": 26920
    },
    {
      "epoch": 0.18579204812793643,
      "grad_norm": 0.0,
      "learning_rate": 8.14325236120789e-06,
      "loss": 0.0,
      "step": 26930
    },
    {
      "epoch": 0.18586103886248076,
      "grad_norm": 0.9882470965385437,
      "learning_rate": 8.142562453862447e-06,
      "loss": 0.0009,
      "step": 26940
    },
    {
      "epoch": 0.18593002959702512,
      "grad_norm": 0.0,
      "learning_rate": 8.141872546517004e-06,
      "loss": 0.0,
      "step": 26950
    },
    {
      "epoch": 0.18599902033156948,
      "grad_norm": 0.0,
      "learning_rate": 8.14118263917156e-06,
      "loss": 0.0,
      "step": 26960
    },
    {
      "epoch": 0.18606801106611381,
      "grad_norm": 0.0002240832254756242,
      "learning_rate": 8.140492731826117e-06,
      "loss": 0.0,
      "step": 26970
    },
    {
      "epoch": 0.18613700180065818,
      "grad_norm": 0.0,
      "learning_rate": 8.139802824480673e-06,
      "loss": 0.0935,
      "step": 26980
    },
    {
      "epoch": 0.18620599253520254,
      "grad_norm": 0.0,
      "learning_rate": 8.13911291713523e-06,
      "loss": 0.0,
      "step": 26990
    },
    {
      "epoch": 0.18627498326974687,
      "grad_norm": 0.0,
      "learning_rate": 8.138423009789786e-06,
      "loss": 0.0,
      "step": 27000
    },
    {
      "epoch": 0.18634397400429123,
      "grad_norm": 0.012020009569823742,
      "learning_rate": 8.137733102444343e-06,
      "loss": 0.0,
      "step": 27010
    },
    {
      "epoch": 0.18641296473883556,
      "grad_norm": 0.0,
      "learning_rate": 8.137043195098899e-06,
      "loss": 0.0,
      "step": 27020
    },
    {
      "epoch": 0.18648195547337992,
      "grad_norm": 0.00021059042774140835,
      "learning_rate": 8.136353287753455e-06,
      "loss": 0.0,
      "step": 27030
    },
    {
      "epoch": 0.18655094620792428,
      "grad_norm": 0.0,
      "learning_rate": 8.135663380408012e-06,
      "loss": 0.0539,
      "step": 27040
    },
    {
      "epoch": 0.18661993694246862,
      "grad_norm": 0.0,
      "learning_rate": 8.134973473062568e-06,
      "loss": 0.0,
      "step": 27050
    },
    {
      "epoch": 0.18668892767701298,
      "grad_norm": 0.0,
      "learning_rate": 8.134283565717125e-06,
      "loss": 0.0,
      "step": 27060
    },
    {
      "epoch": 0.18675791841155734,
      "grad_norm": 0.0,
      "learning_rate": 8.133593658371681e-06,
      "loss": 0.0418,
      "step": 27070
    },
    {
      "epoch": 0.18682690914610167,
      "grad_norm": 5.648707428917987e-06,
      "learning_rate": 8.132903751026238e-06,
      "loss": 0.0536,
      "step": 27080
    },
    {
      "epoch": 0.18689589988064603,
      "grad_norm": 0.021818069741129875,
      "learning_rate": 8.132213843680794e-06,
      "loss": 0.0,
      "step": 27090
    },
    {
      "epoch": 0.1869648906151904,
      "grad_norm": 0.0,
      "learning_rate": 8.131523936335351e-06,
      "loss": 0.0001,
      "step": 27100
    },
    {
      "epoch": 0.18703388134973473,
      "grad_norm": 0.0,
      "learning_rate": 8.130834028989907e-06,
      "loss": 0.0027,
      "step": 27110
    },
    {
      "epoch": 0.1871028720842791,
      "grad_norm": 0.0,
      "learning_rate": 8.130144121644464e-06,
      "loss": 0.0023,
      "step": 27120
    },
    {
      "epoch": 0.18717186281882342,
      "grad_norm": 0.0,
      "learning_rate": 8.12945421429902e-06,
      "loss": 0.0,
      "step": 27130
    },
    {
      "epoch": 0.18724085355336778,
      "grad_norm": 1.878991497505922e-05,
      "learning_rate": 8.128764306953577e-06,
      "loss": 0.0,
      "step": 27140
    },
    {
      "epoch": 0.18730984428791214,
      "grad_norm": 0.0,
      "learning_rate": 8.128074399608133e-06,
      "loss": 0.0032,
      "step": 27150
    },
    {
      "epoch": 0.18737883502245647,
      "grad_norm": 0.0,
      "learning_rate": 8.12738449226269e-06,
      "loss": 0.6387,
      "step": 27160
    },
    {
      "epoch": 0.18744782575700084,
      "grad_norm": 0.0,
      "learning_rate": 8.126694584917246e-06,
      "loss": 0.0,
      "step": 27170
    },
    {
      "epoch": 0.1875168164915452,
      "grad_norm": 0.0,
      "learning_rate": 8.126004677571803e-06,
      "loss": 0.0,
      "step": 27180
    },
    {
      "epoch": 0.18758580722608953,
      "grad_norm": 0.0,
      "learning_rate": 8.12531477022636e-06,
      "loss": 0.1798,
      "step": 27190
    },
    {
      "epoch": 0.1876547979606339,
      "grad_norm": 0.0,
      "learning_rate": 8.124624862880916e-06,
      "loss": 0.0,
      "step": 27200
    },
    {
      "epoch": 0.18772378869517825,
      "grad_norm": 0.0,
      "learning_rate": 8.123934955535472e-06,
      "loss": 0.0,
      "step": 27210
    },
    {
      "epoch": 0.18779277942972258,
      "grad_norm": 0.0,
      "learning_rate": 8.123245048190029e-06,
      "loss": 0.4548,
      "step": 27220
    },
    {
      "epoch": 0.18786177016426694,
      "grad_norm": 0.0,
      "learning_rate": 8.122555140844585e-06,
      "loss": 0.0024,
      "step": 27230
    },
    {
      "epoch": 0.18793076089881128,
      "grad_norm": 0.0,
      "learning_rate": 8.121865233499142e-06,
      "loss": 0.0,
      "step": 27240
    },
    {
      "epoch": 0.18799975163335564,
      "grad_norm": 3.6567162631229166e-09,
      "learning_rate": 8.121175326153698e-06,
      "loss": 0.0007,
      "step": 27250
    },
    {
      "epoch": 0.1880687423679,
      "grad_norm": 0.0,
      "learning_rate": 8.120485418808255e-06,
      "loss": 0.0182,
      "step": 27260
    },
    {
      "epoch": 0.18813773310244433,
      "grad_norm": 0.0,
      "learning_rate": 8.119795511462811e-06,
      "loss": 0.0,
      "step": 27270
    },
    {
      "epoch": 0.1882067238369887,
      "grad_norm": 0.0,
      "learning_rate": 8.119105604117368e-06,
      "loss": 0.0,
      "step": 27280
    },
    {
      "epoch": 0.18827571457153305,
      "grad_norm": 0.0,
      "learning_rate": 8.118415696771924e-06,
      "loss": 0.0,
      "step": 27290
    },
    {
      "epoch": 0.1883447053060774,
      "grad_norm": 0.0,
      "learning_rate": 8.11772578942648e-06,
      "loss": 0.0,
      "step": 27300
    },
    {
      "epoch": 0.18841369604062175,
      "grad_norm": 0.0,
      "learning_rate": 8.117035882081037e-06,
      "loss": 0.0,
      "step": 27310
    },
    {
      "epoch": 0.1884826867751661,
      "grad_norm": 3.6656784274669008e-09,
      "learning_rate": 8.116345974735594e-06,
      "loss": 0.0043,
      "step": 27320
    },
    {
      "epoch": 0.18855167750971044,
      "grad_norm": 0.0,
      "learning_rate": 8.11565606739015e-06,
      "loss": 0.0,
      "step": 27330
    },
    {
      "epoch": 0.1886206682442548,
      "grad_norm": 0.0,
      "learning_rate": 8.114966160044707e-06,
      "loss": 0.0,
      "step": 27340
    },
    {
      "epoch": 0.18868965897879914,
      "grad_norm": 0.0,
      "learning_rate": 8.114276252699263e-06,
      "loss": 0.5191,
      "step": 27350
    },
    {
      "epoch": 0.1887586497133435,
      "grad_norm": 0.0,
      "learning_rate": 8.11358634535382e-06,
      "loss": 0.0001,
      "step": 27360
    },
    {
      "epoch": 0.18882764044788786,
      "grad_norm": 0.0,
      "learning_rate": 8.112896438008376e-06,
      "loss": 0.0,
      "step": 27370
    },
    {
      "epoch": 0.1888966311824322,
      "grad_norm": 0.0,
      "learning_rate": 8.112206530662933e-06,
      "loss": 0.0,
      "step": 27380
    },
    {
      "epoch": 0.18896562191697655,
      "grad_norm": 0.0005686933291144669,
      "learning_rate": 8.111516623317489e-06,
      "loss": 0.0001,
      "step": 27390
    },
    {
      "epoch": 0.1890346126515209,
      "grad_norm": 0.0,
      "learning_rate": 8.110826715972046e-06,
      "loss": 0.0003,
      "step": 27400
    },
    {
      "epoch": 0.18910360338606524,
      "grad_norm": 0.0,
      "learning_rate": 8.110136808626602e-06,
      "loss": 0.0003,
      "step": 27410
    },
    {
      "epoch": 0.1891725941206096,
      "grad_norm": 0.0,
      "learning_rate": 8.109446901281159e-06,
      "loss": 0.0001,
      "step": 27420
    },
    {
      "epoch": 0.18924158485515397,
      "grad_norm": 1.7458006595916231e-06,
      "learning_rate": 8.108756993935715e-06,
      "loss": 0.0013,
      "step": 27430
    },
    {
      "epoch": 0.1893105755896983,
      "grad_norm": 0.0026687169447541237,
      "learning_rate": 8.108067086590272e-06,
      "loss": 0.0,
      "step": 27440
    },
    {
      "epoch": 0.18937956632424266,
      "grad_norm": 0.0,
      "learning_rate": 8.107377179244828e-06,
      "loss": 0.0,
      "step": 27450
    },
    {
      "epoch": 0.189448557058787,
      "grad_norm": 0.0,
      "learning_rate": 8.106687271899385e-06,
      "loss": 0.0,
      "step": 27460
    },
    {
      "epoch": 0.18951754779333135,
      "grad_norm": 3.820101119345054e-05,
      "learning_rate": 8.105997364553941e-06,
      "loss": 0.0,
      "step": 27470
    },
    {
      "epoch": 0.18958653852787571,
      "grad_norm": 1.2474804407247575e-06,
      "learning_rate": 8.105307457208498e-06,
      "loss": 0.0,
      "step": 27480
    },
    {
      "epoch": 0.18965552926242005,
      "grad_norm": 1.9436503517766823e-09,
      "learning_rate": 8.104617549863054e-06,
      "loss": 0.0021,
      "step": 27490
    },
    {
      "epoch": 0.1897245199969644,
      "grad_norm": 0.0,
      "learning_rate": 8.10392764251761e-06,
      "loss": 0.0,
      "step": 27500
    },
    {
      "epoch": 0.18979351073150877,
      "grad_norm": 0.0,
      "learning_rate": 8.103237735172167e-06,
      "loss": 0.0,
      "step": 27510
    },
    {
      "epoch": 0.1898625014660531,
      "grad_norm": 0.0,
      "learning_rate": 8.102547827826724e-06,
      "loss": 0.0,
      "step": 27520
    },
    {
      "epoch": 0.18993149220059746,
      "grad_norm": 0.0,
      "learning_rate": 8.10185792048128e-06,
      "loss": 0.0,
      "step": 27530
    },
    {
      "epoch": 0.19000048293514182,
      "grad_norm": 0.0,
      "learning_rate": 8.101168013135837e-06,
      "loss": 0.0,
      "step": 27540
    },
    {
      "epoch": 0.19006947366968616,
      "grad_norm": 0.0,
      "learning_rate": 8.100478105790393e-06,
      "loss": 0.0,
      "step": 27550
    },
    {
      "epoch": 0.19013846440423052,
      "grad_norm": 0.0,
      "learning_rate": 8.09978819844495e-06,
      "loss": 0.0,
      "step": 27560
    },
    {
      "epoch": 0.19020745513877485,
      "grad_norm": 8.863953127047353e-08,
      "learning_rate": 8.099098291099506e-06,
      "loss": 0.003,
      "step": 27570
    },
    {
      "epoch": 0.1902764458733192,
      "grad_norm": 0.0,
      "learning_rate": 8.098408383754062e-06,
      "loss": 0.0,
      "step": 27580
    },
    {
      "epoch": 0.19034543660786357,
      "grad_norm": 0.0,
      "learning_rate": 8.097718476408619e-06,
      "loss": 0.0,
      "step": 27590
    },
    {
      "epoch": 0.1904144273424079,
      "grad_norm": 0.0,
      "learning_rate": 8.097028569063175e-06,
      "loss": 0.0002,
      "step": 27600
    },
    {
      "epoch": 0.19048341807695227,
      "grad_norm": 0.0,
      "learning_rate": 8.096338661717732e-06,
      "loss": 0.0093,
      "step": 27610
    },
    {
      "epoch": 0.19055240881149663,
      "grad_norm": 0.0,
      "learning_rate": 8.095648754372288e-06,
      "loss": 0.3045,
      "step": 27620
    },
    {
      "epoch": 0.19062139954604096,
      "grad_norm": 0.0,
      "learning_rate": 8.094958847026845e-06,
      "loss": 0.0001,
      "step": 27630
    },
    {
      "epoch": 0.19069039028058532,
      "grad_norm": 1.0063113222713582e-06,
      "learning_rate": 8.094268939681401e-06,
      "loss": 0.0,
      "step": 27640
    },
    {
      "epoch": 0.19075938101512968,
      "grad_norm": 0.0,
      "learning_rate": 8.093579032335958e-06,
      "loss": 0.0,
      "step": 27650
    },
    {
      "epoch": 0.19082837174967401,
      "grad_norm": 0.0,
      "learning_rate": 8.092889124990514e-06,
      "loss": 0.0,
      "step": 27660
    },
    {
      "epoch": 0.19089736248421837,
      "grad_norm": 0.0,
      "learning_rate": 8.092199217645071e-06,
      "loss": 0.0,
      "step": 27670
    },
    {
      "epoch": 0.1909663532187627,
      "grad_norm": 0.0,
      "learning_rate": 8.091509310299627e-06,
      "loss": 0.0,
      "step": 27680
    },
    {
      "epoch": 0.19103534395330707,
      "grad_norm": 0.0,
      "learning_rate": 8.090819402954184e-06,
      "loss": 0.0,
      "step": 27690
    },
    {
      "epoch": 0.19110433468785143,
      "grad_norm": 0.10012844949960709,
      "learning_rate": 8.09012949560874e-06,
      "loss": 0.0,
      "step": 27700
    },
    {
      "epoch": 0.19117332542239576,
      "grad_norm": 662.7427368164062,
      "learning_rate": 8.089439588263297e-06,
      "loss": 0.4266,
      "step": 27710
    },
    {
      "epoch": 0.19124231615694012,
      "grad_norm": 0.0,
      "learning_rate": 8.088749680917853e-06,
      "loss": 0.0,
      "step": 27720
    },
    {
      "epoch": 0.19131130689148448,
      "grad_norm": 0.0,
      "learning_rate": 8.08805977357241e-06,
      "loss": 0.0,
      "step": 27730
    },
    {
      "epoch": 0.19138029762602882,
      "grad_norm": 99.03844451904297,
      "learning_rate": 8.087369866226966e-06,
      "loss": 0.0156,
      "step": 27740
    },
    {
      "epoch": 0.19144928836057318,
      "grad_norm": 0.07214906066656113,
      "learning_rate": 8.086679958881523e-06,
      "loss": 0.0,
      "step": 27750
    },
    {
      "epoch": 0.19151827909511754,
      "grad_norm": 0.029231760650873184,
      "learning_rate": 8.08599005153608e-06,
      "loss": 0.0,
      "step": 27760
    },
    {
      "epoch": 0.19158726982966187,
      "grad_norm": 0.0,
      "learning_rate": 8.085300144190636e-06,
      "loss": 0.001,
      "step": 27770
    },
    {
      "epoch": 0.19165626056420623,
      "grad_norm": 0.0,
      "learning_rate": 8.084610236845192e-06,
      "loss": 0.0005,
      "step": 27780
    },
    {
      "epoch": 0.19172525129875057,
      "grad_norm": 9.33956544031389e-05,
      "learning_rate": 8.083920329499749e-06,
      "loss": 0.6906,
      "step": 27790
    },
    {
      "epoch": 0.19179424203329493,
      "grad_norm": 0.0,
      "learning_rate": 8.083230422154305e-06,
      "loss": 0.0,
      "step": 27800
    },
    {
      "epoch": 0.1918632327678393,
      "grad_norm": 0.2688226103782654,
      "learning_rate": 8.082540514808862e-06,
      "loss": 0.0001,
      "step": 27810
    },
    {
      "epoch": 0.19193222350238362,
      "grad_norm": 2.284170705024735e-06,
      "learning_rate": 8.081850607463418e-06,
      "loss": 0.0,
      "step": 27820
    },
    {
      "epoch": 0.19200121423692798,
      "grad_norm": 0.0,
      "learning_rate": 8.081160700117975e-06,
      "loss": 0.0,
      "step": 27830
    },
    {
      "epoch": 0.19207020497147234,
      "grad_norm": 7.018526275714976e-07,
      "learning_rate": 8.080470792772531e-06,
      "loss": 0.0001,
      "step": 27840
    },
    {
      "epoch": 0.19213919570601667,
      "grad_norm": 0.0,
      "learning_rate": 8.079780885427088e-06,
      "loss": 0.0278,
      "step": 27850
    },
    {
      "epoch": 0.19220818644056104,
      "grad_norm": 0.0002476059889886528,
      "learning_rate": 8.079090978081644e-06,
      "loss": 0.0,
      "step": 27860
    },
    {
      "epoch": 0.1922771771751054,
      "grad_norm": 0.0,
      "learning_rate": 8.0784010707362e-06,
      "loss": 0.0,
      "step": 27870
    },
    {
      "epoch": 0.19234616790964973,
      "grad_norm": 0.0,
      "learning_rate": 8.077711163390757e-06,
      "loss": 0.0,
      "step": 27880
    },
    {
      "epoch": 0.1924151586441941,
      "grad_norm": 0.0,
      "learning_rate": 8.077021256045314e-06,
      "loss": 0.0002,
      "step": 27890
    },
    {
      "epoch": 0.19248414937873842,
      "grad_norm": 0.0,
      "learning_rate": 8.07633134869987e-06,
      "loss": 0.0004,
      "step": 27900
    },
    {
      "epoch": 0.19255314011328278,
      "grad_norm": 0.0,
      "learning_rate": 8.075641441354427e-06,
      "loss": 0.0,
      "step": 27910
    },
    {
      "epoch": 0.19262213084782714,
      "grad_norm": 0.0,
      "learning_rate": 8.074951534008983e-06,
      "loss": 0.0,
      "step": 27920
    },
    {
      "epoch": 0.19269112158237148,
      "grad_norm": 0.0,
      "learning_rate": 8.07426162666354e-06,
      "loss": 0.5188,
      "step": 27930
    },
    {
      "epoch": 0.19276011231691584,
      "grad_norm": 0.0,
      "learning_rate": 8.073571719318096e-06,
      "loss": 0.0,
      "step": 27940
    },
    {
      "epoch": 0.1928291030514602,
      "grad_norm": 0.9008578658103943,
      "learning_rate": 8.072881811972653e-06,
      "loss": 0.0002,
      "step": 27950
    },
    {
      "epoch": 0.19289809378600453,
      "grad_norm": 7.989507139427587e-05,
      "learning_rate": 8.072191904627209e-06,
      "loss": 0.0001,
      "step": 27960
    },
    {
      "epoch": 0.1929670845205489,
      "grad_norm": 0.0,
      "learning_rate": 8.071501997281766e-06,
      "loss": 0.0001,
      "step": 27970
    },
    {
      "epoch": 0.19303607525509325,
      "grad_norm": 0.0,
      "learning_rate": 8.070812089936322e-06,
      "loss": 0.0001,
      "step": 27980
    },
    {
      "epoch": 0.1931050659896376,
      "grad_norm": 0.0,
      "learning_rate": 8.070122182590879e-06,
      "loss": 0.0,
      "step": 27990
    },
    {
      "epoch": 0.19317405672418195,
      "grad_norm": 0.0,
      "learning_rate": 8.069432275245435e-06,
      "loss": 0.0,
      "step": 28000
    },
    {
      "epoch": 0.19324304745872628,
      "grad_norm": 5.028121449868195e-06,
      "learning_rate": 8.068742367899992e-06,
      "loss": 0.0,
      "step": 28010
    },
    {
      "epoch": 0.19331203819327064,
      "grad_norm": 0.0,
      "learning_rate": 8.068052460554548e-06,
      "loss": 0.0001,
      "step": 28020
    },
    {
      "epoch": 0.193381028927815,
      "grad_norm": 0.0,
      "learning_rate": 8.067362553209105e-06,
      "loss": 0.0,
      "step": 28030
    },
    {
      "epoch": 0.19345001966235933,
      "grad_norm": 0.0,
      "learning_rate": 8.066672645863661e-06,
      "loss": 0.0,
      "step": 28040
    },
    {
      "epoch": 0.1935190103969037,
      "grad_norm": 0.0,
      "learning_rate": 8.065982738518218e-06,
      "loss": 0.0,
      "step": 28050
    },
    {
      "epoch": 0.19358800113144806,
      "grad_norm": 0.0,
      "learning_rate": 8.065292831172774e-06,
      "loss": 0.0,
      "step": 28060
    },
    {
      "epoch": 0.1936569918659924,
      "grad_norm": 1.2853544149038498e-06,
      "learning_rate": 8.06460292382733e-06,
      "loss": 0.0,
      "step": 28070
    },
    {
      "epoch": 0.19372598260053675,
      "grad_norm": 0.006968051660805941,
      "learning_rate": 8.063913016481887e-06,
      "loss": 0.0,
      "step": 28080
    },
    {
      "epoch": 0.1937949733350811,
      "grad_norm": 0.0,
      "learning_rate": 8.063223109136443e-06,
      "loss": 0.4004,
      "step": 28090
    },
    {
      "epoch": 0.19386396406962544,
      "grad_norm": 0.006582055706530809,
      "learning_rate": 8.062533201791e-06,
      "loss": 0.0,
      "step": 28100
    },
    {
      "epoch": 0.1939329548041698,
      "grad_norm": 0.0,
      "learning_rate": 8.061843294445556e-06,
      "loss": 0.0179,
      "step": 28110
    },
    {
      "epoch": 0.19400194553871414,
      "grad_norm": 0.0,
      "learning_rate": 8.061153387100113e-06,
      "loss": 0.0283,
      "step": 28120
    },
    {
      "epoch": 0.1940709362732585,
      "grad_norm": 0.0,
      "learning_rate": 8.06046347975467e-06,
      "loss": 0.0,
      "step": 28130
    },
    {
      "epoch": 0.19413992700780286,
      "grad_norm": 0.0,
      "learning_rate": 8.059773572409226e-06,
      "loss": 0.0,
      "step": 28140
    },
    {
      "epoch": 0.1942089177423472,
      "grad_norm": 0.0,
      "learning_rate": 8.059083665063782e-06,
      "loss": 0.0,
      "step": 28150
    },
    {
      "epoch": 0.19427790847689155,
      "grad_norm": 3.1908464431762695,
      "learning_rate": 8.058393757718339e-06,
      "loss": 0.0015,
      "step": 28160
    },
    {
      "epoch": 0.19434689921143591,
      "grad_norm": 0.20757558941841125,
      "learning_rate": 8.057703850372895e-06,
      "loss": 0.0001,
      "step": 28170
    },
    {
      "epoch": 0.19441588994598025,
      "grad_norm": 0.0,
      "learning_rate": 8.057013943027452e-06,
      "loss": 0.021,
      "step": 28180
    },
    {
      "epoch": 0.1944848806805246,
      "grad_norm": 0.0,
      "learning_rate": 8.056324035682008e-06,
      "loss": 0.0,
      "step": 28190
    },
    {
      "epoch": 0.19455387141506897,
      "grad_norm": 0.0,
      "learning_rate": 8.055634128336565e-06,
      "loss": 0.0,
      "step": 28200
    },
    {
      "epoch": 0.1946228621496133,
      "grad_norm": 0.0,
      "learning_rate": 8.054944220991121e-06,
      "loss": 0.0,
      "step": 28210
    },
    {
      "epoch": 0.19469185288415766,
      "grad_norm": 0.0,
      "learning_rate": 8.054254313645678e-06,
      "loss": 0.0,
      "step": 28220
    },
    {
      "epoch": 0.194760843618702,
      "grad_norm": 0.0,
      "learning_rate": 8.053564406300234e-06,
      "loss": 0.0013,
      "step": 28230
    },
    {
      "epoch": 0.19482983435324636,
      "grad_norm": 0.0,
      "learning_rate": 8.052943489689335e-06,
      "loss": 1.6985,
      "step": 28240
    },
    {
      "epoch": 0.19489882508779072,
      "grad_norm": 0.0,
      "learning_rate": 8.052253582343891e-06,
      "loss": 0.0,
      "step": 28250
    },
    {
      "epoch": 0.19496781582233505,
      "grad_norm": 0.0,
      "learning_rate": 8.051563674998448e-06,
      "loss": 0.0,
      "step": 28260
    },
    {
      "epoch": 0.1950368065568794,
      "grad_norm": 0.0,
      "learning_rate": 8.050873767653004e-06,
      "loss": 0.0,
      "step": 28270
    },
    {
      "epoch": 0.19510579729142377,
      "grad_norm": 0.0,
      "learning_rate": 8.050183860307561e-06,
      "loss": 0.0,
      "step": 28280
    },
    {
      "epoch": 0.1951747880259681,
      "grad_norm": 0.0,
      "learning_rate": 8.049493952962117e-06,
      "loss": 0.473,
      "step": 28290
    },
    {
      "epoch": 0.19524377876051247,
      "grad_norm": 32.63266372680664,
      "learning_rate": 8.048804045616674e-06,
      "loss": 0.007,
      "step": 28300
    },
    {
      "epoch": 0.19531276949505683,
      "grad_norm": 0.0,
      "learning_rate": 8.048114138271232e-06,
      "loss": 0.0,
      "step": 28310
    },
    {
      "epoch": 0.19538176022960116,
      "grad_norm": 2.072190909530036e-06,
      "learning_rate": 8.047424230925787e-06,
      "loss": 0.0,
      "step": 28320
    },
    {
      "epoch": 0.19545075096414552,
      "grad_norm": 0.0,
      "learning_rate": 8.046734323580343e-06,
      "loss": 0.1123,
      "step": 28330
    },
    {
      "epoch": 0.19551974169868985,
      "grad_norm": 1.80930851456651e-06,
      "learning_rate": 8.0460444162349e-06,
      "loss": 0.0,
      "step": 28340
    },
    {
      "epoch": 0.1955887324332342,
      "grad_norm": 0.0,
      "learning_rate": 8.045354508889456e-06,
      "loss": 0.0,
      "step": 28350
    },
    {
      "epoch": 0.19565772316777857,
      "grad_norm": 0.0,
      "learning_rate": 8.044664601544013e-06,
      "loss": 0.0,
      "step": 28360
    },
    {
      "epoch": 0.1957267139023229,
      "grad_norm": 0.0,
      "learning_rate": 8.04397469419857e-06,
      "loss": 0.0039,
      "step": 28370
    },
    {
      "epoch": 0.19579570463686727,
      "grad_norm": 0.0,
      "learning_rate": 8.043284786853126e-06,
      "loss": 0.0,
      "step": 28380
    },
    {
      "epoch": 0.19586469537141163,
      "grad_norm": 0.0,
      "learning_rate": 8.042594879507682e-06,
      "loss": 0.0,
      "step": 28390
    },
    {
      "epoch": 0.19593368610595596,
      "grad_norm": 0.00044247324694879353,
      "learning_rate": 8.041904972162239e-06,
      "loss": 0.0,
      "step": 28400
    },
    {
      "epoch": 0.19600267684050032,
      "grad_norm": 0.0,
      "learning_rate": 8.041215064816795e-06,
      "loss": 0.0,
      "step": 28410
    },
    {
      "epoch": 0.19607166757504468,
      "grad_norm": 0.0,
      "learning_rate": 8.040525157471354e-06,
      "loss": 0.0,
      "step": 28420
    },
    {
      "epoch": 0.19614065830958902,
      "grad_norm": 1.752103031549268e-07,
      "learning_rate": 8.039835250125908e-06,
      "loss": 0.0,
      "step": 28430
    },
    {
      "epoch": 0.19620964904413338,
      "grad_norm": 2.3830211830500048e-06,
      "learning_rate": 8.039145342780465e-06,
      "loss": 0.0,
      "step": 28440
    },
    {
      "epoch": 0.1962786397786777,
      "grad_norm": 0.0,
      "learning_rate": 8.038455435435021e-06,
      "loss": 0.0,
      "step": 28450
    },
    {
      "epoch": 0.19634763051322207,
      "grad_norm": 0.0,
      "learning_rate": 8.037765528089578e-06,
      "loss": 0.0,
      "step": 28460
    },
    {
      "epoch": 0.19641662124776643,
      "grad_norm": 0.0,
      "learning_rate": 8.037075620744134e-06,
      "loss": 0.0,
      "step": 28470
    },
    {
      "epoch": 0.19648561198231076,
      "grad_norm": 0.0,
      "learning_rate": 8.03638571339869e-06,
      "loss": 0.0,
      "step": 28480
    },
    {
      "epoch": 0.19655460271685513,
      "grad_norm": 0.0,
      "learning_rate": 8.035695806053247e-06,
      "loss": 0.0083,
      "step": 28490
    },
    {
      "epoch": 0.1966235934513995,
      "grad_norm": 0.0,
      "learning_rate": 8.035005898707804e-06,
      "loss": 0.0004,
      "step": 28500
    },
    {
      "epoch": 0.19669258418594382,
      "grad_norm": 5.137346192896075e-08,
      "learning_rate": 8.03431599136236e-06,
      "loss": 0.0,
      "step": 28510
    },
    {
      "epoch": 0.19676157492048818,
      "grad_norm": 0.0,
      "learning_rate": 8.033626084016917e-06,
      "loss": 0.0,
      "step": 28520
    },
    {
      "epoch": 0.19683056565503254,
      "grad_norm": 0.0,
      "learning_rate": 8.032936176671475e-06,
      "loss": 0.0,
      "step": 28530
    },
    {
      "epoch": 0.19689955638957687,
      "grad_norm": 0.0,
      "learning_rate": 8.032246269326031e-06,
      "loss": 0.001,
      "step": 28540
    },
    {
      "epoch": 0.19696854712412123,
      "grad_norm": 0.0,
      "learning_rate": 8.031556361980586e-06,
      "loss": 0.0,
      "step": 28550
    },
    {
      "epoch": 0.19703753785866557,
      "grad_norm": 0.0,
      "learning_rate": 8.030866454635143e-06,
      "loss": 0.0,
      "step": 28560
    },
    {
      "epoch": 0.19710652859320993,
      "grad_norm": 2.1289222786435857e-05,
      "learning_rate": 8.0301765472897e-06,
      "loss": 0.001,
      "step": 28570
    },
    {
      "epoch": 0.1971755193277543,
      "grad_norm": 0.0,
      "learning_rate": 8.029486639944256e-06,
      "loss": 0.0,
      "step": 28580
    },
    {
      "epoch": 0.19724451006229862,
      "grad_norm": 0.0,
      "learning_rate": 8.028796732598812e-06,
      "loss": 0.0861,
      "step": 28590
    },
    {
      "epoch": 0.19731350079684298,
      "grad_norm": 0.0,
      "learning_rate": 8.028106825253369e-06,
      "loss": 0.0001,
      "step": 28600
    },
    {
      "epoch": 0.19738249153138734,
      "grad_norm": 0.0,
      "learning_rate": 8.027416917907925e-06,
      "loss": 0.0012,
      "step": 28610
    },
    {
      "epoch": 0.19745148226593168,
      "grad_norm": 0.0,
      "learning_rate": 8.026727010562482e-06,
      "loss": 0.2949,
      "step": 28620
    },
    {
      "epoch": 0.19752047300047604,
      "grad_norm": 0.0,
      "learning_rate": 8.026037103217038e-06,
      "loss": 0.0,
      "step": 28630
    },
    {
      "epoch": 0.1975894637350204,
      "grad_norm": 3.5208393001084914e-06,
      "learning_rate": 8.025347195871596e-06,
      "loss": 0.0001,
      "step": 28640
    },
    {
      "epoch": 0.19765845446956473,
      "grad_norm": 0.0,
      "learning_rate": 8.024657288526153e-06,
      "loss": 0.0,
      "step": 28650
    },
    {
      "epoch": 0.1977274452041091,
      "grad_norm": 10.6176118850708,
      "learning_rate": 8.023967381180708e-06,
      "loss": 0.0025,
      "step": 28660
    },
    {
      "epoch": 0.19779643593865343,
      "grad_norm": 0.0,
      "learning_rate": 8.023277473835264e-06,
      "loss": 0.0,
      "step": 28670
    },
    {
      "epoch": 0.1978654266731978,
      "grad_norm": 0.0,
      "learning_rate": 8.02258756648982e-06,
      "loss": 0.0,
      "step": 28680
    },
    {
      "epoch": 0.19793441740774215,
      "grad_norm": 0.0,
      "learning_rate": 8.021897659144377e-06,
      "loss": 0.0029,
      "step": 28690
    },
    {
      "epoch": 0.19800340814228648,
      "grad_norm": 0.0,
      "learning_rate": 8.021207751798934e-06,
      "loss": 0.0007,
      "step": 28700
    },
    {
      "epoch": 0.19807239887683084,
      "grad_norm": 0.0,
      "learning_rate": 8.02051784445349e-06,
      "loss": 0.0,
      "step": 28710
    },
    {
      "epoch": 0.1981413896113752,
      "grad_norm": 0.0002310090058017522,
      "learning_rate": 8.019827937108047e-06,
      "loss": 0.0,
      "step": 28720
    },
    {
      "epoch": 0.19821038034591953,
      "grad_norm": 0.0,
      "learning_rate": 8.019138029762603e-06,
      "loss": 0.0001,
      "step": 28730
    },
    {
      "epoch": 0.1982793710804639,
      "grad_norm": 0.0,
      "learning_rate": 8.01844812241716e-06,
      "loss": 0.0,
      "step": 28740
    },
    {
      "epoch": 0.19834836181500826,
      "grad_norm": 0.0,
      "learning_rate": 8.017758215071718e-06,
      "loss": 0.0,
      "step": 28750
    },
    {
      "epoch": 0.1984173525495526,
      "grad_norm": 0.0,
      "learning_rate": 8.017068307726274e-06,
      "loss": 0.0,
      "step": 28760
    },
    {
      "epoch": 0.19848634328409695,
      "grad_norm": 0.0,
      "learning_rate": 8.016378400380829e-06,
      "loss": 0.0,
      "step": 28770
    },
    {
      "epoch": 0.19855533401864128,
      "grad_norm": 1.0427751817587705e-07,
      "learning_rate": 8.015688493035385e-06,
      "loss": 0.0116,
      "step": 28780
    },
    {
      "epoch": 0.19862432475318564,
      "grad_norm": 0.0,
      "learning_rate": 8.014998585689942e-06,
      "loss": 0.0,
      "step": 28790
    },
    {
      "epoch": 0.19869331548773,
      "grad_norm": 0.0,
      "learning_rate": 8.014308678344498e-06,
      "loss": 0.2629,
      "step": 28800
    },
    {
      "epoch": 0.19876230622227434,
      "grad_norm": 0.0,
      "learning_rate": 8.013618770999055e-06,
      "loss": 0.0,
      "step": 28810
    },
    {
      "epoch": 0.1988312969568187,
      "grad_norm": 0.0,
      "learning_rate": 8.012928863653611e-06,
      "loss": 0.0,
      "step": 28820
    },
    {
      "epoch": 0.19890028769136306,
      "grad_norm": 4.594663778334507e-07,
      "learning_rate": 8.012238956308168e-06,
      "loss": 0.0002,
      "step": 28830
    },
    {
      "epoch": 0.1989692784259074,
      "grad_norm": 0.0002869892923627049,
      "learning_rate": 8.011549048962724e-06,
      "loss": 0.0232,
      "step": 28840
    },
    {
      "epoch": 0.19903826916045175,
      "grad_norm": 0.0,
      "learning_rate": 8.010859141617281e-06,
      "loss": 0.119,
      "step": 28850
    },
    {
      "epoch": 0.1991072598949961,
      "grad_norm": 2.7629999749478884e-05,
      "learning_rate": 8.010169234271839e-06,
      "loss": 0.0,
      "step": 28860
    },
    {
      "epoch": 0.19917625062954045,
      "grad_norm": 0.019157318398356438,
      "learning_rate": 8.009479326926396e-06,
      "loss": 0.0,
      "step": 28870
    },
    {
      "epoch": 0.1992452413640848,
      "grad_norm": 0.0,
      "learning_rate": 8.008789419580952e-06,
      "loss": 0.0001,
      "step": 28880
    },
    {
      "epoch": 0.19931423209862914,
      "grad_norm": 0.0,
      "learning_rate": 8.008099512235507e-06,
      "loss": 0.0,
      "step": 28890
    },
    {
      "epoch": 0.1993832228331735,
      "grad_norm": 0.0,
      "learning_rate": 8.007409604890063e-06,
      "loss": 0.0,
      "step": 28900
    },
    {
      "epoch": 0.19945221356771786,
      "grad_norm": 0.0,
      "learning_rate": 8.00671969754462e-06,
      "loss": 0.0,
      "step": 28910
    },
    {
      "epoch": 0.1995212043022622,
      "grad_norm": 20.248844146728516,
      "learning_rate": 8.006029790199176e-06,
      "loss": 0.0042,
      "step": 28920
    },
    {
      "epoch": 0.19959019503680656,
      "grad_norm": 0.0,
      "learning_rate": 8.005339882853733e-06,
      "loss": 0.0064,
      "step": 28930
    },
    {
      "epoch": 0.19965918577135092,
      "grad_norm": 0.0,
      "learning_rate": 8.00464997550829e-06,
      "loss": 0.0194,
      "step": 28940
    },
    {
      "epoch": 0.19972817650589525,
      "grad_norm": 0.0,
      "learning_rate": 8.003960068162846e-06,
      "loss": 0.0,
      "step": 28950
    },
    {
      "epoch": 0.1997971672404396,
      "grad_norm": 1.489995497649943e-06,
      "learning_rate": 8.003270160817402e-06,
      "loss": 0.0024,
      "step": 28960
    },
    {
      "epoch": 0.19986615797498397,
      "grad_norm": 0.0,
      "learning_rate": 8.00258025347196e-06,
      "loss": 0.0457,
      "step": 28970
    },
    {
      "epoch": 0.1999351487095283,
      "grad_norm": 0.0,
      "learning_rate": 8.001890346126517e-06,
      "loss": 0.0004,
      "step": 28980
    },
    {
      "epoch": 0.20000413944407266,
      "grad_norm": 0.0,
      "learning_rate": 8.001200438781073e-06,
      "loss": 0.0,
      "step": 28990
    },
    {
      "epoch": 0.200073130178617,
      "grad_norm": 0.00046699633821845055,
      "learning_rate": 8.000510531435628e-06,
      "loss": 0.0001,
      "step": 29000
    },
    {
      "epoch": 0.20014212091316136,
      "grad_norm": 0.0,
      "learning_rate": 7.999820624090185e-06,
      "loss": 0.0511,
      "step": 29010
    },
    {
      "epoch": 0.20021111164770572,
      "grad_norm": 0.0,
      "learning_rate": 7.999130716744741e-06,
      "loss": 0.0,
      "step": 29020
    },
    {
      "epoch": 0.20028010238225005,
      "grad_norm": 0.0,
      "learning_rate": 7.998440809399298e-06,
      "loss": 0.0,
      "step": 29030
    },
    {
      "epoch": 0.2003490931167944,
      "grad_norm": 0.0,
      "learning_rate": 7.997750902053854e-06,
      "loss": 0.0,
      "step": 29040
    },
    {
      "epoch": 0.20041808385133877,
      "grad_norm": 0.0,
      "learning_rate": 7.99706099470841e-06,
      "loss": 0.0977,
      "step": 29050
    },
    {
      "epoch": 0.2004870745858831,
      "grad_norm": 0.0,
      "learning_rate": 7.996371087362967e-06,
      "loss": 0.0,
      "step": 29060
    },
    {
      "epoch": 0.20055606532042747,
      "grad_norm": 0.0001643638388486579,
      "learning_rate": 7.995681180017524e-06,
      "loss": 0.548,
      "step": 29070
    },
    {
      "epoch": 0.20062505605497183,
      "grad_norm": 0.0,
      "learning_rate": 7.994991272672082e-06,
      "loss": 0.0,
      "step": 29080
    },
    {
      "epoch": 0.20069404678951616,
      "grad_norm": 0.006344607565551996,
      "learning_rate": 7.994301365326638e-06,
      "loss": 0.0,
      "step": 29090
    },
    {
      "epoch": 0.20076303752406052,
      "grad_norm": 0.0,
      "learning_rate": 7.993611457981195e-06,
      "loss": 0.0,
      "step": 29100
    },
    {
      "epoch": 0.20083202825860486,
      "grad_norm": 5.787363079434726e-06,
      "learning_rate": 7.99292155063575e-06,
      "loss": 0.0207,
      "step": 29110
    },
    {
      "epoch": 0.20090101899314922,
      "grad_norm": 0.0,
      "learning_rate": 7.992231643290306e-06,
      "loss": 0.0,
      "step": 29120
    },
    {
      "epoch": 0.20097000972769358,
      "grad_norm": 0.0,
      "learning_rate": 7.991541735944863e-06,
      "loss": 0.0,
      "step": 29130
    },
    {
      "epoch": 0.2010390004622379,
      "grad_norm": 0.0,
      "learning_rate": 7.990851828599419e-06,
      "loss": 0.0001,
      "step": 29140
    },
    {
      "epoch": 0.20110799119678227,
      "grad_norm": 7.075677422108129e-05,
      "learning_rate": 7.990161921253976e-06,
      "loss": 0.0,
      "step": 29150
    },
    {
      "epoch": 0.20117698193132663,
      "grad_norm": 0.0,
      "learning_rate": 7.989472013908532e-06,
      "loss": 0.0,
      "step": 29160
    },
    {
      "epoch": 0.20124597266587096,
      "grad_norm": 0.0,
      "learning_rate": 7.988782106563089e-06,
      "loss": 0.0005,
      "step": 29170
    },
    {
      "epoch": 0.20131496340041533,
      "grad_norm": 0.0,
      "learning_rate": 7.988092199217645e-06,
      "loss": 0.0,
      "step": 29180
    },
    {
      "epoch": 0.2013839541349597,
      "grad_norm": 0.0,
      "learning_rate": 7.987402291872203e-06,
      "loss": 0.0,
      "step": 29190
    },
    {
      "epoch": 0.20145294486950402,
      "grad_norm": 0.0,
      "learning_rate": 7.98671238452676e-06,
      "loss": 0.0,
      "step": 29200
    },
    {
      "epoch": 0.20152193560404838,
      "grad_norm": 0.0,
      "learning_rate": 7.986022477181316e-06,
      "loss": 0.0,
      "step": 29210
    },
    {
      "epoch": 0.20159092633859274,
      "grad_norm": 0.0,
      "learning_rate": 7.985332569835871e-06,
      "loss": 0.012,
      "step": 29220
    },
    {
      "epoch": 0.20165991707313707,
      "grad_norm": 0.0,
      "learning_rate": 7.984642662490428e-06,
      "loss": 0.0,
      "step": 29230
    },
    {
      "epoch": 0.20172890780768143,
      "grad_norm": 2.1180779796736715e-08,
      "learning_rate": 7.983952755144984e-06,
      "loss": 0.0,
      "step": 29240
    },
    {
      "epoch": 0.20179789854222577,
      "grad_norm": 0.0,
      "learning_rate": 7.98326284779954e-06,
      "loss": 0.0,
      "step": 29250
    },
    {
      "epoch": 0.20186688927677013,
      "grad_norm": 0.0,
      "learning_rate": 7.982572940454097e-06,
      "loss": 0.0,
      "step": 29260
    },
    {
      "epoch": 0.2019358800113145,
      "grad_norm": 0.0,
      "learning_rate": 7.981883033108654e-06,
      "loss": 0.0,
      "step": 29270
    },
    {
      "epoch": 0.20200487074585882,
      "grad_norm": 0.0,
      "learning_rate": 7.98119312576321e-06,
      "loss": 0.0,
      "step": 29280
    },
    {
      "epoch": 0.20207386148040318,
      "grad_norm": 0.0,
      "learning_rate": 7.980503218417766e-06,
      "loss": 0.0,
      "step": 29290
    },
    {
      "epoch": 0.20214285221494754,
      "grad_norm": 0.0,
      "learning_rate": 7.979813311072325e-06,
      "loss": 0.0,
      "step": 29300
    },
    {
      "epoch": 0.20221184294949188,
      "grad_norm": 0.0,
      "learning_rate": 7.979123403726881e-06,
      "loss": 0.0,
      "step": 29310
    },
    {
      "epoch": 0.20228083368403624,
      "grad_norm": 0.0,
      "learning_rate": 7.978433496381438e-06,
      "loss": 0.0,
      "step": 29320
    },
    {
      "epoch": 0.2023498244185806,
      "grad_norm": 0.16354624927043915,
      "learning_rate": 7.977743589035994e-06,
      "loss": 0.0,
      "step": 29330
    },
    {
      "epoch": 0.20241881515312493,
      "grad_norm": 0.0,
      "learning_rate": 7.977053681690549e-06,
      "loss": 0.0,
      "step": 29340
    },
    {
      "epoch": 0.2024878058876693,
      "grad_norm": 6.709145054628607e-06,
      "learning_rate": 7.976363774345105e-06,
      "loss": 0.0,
      "step": 29350
    },
    {
      "epoch": 0.20255679662221363,
      "grad_norm": 8.35243336041458e-05,
      "learning_rate": 7.975673866999662e-06,
      "loss": 0.0,
      "step": 29360
    },
    {
      "epoch": 0.20262578735675799,
      "grad_norm": 0.005395881831645966,
      "learning_rate": 7.974983959654218e-06,
      "loss": 0.0058,
      "step": 29370
    },
    {
      "epoch": 0.20269477809130235,
      "grad_norm": 0.0,
      "learning_rate": 7.974294052308775e-06,
      "loss": 0.0002,
      "step": 29380
    },
    {
      "epoch": 0.20276376882584668,
      "grad_norm": 3.590628017136055e-09,
      "learning_rate": 7.973604144963331e-06,
      "loss": 0.0,
      "step": 29390
    },
    {
      "epoch": 0.20283275956039104,
      "grad_norm": 0.0029196913819760084,
      "learning_rate": 7.972914237617888e-06,
      "loss": 0.0,
      "step": 29400
    },
    {
      "epoch": 0.2029017502949354,
      "grad_norm": 0.0,
      "learning_rate": 7.972224330272446e-06,
      "loss": 0.0,
      "step": 29410
    },
    {
      "epoch": 0.20297074102947973,
      "grad_norm": 2.9109278329997323e-05,
      "learning_rate": 7.971534422927003e-06,
      "loss": 0.0346,
      "step": 29420
    },
    {
      "epoch": 0.2030397317640241,
      "grad_norm": 0.0,
      "learning_rate": 7.970844515581559e-06,
      "loss": 0.0,
      "step": 29430
    },
    {
      "epoch": 0.20310872249856846,
      "grad_norm": 0.0,
      "learning_rate": 7.970154608236116e-06,
      "loss": 0.031,
      "step": 29440
    },
    {
      "epoch": 0.2031777132331128,
      "grad_norm": 0.0015341612743213773,
      "learning_rate": 7.96946470089067e-06,
      "loss": 0.0,
      "step": 29450
    },
    {
      "epoch": 0.20324670396765715,
      "grad_norm": 0.0,
      "learning_rate": 7.968774793545227e-06,
      "loss": 0.0001,
      "step": 29460
    },
    {
      "epoch": 0.20331569470220148,
      "grad_norm": 3.71329201698245e-06,
      "learning_rate": 7.968084886199783e-06,
      "loss": 0.0,
      "step": 29470
    },
    {
      "epoch": 0.20338468543674584,
      "grad_norm": 0.0,
      "learning_rate": 7.96739497885434e-06,
      "loss": 0.0,
      "step": 29480
    },
    {
      "epoch": 0.2034536761712902,
      "grad_norm": 0.0,
      "learning_rate": 7.966705071508896e-06,
      "loss": 0.0,
      "step": 29490
    },
    {
      "epoch": 0.20352266690583454,
      "grad_norm": 0.0,
      "learning_rate": 7.966015164163453e-06,
      "loss": 0.0,
      "step": 29500
    },
    {
      "epoch": 0.2035916576403789,
      "grad_norm": 0.009508379735052586,
      "learning_rate": 7.96532525681801e-06,
      "loss": 0.0,
      "step": 29510
    },
    {
      "epoch": 0.20366064837492326,
      "grad_norm": 0.0,
      "learning_rate": 7.964635349472567e-06,
      "loss": 0.0,
      "step": 29520
    },
    {
      "epoch": 0.2037296391094676,
      "grad_norm": 0.00326364953070879,
      "learning_rate": 7.963945442127124e-06,
      "loss": 0.0003,
      "step": 29530
    },
    {
      "epoch": 0.20379862984401195,
      "grad_norm": 0.0,
      "learning_rate": 7.96325553478168e-06,
      "loss": 0.0,
      "step": 29540
    },
    {
      "epoch": 0.2038676205785563,
      "grad_norm": 0.0,
      "learning_rate": 7.962565627436237e-06,
      "loss": 0.0,
      "step": 29550
    },
    {
      "epoch": 0.20393661131310065,
      "grad_norm": 0.0,
      "learning_rate": 7.961875720090792e-06,
      "loss": 0.0,
      "step": 29560
    },
    {
      "epoch": 0.204005602047645,
      "grad_norm": 5.616179166167967e-08,
      "learning_rate": 7.961185812745348e-06,
      "loss": 0.0,
      "step": 29570
    },
    {
      "epoch": 0.20407459278218934,
      "grad_norm": 0.0,
      "learning_rate": 7.960495905399905e-06,
      "loss": 0.0,
      "step": 29580
    },
    {
      "epoch": 0.2041435835167337,
      "grad_norm": 1.4619402463722508e-05,
      "learning_rate": 7.959805998054461e-06,
      "loss": 0.0,
      "step": 29590
    },
    {
      "epoch": 0.20421257425127806,
      "grad_norm": 0.0,
      "learning_rate": 7.959116090709018e-06,
      "loss": 0.0065,
      "step": 29600
    },
    {
      "epoch": 0.2042815649858224,
      "grad_norm": 0.0,
      "learning_rate": 7.958426183363574e-06,
      "loss": 0.0005,
      "step": 29610
    },
    {
      "epoch": 0.20435055572036676,
      "grad_norm": 0.0,
      "learning_rate": 7.95773627601813e-06,
      "loss": 0.0002,
      "step": 29620
    },
    {
      "epoch": 0.20441954645491112,
      "grad_norm": 0.0,
      "learning_rate": 7.957046368672689e-06,
      "loss": 0.0,
      "step": 29630
    },
    {
      "epoch": 0.20448853718945545,
      "grad_norm": 0.0,
      "learning_rate": 7.956356461327245e-06,
      "loss": 0.0002,
      "step": 29640
    },
    {
      "epoch": 0.2045575279239998,
      "grad_norm": 1.8629722209340116e-09,
      "learning_rate": 7.955666553981802e-06,
      "loss": 0.0,
      "step": 29650
    },
    {
      "epoch": 0.20462651865854417,
      "grad_norm": 0.0,
      "learning_rate": 7.954976646636358e-06,
      "loss": 0.0,
      "step": 29660
    },
    {
      "epoch": 0.2046955093930885,
      "grad_norm": 0.0,
      "learning_rate": 7.954286739290915e-06,
      "loss": 0.0,
      "step": 29670
    },
    {
      "epoch": 0.20476450012763286,
      "grad_norm": 0.0,
      "learning_rate": 7.95359683194547e-06,
      "loss": 0.0001,
      "step": 29680
    },
    {
      "epoch": 0.2048334908621772,
      "grad_norm": 0.0,
      "learning_rate": 7.952906924600026e-06,
      "loss": 0.0,
      "step": 29690
    },
    {
      "epoch": 0.20490248159672156,
      "grad_norm": 0.0,
      "learning_rate": 7.952217017254583e-06,
      "loss": 0.0,
      "step": 29700
    },
    {
      "epoch": 0.20497147233126592,
      "grad_norm": 0.0,
      "learning_rate": 7.951527109909139e-06,
      "loss": 0.0,
      "step": 29710
    },
    {
      "epoch": 0.20504046306581025,
      "grad_norm": 0.008679023943841457,
      "learning_rate": 7.950837202563696e-06,
      "loss": 0.0,
      "step": 29720
    },
    {
      "epoch": 0.2051094538003546,
      "grad_norm": 0.11770489811897278,
      "learning_rate": 7.950147295218252e-06,
      "loss": 0.0,
      "step": 29730
    },
    {
      "epoch": 0.20517844453489897,
      "grad_norm": 0.17468474805355072,
      "learning_rate": 7.94945738787281e-06,
      "loss": 0.0,
      "step": 29740
    },
    {
      "epoch": 0.2052474352694433,
      "grad_norm": 0.0,
      "learning_rate": 7.948767480527367e-06,
      "loss": 0.0,
      "step": 29750
    },
    {
      "epoch": 0.20531642600398767,
      "grad_norm": 533.59765625,
      "learning_rate": 7.948077573181923e-06,
      "loss": 0.602,
      "step": 29760
    },
    {
      "epoch": 0.20538541673853203,
      "grad_norm": 0.0,
      "learning_rate": 7.94738766583648e-06,
      "loss": 0.0,
      "step": 29770
    },
    {
      "epoch": 0.20545440747307636,
      "grad_norm": 0.0,
      "learning_rate": 7.946697758491036e-06,
      "loss": 0.004,
      "step": 29780
    },
    {
      "epoch": 0.20552339820762072,
      "grad_norm": 0.0,
      "learning_rate": 7.946007851145591e-06,
      "loss": 0.0,
      "step": 29790
    },
    {
      "epoch": 0.20559238894216506,
      "grad_norm": 0.0,
      "learning_rate": 7.945317943800148e-06,
      "loss": 0.0003,
      "step": 29800
    },
    {
      "epoch": 0.20566137967670942,
      "grad_norm": 0.0,
      "learning_rate": 7.944628036454704e-06,
      "loss": 0.0001,
      "step": 29810
    },
    {
      "epoch": 0.20573037041125378,
      "grad_norm": 1.8241812504982136e-09,
      "learning_rate": 7.94393812910926e-06,
      "loss": 0.0591,
      "step": 29820
    },
    {
      "epoch": 0.2057993611457981,
      "grad_norm": 4.153274858254008e-06,
      "learning_rate": 7.943248221763817e-06,
      "loss": 0.048,
      "step": 29830
    },
    {
      "epoch": 0.20586835188034247,
      "grad_norm": 0.0,
      "learning_rate": 7.942558314418373e-06,
      "loss": 0.0,
      "step": 29840
    },
    {
      "epoch": 0.20593734261488683,
      "grad_norm": 0.0,
      "learning_rate": 7.941868407072932e-06,
      "loss": 0.0,
      "step": 29850
    },
    {
      "epoch": 0.20600633334943116,
      "grad_norm": 0.0,
      "learning_rate": 7.941178499727488e-06,
      "loss": 0.0039,
      "step": 29860
    },
    {
      "epoch": 0.20607532408397553,
      "grad_norm": 0.0,
      "learning_rate": 7.940488592382045e-06,
      "loss": 0.0,
      "step": 29870
    },
    {
      "epoch": 0.20614431481851989,
      "grad_norm": 0.0,
      "learning_rate": 7.939798685036601e-06,
      "loss": 0.0,
      "step": 29880
    },
    {
      "epoch": 0.20621330555306422,
      "grad_norm": 0.0,
      "learning_rate": 7.939108777691158e-06,
      "loss": 0.0,
      "step": 29890
    },
    {
      "epoch": 0.20628229628760858,
      "grad_norm": 0.0,
      "learning_rate": 7.938418870345712e-06,
      "loss": 0.0,
      "step": 29900
    },
    {
      "epoch": 0.2063512870221529,
      "grad_norm": 4.2027363811314444e-09,
      "learning_rate": 7.937728963000269e-06,
      "loss": 0.0,
      "step": 29910
    },
    {
      "epoch": 0.20642027775669727,
      "grad_norm": 0.0,
      "learning_rate": 7.937039055654825e-06,
      "loss": 0.0,
      "step": 29920
    },
    {
      "epoch": 0.20648926849124163,
      "grad_norm": 0.0,
      "learning_rate": 7.936349148309382e-06,
      "loss": 0.9637,
      "step": 29930
    },
    {
      "epoch": 0.20655825922578597,
      "grad_norm": 0.0,
      "learning_rate": 7.935659240963938e-06,
      "loss": 0.0,
      "step": 29940
    },
    {
      "epoch": 0.20662724996033033,
      "grad_norm": 2.95427685159666e-06,
      "learning_rate": 7.934969333618495e-06,
      "loss": 0.0,
      "step": 29950
    },
    {
      "epoch": 0.2066962406948747,
      "grad_norm": 0.0,
      "learning_rate": 7.934279426273053e-06,
      "loss": 0.0,
      "step": 29960
    },
    {
      "epoch": 0.20676523142941902,
      "grad_norm": 0.0,
      "learning_rate": 7.93358951892761e-06,
      "loss": 0.0,
      "step": 29970
    },
    {
      "epoch": 0.20683422216396338,
      "grad_norm": 0.0,
      "learning_rate": 7.932899611582166e-06,
      "loss": 0.0,
      "step": 29980
    },
    {
      "epoch": 0.20690321289850774,
      "grad_norm": 0.0,
      "learning_rate": 7.932209704236723e-06,
      "loss": 0.0082,
      "step": 29990
    },
    {
      "epoch": 0.20697220363305208,
      "grad_norm": 0.0,
      "learning_rate": 7.931519796891279e-06,
      "loss": 0.0,
      "step": 30000
    },
    {
      "epoch": 0.20704119436759644,
      "grad_norm": 2.69940727548601e-08,
      "learning_rate": 7.930829889545835e-06,
      "loss": 0.0,
      "step": 30010
    },
    {
      "epoch": 0.20711018510214077,
      "grad_norm": 0.0,
      "learning_rate": 7.93013998220039e-06,
      "loss": 0.0,
      "step": 30020
    },
    {
      "epoch": 0.20717917583668513,
      "grad_norm": 0.0,
      "learning_rate": 7.929450074854947e-06,
      "loss": 0.0,
      "step": 30030
    },
    {
      "epoch": 0.2072481665712295,
      "grad_norm": 0.005969700403511524,
      "learning_rate": 7.928760167509503e-06,
      "loss": 0.0,
      "step": 30040
    },
    {
      "epoch": 0.20731715730577382,
      "grad_norm": 0.02649962157011032,
      "learning_rate": 7.92807026016406e-06,
      "loss": 0.0,
      "step": 30050
    },
    {
      "epoch": 0.20738614804031819,
      "grad_norm": 0.0,
      "learning_rate": 7.927380352818616e-06,
      "loss": 0.0,
      "step": 30060
    },
    {
      "epoch": 0.20745513877486255,
      "grad_norm": 0.0,
      "learning_rate": 7.926690445473174e-06,
      "loss": 0.0,
      "step": 30070
    },
    {
      "epoch": 0.20752412950940688,
      "grad_norm": 0.0,
      "learning_rate": 7.926000538127731e-06,
      "loss": 0.0,
      "step": 30080
    },
    {
      "epoch": 0.20759312024395124,
      "grad_norm": 0.0,
      "learning_rate": 7.925310630782287e-06,
      "loss": 0.0,
      "step": 30090
    },
    {
      "epoch": 0.2076621109784956,
      "grad_norm": 587.1788330078125,
      "learning_rate": 7.924620723436844e-06,
      "loss": 0.4207,
      "step": 30100
    },
    {
      "epoch": 0.20773110171303993,
      "grad_norm": 3.318666008667037e-09,
      "learning_rate": 7.9239308160914e-06,
      "loss": 0.0001,
      "step": 30110
    },
    {
      "epoch": 0.2078000924475843,
      "grad_norm": 0.0,
      "learning_rate": 7.923240908745957e-06,
      "loss": 0.0907,
      "step": 30120
    },
    {
      "epoch": 0.20786908318212863,
      "grad_norm": 0.0,
      "learning_rate": 7.922551001400512e-06,
      "loss": 0.0,
      "step": 30130
    },
    {
      "epoch": 0.207938073916673,
      "grad_norm": 0.0,
      "learning_rate": 7.921861094055068e-06,
      "loss": 0.0003,
      "step": 30140
    },
    {
      "epoch": 0.20800706465121735,
      "grad_norm": 0.0,
      "learning_rate": 7.921171186709625e-06,
      "loss": 0.0001,
      "step": 30150
    },
    {
      "epoch": 0.20807605538576168,
      "grad_norm": 0.012880227528512478,
      "learning_rate": 7.920481279364181e-06,
      "loss": 0.0,
      "step": 30160
    },
    {
      "epoch": 0.20814504612030604,
      "grad_norm": 0.0,
      "learning_rate": 7.919791372018738e-06,
      "loss": 0.0,
      "step": 30170
    },
    {
      "epoch": 0.2082140368548504,
      "grad_norm": 0.0,
      "learning_rate": 7.919101464673296e-06,
      "loss": 0.0002,
      "step": 30180
    },
    {
      "epoch": 0.20828302758939474,
      "grad_norm": 0.0,
      "learning_rate": 7.918411557327852e-06,
      "loss": 0.0323,
      "step": 30190
    },
    {
      "epoch": 0.2083520183239391,
      "grad_norm": 0.0,
      "learning_rate": 7.917721649982409e-06,
      "loss": 0.0,
      "step": 30200
    },
    {
      "epoch": 0.20842100905848346,
      "grad_norm": 0.0,
      "learning_rate": 7.917031742636965e-06,
      "loss": 0.0042,
      "step": 30210
    },
    {
      "epoch": 0.2084899997930278,
      "grad_norm": 0.5870230197906494,
      "learning_rate": 7.916341835291522e-06,
      "loss": 0.0001,
      "step": 30220
    },
    {
      "epoch": 0.20855899052757215,
      "grad_norm": 7.398648449452594e-05,
      "learning_rate": 7.915651927946078e-06,
      "loss": 0.0,
      "step": 30230
    },
    {
      "epoch": 0.20862798126211649,
      "grad_norm": 6.213717824721243e-06,
      "learning_rate": 7.914962020600633e-06,
      "loss": 0.0,
      "step": 30240
    },
    {
      "epoch": 0.20869697199666085,
      "grad_norm": 0.0,
      "learning_rate": 7.91427211325519e-06,
      "loss": 0.0,
      "step": 30250
    },
    {
      "epoch": 0.2087659627312052,
      "grad_norm": 0.0,
      "learning_rate": 7.913582205909746e-06,
      "loss": 0.0001,
      "step": 30260
    },
    {
      "epoch": 0.20883495346574954,
      "grad_norm": 1.6718652018354874e-09,
      "learning_rate": 7.912892298564303e-06,
      "loss": 0.0,
      "step": 30270
    },
    {
      "epoch": 0.2089039442002939,
      "grad_norm": 0.0,
      "learning_rate": 7.912202391218859e-06,
      "loss": 0.0,
      "step": 30280
    },
    {
      "epoch": 0.20897293493483826,
      "grad_norm": 0.0,
      "learning_rate": 7.911512483873417e-06,
      "loss": 0.0,
      "step": 30290
    },
    {
      "epoch": 0.2090419256693826,
      "grad_norm": 0.0,
      "learning_rate": 7.910822576527974e-06,
      "loss": 0.0,
      "step": 30300
    },
    {
      "epoch": 0.20911091640392696,
      "grad_norm": 0.0,
      "learning_rate": 7.91013266918253e-06,
      "loss": 0.0,
      "step": 30310
    },
    {
      "epoch": 0.20917990713847132,
      "grad_norm": 0.01005993876606226,
      "learning_rate": 7.909442761837087e-06,
      "loss": 0.0,
      "step": 30320
    },
    {
      "epoch": 0.20924889787301565,
      "grad_norm": 0.00029130533221177757,
      "learning_rate": 7.908752854491643e-06,
      "loss": 0.0004,
      "step": 30330
    },
    {
      "epoch": 0.20931788860756,
      "grad_norm": 0.0,
      "learning_rate": 7.9080629471462e-06,
      "loss": 0.0,
      "step": 30340
    },
    {
      "epoch": 0.20938687934210434,
      "grad_norm": 154.89215087890625,
      "learning_rate": 7.907373039800756e-06,
      "loss": 0.0574,
      "step": 30350
    },
    {
      "epoch": 0.2094558700766487,
      "grad_norm": 0.0031173587776720524,
      "learning_rate": 7.906683132455311e-06,
      "loss": 0.2479,
      "step": 30360
    },
    {
      "epoch": 0.20952486081119306,
      "grad_norm": 0.0,
      "learning_rate": 7.905993225109867e-06,
      "loss": 0.0196,
      "step": 30370
    },
    {
      "epoch": 0.2095938515457374,
      "grad_norm": 1.9878325474564917e-05,
      "learning_rate": 7.905303317764424e-06,
      "loss": 0.0,
      "step": 30380
    },
    {
      "epoch": 0.20966284228028176,
      "grad_norm": 0.00028508290415629745,
      "learning_rate": 7.90461341041898e-06,
      "loss": 0.0,
      "step": 30390
    },
    {
      "epoch": 0.20973183301482612,
      "grad_norm": 3.909709000993189e-09,
      "learning_rate": 7.903923503073539e-06,
      "loss": 0.0,
      "step": 30400
    },
    {
      "epoch": 0.20980082374937045,
      "grad_norm": 0.0,
      "learning_rate": 7.903233595728095e-06,
      "loss": 0.0,
      "step": 30410
    },
    {
      "epoch": 0.2098698144839148,
      "grad_norm": 1.0118680338067065e-09,
      "learning_rate": 7.902543688382652e-06,
      "loss": 0.0,
      "step": 30420
    },
    {
      "epoch": 0.20993880521845917,
      "grad_norm": 0.022737300023436546,
      "learning_rate": 7.901853781037208e-06,
      "loss": 0.0,
      "step": 30430
    },
    {
      "epoch": 0.2100077959530035,
      "grad_norm": 0.002234002575278282,
      "learning_rate": 7.901163873691765e-06,
      "loss": 0.0055,
      "step": 30440
    },
    {
      "epoch": 0.21007678668754787,
      "grad_norm": 0.0017814544262364507,
      "learning_rate": 7.900473966346321e-06,
      "loss": 0.0052,
      "step": 30450
    },
    {
      "epoch": 0.2101457774220922,
      "grad_norm": 0.0,
      "learning_rate": 7.899784059000878e-06,
      "loss": 0.0,
      "step": 30460
    },
    {
      "epoch": 0.21021476815663656,
      "grad_norm": 1.030958890914917,
      "learning_rate": 7.899094151655432e-06,
      "loss": 0.0003,
      "step": 30470
    },
    {
      "epoch": 0.21028375889118092,
      "grad_norm": 8.652215610638336e-10,
      "learning_rate": 7.898404244309989e-06,
      "loss": 0.0,
      "step": 30480
    },
    {
      "epoch": 0.21035274962572525,
      "grad_norm": 0.0,
      "learning_rate": 7.897714336964545e-06,
      "loss": 0.0017,
      "step": 30490
    },
    {
      "epoch": 0.21042174036026962,
      "grad_norm": 0.1631506383419037,
      "learning_rate": 7.897024429619102e-06,
      "loss": 0.0,
      "step": 30500
    },
    {
      "epoch": 0.21049073109481398,
      "grad_norm": 0.009636785835027695,
      "learning_rate": 7.89633452227366e-06,
      "loss": 0.0,
      "step": 30510
    },
    {
      "epoch": 0.2105597218293583,
      "grad_norm": 1.7102615990083336e-09,
      "learning_rate": 7.895644614928217e-06,
      "loss": 0.0,
      "step": 30520
    },
    {
      "epoch": 0.21062871256390267,
      "grad_norm": 0.0,
      "learning_rate": 7.894954707582773e-06,
      "loss": 0.0,
      "step": 30530
    },
    {
      "epoch": 0.21069770329844703,
      "grad_norm": 0.0049896929413080215,
      "learning_rate": 7.89426480023733e-06,
      "loss": 0.0,
      "step": 30540
    },
    {
      "epoch": 0.21076669403299136,
      "grad_norm": 0.0,
      "learning_rate": 7.893574892891886e-06,
      "loss": 0.0,
      "step": 30550
    },
    {
      "epoch": 0.21083568476753572,
      "grad_norm": 0.0,
      "learning_rate": 7.892884985546442e-06,
      "loss": 0.0,
      "step": 30560
    },
    {
      "epoch": 0.21090467550208006,
      "grad_norm": 0.0,
      "learning_rate": 7.892195078200999e-06,
      "loss": 0.3744,
      "step": 30570
    },
    {
      "epoch": 0.21097366623662442,
      "grad_norm": 1.631396884249625e-08,
      "learning_rate": 7.891505170855554e-06,
      "loss": 0.0,
      "step": 30580
    },
    {
      "epoch": 0.21104265697116878,
      "grad_norm": 0.0,
      "learning_rate": 7.89081526351011e-06,
      "loss": 0.0,
      "step": 30590
    },
    {
      "epoch": 0.2111116477057131,
      "grad_norm": 0.0,
      "learning_rate": 7.890125356164667e-06,
      "loss": 0.0,
      "step": 30600
    },
    {
      "epoch": 0.21118063844025747,
      "grad_norm": 0.0,
      "learning_rate": 7.889435448819223e-06,
      "loss": 0.0,
      "step": 30610
    },
    {
      "epoch": 0.21124962917480183,
      "grad_norm": 0.0,
      "learning_rate": 7.888745541473781e-06,
      "loss": 0.0,
      "step": 30620
    },
    {
      "epoch": 0.21131861990934617,
      "grad_norm": 0.0,
      "learning_rate": 7.888055634128338e-06,
      "loss": 0.0,
      "step": 30630
    },
    {
      "epoch": 0.21138761064389053,
      "grad_norm": 0.0,
      "learning_rate": 7.887365726782894e-06,
      "loss": 0.0002,
      "step": 30640
    },
    {
      "epoch": 0.2114566013784349,
      "grad_norm": 4.134491064178292e-06,
      "learning_rate": 7.886675819437451e-06,
      "loss": 0.0002,
      "step": 30650
    },
    {
      "epoch": 0.21152559211297922,
      "grad_norm": 0.0,
      "learning_rate": 7.885985912092007e-06,
      "loss": 0.0,
      "step": 30660
    },
    {
      "epoch": 0.21159458284752358,
      "grad_norm": 0.0,
      "learning_rate": 7.885296004746564e-06,
      "loss": 0.0,
      "step": 30670
    },
    {
      "epoch": 0.21166357358206792,
      "grad_norm": 0.004317484796047211,
      "learning_rate": 7.88460609740112e-06,
      "loss": 0.0,
      "step": 30680
    },
    {
      "epoch": 0.21173256431661228,
      "grad_norm": 0.0,
      "learning_rate": 7.883916190055675e-06,
      "loss": 0.0001,
      "step": 30690
    },
    {
      "epoch": 0.21180155505115664,
      "grad_norm": 0.0,
      "learning_rate": 7.883226282710232e-06,
      "loss": 0.0,
      "step": 30700
    },
    {
      "epoch": 0.21187054578570097,
      "grad_norm": 3.0680889722134452e-06,
      "learning_rate": 7.882536375364788e-06,
      "loss": 0.0,
      "step": 30710
    },
    {
      "epoch": 0.21193953652024533,
      "grad_norm": 3.896599487518415e-09,
      "learning_rate": 7.881846468019345e-06,
      "loss": 0.0,
      "step": 30720
    },
    {
      "epoch": 0.2120085272547897,
      "grad_norm": 0.035693928599357605,
      "learning_rate": 7.881156560673903e-06,
      "loss": 0.0,
      "step": 30730
    },
    {
      "epoch": 0.21207751798933402,
      "grad_norm": 4.6698460209881887e-07,
      "learning_rate": 7.88046665332846e-06,
      "loss": 0.0008,
      "step": 30740
    },
    {
      "epoch": 0.21214650872387839,
      "grad_norm": 0.0,
      "learning_rate": 7.879776745983016e-06,
      "loss": 0.0,
      "step": 30750
    },
    {
      "epoch": 0.21221549945842275,
      "grad_norm": 0.0,
      "learning_rate": 7.879086838637572e-06,
      "loss": 0.0,
      "step": 30760
    },
    {
      "epoch": 0.21228449019296708,
      "grad_norm": 0.0,
      "learning_rate": 7.878396931292129e-06,
      "loss": 0.0,
      "step": 30770
    },
    {
      "epoch": 0.21235348092751144,
      "grad_norm": 0.0,
      "learning_rate": 7.877707023946685e-06,
      "loss": 0.0003,
      "step": 30780
    },
    {
      "epoch": 0.21242247166205577,
      "grad_norm": 0.0,
      "learning_rate": 7.877017116601242e-06,
      "loss": 0.0006,
      "step": 30790
    },
    {
      "epoch": 0.21249146239660013,
      "grad_norm": 3.0766545933147427e-06,
      "learning_rate": 7.876327209255798e-06,
      "loss": 0.0,
      "step": 30800
    },
    {
      "epoch": 0.2125604531311445,
      "grad_norm": 0.0,
      "learning_rate": 7.875637301910353e-06,
      "loss": 0.0001,
      "step": 30810
    },
    {
      "epoch": 0.21262944386568883,
      "grad_norm": 3.128637672489276e-06,
      "learning_rate": 7.87494739456491e-06,
      "loss": 0.0,
      "step": 30820
    },
    {
      "epoch": 0.2126984346002332,
      "grad_norm": 0.5101665258407593,
      "learning_rate": 7.874257487219466e-06,
      "loss": 0.0001,
      "step": 30830
    },
    {
      "epoch": 0.21276742533477755,
      "grad_norm": 0.0,
      "learning_rate": 7.873567579874024e-06,
      "loss": 0.0188,
      "step": 30840
    },
    {
      "epoch": 0.21283641606932188,
      "grad_norm": 0.0,
      "learning_rate": 7.87287767252858e-06,
      "loss": 0.0007,
      "step": 30850
    },
    {
      "epoch": 0.21290540680386624,
      "grad_norm": 0.0,
      "learning_rate": 7.872187765183137e-06,
      "loss": 0.0,
      "step": 30860
    },
    {
      "epoch": 0.2129743975384106,
      "grad_norm": 0.00019063099171034992,
      "learning_rate": 7.871497857837694e-06,
      "loss": 0.0,
      "step": 30870
    },
    {
      "epoch": 0.21304338827295494,
      "grad_norm": 0.0,
      "learning_rate": 7.87080795049225e-06,
      "loss": 0.0,
      "step": 30880
    },
    {
      "epoch": 0.2131123790074993,
      "grad_norm": 0.0,
      "learning_rate": 7.870118043146807e-06,
      "loss": 0.0005,
      "step": 30890
    },
    {
      "epoch": 0.21318136974204363,
      "grad_norm": 0.0,
      "learning_rate": 7.869428135801363e-06,
      "loss": 0.0,
      "step": 30900
    },
    {
      "epoch": 0.213250360476588,
      "grad_norm": 0.0,
      "learning_rate": 7.86873822845592e-06,
      "loss": 0.0,
      "step": 30910
    },
    {
      "epoch": 0.21331935121113235,
      "grad_norm": 0.0,
      "learning_rate": 7.868048321110474e-06,
      "loss": 0.0,
      "step": 30920
    },
    {
      "epoch": 0.21338834194567669,
      "grad_norm": 0.0,
      "learning_rate": 7.867358413765031e-06,
      "loss": 0.0,
      "step": 30930
    },
    {
      "epoch": 0.21345733268022105,
      "grad_norm": 0.0,
      "learning_rate": 7.866668506419587e-06,
      "loss": 0.1472,
      "step": 30940
    },
    {
      "epoch": 0.2135263234147654,
      "grad_norm": 0.0,
      "learning_rate": 7.865978599074146e-06,
      "loss": 0.0027,
      "step": 30950
    },
    {
      "epoch": 0.21359531414930974,
      "grad_norm": 0.0,
      "learning_rate": 7.865288691728702e-06,
      "loss": 0.0,
      "step": 30960
    },
    {
      "epoch": 0.2136643048838541,
      "grad_norm": 0.0,
      "learning_rate": 7.864598784383259e-06,
      "loss": 0.0,
      "step": 30970
    },
    {
      "epoch": 0.21373329561839846,
      "grad_norm": 0.0,
      "learning_rate": 7.863908877037815e-06,
      "loss": 0.0,
      "step": 30980
    },
    {
      "epoch": 0.2138022863529428,
      "grad_norm": 0.0,
      "learning_rate": 7.863218969692372e-06,
      "loss": 0.0,
      "step": 30990
    },
    {
      "epoch": 0.21387127708748715,
      "grad_norm": 0.0,
      "learning_rate": 7.862529062346928e-06,
      "loss": 0.0633,
      "step": 31000
    },
    {
      "epoch": 0.2139402678220315,
      "grad_norm": 0.0,
      "learning_rate": 7.861839155001485e-06,
      "loss": 0.0,
      "step": 31010
    },
    {
      "epoch": 0.21400925855657585,
      "grad_norm": 0.0,
      "learning_rate": 7.861149247656041e-06,
      "loss": 0.0,
      "step": 31020
    },
    {
      "epoch": 0.2140782492911202,
      "grad_norm": 0.0,
      "learning_rate": 7.860459340310596e-06,
      "loss": 0.0003,
      "step": 31030
    },
    {
      "epoch": 0.21414724002566454,
      "grad_norm": 0.0,
      "learning_rate": 7.859769432965152e-06,
      "loss": 0.0,
      "step": 31040
    },
    {
      "epoch": 0.2142162307602089,
      "grad_norm": 4.306132177589461e-05,
      "learning_rate": 7.859079525619709e-06,
      "loss": 0.0,
      "step": 31050
    },
    {
      "epoch": 0.21428522149475326,
      "grad_norm": 0.0,
      "learning_rate": 7.858389618274267e-06,
      "loss": 0.0005,
      "step": 31060
    },
    {
      "epoch": 0.2143542122292976,
      "grad_norm": 0.0,
      "learning_rate": 7.857699710928823e-06,
      "loss": 0.0,
      "step": 31070
    },
    {
      "epoch": 0.21442320296384196,
      "grad_norm": 1.4645641385868657e-05,
      "learning_rate": 7.85700980358338e-06,
      "loss": 2.0531,
      "step": 31080
    },
    {
      "epoch": 0.21449219369838632,
      "grad_norm": 0.0,
      "learning_rate": 7.856319896237936e-06,
      "loss": 0.0,
      "step": 31090
    },
    {
      "epoch": 0.21456118443293065,
      "grad_norm": 0.00010842143092304468,
      "learning_rate": 7.855629988892493e-06,
      "loss": 0.0,
      "step": 31100
    },
    {
      "epoch": 0.214630175167475,
      "grad_norm": 0.0,
      "learning_rate": 7.85494008154705e-06,
      "loss": 0.0,
      "step": 31110
    },
    {
      "epoch": 0.21469916590201935,
      "grad_norm": 0.0,
      "learning_rate": 7.854250174201606e-06,
      "loss": 0.0001,
      "step": 31120
    },
    {
      "epoch": 0.2147681566365637,
      "grad_norm": 0.00031385431066155434,
      "learning_rate": 7.853560266856162e-06,
      "loss": 0.0,
      "step": 31130
    },
    {
      "epoch": 0.21483714737110807,
      "grad_norm": 0.0,
      "learning_rate": 7.852870359510719e-06,
      "loss": 0.0,
      "step": 31140
    },
    {
      "epoch": 0.2149061381056524,
      "grad_norm": 0.0,
      "learning_rate": 7.852180452165274e-06,
      "loss": 0.0,
      "step": 31150
    },
    {
      "epoch": 0.21497512884019676,
      "grad_norm": 0.0,
      "learning_rate": 7.85149054481983e-06,
      "loss": 0.0,
      "step": 31160
    },
    {
      "epoch": 0.21504411957474112,
      "grad_norm": 1.1168711353093386e-05,
      "learning_rate": 7.850800637474388e-06,
      "loss": 0.0,
      "step": 31170
    },
    {
      "epoch": 0.21511311030928545,
      "grad_norm": 0.0,
      "learning_rate": 7.850110730128945e-06,
      "loss": 0.0,
      "step": 31180
    },
    {
      "epoch": 0.21518210104382982,
      "grad_norm": 0.0,
      "learning_rate": 7.849420822783501e-06,
      "loss": 0.0,
      "step": 31190
    },
    {
      "epoch": 0.21525109177837418,
      "grad_norm": 0.0,
      "learning_rate": 7.848730915438058e-06,
      "loss": 0.0,
      "step": 31200
    },
    {
      "epoch": 0.2153200825129185,
      "grad_norm": 0.0,
      "learning_rate": 7.848041008092614e-06,
      "loss": 0.0,
      "step": 31210
    },
    {
      "epoch": 0.21538907324746287,
      "grad_norm": 0.0,
      "learning_rate": 7.84735110074717e-06,
      "loss": 0.0,
      "step": 31220
    },
    {
      "epoch": 0.2154580639820072,
      "grad_norm": 4.262914021779807e-09,
      "learning_rate": 7.846661193401727e-06,
      "loss": 0.2951,
      "step": 31230
    },
    {
      "epoch": 0.21552705471655156,
      "grad_norm": 0.0,
      "learning_rate": 7.845971286056284e-06,
      "loss": 0.0648,
      "step": 31240
    },
    {
      "epoch": 0.21559604545109592,
      "grad_norm": 0.0,
      "learning_rate": 7.84528137871084e-06,
      "loss": 0.0,
      "step": 31250
    },
    {
      "epoch": 0.21566503618564026,
      "grad_norm": 0.0,
      "learning_rate": 7.844591471365395e-06,
      "loss": 0.0,
      "step": 31260
    },
    {
      "epoch": 0.21573402692018462,
      "grad_norm": 353.7784423828125,
      "learning_rate": 7.843901564019952e-06,
      "loss": 0.0671,
      "step": 31270
    },
    {
      "epoch": 0.21580301765472898,
      "grad_norm": 2.469707993668635e-08,
      "learning_rate": 7.84321165667451e-06,
      "loss": 0.0,
      "step": 31280
    },
    {
      "epoch": 0.2158720083892733,
      "grad_norm": 0.0,
      "learning_rate": 7.842521749329066e-06,
      "loss": 0.0,
      "step": 31290
    },
    {
      "epoch": 0.21594099912381767,
      "grad_norm": 113.75282287597656,
      "learning_rate": 7.841831841983623e-06,
      "loss": 0.0248,
      "step": 31300
    },
    {
      "epoch": 0.21600998985836203,
      "grad_norm": 1.564249032526277e-05,
      "learning_rate": 7.84114193463818e-06,
      "loss": 0.0,
      "step": 31310
    },
    {
      "epoch": 0.21607898059290637,
      "grad_norm": 2.0569549974425172e-07,
      "learning_rate": 7.840452027292736e-06,
      "loss": 0.0007,
      "step": 31320
    },
    {
      "epoch": 0.21614797132745073,
      "grad_norm": 0.0845765545964241,
      "learning_rate": 7.839762119947292e-06,
      "loss": 0.0,
      "step": 31330
    },
    {
      "epoch": 0.21621696206199506,
      "grad_norm": 0.00021235295571386814,
      "learning_rate": 7.839072212601849e-06,
      "loss": 0.0,
      "step": 31340
    },
    {
      "epoch": 0.21628595279653942,
      "grad_norm": 0.0,
      "learning_rate": 7.838382305256405e-06,
      "loss": 0.0,
      "step": 31350
    },
    {
      "epoch": 0.21635494353108378,
      "grad_norm": 0.0,
      "learning_rate": 7.837692397910962e-06,
      "loss": 0.0,
      "step": 31360
    },
    {
      "epoch": 0.21642393426562812,
      "grad_norm": 0.0,
      "learning_rate": 7.837002490565516e-06,
      "loss": 0.0002,
      "step": 31370
    },
    {
      "epoch": 0.21649292500017248,
      "grad_norm": 0.0,
      "learning_rate": 7.836312583220073e-06,
      "loss": 0.0,
      "step": 31380
    },
    {
      "epoch": 0.21656191573471684,
      "grad_norm": 1.364437821393949e-06,
      "learning_rate": 7.835622675874631e-06,
      "loss": 0.0001,
      "step": 31390
    },
    {
      "epoch": 0.21663090646926117,
      "grad_norm": 0.0,
      "learning_rate": 7.834932768529188e-06,
      "loss": 0.0,
      "step": 31400
    },
    {
      "epoch": 0.21669989720380553,
      "grad_norm": 2.233626219094731e-06,
      "learning_rate": 7.834242861183744e-06,
      "loss": 0.0,
      "step": 31410
    },
    {
      "epoch": 0.2167688879383499,
      "grad_norm": 0.0,
      "learning_rate": 7.8335529538383e-06,
      "loss": 0.0,
      "step": 31420
    },
    {
      "epoch": 0.21683787867289422,
      "grad_norm": 0.3814917206764221,
      "learning_rate": 7.832863046492857e-06,
      "loss": 0.0002,
      "step": 31430
    },
    {
      "epoch": 0.21690686940743859,
      "grad_norm": 0.0,
      "learning_rate": 7.832173139147414e-06,
      "loss": 0.0,
      "step": 31440
    },
    {
      "epoch": 0.21697586014198292,
      "grad_norm": 0.0,
      "learning_rate": 7.83148323180197e-06,
      "loss": 0.0,
      "step": 31450
    },
    {
      "epoch": 0.21704485087652728,
      "grad_norm": 0.0,
      "learning_rate": 7.830793324456527e-06,
      "loss": 0.0,
      "step": 31460
    },
    {
      "epoch": 0.21711384161107164,
      "grad_norm": 0.0,
      "learning_rate": 7.830103417111083e-06,
      "loss": 0.0,
      "step": 31470
    },
    {
      "epoch": 0.21718283234561597,
      "grad_norm": 3.546709592683328e-09,
      "learning_rate": 7.82941350976564e-06,
      "loss": 0.0,
      "step": 31480
    },
    {
      "epoch": 0.21725182308016033,
      "grad_norm": 0.0007365705096162856,
      "learning_rate": 7.828723602420194e-06,
      "loss": 0.012,
      "step": 31490
    },
    {
      "epoch": 0.2173208138147047,
      "grad_norm": 8.523781125369112e-10,
      "learning_rate": 7.828033695074753e-06,
      "loss": 0.0,
      "step": 31500
    },
    {
      "epoch": 0.21738980454924903,
      "grad_norm": 0.0,
      "learning_rate": 7.827343787729309e-06,
      "loss": 0.293,
      "step": 31510
    },
    {
      "epoch": 0.2174587952837934,
      "grad_norm": 2.8272589247535507e-07,
      "learning_rate": 7.826653880383866e-06,
      "loss": 0.0001,
      "step": 31520
    },
    {
      "epoch": 0.21752778601833775,
      "grad_norm": 0.0,
      "learning_rate": 7.825963973038422e-06,
      "loss": 0.0007,
      "step": 31530
    },
    {
      "epoch": 0.21759677675288208,
      "grad_norm": 0.0,
      "learning_rate": 7.825274065692979e-06,
      "loss": 0.0,
      "step": 31540
    },
    {
      "epoch": 0.21766576748742644,
      "grad_norm": 3.286144760750176e-07,
      "learning_rate": 7.824584158347535e-06,
      "loss": 0.0025,
      "step": 31550
    },
    {
      "epoch": 0.21773475822197078,
      "grad_norm": 1.3250197525849217e-06,
      "learning_rate": 7.823894251002092e-06,
      "loss": 0.0001,
      "step": 31560
    },
    {
      "epoch": 0.21780374895651514,
      "grad_norm": 0.0,
      "learning_rate": 7.823204343656648e-06,
      "loss": 0.0113,
      "step": 31570
    },
    {
      "epoch": 0.2178727396910595,
      "grad_norm": 0.0,
      "learning_rate": 7.822514436311204e-06,
      "loss": 0.0,
      "step": 31580
    },
    {
      "epoch": 0.21794173042560383,
      "grad_norm": 0.0,
      "learning_rate": 7.821824528965761e-06,
      "loss": 0.0,
      "step": 31590
    },
    {
      "epoch": 0.2180107211601482,
      "grad_norm": 0.0,
      "learning_rate": 7.821134621620316e-06,
      "loss": 0.0,
      "step": 31600
    },
    {
      "epoch": 0.21807971189469255,
      "grad_norm": 0.0,
      "learning_rate": 7.820444714274874e-06,
      "loss": 0.0693,
      "step": 31610
    },
    {
      "epoch": 0.21814870262923688,
      "grad_norm": 3.377792381797917e-05,
      "learning_rate": 7.81975480692943e-06,
      "loss": 0.0,
      "step": 31620
    },
    {
      "epoch": 0.21821769336378125,
      "grad_norm": 0.0,
      "learning_rate": 7.819064899583987e-06,
      "loss": 0.0,
      "step": 31630
    },
    {
      "epoch": 0.2182866840983256,
      "grad_norm": 0.00022295626695267856,
      "learning_rate": 7.818374992238543e-06,
      "loss": 0.0,
      "step": 31640
    },
    {
      "epoch": 0.21835567483286994,
      "grad_norm": 0.12362496554851532,
      "learning_rate": 7.8176850848931e-06,
      "loss": 0.0,
      "step": 31650
    },
    {
      "epoch": 0.2184246655674143,
      "grad_norm": 0.0,
      "learning_rate": 7.816995177547656e-06,
      "loss": 0.0,
      "step": 31660
    },
    {
      "epoch": 0.21849365630195863,
      "grad_norm": 0.0,
      "learning_rate": 7.816305270202213e-06,
      "loss": 0.0,
      "step": 31670
    },
    {
      "epoch": 0.218562647036503,
      "grad_norm": 0.0,
      "learning_rate": 7.81561536285677e-06,
      "loss": 0.0,
      "step": 31680
    },
    {
      "epoch": 0.21863163777104735,
      "grad_norm": 0.00035919054062105715,
      "learning_rate": 7.814925455511326e-06,
      "loss": 0.07,
      "step": 31690
    },
    {
      "epoch": 0.2187006285055917,
      "grad_norm": 0.0,
      "learning_rate": 7.814235548165882e-06,
      "loss": 0.0,
      "step": 31700
    },
    {
      "epoch": 0.21876961924013605,
      "grad_norm": 8.322115172632039e-05,
      "learning_rate": 7.813545640820437e-06,
      "loss": 0.0,
      "step": 31710
    },
    {
      "epoch": 0.2188386099746804,
      "grad_norm": 0.0,
      "learning_rate": 7.812855733474995e-06,
      "loss": 0.0,
      "step": 31720
    },
    {
      "epoch": 0.21890760070922474,
      "grad_norm": 0.0,
      "learning_rate": 7.812165826129552e-06,
      "loss": 0.0,
      "step": 31730
    },
    {
      "epoch": 0.2189765914437691,
      "grad_norm": 0.0,
      "learning_rate": 7.811475918784108e-06,
      "loss": 0.0,
      "step": 31740
    },
    {
      "epoch": 0.21904558217831346,
      "grad_norm": 5.785098892374663e-06,
      "learning_rate": 7.810786011438665e-06,
      "loss": 0.0028,
      "step": 31750
    },
    {
      "epoch": 0.2191145729128578,
      "grad_norm": 0.0,
      "learning_rate": 7.810096104093221e-06,
      "loss": 0.0,
      "step": 31760
    },
    {
      "epoch": 0.21918356364740216,
      "grad_norm": 0.0,
      "learning_rate": 7.809406196747778e-06,
      "loss": 0.0,
      "step": 31770
    },
    {
      "epoch": 0.21925255438194652,
      "grad_norm": 0.0,
      "learning_rate": 7.808716289402334e-06,
      "loss": 0.0,
      "step": 31780
    },
    {
      "epoch": 0.21932154511649085,
      "grad_norm": 0.0,
      "learning_rate": 7.80802638205689e-06,
      "loss": 0.0,
      "step": 31790
    },
    {
      "epoch": 0.2193905358510352,
      "grad_norm": 0.0017589419148862362,
      "learning_rate": 7.807336474711447e-06,
      "loss": 0.0,
      "step": 31800
    },
    {
      "epoch": 0.21945952658557955,
      "grad_norm": 0.0,
      "learning_rate": 7.806646567366004e-06,
      "loss": 0.0,
      "step": 31810
    },
    {
      "epoch": 0.2195285173201239,
      "grad_norm": 8.556105268731073e-10,
      "learning_rate": 7.80595666002056e-06,
      "loss": 0.0001,
      "step": 31820
    },
    {
      "epoch": 0.21959750805466827,
      "grad_norm": 0.0034977784380316734,
      "learning_rate": 7.805266752675117e-06,
      "loss": 0.0,
      "step": 31830
    },
    {
      "epoch": 0.2196664987892126,
      "grad_norm": 0.0,
      "learning_rate": 7.804576845329673e-06,
      "loss": 0.0,
      "step": 31840
    },
    {
      "epoch": 0.21973548952375696,
      "grad_norm": 0.0001161557956947945,
      "learning_rate": 7.80388693798423e-06,
      "loss": 0.0,
      "step": 31850
    },
    {
      "epoch": 0.21980448025830132,
      "grad_norm": 0.0,
      "learning_rate": 7.803197030638786e-06,
      "loss": 0.0003,
      "step": 31860
    },
    {
      "epoch": 0.21987347099284565,
      "grad_norm": 0.0,
      "learning_rate": 7.802507123293343e-06,
      "loss": 0.0,
      "step": 31870
    },
    {
      "epoch": 0.21994246172739002,
      "grad_norm": 0.0,
      "learning_rate": 7.8018172159479e-06,
      "loss": 0.0001,
      "step": 31880
    },
    {
      "epoch": 0.22001145246193438,
      "grad_norm": 0.07350252568721771,
      "learning_rate": 7.801127308602456e-06,
      "loss": 0.0161,
      "step": 31890
    },
    {
      "epoch": 0.2200804431964787,
      "grad_norm": 9.503233755481233e-10,
      "learning_rate": 7.800437401257012e-06,
      "loss": 0.0028,
      "step": 31900
    },
    {
      "epoch": 0.22014943393102307,
      "grad_norm": 4.8008246267272625e-06,
      "learning_rate": 7.799747493911569e-06,
      "loss": 0.0008,
      "step": 31910
    },
    {
      "epoch": 0.2202184246655674,
      "grad_norm": 3.972463247237101e-09,
      "learning_rate": 7.799057586566125e-06,
      "loss": 0.0,
      "step": 31920
    },
    {
      "epoch": 0.22028741540011176,
      "grad_norm": 0.0,
      "learning_rate": 7.798367679220682e-06,
      "loss": 0.0,
      "step": 31930
    },
    {
      "epoch": 0.22035640613465612,
      "grad_norm": 0.0,
      "learning_rate": 7.797677771875238e-06,
      "loss": 0.0,
      "step": 31940
    },
    {
      "epoch": 0.22042539686920046,
      "grad_norm": 0.0,
      "learning_rate": 7.796987864529795e-06,
      "loss": 0.0,
      "step": 31950
    },
    {
      "epoch": 0.22049438760374482,
      "grad_norm": 0.0,
      "learning_rate": 7.796297957184351e-06,
      "loss": 0.0,
      "step": 31960
    },
    {
      "epoch": 0.22056337833828918,
      "grad_norm": 0.00018339813686907291,
      "learning_rate": 7.795608049838908e-06,
      "loss": 0.0,
      "step": 31970
    },
    {
      "epoch": 0.2206323690728335,
      "grad_norm": 1.0926111571052388e-07,
      "learning_rate": 7.794918142493464e-06,
      "loss": 0.0,
      "step": 31980
    },
    {
      "epoch": 0.22070135980737787,
      "grad_norm": 0.0,
      "learning_rate": 7.79422823514802e-06,
      "loss": 0.0,
      "step": 31990
    },
    {
      "epoch": 0.22077035054192223,
      "grad_norm": 8.809608698356897e-05,
      "learning_rate": 7.793538327802577e-06,
      "loss": 0.0,
      "step": 32000
    },
    {
      "epoch": 0.22083934127646657,
      "grad_norm": 3.545181357367255e-07,
      "learning_rate": 7.792848420457134e-06,
      "loss": 0.0,
      "step": 32010
    },
    {
      "epoch": 0.22090833201101093,
      "grad_norm": 0.0,
      "learning_rate": 7.79215851311169e-06,
      "loss": 0.0017,
      "step": 32020
    },
    {
      "epoch": 0.22097732274555526,
      "grad_norm": 0.09344573318958282,
      "learning_rate": 7.791468605766247e-06,
      "loss": 0.0,
      "step": 32030
    },
    {
      "epoch": 0.22104631348009962,
      "grad_norm": 0.0,
      "learning_rate": 7.790778698420803e-06,
      "loss": 0.0,
      "step": 32040
    },
    {
      "epoch": 0.22111530421464398,
      "grad_norm": 0.0,
      "learning_rate": 7.79008879107536e-06,
      "loss": 0.0,
      "step": 32050
    },
    {
      "epoch": 0.22118429494918831,
      "grad_norm": 0.0,
      "learning_rate": 7.789398883729916e-06,
      "loss": 0.0,
      "step": 32060
    },
    {
      "epoch": 0.22125328568373268,
      "grad_norm": 6.706374645233154,
      "learning_rate": 7.788708976384473e-06,
      "loss": 0.0011,
      "step": 32070
    },
    {
      "epoch": 0.22132227641827704,
      "grad_norm": 0.0,
      "learning_rate": 7.788019069039029e-06,
      "loss": 0.0,
      "step": 32080
    },
    {
      "epoch": 0.22139126715282137,
      "grad_norm": 0.0,
      "learning_rate": 7.787329161693585e-06,
      "loss": 0.0,
      "step": 32090
    },
    {
      "epoch": 0.22146025788736573,
      "grad_norm": 0.0,
      "learning_rate": 7.786639254348142e-06,
      "loss": 0.0,
      "step": 32100
    },
    {
      "epoch": 0.2215292486219101,
      "grad_norm": 0.0,
      "learning_rate": 7.785949347002698e-06,
      "loss": 0.0,
      "step": 32110
    },
    {
      "epoch": 0.22159823935645442,
      "grad_norm": 0.0,
      "learning_rate": 7.785259439657255e-06,
      "loss": 0.0001,
      "step": 32120
    },
    {
      "epoch": 0.22166723009099878,
      "grad_norm": 0.0,
      "learning_rate": 7.784569532311811e-06,
      "loss": 0.0013,
      "step": 32130
    },
    {
      "epoch": 0.22173622082554312,
      "grad_norm": 0.0,
      "learning_rate": 7.783879624966368e-06,
      "loss": 0.0,
      "step": 32140
    },
    {
      "epoch": 0.22180521156008748,
      "grad_norm": 0.0,
      "learning_rate": 7.783189717620924e-06,
      "loss": 0.0,
      "step": 32150
    },
    {
      "epoch": 0.22187420229463184,
      "grad_norm": 0.0,
      "learning_rate": 7.782499810275481e-06,
      "loss": 0.0389,
      "step": 32160
    },
    {
      "epoch": 0.22194319302917617,
      "grad_norm": 0.0,
      "learning_rate": 7.781809902930037e-06,
      "loss": 0.0,
      "step": 32170
    },
    {
      "epoch": 0.22201218376372053,
      "grad_norm": 0.0,
      "learning_rate": 7.781119995584594e-06,
      "loss": 0.0,
      "step": 32180
    },
    {
      "epoch": 0.2220811744982649,
      "grad_norm": 0.0,
      "learning_rate": 7.78043008823915e-06,
      "loss": 0.0,
      "step": 32190
    },
    {
      "epoch": 0.22215016523280923,
      "grad_norm": 0.0,
      "learning_rate": 7.779740180893707e-06,
      "loss": 0.0,
      "step": 32200
    },
    {
      "epoch": 0.2222191559673536,
      "grad_norm": 0.0,
      "learning_rate": 7.779050273548263e-06,
      "loss": 0.0008,
      "step": 32210
    },
    {
      "epoch": 0.22228814670189795,
      "grad_norm": 0.0,
      "learning_rate": 7.77836036620282e-06,
      "loss": 0.0,
      "step": 32220
    },
    {
      "epoch": 0.22235713743644228,
      "grad_norm": 0.0,
      "learning_rate": 7.777670458857376e-06,
      "loss": 0.0002,
      "step": 32230
    },
    {
      "epoch": 0.22242612817098664,
      "grad_norm": 0.0,
      "learning_rate": 7.776980551511933e-06,
      "loss": 0.0,
      "step": 32240
    },
    {
      "epoch": 0.22249511890553098,
      "grad_norm": 0.0,
      "learning_rate": 7.77629064416649e-06,
      "loss": 0.0,
      "step": 32250
    },
    {
      "epoch": 0.22256410964007534,
      "grad_norm": 0.0,
      "learning_rate": 7.775600736821046e-06,
      "loss": 0.0003,
      "step": 32260
    },
    {
      "epoch": 0.2226331003746197,
      "grad_norm": 0.00026333125424571335,
      "learning_rate": 7.774910829475602e-06,
      "loss": 0.0,
      "step": 32270
    },
    {
      "epoch": 0.22270209110916403,
      "grad_norm": 0.0,
      "learning_rate": 7.774220922130159e-06,
      "loss": 0.0,
      "step": 32280
    },
    {
      "epoch": 0.2227710818437084,
      "grad_norm": 0.0,
      "learning_rate": 7.773531014784715e-06,
      "loss": 0.0123,
      "step": 32290
    },
    {
      "epoch": 0.22284007257825275,
      "grad_norm": 1.7667003703536466e-05,
      "learning_rate": 7.772841107439272e-06,
      "loss": 0.0,
      "step": 32300
    },
    {
      "epoch": 0.22290906331279708,
      "grad_norm": 591.3378295898438,
      "learning_rate": 7.772151200093828e-06,
      "loss": 0.1389,
      "step": 32310
    },
    {
      "epoch": 0.22297805404734145,
      "grad_norm": 0.0,
      "learning_rate": 7.771461292748385e-06,
      "loss": 0.0746,
      "step": 32320
    },
    {
      "epoch": 0.2230470447818858,
      "grad_norm": 0.0,
      "learning_rate": 7.770771385402941e-06,
      "loss": 0.0317,
      "step": 32330
    },
    {
      "epoch": 0.22311603551643014,
      "grad_norm": 0.0005945048178546131,
      "learning_rate": 7.770081478057498e-06,
      "loss": 0.0,
      "step": 32340
    },
    {
      "epoch": 0.2231850262509745,
      "grad_norm": 0.0,
      "learning_rate": 7.769391570712054e-06,
      "loss": 0.0,
      "step": 32350
    },
    {
      "epoch": 0.22325401698551883,
      "grad_norm": 0.02066963165998459,
      "learning_rate": 7.76870166336661e-06,
      "loss": 0.0,
      "step": 32360
    },
    {
      "epoch": 0.2233230077200632,
      "grad_norm": 0.0,
      "learning_rate": 7.768011756021167e-06,
      "loss": 0.0006,
      "step": 32370
    },
    {
      "epoch": 0.22339199845460755,
      "grad_norm": 0.0,
      "learning_rate": 7.767321848675724e-06,
      "loss": 0.0001,
      "step": 32380
    },
    {
      "epoch": 0.2234609891891519,
      "grad_norm": 0.0,
      "learning_rate": 7.76663194133028e-06,
      "loss": 0.001,
      "step": 32390
    },
    {
      "epoch": 0.22352997992369625,
      "grad_norm": 0.7026162147521973,
      "learning_rate": 7.765942033984837e-06,
      "loss": 0.0011,
      "step": 32400
    },
    {
      "epoch": 0.2235989706582406,
      "grad_norm": 0.0,
      "learning_rate": 7.765252126639393e-06,
      "loss": 0.0,
      "step": 32410
    },
    {
      "epoch": 0.22366796139278494,
      "grad_norm": 0.0,
      "learning_rate": 7.76456221929395e-06,
      "loss": 0.0,
      "step": 32420
    },
    {
      "epoch": 0.2237369521273293,
      "grad_norm": 1.3782530913886148e-05,
      "learning_rate": 7.763872311948506e-06,
      "loss": 0.0,
      "step": 32430
    },
    {
      "epoch": 0.22380594286187366,
      "grad_norm": 0.0,
      "learning_rate": 7.763182404603063e-06,
      "loss": 0.0001,
      "step": 32440
    },
    {
      "epoch": 0.223874933596418,
      "grad_norm": 0.0,
      "learning_rate": 7.762492497257619e-06,
      "loss": 0.004,
      "step": 32450
    },
    {
      "epoch": 0.22394392433096236,
      "grad_norm": 7.390298378595617e-06,
      "learning_rate": 7.761802589912176e-06,
      "loss": 0.0,
      "step": 32460
    },
    {
      "epoch": 0.2240129150655067,
      "grad_norm": 2.8584359768046852e-08,
      "learning_rate": 7.761112682566732e-06,
      "loss": 0.0,
      "step": 32470
    },
    {
      "epoch": 0.22408190580005105,
      "grad_norm": 0.0,
      "learning_rate": 7.760422775221289e-06,
      "loss": 0.0,
      "step": 32480
    },
    {
      "epoch": 0.2241508965345954,
      "grad_norm": 0.0,
      "learning_rate": 7.759732867875845e-06,
      "loss": 0.0,
      "step": 32490
    },
    {
      "epoch": 0.22421988726913974,
      "grad_norm": 0.0,
      "learning_rate": 7.759042960530402e-06,
      "loss": 0.0001,
      "step": 32500
    },
    {
      "epoch": 0.2242888780036841,
      "grad_norm": 0.0,
      "learning_rate": 7.758353053184958e-06,
      "loss": 0.0,
      "step": 32510
    },
    {
      "epoch": 0.22435786873822847,
      "grad_norm": 0.0,
      "learning_rate": 7.757663145839515e-06,
      "loss": 0.0,
      "step": 32520
    },
    {
      "epoch": 0.2244268594727728,
      "grad_norm": 0.0,
      "learning_rate": 7.756973238494071e-06,
      "loss": 0.0,
      "step": 32530
    },
    {
      "epoch": 0.22449585020731716,
      "grad_norm": 0.0,
      "learning_rate": 7.756283331148628e-06,
      "loss": 0.0,
      "step": 32540
    },
    {
      "epoch": 0.22456484094186152,
      "grad_norm": 0.0,
      "learning_rate": 7.755593423803184e-06,
      "loss": 0.0,
      "step": 32550
    },
    {
      "epoch": 0.22463383167640585,
      "grad_norm": 0.0,
      "learning_rate": 7.75490351645774e-06,
      "loss": 0.0003,
      "step": 32560
    },
    {
      "epoch": 0.22470282241095021,
      "grad_norm": 0.0,
      "learning_rate": 7.754213609112297e-06,
      "loss": 0.0,
      "step": 32570
    },
    {
      "epoch": 0.22477181314549455,
      "grad_norm": 0.0,
      "learning_rate": 7.753523701766854e-06,
      "loss": 0.0,
      "step": 32580
    },
    {
      "epoch": 0.2248408038800389,
      "grad_norm": 0.0,
      "learning_rate": 7.75283379442141e-06,
      "loss": 0.0,
      "step": 32590
    },
    {
      "epoch": 0.22490979461458327,
      "grad_norm": 0.0,
      "learning_rate": 7.752143887075966e-06,
      "loss": 0.0,
      "step": 32600
    },
    {
      "epoch": 0.2249787853491276,
      "grad_norm": 0.0009406035533174872,
      "learning_rate": 7.751453979730523e-06,
      "loss": 0.1137,
      "step": 32610
    },
    {
      "epoch": 0.22504777608367196,
      "grad_norm": 0.0,
      "learning_rate": 7.75076407238508e-06,
      "loss": 0.0,
      "step": 32620
    },
    {
      "epoch": 0.22511676681821632,
      "grad_norm": 0.0,
      "learning_rate": 7.750074165039636e-06,
      "loss": 0.0,
      "step": 32630
    },
    {
      "epoch": 0.22518575755276066,
      "grad_norm": 0.4979203939437866,
      "learning_rate": 7.749384257694192e-06,
      "loss": 0.0001,
      "step": 32640
    },
    {
      "epoch": 0.22525474828730502,
      "grad_norm": 0.0,
      "learning_rate": 7.748694350348749e-06,
      "loss": 0.0,
      "step": 32650
    },
    {
      "epoch": 0.22532373902184938,
      "grad_norm": 0.0,
      "learning_rate": 7.748004443003305e-06,
      "loss": 0.0,
      "step": 32660
    },
    {
      "epoch": 0.2253927297563937,
      "grad_norm": 0.00028845269116573036,
      "learning_rate": 7.747314535657862e-06,
      "loss": 0.0,
      "step": 32670
    },
    {
      "epoch": 0.22546172049093807,
      "grad_norm": 0.0,
      "learning_rate": 7.746624628312418e-06,
      "loss": 0.0001,
      "step": 32680
    },
    {
      "epoch": 0.2255307112254824,
      "grad_norm": 0.0,
      "learning_rate": 7.745934720966975e-06,
      "loss": 0.0,
      "step": 32690
    },
    {
      "epoch": 0.22559970196002677,
      "grad_norm": 0.0,
      "learning_rate": 7.745244813621531e-06,
      "loss": 0.0,
      "step": 32700
    },
    {
      "epoch": 0.22566869269457113,
      "grad_norm": 2.1430544450140587e-07,
      "learning_rate": 7.744554906276088e-06,
      "loss": 0.0,
      "step": 32710
    },
    {
      "epoch": 0.22573768342911546,
      "grad_norm": 0.0,
      "learning_rate": 7.743864998930644e-06,
      "loss": 0.0,
      "step": 32720
    },
    {
      "epoch": 0.22580667416365982,
      "grad_norm": 0.0,
      "learning_rate": 7.743175091585201e-06,
      "loss": 0.0861,
      "step": 32730
    },
    {
      "epoch": 0.22587566489820418,
      "grad_norm": 0.0,
      "learning_rate": 7.742485184239757e-06,
      "loss": 0.0,
      "step": 32740
    },
    {
      "epoch": 0.22594465563274851,
      "grad_norm": 0.0,
      "learning_rate": 7.741795276894314e-06,
      "loss": 0.0,
      "step": 32750
    },
    {
      "epoch": 0.22601364636729288,
      "grad_norm": 1.7973080801425567e-08,
      "learning_rate": 7.74110536954887e-06,
      "loss": 0.088,
      "step": 32760
    },
    {
      "epoch": 0.22608263710183724,
      "grad_norm": 0.0,
      "learning_rate": 7.740415462203427e-06,
      "loss": 0.0337,
      "step": 32770
    },
    {
      "epoch": 0.22615162783638157,
      "grad_norm": 6.393340612476095e-08,
      "learning_rate": 7.739725554857983e-06,
      "loss": 0.0,
      "step": 32780
    },
    {
      "epoch": 0.22622061857092593,
      "grad_norm": 0.0,
      "learning_rate": 7.73903564751254e-06,
      "loss": 0.0,
      "step": 32790
    },
    {
      "epoch": 0.22628960930547026,
      "grad_norm": 0.0,
      "learning_rate": 7.738345740167096e-06,
      "loss": 0.0136,
      "step": 32800
    },
    {
      "epoch": 0.22635860004001462,
      "grad_norm": 0.014421519823372364,
      "learning_rate": 7.737655832821653e-06,
      "loss": 0.0,
      "step": 32810
    },
    {
      "epoch": 0.22642759077455898,
      "grad_norm": 0.0,
      "learning_rate": 7.73696592547621e-06,
      "loss": 0.0,
      "step": 32820
    },
    {
      "epoch": 0.22649658150910332,
      "grad_norm": 0.0,
      "learning_rate": 7.736276018130766e-06,
      "loss": 0.0,
      "step": 32830
    },
    {
      "epoch": 0.22656557224364768,
      "grad_norm": 0.0037522935308516026,
      "learning_rate": 7.735586110785322e-06,
      "loss": 0.001,
      "step": 32840
    },
    {
      "epoch": 0.22663456297819204,
      "grad_norm": 0.0,
      "learning_rate": 7.734896203439879e-06,
      "loss": 0.0,
      "step": 32850
    },
    {
      "epoch": 0.22670355371273637,
      "grad_norm": 0.0,
      "learning_rate": 7.734206296094435e-06,
      "loss": 0.0,
      "step": 32860
    },
    {
      "epoch": 0.22677254444728073,
      "grad_norm": 0.00032297128927893937,
      "learning_rate": 7.733516388748992e-06,
      "loss": 0.0,
      "step": 32870
    },
    {
      "epoch": 0.2268415351818251,
      "grad_norm": 6.696091531921411e-06,
      "learning_rate": 7.732826481403548e-06,
      "loss": 0.0,
      "step": 32880
    },
    {
      "epoch": 0.22691052591636943,
      "grad_norm": 0.0,
      "learning_rate": 7.732136574058105e-06,
      "loss": 0.0,
      "step": 32890
    },
    {
      "epoch": 0.2269795166509138,
      "grad_norm": 0.0,
      "learning_rate": 7.731446666712661e-06,
      "loss": 0.0,
      "step": 32900
    },
    {
      "epoch": 0.22704850738545812,
      "grad_norm": 0.0030558868311345577,
      "learning_rate": 7.730756759367218e-06,
      "loss": 0.0118,
      "step": 32910
    },
    {
      "epoch": 0.22711749812000248,
      "grad_norm": 0.0,
      "learning_rate": 7.730066852021774e-06,
      "loss": 0.0,
      "step": 32920
    },
    {
      "epoch": 0.22718648885454684,
      "grad_norm": 0.0,
      "learning_rate": 7.72937694467633e-06,
      "loss": 0.0,
      "step": 32930
    },
    {
      "epoch": 0.22725547958909117,
      "grad_norm": 8.771455213718582e-06,
      "learning_rate": 7.728687037330887e-06,
      "loss": 0.0,
      "step": 32940
    },
    {
      "epoch": 0.22732447032363554,
      "grad_norm": 0.0024318480864167213,
      "learning_rate": 7.727997129985444e-06,
      "loss": 0.0,
      "step": 32950
    },
    {
      "epoch": 0.2273934610581799,
      "grad_norm": 0.0,
      "learning_rate": 7.72730722264e-06,
      "loss": 0.0,
      "step": 32960
    },
    {
      "epoch": 0.22746245179272423,
      "grad_norm": 0.0,
      "learning_rate": 7.726617315294557e-06,
      "loss": 0.0,
      "step": 32970
    },
    {
      "epoch": 0.2275314425272686,
      "grad_norm": 0.0,
      "learning_rate": 7.725927407949113e-06,
      "loss": 0.4219,
      "step": 32980
    },
    {
      "epoch": 0.22760043326181295,
      "grad_norm": 0.0,
      "learning_rate": 7.72523750060367e-06,
      "loss": 0.0001,
      "step": 32990
    },
    {
      "epoch": 0.22766942399635728,
      "grad_norm": 0.0,
      "learning_rate": 7.724547593258226e-06,
      "loss": 0.0002,
      "step": 33000
    },
    {
      "epoch": 0.22773841473090164,
      "grad_norm": 0.00014790760178584605,
      "learning_rate": 7.723857685912783e-06,
      "loss": 0.0001,
      "step": 33010
    },
    {
      "epoch": 0.22780740546544598,
      "grad_norm": 0.0,
      "learning_rate": 7.723167778567339e-06,
      "loss": 0.0,
      "step": 33020
    },
    {
      "epoch": 0.22787639619999034,
      "grad_norm": 0.0,
      "learning_rate": 7.722477871221896e-06,
      "loss": 0.0002,
      "step": 33030
    },
    {
      "epoch": 0.2279453869345347,
      "grad_norm": 475.7007751464844,
      "learning_rate": 7.721787963876452e-06,
      "loss": 0.3242,
      "step": 33040
    },
    {
      "epoch": 0.22801437766907903,
      "grad_norm": 0.0,
      "learning_rate": 7.721098056531009e-06,
      "loss": 0.0,
      "step": 33050
    },
    {
      "epoch": 0.2280833684036234,
      "grad_norm": 0.0,
      "learning_rate": 7.720408149185565e-06,
      "loss": 0.0,
      "step": 33060
    },
    {
      "epoch": 0.22815235913816775,
      "grad_norm": 0.0,
      "learning_rate": 7.719718241840122e-06,
      "loss": 0.0,
      "step": 33070
    },
    {
      "epoch": 0.2282213498727121,
      "grad_norm": 0.0,
      "learning_rate": 7.719028334494678e-06,
      "loss": 0.0,
      "step": 33080
    },
    {
      "epoch": 0.22829034060725645,
      "grad_norm": 0.0,
      "learning_rate": 7.718338427149235e-06,
      "loss": 0.3926,
      "step": 33090
    },
    {
      "epoch": 0.2283593313418008,
      "grad_norm": 0.0,
      "learning_rate": 7.717648519803791e-06,
      "loss": 0.0,
      "step": 33100
    },
    {
      "epoch": 0.22842832207634514,
      "grad_norm": 0.0,
      "learning_rate": 7.716958612458348e-06,
      "loss": 0.0028,
      "step": 33110
    },
    {
      "epoch": 0.2284973128108895,
      "grad_norm": 9.887799468089042e-10,
      "learning_rate": 7.716268705112904e-06,
      "loss": 0.0,
      "step": 33120
    },
    {
      "epoch": 0.22856630354543384,
      "grad_norm": 0.0,
      "learning_rate": 7.71557879776746e-06,
      "loss": 0.0,
      "step": 33130
    },
    {
      "epoch": 0.2286352942799782,
      "grad_norm": 0.0,
      "learning_rate": 7.714888890422017e-06,
      "loss": 0.0001,
      "step": 33140
    },
    {
      "epoch": 0.22870428501452256,
      "grad_norm": 2.5723660979082297e-09,
      "learning_rate": 7.714198983076573e-06,
      "loss": 0.0003,
      "step": 33150
    },
    {
      "epoch": 0.2287732757490669,
      "grad_norm": 0.0,
      "learning_rate": 7.71350907573113e-06,
      "loss": 0.0,
      "step": 33160
    },
    {
      "epoch": 0.22884226648361125,
      "grad_norm": 2.1958415530320963e-08,
      "learning_rate": 7.712819168385686e-06,
      "loss": 0.0,
      "step": 33170
    },
    {
      "epoch": 0.2289112572181556,
      "grad_norm": 310.5282897949219,
      "learning_rate": 7.712129261040243e-06,
      "loss": 0.0534,
      "step": 33180
    },
    {
      "epoch": 0.22898024795269994,
      "grad_norm": 0.0,
      "learning_rate": 7.7114393536948e-06,
      "loss": 0.0,
      "step": 33190
    },
    {
      "epoch": 0.2290492386872443,
      "grad_norm": 0.0,
      "learning_rate": 7.710749446349356e-06,
      "loss": 0.0,
      "step": 33200
    },
    {
      "epoch": 0.22911822942178867,
      "grad_norm": 0.0,
      "learning_rate": 7.710059539003912e-06,
      "loss": 0.0,
      "step": 33210
    },
    {
      "epoch": 0.229187220156333,
      "grad_norm": 0.0,
      "learning_rate": 7.709369631658469e-06,
      "loss": 0.0,
      "step": 33220
    },
    {
      "epoch": 0.22925621089087736,
      "grad_norm": 0.0,
      "learning_rate": 7.708679724313025e-06,
      "loss": 0.0,
      "step": 33230
    },
    {
      "epoch": 0.2293252016254217,
      "grad_norm": 0.0,
      "learning_rate": 7.707989816967582e-06,
      "loss": 0.0,
      "step": 33240
    },
    {
      "epoch": 0.22939419235996605,
      "grad_norm": 0.0,
      "learning_rate": 7.707299909622138e-06,
      "loss": 0.0002,
      "step": 33250
    },
    {
      "epoch": 0.22946318309451041,
      "grad_norm": 0.0,
      "learning_rate": 7.706610002276695e-06,
      "loss": 0.0,
      "step": 33260
    },
    {
      "epoch": 0.22953217382905475,
      "grad_norm": 8.9177415452113e-10,
      "learning_rate": 7.705920094931251e-06,
      "loss": 0.0038,
      "step": 33270
    },
    {
      "epoch": 0.2296011645635991,
      "grad_norm": 3.8129115864649066e-08,
      "learning_rate": 7.705230187585808e-06,
      "loss": 0.0,
      "step": 33280
    },
    {
      "epoch": 0.22967015529814347,
      "grad_norm": 0.000523974362295121,
      "learning_rate": 7.704540280240364e-06,
      "loss": 0.048,
      "step": 33290
    },
    {
      "epoch": 0.2297391460326878,
      "grad_norm": 0.0,
      "learning_rate": 7.70385037289492e-06,
      "loss": 0.0022,
      "step": 33300
    },
    {
      "epoch": 0.22980813676723216,
      "grad_norm": 0.0,
      "learning_rate": 7.703160465549477e-06,
      "loss": 0.0,
      "step": 33310
    },
    {
      "epoch": 0.22987712750177652,
      "grad_norm": 3.273544564308395e-07,
      "learning_rate": 7.702470558204034e-06,
      "loss": 0.0001,
      "step": 33320
    },
    {
      "epoch": 0.22994611823632086,
      "grad_norm": 0.0,
      "learning_rate": 7.70178065085859e-06,
      "loss": 0.0,
      "step": 33330
    },
    {
      "epoch": 0.23001510897086522,
      "grad_norm": 3.958035904361168e-06,
      "learning_rate": 7.701090743513147e-06,
      "loss": 0.0,
      "step": 33340
    },
    {
      "epoch": 0.23008409970540955,
      "grad_norm": 0.0,
      "learning_rate": 7.700400836167703e-06,
      "loss": 0.0001,
      "step": 33350
    },
    {
      "epoch": 0.2301530904399539,
      "grad_norm": 0.0,
      "learning_rate": 7.69971092882226e-06,
      "loss": 0.0,
      "step": 33360
    },
    {
      "epoch": 0.23022208117449827,
      "grad_norm": 0.0,
      "learning_rate": 7.699021021476816e-06,
      "loss": 0.124,
      "step": 33370
    },
    {
      "epoch": 0.2302910719090426,
      "grad_norm": 0.0009574118885211647,
      "learning_rate": 7.698400104865917e-06,
      "loss": 0.1053,
      "step": 33380
    },
    {
      "epoch": 0.23036006264358697,
      "grad_norm": 3.641016732558455e-08,
      "learning_rate": 7.697710197520473e-06,
      "loss": 0.0,
      "step": 33390
    },
    {
      "epoch": 0.23042905337813133,
      "grad_norm": 2.59740762231786e-09,
      "learning_rate": 7.69702029017503e-06,
      "loss": 0.0,
      "step": 33400
    },
    {
      "epoch": 0.23049804411267566,
      "grad_norm": 0.0,
      "learning_rate": 7.696330382829586e-06,
      "loss": 0.0,
      "step": 33410
    },
    {
      "epoch": 0.23056703484722002,
      "grad_norm": 1.5054170944495127e-05,
      "learning_rate": 7.695640475484143e-06,
      "loss": 0.0,
      "step": 33420
    },
    {
      "epoch": 0.23063602558176438,
      "grad_norm": 0.0,
      "learning_rate": 7.6949505681387e-06,
      "loss": 0.0001,
      "step": 33430
    },
    {
      "epoch": 0.23070501631630871,
      "grad_norm": 0.0,
      "learning_rate": 7.694260660793256e-06,
      "loss": 0.0,
      "step": 33440
    },
    {
      "epoch": 0.23077400705085307,
      "grad_norm": 0.0,
      "learning_rate": 7.693570753447812e-06,
      "loss": 0.0,
      "step": 33450
    },
    {
      "epoch": 0.2308429977853974,
      "grad_norm": 0.0,
      "learning_rate": 7.692880846102369e-06,
      "loss": 0.0,
      "step": 33460
    },
    {
      "epoch": 0.23091198851994177,
      "grad_norm": 0.00023554077779408544,
      "learning_rate": 7.692190938756925e-06,
      "loss": 0.0,
      "step": 33470
    },
    {
      "epoch": 0.23098097925448613,
      "grad_norm": 0.0,
      "learning_rate": 7.691501031411482e-06,
      "loss": 0.0,
      "step": 33480
    },
    {
      "epoch": 0.23104996998903046,
      "grad_norm": 0.0,
      "learning_rate": 7.690811124066038e-06,
      "loss": 0.0,
      "step": 33490
    },
    {
      "epoch": 0.23111896072357482,
      "grad_norm": 0.0,
      "learning_rate": 7.690121216720595e-06,
      "loss": 0.0,
      "step": 33500
    },
    {
      "epoch": 0.23118795145811918,
      "grad_norm": 1.8539085431257263e-05,
      "learning_rate": 7.689431309375151e-06,
      "loss": 0.0,
      "step": 33510
    },
    {
      "epoch": 0.23125694219266352,
      "grad_norm": 0.0,
      "learning_rate": 7.688741402029708e-06,
      "loss": 0.0331,
      "step": 33520
    },
    {
      "epoch": 0.23132593292720788,
      "grad_norm": 0.0,
      "learning_rate": 7.688051494684264e-06,
      "loss": 0.0,
      "step": 33530
    },
    {
      "epoch": 0.23139492366175224,
      "grad_norm": 0.0,
      "learning_rate": 7.68736158733882e-06,
      "loss": 0.0061,
      "step": 33540
    },
    {
      "epoch": 0.23146391439629657,
      "grad_norm": 0.0,
      "learning_rate": 7.686671679993377e-06,
      "loss": 0.0213,
      "step": 33550
    },
    {
      "epoch": 0.23153290513084093,
      "grad_norm": 9.179851878649004e-10,
      "learning_rate": 7.685981772647934e-06,
      "loss": 0.0,
      "step": 33560
    },
    {
      "epoch": 0.23160189586538527,
      "grad_norm": 3.3695576462378085e-07,
      "learning_rate": 7.68529186530249e-06,
      "loss": 0.0,
      "step": 33570
    },
    {
      "epoch": 0.23167088659992963,
      "grad_norm": 0.0,
      "learning_rate": 7.684601957957047e-06,
      "loss": 0.0,
      "step": 33580
    },
    {
      "epoch": 0.231739877334474,
      "grad_norm": 0.0,
      "learning_rate": 7.683912050611603e-06,
      "loss": 0.0001,
      "step": 33590
    },
    {
      "epoch": 0.23180886806901832,
      "grad_norm": 0.0,
      "learning_rate": 7.68322214326616e-06,
      "loss": 0.0,
      "step": 33600
    },
    {
      "epoch": 0.23187785880356268,
      "grad_norm": 0.0,
      "learning_rate": 7.682532235920716e-06,
      "loss": 0.0,
      "step": 33610
    },
    {
      "epoch": 0.23194684953810704,
      "grad_norm": 0.013300956226885319,
      "learning_rate": 7.681842328575273e-06,
      "loss": 0.0,
      "step": 33620
    },
    {
      "epoch": 0.23201584027265137,
      "grad_norm": 0.0,
      "learning_rate": 7.68115242122983e-06,
      "loss": 0.0,
      "step": 33630
    },
    {
      "epoch": 0.23208483100719574,
      "grad_norm": 0.0,
      "learning_rate": 7.680462513884386e-06,
      "loss": 0.0311,
      "step": 33640
    },
    {
      "epoch": 0.2321538217417401,
      "grad_norm": 0.0,
      "learning_rate": 7.679772606538942e-06,
      "loss": 0.0,
      "step": 33650
    },
    {
      "epoch": 0.23222281247628443,
      "grad_norm": 0.0,
      "learning_rate": 7.679082699193499e-06,
      "loss": 0.0,
      "step": 33660
    },
    {
      "epoch": 0.2322918032108288,
      "grad_norm": 0.0,
      "learning_rate": 7.678392791848055e-06,
      "loss": 0.0012,
      "step": 33670
    },
    {
      "epoch": 0.23236079394537312,
      "grad_norm": 0.0,
      "learning_rate": 7.677702884502612e-06,
      "loss": 0.0,
      "step": 33680
    },
    {
      "epoch": 0.23242978467991748,
      "grad_norm": 2.088952344436734e-09,
      "learning_rate": 7.677012977157168e-06,
      "loss": 0.0,
      "step": 33690
    },
    {
      "epoch": 0.23249877541446184,
      "grad_norm": 0.0,
      "learning_rate": 7.676323069811725e-06,
      "loss": 0.0,
      "step": 33700
    },
    {
      "epoch": 0.23256776614900618,
      "grad_norm": 0.0,
      "learning_rate": 7.675633162466281e-06,
      "loss": 0.0,
      "step": 33710
    },
    {
      "epoch": 0.23263675688355054,
      "grad_norm": 8.314611221749146e-10,
      "learning_rate": 7.674943255120838e-06,
      "loss": 0.0,
      "step": 33720
    },
    {
      "epoch": 0.2327057476180949,
      "grad_norm": 0.0,
      "learning_rate": 7.674253347775394e-06,
      "loss": 0.0002,
      "step": 33730
    },
    {
      "epoch": 0.23277473835263923,
      "grad_norm": 0.0,
      "learning_rate": 7.67356344042995e-06,
      "loss": 0.0,
      "step": 33740
    },
    {
      "epoch": 0.2328437290871836,
      "grad_norm": 8.990783673112901e-10,
      "learning_rate": 7.672873533084507e-06,
      "loss": 0.0,
      "step": 33750
    },
    {
      "epoch": 0.23291271982172795,
      "grad_norm": 0.00036688672844320536,
      "learning_rate": 7.672183625739064e-06,
      "loss": 0.3103,
      "step": 33760
    },
    {
      "epoch": 0.2329817105562723,
      "grad_norm": 1.7620725989431207e-09,
      "learning_rate": 7.67149371839362e-06,
      "loss": 0.0,
      "step": 33770
    },
    {
      "epoch": 0.23305070129081665,
      "grad_norm": 0.0,
      "learning_rate": 7.670803811048177e-06,
      "loss": 0.0,
      "step": 33780
    },
    {
      "epoch": 0.23311969202536098,
      "grad_norm": 6.612060019506316e-07,
      "learning_rate": 7.670113903702735e-06,
      "loss": 0.0,
      "step": 33790
    },
    {
      "epoch": 0.23318868275990534,
      "grad_norm": 0.0,
      "learning_rate": 7.66942399635729e-06,
      "loss": 0.0172,
      "step": 33800
    },
    {
      "epoch": 0.2332576734944497,
      "grad_norm": 0.0,
      "learning_rate": 7.668734089011846e-06,
      "loss": 0.0,
      "step": 33810
    },
    {
      "epoch": 0.23332666422899404,
      "grad_norm": 0.0,
      "learning_rate": 7.668044181666403e-06,
      "loss": 0.2999,
      "step": 33820
    },
    {
      "epoch": 0.2333956549635384,
      "grad_norm": 0.0,
      "learning_rate": 7.667354274320959e-06,
      "loss": 0.0,
      "step": 33830
    },
    {
      "epoch": 0.23346464569808276,
      "grad_norm": 0.0,
      "learning_rate": 7.666664366975515e-06,
      "loss": 0.0,
      "step": 33840
    },
    {
      "epoch": 0.2335336364326271,
      "grad_norm": 34.06905746459961,
      "learning_rate": 7.665974459630072e-06,
      "loss": 0.0141,
      "step": 33850
    },
    {
      "epoch": 0.23360262716717145,
      "grad_norm": 0.00011056315997848287,
      "learning_rate": 7.665284552284628e-06,
      "loss": 0.0514,
      "step": 33860
    },
    {
      "epoch": 0.2336716179017158,
      "grad_norm": 0.0,
      "learning_rate": 7.664594644939185e-06,
      "loss": 0.0,
      "step": 33870
    },
    {
      "epoch": 0.23374060863626014,
      "grad_norm": 0.0,
      "learning_rate": 7.663904737593741e-06,
      "loss": 0.012,
      "step": 33880
    },
    {
      "epoch": 0.2338095993708045,
      "grad_norm": 5.0882640323379746e-08,
      "learning_rate": 7.663214830248298e-06,
      "loss": 0.0,
      "step": 33890
    },
    {
      "epoch": 0.23387859010534884,
      "grad_norm": 0.0,
      "learning_rate": 7.662524922902856e-06,
      "loss": 0.0008,
      "step": 33900
    },
    {
      "epoch": 0.2339475808398932,
      "grad_norm": 0.3880530893802643,
      "learning_rate": 7.661835015557411e-06,
      "loss": 0.0001,
      "step": 33910
    },
    {
      "epoch": 0.23401657157443756,
      "grad_norm": 8.318660547956824e-05,
      "learning_rate": 7.661145108211967e-06,
      "loss": 0.0018,
      "step": 33920
    },
    {
      "epoch": 0.2340855623089819,
      "grad_norm": 4.030636773677543e-06,
      "learning_rate": 7.660455200866524e-06,
      "loss": 0.0,
      "step": 33930
    },
    {
      "epoch": 0.23415455304352625,
      "grad_norm": 0.011503039859235287,
      "learning_rate": 7.65976529352108e-06,
      "loss": 0.0,
      "step": 33940
    },
    {
      "epoch": 0.23422354377807061,
      "grad_norm": 0.0,
      "learning_rate": 7.659075386175637e-06,
      "loss": 0.0,
      "step": 33950
    },
    {
      "epoch": 0.23429253451261495,
      "grad_norm": 0.0,
      "learning_rate": 7.658385478830193e-06,
      "loss": 0.0001,
      "step": 33960
    },
    {
      "epoch": 0.2343615252471593,
      "grad_norm": 0.0002538344415370375,
      "learning_rate": 7.65769557148475e-06,
      "loss": 0.0,
      "step": 33970
    },
    {
      "epoch": 0.23443051598170367,
      "grad_norm": 0.0,
      "learning_rate": 7.657005664139306e-06,
      "loss": 0.0,
      "step": 33980
    },
    {
      "epoch": 0.234499506716248,
      "grad_norm": 2.225191354751587,
      "learning_rate": 7.656315756793863e-06,
      "loss": 0.0004,
      "step": 33990
    },
    {
      "epoch": 0.23456849745079236,
      "grad_norm": 0.0,
      "learning_rate": 7.65562584944842e-06,
      "loss": 0.0005,
      "step": 34000
    },
    {
      "epoch": 0.2346374881853367,
      "grad_norm": 0.0,
      "learning_rate": 7.654935942102978e-06,
      "loss": 0.0,
      "step": 34010
    },
    {
      "epoch": 0.23470647891988106,
      "grad_norm": 0.0009202035143971443,
      "learning_rate": 7.654246034757532e-06,
      "loss": 0.0,
      "step": 34020
    },
    {
      "epoch": 0.23477546965442542,
      "grad_norm": 0.0,
      "learning_rate": 7.653556127412089e-06,
      "loss": 0.0,
      "step": 34030
    },
    {
      "epoch": 0.23484446038896975,
      "grad_norm": 0.0,
      "learning_rate": 7.652866220066645e-06,
      "loss": 0.0,
      "step": 34040
    },
    {
      "epoch": 0.2349134511235141,
      "grad_norm": 9.138249046358737e-10,
      "learning_rate": 7.652176312721202e-06,
      "loss": 0.0,
      "step": 34050
    },
    {
      "epoch": 0.23498244185805847,
      "grad_norm": 0.0,
      "learning_rate": 7.651486405375758e-06,
      "loss": 0.0,
      "step": 34060
    },
    {
      "epoch": 0.2350514325926028,
      "grad_norm": 0.0,
      "learning_rate": 7.650796498030315e-06,
      "loss": 0.0,
      "step": 34070
    },
    {
      "epoch": 0.23512042332714717,
      "grad_norm": 0.0,
      "learning_rate": 7.650106590684871e-06,
      "loss": 0.0,
      "step": 34080
    },
    {
      "epoch": 0.23518941406169153,
      "grad_norm": 0.26287323236465454,
      "learning_rate": 7.649416683339428e-06,
      "loss": 0.0163,
      "step": 34090
    },
    {
      "epoch": 0.23525840479623586,
      "grad_norm": 0.0,
      "learning_rate": 7.648726775993984e-06,
      "loss": 0.0,
      "step": 34100
    },
    {
      "epoch": 0.23532739553078022,
      "grad_norm": 0.0005133957020007074,
      "learning_rate": 7.64803686864854e-06,
      "loss": 0.0,
      "step": 34110
    },
    {
      "epoch": 0.23539638626532455,
      "grad_norm": 0.0,
      "learning_rate": 7.647346961303099e-06,
      "loss": 0.0,
      "step": 34120
    },
    {
      "epoch": 0.23546537699986891,
      "grad_norm": 1.8666461301108939e-06,
      "learning_rate": 7.646657053957655e-06,
      "loss": 0.0,
      "step": 34130
    },
    {
      "epoch": 0.23553436773441327,
      "grad_norm": 0.0,
      "learning_rate": 7.64596714661221e-06,
      "loss": 0.0,
      "step": 34140
    },
    {
      "epoch": 0.2356033584689576,
      "grad_norm": 0.012658294290304184,
      "learning_rate": 7.645277239266767e-06,
      "loss": 0.0005,
      "step": 34150
    },
    {
      "epoch": 0.23567234920350197,
      "grad_norm": 0.0,
      "learning_rate": 7.644587331921323e-06,
      "loss": 0.0002,
      "step": 34160
    },
    {
      "epoch": 0.23574133993804633,
      "grad_norm": 0.0,
      "learning_rate": 7.64389742457588e-06,
      "loss": 0.0,
      "step": 34170
    },
    {
      "epoch": 0.23581033067259066,
      "grad_norm": 0.0,
      "learning_rate": 7.643207517230436e-06,
      "loss": 0.0,
      "step": 34180
    },
    {
      "epoch": 0.23587932140713502,
      "grad_norm": 0.0,
      "learning_rate": 7.642517609884993e-06,
      "loss": 0.001,
      "step": 34190
    },
    {
      "epoch": 0.23594831214167938,
      "grad_norm": 0.0,
      "learning_rate": 7.641827702539549e-06,
      "loss": 0.0,
      "step": 34200
    },
    {
      "epoch": 0.23601730287622372,
      "grad_norm": 2.0044219493865967,
      "learning_rate": 7.641137795194106e-06,
      "loss": 0.0002,
      "step": 34210
    },
    {
      "epoch": 0.23608629361076808,
      "grad_norm": 0.0,
      "learning_rate": 7.640447887848662e-06,
      "loss": 0.0,
      "step": 34220
    },
    {
      "epoch": 0.23615528434531244,
      "grad_norm": 0.0,
      "learning_rate": 7.63975798050322e-06,
      "loss": 0.0,
      "step": 34230
    },
    {
      "epoch": 0.23622427507985677,
      "grad_norm": 0.0,
      "learning_rate": 7.639068073157777e-06,
      "loss": 0.0,
      "step": 34240
    },
    {
      "epoch": 0.23629326581440113,
      "grad_norm": 0.0,
      "learning_rate": 7.638378165812332e-06,
      "loss": 0.0,
      "step": 34250
    },
    {
      "epoch": 0.23636225654894547,
      "grad_norm": 0.0,
      "learning_rate": 7.637688258466888e-06,
      "loss": 0.0016,
      "step": 34260
    },
    {
      "epoch": 0.23643124728348983,
      "grad_norm": 0.0,
      "learning_rate": 7.636998351121445e-06,
      "loss": 0.0,
      "step": 34270
    },
    {
      "epoch": 0.2365002380180342,
      "grad_norm": 9.605948259050479e-10,
      "learning_rate": 7.636308443776001e-06,
      "loss": 0.0,
      "step": 34280
    },
    {
      "epoch": 0.23656922875257852,
      "grad_norm": 0.0,
      "learning_rate": 7.635618536430558e-06,
      "loss": 0.0,
      "step": 34290
    },
    {
      "epoch": 0.23663821948712288,
      "grad_norm": 2.862710957174386e-08,
      "learning_rate": 7.634928629085114e-06,
      "loss": 0.0,
      "step": 34300
    },
    {
      "epoch": 0.23670721022166724,
      "grad_norm": 0.0,
      "learning_rate": 7.63423872173967e-06,
      "loss": 0.0,
      "step": 34310
    },
    {
      "epoch": 0.23677620095621157,
      "grad_norm": 0.0,
      "learning_rate": 7.633548814394227e-06,
      "loss": 0.1395,
      "step": 34320
    },
    {
      "epoch": 0.23684519169075594,
      "grad_norm": 0.0,
      "learning_rate": 7.632858907048784e-06,
      "loss": 0.0,
      "step": 34330
    },
    {
      "epoch": 0.2369141824253003,
      "grad_norm": 0.0004049877170473337,
      "learning_rate": 7.632168999703342e-06,
      "loss": 0.0,
      "step": 34340
    },
    {
      "epoch": 0.23698317315984463,
      "grad_norm": 1.4445790839090478e-06,
      "learning_rate": 7.631479092357898e-06,
      "loss": 0.0,
      "step": 34350
    },
    {
      "epoch": 0.237052163894389,
      "grad_norm": 0.0,
      "learning_rate": 7.630789185012453e-06,
      "loss": 0.0,
      "step": 34360
    },
    {
      "epoch": 0.23712115462893332,
      "grad_norm": 0.00010177328658755869,
      "learning_rate": 7.63009927766701e-06,
      "loss": 0.0,
      "step": 34370
    },
    {
      "epoch": 0.23719014536347768,
      "grad_norm": 0.0,
      "learning_rate": 7.629409370321566e-06,
      "loss": 0.0,
      "step": 34380
    },
    {
      "epoch": 0.23725913609802204,
      "grad_norm": 0.0,
      "learning_rate": 7.628719462976123e-06,
      "loss": 0.0,
      "step": 34390
    },
    {
      "epoch": 0.23732812683256638,
      "grad_norm": 0.0,
      "learning_rate": 7.62802955563068e-06,
      "loss": 0.0,
      "step": 34400
    },
    {
      "epoch": 0.23739711756711074,
      "grad_norm": 0.0,
      "learning_rate": 7.6273396482852354e-06,
      "loss": 0.0,
      "step": 34410
    },
    {
      "epoch": 0.2374661083016551,
      "grad_norm": 0.0,
      "learning_rate": 7.626649740939792e-06,
      "loss": 0.0,
      "step": 34420
    },
    {
      "epoch": 0.23753509903619943,
      "grad_norm": 3.396797183086164e-05,
      "learning_rate": 7.625959833594348e-06,
      "loss": 0.0,
      "step": 34430
    },
    {
      "epoch": 0.2376040897707438,
      "grad_norm": 0.0,
      "learning_rate": 7.625269926248905e-06,
      "loss": 0.0,
      "step": 34440
    },
    {
      "epoch": 0.23767308050528815,
      "grad_norm": 0.0,
      "learning_rate": 7.624580018903462e-06,
      "loss": 0.0003,
      "step": 34450
    },
    {
      "epoch": 0.2377420712398325,
      "grad_norm": 0.0,
      "learning_rate": 7.623890111558019e-06,
      "loss": 0.0,
      "step": 34460
    },
    {
      "epoch": 0.23781106197437685,
      "grad_norm": 0.0,
      "learning_rate": 7.623200204212575e-06,
      "loss": 0.0,
      "step": 34470
    },
    {
      "epoch": 0.23788005270892118,
      "grad_norm": 0.0,
      "learning_rate": 7.622510296867132e-06,
      "loss": 0.0,
      "step": 34480
    },
    {
      "epoch": 0.23794904344346554,
      "grad_norm": 7.37014289597937e-08,
      "learning_rate": 7.621820389521688e-06,
      "loss": 0.0,
      "step": 34490
    },
    {
      "epoch": 0.2380180341780099,
      "grad_norm": 0.0,
      "learning_rate": 7.621130482176245e-06,
      "loss": 0.0,
      "step": 34500
    },
    {
      "epoch": 0.23808702491255423,
      "grad_norm": 0.0,
      "learning_rate": 7.620440574830801e-06,
      "loss": 0.0,
      "step": 34510
    },
    {
      "epoch": 0.2381560156470986,
      "grad_norm": 2.867292881011963,
      "learning_rate": 7.619750667485357e-06,
      "loss": 0.0006,
      "step": 34520
    },
    {
      "epoch": 0.23822500638164296,
      "grad_norm": 0.0,
      "learning_rate": 7.619060760139913e-06,
      "loss": 0.0,
      "step": 34530
    },
    {
      "epoch": 0.2382939971161873,
      "grad_norm": 0.0,
      "learning_rate": 7.61837085279447e-06,
      "loss": 0.0,
      "step": 34540
    },
    {
      "epoch": 0.23836298785073165,
      "grad_norm": 0.0,
      "learning_rate": 7.617680945449026e-06,
      "loss": 0.0,
      "step": 34550
    },
    {
      "epoch": 0.238431978585276,
      "grad_norm": 0.0,
      "learning_rate": 7.616991038103583e-06,
      "loss": 0.0003,
      "step": 34560
    },
    {
      "epoch": 0.23850096931982034,
      "grad_norm": 0.0,
      "learning_rate": 7.61630113075814e-06,
      "loss": 0.0,
      "step": 34570
    },
    {
      "epoch": 0.2385699600543647,
      "grad_norm": 0.0,
      "learning_rate": 7.615611223412697e-06,
      "loss": 0.0,
      "step": 34580
    },
    {
      "epoch": 0.23863895078890904,
      "grad_norm": 0.0,
      "learning_rate": 7.614921316067253e-06,
      "loss": 0.3754,
      "step": 34590
    },
    {
      "epoch": 0.2387079415234534,
      "grad_norm": 0.0060450066812336445,
      "learning_rate": 7.61423140872181e-06,
      "loss": 0.0,
      "step": 34600
    },
    {
      "epoch": 0.23877693225799776,
      "grad_norm": 0.0,
      "learning_rate": 7.613541501376366e-06,
      "loss": 0.0,
      "step": 34610
    },
    {
      "epoch": 0.2388459229925421,
      "grad_norm": 0.0,
      "learning_rate": 7.612851594030923e-06,
      "loss": 0.0,
      "step": 34620
    },
    {
      "epoch": 0.23891491372708645,
      "grad_norm": 0.0,
      "learning_rate": 7.612161686685478e-06,
      "loss": 0.0001,
      "step": 34630
    },
    {
      "epoch": 0.23898390446163081,
      "grad_norm": 0.0,
      "learning_rate": 7.611471779340035e-06,
      "loss": 1.5211,
      "step": 34640
    },
    {
      "epoch": 0.23905289519617515,
      "grad_norm": 0.0,
      "learning_rate": 7.610781871994591e-06,
      "loss": 0.0,
      "step": 34650
    },
    {
      "epoch": 0.2391218859307195,
      "grad_norm": 5.226797384239035e-06,
      "learning_rate": 7.610091964649148e-06,
      "loss": 0.0001,
      "step": 34660
    },
    {
      "epoch": 0.23919087666526387,
      "grad_norm": 0.0,
      "learning_rate": 7.609402057303704e-06,
      "loss": 0.0,
      "step": 34670
    },
    {
      "epoch": 0.2392598673998082,
      "grad_norm": 0.0,
      "learning_rate": 7.6087121499582615e-06,
      "loss": 0.0,
      "step": 34680
    },
    {
      "epoch": 0.23932885813435256,
      "grad_norm": 0.0,
      "learning_rate": 7.608022242612818e-06,
      "loss": 0.0,
      "step": 34690
    },
    {
      "epoch": 0.2393978488688969,
      "grad_norm": 2.01933944481425e-05,
      "learning_rate": 7.6073323352673745e-06,
      "loss": 0.0001,
      "step": 34700
    },
    {
      "epoch": 0.23946683960344126,
      "grad_norm": 0.0,
      "learning_rate": 7.606642427921931e-06,
      "loss": 0.0001,
      "step": 34710
    },
    {
      "epoch": 0.23953583033798562,
      "grad_norm": 0.0,
      "learning_rate": 7.6059525205764875e-06,
      "loss": 0.0,
      "step": 34720
    },
    {
      "epoch": 0.23960482107252995,
      "grad_norm": 0.0,
      "learning_rate": 7.605262613231044e-06,
      "loss": 0.0001,
      "step": 34730
    },
    {
      "epoch": 0.2396738118070743,
      "grad_norm": 0.0,
      "learning_rate": 7.6045727058856e-06,
      "loss": 0.0,
      "step": 34740
    },
    {
      "epoch": 0.23974280254161867,
      "grad_norm": 0.0,
      "learning_rate": 7.603882798540156e-06,
      "loss": 0.0,
      "step": 34750
    },
    {
      "epoch": 0.239811793276163,
      "grad_norm": 0.0,
      "learning_rate": 7.603192891194713e-06,
      "loss": 0.0893,
      "step": 34760
    },
    {
      "epoch": 0.23988078401070737,
      "grad_norm": 0.0,
      "learning_rate": 7.602502983849269e-06,
      "loss": 0.0013,
      "step": 34770
    },
    {
      "epoch": 0.23994977474525173,
      "grad_norm": 0.0,
      "learning_rate": 7.6018130765038256e-06,
      "loss": 0.0129,
      "step": 34780
    },
    {
      "epoch": 0.24001876547979606,
      "grad_norm": 0.0,
      "learning_rate": 7.601123169158383e-06,
      "loss": 0.0,
      "step": 34790
    },
    {
      "epoch": 0.24008775621434042,
      "grad_norm": 0.0,
      "learning_rate": 7.600433261812939e-06,
      "loss": 0.0,
      "step": 34800
    },
    {
      "epoch": 0.24015674694888475,
      "grad_norm": 0.0,
      "learning_rate": 7.599743354467496e-06,
      "loss": 0.0,
      "step": 34810
    },
    {
      "epoch": 0.2402257376834291,
      "grad_norm": 7.96481799625326e-06,
      "learning_rate": 7.599053447122052e-06,
      "loss": 0.0,
      "step": 34820
    },
    {
      "epoch": 0.24029472841797347,
      "grad_norm": 27.987932205200195,
      "learning_rate": 7.598363539776609e-06,
      "loss": 0.0076,
      "step": 34830
    },
    {
      "epoch": 0.2403637191525178,
      "grad_norm": 0.01797497272491455,
      "learning_rate": 7.597673632431165e-06,
      "loss": 0.0,
      "step": 34840
    },
    {
      "epoch": 0.24043270988706217,
      "grad_norm": 0.0,
      "learning_rate": 7.596983725085722e-06,
      "loss": 0.0,
      "step": 34850
    },
    {
      "epoch": 0.24050170062160653,
      "grad_norm": 0.0,
      "learning_rate": 7.5962938177402775e-06,
      "loss": 0.0,
      "step": 34860
    },
    {
      "epoch": 0.24057069135615086,
      "grad_norm": 0.0,
      "learning_rate": 7.595603910394834e-06,
      "loss": 0.0,
      "step": 34870
    },
    {
      "epoch": 0.24063968209069522,
      "grad_norm": 0.0,
      "learning_rate": 7.5949140030493905e-06,
      "loss": 0.7873,
      "step": 34880
    },
    {
      "epoch": 0.24070867282523958,
      "grad_norm": 0.0,
      "learning_rate": 7.594224095703947e-06,
      "loss": 0.0,
      "step": 34890
    },
    {
      "epoch": 0.24077766355978392,
      "grad_norm": 0.0,
      "learning_rate": 7.593534188358504e-06,
      "loss": 0.0,
      "step": 34900
    },
    {
      "epoch": 0.24084665429432828,
      "grad_norm": 0.0,
      "learning_rate": 7.592844281013061e-06,
      "loss": 0.0,
      "step": 34910
    },
    {
      "epoch": 0.2409156450288726,
      "grad_norm": 0.021901410073041916,
      "learning_rate": 7.592154373667617e-06,
      "loss": 0.0,
      "step": 34920
    },
    {
      "epoch": 0.24098463576341697,
      "grad_norm": 0.0,
      "learning_rate": 7.591464466322174e-06,
      "loss": 0.2842,
      "step": 34930
    },
    {
      "epoch": 0.24105362649796133,
      "grad_norm": 7.792269752826542e-05,
      "learning_rate": 7.59077455897673e-06,
      "loss": 0.0,
      "step": 34940
    },
    {
      "epoch": 0.24112261723250566,
      "grad_norm": 0.006543013267219067,
      "learning_rate": 7.590084651631287e-06,
      "loss": 0.0009,
      "step": 34950
    },
    {
      "epoch": 0.24119160796705003,
      "grad_norm": 0.0,
      "learning_rate": 7.589394744285843e-06,
      "loss": 0.0,
      "step": 34960
    },
    {
      "epoch": 0.2412605987015944,
      "grad_norm": 0.0,
      "learning_rate": 7.588704836940399e-06,
      "loss": 0.0,
      "step": 34970
    },
    {
      "epoch": 0.24132958943613872,
      "grad_norm": 0.04873797297477722,
      "learning_rate": 7.588014929594955e-06,
      "loss": 0.0,
      "step": 34980
    },
    {
      "epoch": 0.24139858017068308,
      "grad_norm": 8.364700598839647e-10,
      "learning_rate": 7.587325022249512e-06,
      "loss": 0.0031,
      "step": 34990
    },
    {
      "epoch": 0.24146757090522744,
      "grad_norm": 0.0,
      "learning_rate": 7.586635114904068e-06,
      "loss": 0.0,
      "step": 35000
    },
    {
      "epoch": 0.24153656163977177,
      "grad_norm": 0.0,
      "learning_rate": 7.585945207558626e-06,
      "loss": 0.0,
      "step": 35010
    },
    {
      "epoch": 0.24160555237431613,
      "grad_norm": 0.0013921007048338652,
      "learning_rate": 7.585255300213182e-06,
      "loss": 0.0,
      "step": 35020
    },
    {
      "epoch": 0.24167454310886047,
      "grad_norm": 7.703089011101838e-08,
      "learning_rate": 7.584565392867739e-06,
      "loss": 0.0,
      "step": 35030
    },
    {
      "epoch": 0.24174353384340483,
      "grad_norm": 0.0001143598128692247,
      "learning_rate": 7.583875485522295e-06,
      "loss": 0.0,
      "step": 35040
    },
    {
      "epoch": 0.2418125245779492,
      "grad_norm": 0.0,
      "learning_rate": 7.583185578176852e-06,
      "loss": 0.0,
      "step": 35050
    },
    {
      "epoch": 0.24188151531249352,
      "grad_norm": 0.0037796262186020613,
      "learning_rate": 7.582495670831408e-06,
      "loss": 0.1351,
      "step": 35060
    },
    {
      "epoch": 0.24195050604703788,
      "grad_norm": 0.0,
      "learning_rate": 7.581805763485965e-06,
      "loss": 0.0,
      "step": 35070
    },
    {
      "epoch": 0.24201949678158224,
      "grad_norm": 0.00025030432152561843,
      "learning_rate": 7.58111585614052e-06,
      "loss": 0.0,
      "step": 35080
    },
    {
      "epoch": 0.24208848751612658,
      "grad_norm": 0.0,
      "learning_rate": 7.580425948795077e-06,
      "loss": 0.0001,
      "step": 35090
    },
    {
      "epoch": 0.24215747825067094,
      "grad_norm": 0.0,
      "learning_rate": 7.579736041449633e-06,
      "loss": 0.0,
      "step": 35100
    },
    {
      "epoch": 0.2422264689852153,
      "grad_norm": 0.003756160382181406,
      "learning_rate": 7.57904613410419e-06,
      "loss": 0.0,
      "step": 35110
    },
    {
      "epoch": 0.24229545971975963,
      "grad_norm": 2.1044198206254805e-07,
      "learning_rate": 7.578356226758747e-06,
      "loss": 0.2871,
      "step": 35120
    },
    {
      "epoch": 0.242364450454304,
      "grad_norm": 0.0,
      "learning_rate": 7.577666319413304e-06,
      "loss": 0.0001,
      "step": 35130
    },
    {
      "epoch": 0.24243344118884833,
      "grad_norm": 0.0,
      "learning_rate": 7.57697641206786e-06,
      "loss": 0.0389,
      "step": 35140
    },
    {
      "epoch": 0.2425024319233927,
      "grad_norm": 0.0,
      "learning_rate": 7.5762865047224166e-06,
      "loss": 0.0,
      "step": 35150
    },
    {
      "epoch": 0.24257142265793705,
      "grad_norm": 0.0,
      "learning_rate": 7.575596597376973e-06,
      "loss": 0.0,
      "step": 35160
    },
    {
      "epoch": 0.24264041339248138,
      "grad_norm": 3.31876170989176e-09,
      "learning_rate": 7.5749066900315295e-06,
      "loss": 0.0,
      "step": 35170
    },
    {
      "epoch": 0.24270940412702574,
      "grad_norm": 0.0,
      "learning_rate": 7.574216782686086e-06,
      "loss": 0.0,
      "step": 35180
    },
    {
      "epoch": 0.2427783948615701,
      "grad_norm": 0.0,
      "learning_rate": 7.5735268753406425e-06,
      "loss": 0.0,
      "step": 35190
    },
    {
      "epoch": 0.24284738559611443,
      "grad_norm": 0.0,
      "learning_rate": 7.572836967995198e-06,
      "loss": 0.0,
      "step": 35200
    },
    {
      "epoch": 0.2429163763306588,
      "grad_norm": 1.3164053598302417e-05,
      "learning_rate": 7.572147060649755e-06,
      "loss": 0.0,
      "step": 35210
    },
    {
      "epoch": 0.24298536706520316,
      "grad_norm": 0.0,
      "learning_rate": 7.571457153304311e-06,
      "loss": 0.0,
      "step": 35220
    },
    {
      "epoch": 0.2430543577997475,
      "grad_norm": 0.0,
      "learning_rate": 7.5707672459588685e-06,
      "loss": 0.0,
      "step": 35230
    },
    {
      "epoch": 0.24312334853429185,
      "grad_norm": 0.0,
      "learning_rate": 7.570077338613425e-06,
      "loss": 0.0,
      "step": 35240
    },
    {
      "epoch": 0.24319233926883618,
      "grad_norm": 5.7123459555441514e-05,
      "learning_rate": 7.5693874312679815e-06,
      "loss": 0.0001,
      "step": 35250
    },
    {
      "epoch": 0.24326133000338054,
      "grad_norm": 0.0,
      "learning_rate": 7.568697523922538e-06,
      "loss": 0.0,
      "step": 35260
    },
    {
      "epoch": 0.2433303207379249,
      "grad_norm": 7.3790483474731445,
      "learning_rate": 7.5680076165770944e-06,
      "loss": 0.0013,
      "step": 35270
    },
    {
      "epoch": 0.24339931147246924,
      "grad_norm": 0.011852694675326347,
      "learning_rate": 7.567317709231651e-06,
      "loss": 0.0001,
      "step": 35280
    },
    {
      "epoch": 0.2434683022070136,
      "grad_norm": 0.0,
      "learning_rate": 7.5666278018862074e-06,
      "loss": 0.0,
      "step": 35290
    },
    {
      "epoch": 0.24353729294155796,
      "grad_norm": 0.0,
      "learning_rate": 7.565937894540764e-06,
      "loss": 0.0,
      "step": 35300
    },
    {
      "epoch": 0.2436062836761023,
      "grad_norm": 0.0,
      "learning_rate": 7.5652479871953196e-06,
      "loss": 0.1551,
      "step": 35310
    },
    {
      "epoch": 0.24367527441064665,
      "grad_norm": 1.3651856534124818e-07,
      "learning_rate": 7.564558079849876e-06,
      "loss": 0.0,
      "step": 35320
    },
    {
      "epoch": 0.243744265145191,
      "grad_norm": 0.0,
      "learning_rate": 7.5638681725044325e-06,
      "loss": 0.0,
      "step": 35330
    },
    {
      "epoch": 0.24381325587973535,
      "grad_norm": 0.0,
      "learning_rate": 7.56317826515899e-06,
      "loss": 0.0045,
      "step": 35340
    },
    {
      "epoch": 0.2438822466142797,
      "grad_norm": 0.0,
      "learning_rate": 7.562488357813546e-06,
      "loss": 0.0,
      "step": 35350
    },
    {
      "epoch": 0.24395123734882404,
      "grad_norm": 0.0,
      "learning_rate": 7.561798450468103e-06,
      "loss": 0.0,
      "step": 35360
    },
    {
      "epoch": 0.2440202280833684,
      "grad_norm": 0.0,
      "learning_rate": 7.561108543122659e-06,
      "loss": 0.0,
      "step": 35370
    },
    {
      "epoch": 0.24408921881791276,
      "grad_norm": 0.0,
      "learning_rate": 7.560418635777216e-06,
      "loss": 0.0,
      "step": 35380
    },
    {
      "epoch": 0.2441582095524571,
      "grad_norm": 0.0,
      "learning_rate": 7.559728728431772e-06,
      "loss": 0.0,
      "step": 35390
    },
    {
      "epoch": 0.24422720028700146,
      "grad_norm": 0.0,
      "learning_rate": 7.559038821086329e-06,
      "loss": 0.0,
      "step": 35400
    },
    {
      "epoch": 0.24429619102154582,
      "grad_norm": 0.0,
      "learning_rate": 7.558348913740885e-06,
      "loss": 0.0001,
      "step": 35410
    },
    {
      "epoch": 0.24436518175609015,
      "grad_norm": 0.0,
      "learning_rate": 7.557659006395441e-06,
      "loss": 0.0,
      "step": 35420
    },
    {
      "epoch": 0.2444341724906345,
      "grad_norm": 1.8869333267211914,
      "learning_rate": 7.5569690990499974e-06,
      "loss": 0.0005,
      "step": 35430
    },
    {
      "epoch": 0.24450316322517887,
      "grad_norm": 0.0,
      "learning_rate": 7.556279191704554e-06,
      "loss": 0.0,
      "step": 35440
    },
    {
      "epoch": 0.2445721539597232,
      "grad_norm": 0.0,
      "learning_rate": 7.555589284359111e-06,
      "loss": 0.0,
      "step": 35450
    },
    {
      "epoch": 0.24464114469426756,
      "grad_norm": 0.0,
      "learning_rate": 7.554899377013668e-06,
      "loss": 0.0,
      "step": 35460
    },
    {
      "epoch": 0.2447101354288119,
      "grad_norm": 0.0,
      "learning_rate": 7.554209469668224e-06,
      "loss": 0.0,
      "step": 35470
    },
    {
      "epoch": 0.24477912616335626,
      "grad_norm": 0.0,
      "learning_rate": 7.553519562322781e-06,
      "loss": 0.0,
      "step": 35480
    },
    {
      "epoch": 0.24484811689790062,
      "grad_norm": 3.905999183654785,
      "learning_rate": 7.552829654977337e-06,
      "loss": 0.0009,
      "step": 35490
    },
    {
      "epoch": 0.24491710763244495,
      "grad_norm": 0.0,
      "learning_rate": 7.552139747631894e-06,
      "loss": 0.0,
      "step": 35500
    },
    {
      "epoch": 0.2449860983669893,
      "grad_norm": 0.0,
      "learning_rate": 7.55144984028645e-06,
      "loss": 0.0044,
      "step": 35510
    },
    {
      "epoch": 0.24505508910153367,
      "grad_norm": 0.0,
      "learning_rate": 7.550759932941007e-06,
      "loss": 0.0,
      "step": 35520
    },
    {
      "epoch": 0.245124079836078,
      "grad_norm": 0.0,
      "learning_rate": 7.550070025595563e-06,
      "loss": 0.0054,
      "step": 35530
    },
    {
      "epoch": 0.24519307057062237,
      "grad_norm": 0.0,
      "learning_rate": 7.549380118250119e-06,
      "loss": 0.0,
      "step": 35540
    },
    {
      "epoch": 0.24526206130516673,
      "grad_norm": 0.0,
      "learning_rate": 7.548690210904675e-06,
      "loss": 0.0068,
      "step": 35550
    },
    {
      "epoch": 0.24533105203971106,
      "grad_norm": 0.0,
      "learning_rate": 7.548000303559233e-06,
      "loss": 0.0,
      "step": 35560
    },
    {
      "epoch": 0.24540004277425542,
      "grad_norm": 0.0,
      "learning_rate": 7.547310396213789e-06,
      "loss": 0.0,
      "step": 35570
    },
    {
      "epoch": 0.24546903350879976,
      "grad_norm": 9.96428184407705e-08,
      "learning_rate": 7.546620488868346e-06,
      "loss": 0.0,
      "step": 35580
    },
    {
      "epoch": 0.24553802424334412,
      "grad_norm": 0.0,
      "learning_rate": 7.545930581522902e-06,
      "loss": 0.0,
      "step": 35590
    },
    {
      "epoch": 0.24560701497788848,
      "grad_norm": 0.0,
      "learning_rate": 7.545240674177459e-06,
      "loss": 0.0,
      "step": 35600
    },
    {
      "epoch": 0.2456760057124328,
      "grad_norm": 4.5747855503819324e-10,
      "learning_rate": 7.544550766832015e-06,
      "loss": 0.0,
      "step": 35610
    },
    {
      "epoch": 0.24574499644697717,
      "grad_norm": 0.0,
      "learning_rate": 7.543860859486572e-06,
      "loss": 0.0033,
      "step": 35620
    },
    {
      "epoch": 0.24581398718152153,
      "grad_norm": 2.5132450787168636e-08,
      "learning_rate": 7.543170952141128e-06,
      "loss": 0.0,
      "step": 35630
    },
    {
      "epoch": 0.24588297791606586,
      "grad_norm": 0.0,
      "learning_rate": 7.542481044795685e-06,
      "loss": 0.0,
      "step": 35640
    },
    {
      "epoch": 0.24595196865061023,
      "grad_norm": 0.0,
      "learning_rate": 7.54179113745024e-06,
      "loss": 0.0,
      "step": 35650
    },
    {
      "epoch": 0.2460209593851546,
      "grad_norm": 0.0,
      "learning_rate": 7.541101230104797e-06,
      "loss": 0.0012,
      "step": 35660
    },
    {
      "epoch": 0.24608995011969892,
      "grad_norm": 0.0,
      "learning_rate": 7.540411322759355e-06,
      "loss": 0.0,
      "step": 35670
    },
    {
      "epoch": 0.24615894085424328,
      "grad_norm": 0.02591627463698387,
      "learning_rate": 7.5397214154139106e-06,
      "loss": 0.0,
      "step": 35680
    },
    {
      "epoch": 0.2462279315887876,
      "grad_norm": 0.0,
      "learning_rate": 7.539031508068467e-06,
      "loss": 0.0,
      "step": 35690
    },
    {
      "epoch": 0.24629692232333197,
      "grad_norm": 0.0,
      "learning_rate": 7.5383416007230235e-06,
      "loss": 0.0,
      "step": 35700
    },
    {
      "epoch": 0.24636591305787633,
      "grad_norm": 0.0,
      "learning_rate": 7.53765169337758e-06,
      "loss": 0.0,
      "step": 35710
    },
    {
      "epoch": 0.24643490379242067,
      "grad_norm": 0.022486288100481033,
      "learning_rate": 7.5369617860321365e-06,
      "loss": 0.0013,
      "step": 35720
    },
    {
      "epoch": 0.24650389452696503,
      "grad_norm": 0.0,
      "learning_rate": 7.536271878686693e-06,
      "loss": 0.0,
      "step": 35730
    },
    {
      "epoch": 0.2465728852615094,
      "grad_norm": 4.3531397930784976e-10,
      "learning_rate": 7.5355819713412495e-06,
      "loss": 0.0,
      "step": 35740
    },
    {
      "epoch": 0.24664187599605372,
      "grad_norm": 0.0,
      "learning_rate": 7.534892063995806e-06,
      "loss": 0.0018,
      "step": 35750
    },
    {
      "epoch": 0.24671086673059808,
      "grad_norm": 0.0,
      "learning_rate": 7.534202156650362e-06,
      "loss": 0.0,
      "step": 35760
    },
    {
      "epoch": 0.24677985746514244,
      "grad_norm": 0.0,
      "learning_rate": 7.533512249304918e-06,
      "loss": 0.0017,
      "step": 35770
    },
    {
      "epoch": 0.24684884819968678,
      "grad_norm": 0.0,
      "learning_rate": 7.532822341959476e-06,
      "loss": 0.0089,
      "step": 35780
    },
    {
      "epoch": 0.24691783893423114,
      "grad_norm": 0.0,
      "learning_rate": 7.532132434614032e-06,
      "loss": 0.0001,
      "step": 35790
    },
    {
      "epoch": 0.24698682966877547,
      "grad_norm": 0.0,
      "learning_rate": 7.5314425272685884e-06,
      "loss": 0.0,
      "step": 35800
    },
    {
      "epoch": 0.24705582040331983,
      "grad_norm": 0.0,
      "learning_rate": 7.530752619923145e-06,
      "loss": 0.0,
      "step": 35810
    },
    {
      "epoch": 0.2471248111378642,
      "grad_norm": 0.0,
      "learning_rate": 7.530062712577701e-06,
      "loss": 0.1126,
      "step": 35820
    },
    {
      "epoch": 0.24719380187240853,
      "grad_norm": 0.0,
      "learning_rate": 7.529372805232258e-06,
      "loss": 0.0002,
      "step": 35830
    },
    {
      "epoch": 0.24726279260695289,
      "grad_norm": 0.0,
      "learning_rate": 7.528682897886814e-06,
      "loss": 0.0,
      "step": 35840
    },
    {
      "epoch": 0.24733178334149725,
      "grad_norm": 0.0,
      "learning_rate": 7.527992990541371e-06,
      "loss": 0.0002,
      "step": 35850
    },
    {
      "epoch": 0.24740077407604158,
      "grad_norm": 0.0,
      "learning_rate": 7.527303083195927e-06,
      "loss": 0.0235,
      "step": 35860
    },
    {
      "epoch": 0.24746976481058594,
      "grad_norm": 9.46808941080235e-05,
      "learning_rate": 7.526613175850484e-06,
      "loss": 0.0,
      "step": 35870
    },
    {
      "epoch": 0.2475387555451303,
      "grad_norm": 0.0,
      "learning_rate": 7.5259232685050395e-06,
      "loss": 0.0001,
      "step": 35880
    },
    {
      "epoch": 0.24760774627967463,
      "grad_norm": 1.8822514903149568e-05,
      "learning_rate": 7.525233361159598e-06,
      "loss": 0.0036,
      "step": 35890
    },
    {
      "epoch": 0.247676737014219,
      "grad_norm": 0.0,
      "learning_rate": 7.524543453814153e-06,
      "loss": 0.0,
      "step": 35900
    },
    {
      "epoch": 0.24774572774876333,
      "grad_norm": 0.0,
      "learning_rate": 7.52385354646871e-06,
      "loss": 0.0,
      "step": 35910
    },
    {
      "epoch": 0.2478147184833077,
      "grad_norm": 0.0,
      "learning_rate": 7.523163639123266e-06,
      "loss": 0.0,
      "step": 35920
    },
    {
      "epoch": 0.24788370921785205,
      "grad_norm": 0.0,
      "learning_rate": 7.522473731777823e-06,
      "loss": 0.0,
      "step": 35930
    },
    {
      "epoch": 0.24795269995239638,
      "grad_norm": 0.0,
      "learning_rate": 7.521783824432379e-06,
      "loss": 0.0,
      "step": 35940
    },
    {
      "epoch": 0.24802169068694074,
      "grad_norm": 16.917089462280273,
      "learning_rate": 7.521093917086936e-06,
      "loss": 0.0035,
      "step": 35950
    },
    {
      "epoch": 0.2480906814214851,
      "grad_norm": 0.0,
      "learning_rate": 7.520404009741492e-06,
      "loss": 0.0,
      "step": 35960
    },
    {
      "epoch": 0.24815967215602944,
      "grad_norm": 0.0,
      "learning_rate": 7.519714102396049e-06,
      "loss": 0.0003,
      "step": 35970
    },
    {
      "epoch": 0.2482286628905738,
      "grad_norm": 0.0,
      "learning_rate": 7.519024195050605e-06,
      "loss": 0.0,
      "step": 35980
    },
    {
      "epoch": 0.24829765362511816,
      "grad_norm": 0.0,
      "learning_rate": 7.518334287705161e-06,
      "loss": 0.0,
      "step": 35990
    },
    {
      "epoch": 0.2483666443596625,
      "grad_norm": 0.0015563899651169777,
      "learning_rate": 7.517644380359719e-06,
      "loss": 0.0,
      "step": 36000
    },
    {
      "epoch": 0.24843563509420685,
      "grad_norm": 6.156587460282026e-06,
      "learning_rate": 7.516954473014276e-06,
      "loss": 0.0,
      "step": 36010
    },
    {
      "epoch": 0.24850462582875119,
      "grad_norm": 0.0,
      "learning_rate": 7.516264565668831e-06,
      "loss": 0.0,
      "step": 36020
    },
    {
      "epoch": 0.24857361656329555,
      "grad_norm": 0.0,
      "learning_rate": 7.515574658323388e-06,
      "loss": 0.0,
      "step": 36030
    },
    {
      "epoch": 0.2486426072978399,
      "grad_norm": 7.864497456466779e-05,
      "learning_rate": 7.514884750977944e-06,
      "loss": 0.0,
      "step": 36040
    },
    {
      "epoch": 0.24871159803238424,
      "grad_norm": 0.0,
      "learning_rate": 7.514194843632501e-06,
      "loss": 0.0,
      "step": 36050
    },
    {
      "epoch": 0.2487805887669286,
      "grad_norm": 0.0,
      "learning_rate": 7.513504936287057e-06,
      "loss": 0.0001,
      "step": 36060
    },
    {
      "epoch": 0.24884957950147296,
      "grad_norm": 4.391550179061454e-10,
      "learning_rate": 7.512815028941614e-06,
      "loss": 0.0,
      "step": 36070
    },
    {
      "epoch": 0.2489185702360173,
      "grad_norm": 0.0,
      "learning_rate": 7.51212512159617e-06,
      "loss": 0.0,
      "step": 36080
    },
    {
      "epoch": 0.24898756097056166,
      "grad_norm": 0.0,
      "learning_rate": 7.511435214250727e-06,
      "loss": 0.0,
      "step": 36090
    },
    {
      "epoch": 0.24905655170510602,
      "grad_norm": 0.0,
      "learning_rate": 7.510745306905282e-06,
      "loss": 0.0,
      "step": 36100
    },
    {
      "epoch": 0.24912554243965035,
      "grad_norm": 0.0,
      "learning_rate": 7.5100553995598405e-06,
      "loss": 0.0002,
      "step": 36110
    },
    {
      "epoch": 0.2491945331741947,
      "grad_norm": 2.0434224268228718e-07,
      "learning_rate": 7.509365492214397e-06,
      "loss": 0.0,
      "step": 36120
    },
    {
      "epoch": 0.24926352390873904,
      "grad_norm": 9.190723358187824e-05,
      "learning_rate": 7.508675584868953e-06,
      "loss": 0.0,
      "step": 36130
    },
    {
      "epoch": 0.2493325146432834,
      "grad_norm": 0.0,
      "learning_rate": 7.507985677523509e-06,
      "loss": 0.0,
      "step": 36140
    },
    {
      "epoch": 0.24940150537782776,
      "grad_norm": 0.13615694642066956,
      "learning_rate": 7.507295770178066e-06,
      "loss": 0.007,
      "step": 36150
    },
    {
      "epoch": 0.2494704961123721,
      "grad_norm": 0.0,
      "learning_rate": 7.506605862832622e-06,
      "loss": 0.0,
      "step": 36160
    },
    {
      "epoch": 0.24953948684691646,
      "grad_norm": 0.0,
      "learning_rate": 7.505915955487179e-06,
      "loss": 0.0,
      "step": 36170
    },
    {
      "epoch": 0.24960847758146082,
      "grad_norm": 0.0,
      "learning_rate": 7.505226048141735e-06,
      "loss": 0.0,
      "step": 36180
    },
    {
      "epoch": 0.24967746831600515,
      "grad_norm": 0.0,
      "learning_rate": 7.5045361407962916e-06,
      "loss": 0.0,
      "step": 36190
    },
    {
      "epoch": 0.2497464590505495,
      "grad_norm": 0.02043294906616211,
      "learning_rate": 7.503846233450848e-06,
      "loss": 0.0177,
      "step": 36200
    },
    {
      "epoch": 0.24981544978509387,
      "grad_norm": 0.0,
      "learning_rate": 7.5031563261054045e-06,
      "loss": 0.0,
      "step": 36210
    },
    {
      "epoch": 0.2498844405196382,
      "grad_norm": 0.0,
      "learning_rate": 7.502466418759962e-06,
      "loss": 0.0,
      "step": 36220
    },
    {
      "epoch": 0.24995343125418257,
      "grad_norm": 0.2231520265340805,
      "learning_rate": 7.501776511414518e-06,
      "loss": 0.0002,
      "step": 36230
    },
    {
      "epoch": 0.25002242198872693,
      "grad_norm": 0.0,
      "learning_rate": 7.501086604069074e-06,
      "loss": 0.0,
      "step": 36240
    },
    {
      "epoch": 0.2500914127232713,
      "grad_norm": 1.3483792145052576e-06,
      "learning_rate": 7.5003966967236305e-06,
      "loss": 0.0088,
      "step": 36250
    },
    {
      "epoch": 0.2501604034578156,
      "grad_norm": 0.0,
      "learning_rate": 7.499706789378187e-06,
      "loss": 0.0006,
      "step": 36260
    },
    {
      "epoch": 0.25022939419235996,
      "grad_norm": 4.065394065833061e-08,
      "learning_rate": 7.4990168820327435e-06,
      "loss": 0.0,
      "step": 36270
    },
    {
      "epoch": 0.2502983849269043,
      "grad_norm": 0.0,
      "learning_rate": 7.4983269746873e-06,
      "loss": 0.0,
      "step": 36280
    },
    {
      "epoch": 0.2503673756614487,
      "grad_norm": 0.0,
      "learning_rate": 7.4976370673418565e-06,
      "loss": 0.0,
      "step": 36290
    },
    {
      "epoch": 0.25043636639599304,
      "grad_norm": 0.0,
      "learning_rate": 7.496947159996413e-06,
      "loss": 0.0,
      "step": 36300
    },
    {
      "epoch": 0.25050535713053734,
      "grad_norm": 0.0,
      "learning_rate": 7.4962572526509694e-06,
      "loss": 0.001,
      "step": 36310
    },
    {
      "epoch": 0.2505743478650817,
      "grad_norm": 0.0,
      "learning_rate": 7.495567345305526e-06,
      "loss": 0.1296,
      "step": 36320
    },
    {
      "epoch": 0.25064333859962606,
      "grad_norm": 0.0,
      "learning_rate": 7.494877437960083e-06,
      "loss": 0.0,
      "step": 36330
    },
    {
      "epoch": 0.2507123293341704,
      "grad_norm": 0.0,
      "learning_rate": 7.49418753061464e-06,
      "loss": 0.0,
      "step": 36340
    },
    {
      "epoch": 0.2507813200687148,
      "grad_norm": 0.0,
      "learning_rate": 7.493497623269195e-06,
      "loss": 0.0,
      "step": 36350
    },
    {
      "epoch": 0.25085031080325915,
      "grad_norm": 0.0,
      "learning_rate": 7.492807715923752e-06,
      "loss": 0.0001,
      "step": 36360
    },
    {
      "epoch": 0.25091930153780345,
      "grad_norm": 0.0,
      "learning_rate": 7.492117808578308e-06,
      "loss": 0.0003,
      "step": 36370
    },
    {
      "epoch": 0.2509882922723478,
      "grad_norm": 0.0,
      "learning_rate": 7.491427901232865e-06,
      "loss": 0.0,
      "step": 36380
    },
    {
      "epoch": 0.2510572830068922,
      "grad_norm": 0.0,
      "learning_rate": 7.490737993887421e-06,
      "loss": 0.0,
      "step": 36390
    },
    {
      "epoch": 0.25112627374143653,
      "grad_norm": 0.0,
      "learning_rate": 7.490048086541978e-06,
      "loss": 0.0,
      "step": 36400
    },
    {
      "epoch": 0.2511952644759809,
      "grad_norm": 0.0,
      "learning_rate": 7.489358179196534e-06,
      "loss": 0.0,
      "step": 36410
    },
    {
      "epoch": 0.2512642552105252,
      "grad_norm": 56.675270080566406,
      "learning_rate": 7.488668271851091e-06,
      "loss": 0.0097,
      "step": 36420
    },
    {
      "epoch": 0.25133324594506956,
      "grad_norm": 0.0,
      "learning_rate": 7.487978364505647e-06,
      "loss": 0.0,
      "step": 36430
    },
    {
      "epoch": 0.2514022366796139,
      "grad_norm": 0.0,
      "learning_rate": 7.487288457160205e-06,
      "loss": 0.0017,
      "step": 36440
    },
    {
      "epoch": 0.2514712274141583,
      "grad_norm": 0.0,
      "learning_rate": 7.486598549814761e-06,
      "loss": 0.0007,
      "step": 36450
    },
    {
      "epoch": 0.25154021814870264,
      "grad_norm": 0.03382086753845215,
      "learning_rate": 7.485908642469318e-06,
      "loss": 0.1864,
      "step": 36460
    },
    {
      "epoch": 0.251609208883247,
      "grad_norm": 0.0015447885962203145,
      "learning_rate": 7.485218735123873e-06,
      "loss": 0.0,
      "step": 36470
    },
    {
      "epoch": 0.2516781996177913,
      "grad_norm": 0.0,
      "learning_rate": 7.48452882777843e-06,
      "loss": 0.0,
      "step": 36480
    },
    {
      "epoch": 0.25174719035233567,
      "grad_norm": 0.0,
      "learning_rate": 7.483838920432986e-06,
      "loss": 0.0,
      "step": 36490
    },
    {
      "epoch": 0.25181618108688003,
      "grad_norm": 0.0,
      "learning_rate": 7.483149013087543e-06,
      "loss": 0.0002,
      "step": 36500
    },
    {
      "epoch": 0.2518851718214244,
      "grad_norm": 0.0,
      "learning_rate": 7.482459105742099e-06,
      "loss": 0.0006,
      "step": 36510
    },
    {
      "epoch": 0.25195416255596875,
      "grad_norm": 0.0,
      "learning_rate": 7.481769198396656e-06,
      "loss": 0.0,
      "step": 36520
    },
    {
      "epoch": 0.25202315329051306,
      "grad_norm": 0.0,
      "learning_rate": 7.481079291051212e-06,
      "loss": 0.0,
      "step": 36530
    },
    {
      "epoch": 0.2520921440250574,
      "grad_norm": 0.0,
      "learning_rate": 7.480389383705769e-06,
      "loss": 0.0,
      "step": 36540
    },
    {
      "epoch": 0.2521611347596018,
      "grad_norm": 0.0004022032371722162,
      "learning_rate": 7.479699476360326e-06,
      "loss": 0.0333,
      "step": 36550
    },
    {
      "epoch": 0.25223012549414614,
      "grad_norm": 0.0,
      "learning_rate": 7.4790095690148826e-06,
      "loss": 0.0,
      "step": 36560
    },
    {
      "epoch": 0.2522991162286905,
      "grad_norm": 0.0,
      "learning_rate": 7.478319661669439e-06,
      "loss": 0.0,
      "step": 36570
    },
    {
      "epoch": 0.25236810696323486,
      "grad_norm": 5.063108332592492e-08,
      "learning_rate": 7.477629754323995e-06,
      "loss": 0.0001,
      "step": 36580
    },
    {
      "epoch": 0.25243709769777917,
      "grad_norm": 0.0,
      "learning_rate": 7.476939846978551e-06,
      "loss": 0.0,
      "step": 36590
    },
    {
      "epoch": 0.2525060884323235,
      "grad_norm": 0.00015263649402186275,
      "learning_rate": 7.476249939633108e-06,
      "loss": 0.0,
      "step": 36600
    },
    {
      "epoch": 0.2525750791668679,
      "grad_norm": 0.0,
      "learning_rate": 7.475560032287664e-06,
      "loss": 0.0109,
      "step": 36610
    },
    {
      "epoch": 0.25264406990141225,
      "grad_norm": 0.0,
      "learning_rate": 7.474870124942221e-06,
      "loss": 0.0,
      "step": 36620
    },
    {
      "epoch": 0.2527130606359566,
      "grad_norm": 0.0,
      "learning_rate": 7.474180217596777e-06,
      "loss": 0.0,
      "step": 36630
    },
    {
      "epoch": 0.2527820513705009,
      "grad_norm": 0.0,
      "learning_rate": 7.473490310251334e-06,
      "loss": 0.0,
      "step": 36640
    },
    {
      "epoch": 0.2528510421050453,
      "grad_norm": 0.0,
      "learning_rate": 7.47280040290589e-06,
      "loss": 0.001,
      "step": 36650
    },
    {
      "epoch": 0.25292003283958964,
      "grad_norm": 5.780433980362432e-07,
      "learning_rate": 7.4721104955604475e-06,
      "loss": 0.0,
      "step": 36660
    },
    {
      "epoch": 0.252989023574134,
      "grad_norm": 4.935407083017651e-10,
      "learning_rate": 7.471420588215004e-06,
      "loss": 0.0,
      "step": 36670
    },
    {
      "epoch": 0.25305801430867836,
      "grad_norm": 0.0,
      "learning_rate": 7.4707306808695604e-06,
      "loss": 0.0,
      "step": 36680
    },
    {
      "epoch": 0.2531270050432227,
      "grad_norm": 0.0,
      "learning_rate": 7.470040773524116e-06,
      "loss": 0.0,
      "step": 36690
    },
    {
      "epoch": 0.253195995777767,
      "grad_norm": 0.0,
      "learning_rate": 7.4693508661786726e-06,
      "loss": 0.0,
      "step": 36700
    },
    {
      "epoch": 0.2532649865123114,
      "grad_norm": 0.0,
      "learning_rate": 7.468660958833229e-06,
      "loss": 0.0,
      "step": 36710
    },
    {
      "epoch": 0.25333397724685575,
      "grad_norm": 0.08687032014131546,
      "learning_rate": 7.4679710514877856e-06,
      "loss": 0.0038,
      "step": 36720
    },
    {
      "epoch": 0.2534029679814001,
      "grad_norm": 9.497213113718317e-07,
      "learning_rate": 7.467281144142342e-06,
      "loss": 0.0,
      "step": 36730
    },
    {
      "epoch": 0.25347195871594447,
      "grad_norm": 11.538050651550293,
      "learning_rate": 7.4665912367968985e-06,
      "loss": 0.0027,
      "step": 36740
    },
    {
      "epoch": 0.2535409494504888,
      "grad_norm": 0.0,
      "learning_rate": 7.465901329451455e-06,
      "loss": 0.0,
      "step": 36750
    },
    {
      "epoch": 0.25360994018503313,
      "grad_norm": 1.7542208796683667e-09,
      "learning_rate": 7.4652114221060115e-06,
      "loss": 0.0,
      "step": 36760
    },
    {
      "epoch": 0.2536789309195775,
      "grad_norm": 0.0,
      "learning_rate": 7.464521514760569e-06,
      "loss": 0.0,
      "step": 36770
    },
    {
      "epoch": 0.25374792165412186,
      "grad_norm": 0.0061941323801875114,
      "learning_rate": 7.463831607415125e-06,
      "loss": 0.0,
      "step": 36780
    },
    {
      "epoch": 0.2538169123886662,
      "grad_norm": 0.0005129583296366036,
      "learning_rate": 7.463141700069682e-06,
      "loss": 0.0,
      "step": 36790
    },
    {
      "epoch": 0.2538859031232106,
      "grad_norm": 0.0,
      "learning_rate": 7.462451792724238e-06,
      "loss": 0.0,
      "step": 36800
    },
    {
      "epoch": 0.2539548938577549,
      "grad_norm": 0.0,
      "learning_rate": 7.461761885378794e-06,
      "loss": 0.0,
      "step": 36810
    },
    {
      "epoch": 0.25402388459229924,
      "grad_norm": 0.0,
      "learning_rate": 7.4610719780333505e-06,
      "loss": 0.0015,
      "step": 36820
    },
    {
      "epoch": 0.2540928753268436,
      "grad_norm": 4.33061864146822e-10,
      "learning_rate": 7.460382070687907e-06,
      "loss": 0.0,
      "step": 36830
    },
    {
      "epoch": 0.25416186606138796,
      "grad_norm": 0.0,
      "learning_rate": 7.4596921633424634e-06,
      "loss": 0.0,
      "step": 36840
    },
    {
      "epoch": 0.2542308567959323,
      "grad_norm": 0.0,
      "learning_rate": 7.45900225599702e-06,
      "loss": 0.0,
      "step": 36850
    },
    {
      "epoch": 0.25429984753047663,
      "grad_norm": 0.0,
      "learning_rate": 7.458312348651576e-06,
      "loss": 0.0,
      "step": 36860
    },
    {
      "epoch": 0.254368838265021,
      "grad_norm": 0.0,
      "learning_rate": 7.457622441306133e-06,
      "loss": 0.2332,
      "step": 36870
    },
    {
      "epoch": 0.25443782899956535,
      "grad_norm": 0.0,
      "learning_rate": 7.45693253396069e-06,
      "loss": 0.0002,
      "step": 36880
    },
    {
      "epoch": 0.2545068197341097,
      "grad_norm": 0.006251172628253698,
      "learning_rate": 7.456242626615247e-06,
      "loss": 0.191,
      "step": 36890
    },
    {
      "epoch": 0.2545758104686541,
      "grad_norm": 4.6053352797947866e-10,
      "learning_rate": 7.455552719269803e-06,
      "loss": 0.0,
      "step": 36900
    },
    {
      "epoch": 0.25464480120319843,
      "grad_norm": 0.010192696005105972,
      "learning_rate": 7.45486281192436e-06,
      "loss": 0.0,
      "step": 36910
    },
    {
      "epoch": 0.25471379193774274,
      "grad_norm": 4.805706943500354e-10,
      "learning_rate": 7.454172904578915e-06,
      "loss": 0.0,
      "step": 36920
    },
    {
      "epoch": 0.2547827826722871,
      "grad_norm": 1.4466934317169944e-06,
      "learning_rate": 7.453482997233472e-06,
      "loss": 0.0,
      "step": 36930
    },
    {
      "epoch": 0.25485177340683146,
      "grad_norm": 1.3884893235172058e-08,
      "learning_rate": 7.452793089888028e-06,
      "loss": 0.0,
      "step": 36940
    },
    {
      "epoch": 0.2549207641413758,
      "grad_norm": 0.0,
      "learning_rate": 7.452103182542585e-06,
      "loss": 0.0,
      "step": 36950
    },
    {
      "epoch": 0.2549897548759202,
      "grad_norm": 5.604390906910339e-08,
      "learning_rate": 7.451413275197141e-06,
      "loss": 0.0,
      "step": 36960
    },
    {
      "epoch": 0.2550587456104645,
      "grad_norm": 0.0015580153558403254,
      "learning_rate": 7.450723367851698e-06,
      "loss": 0.0,
      "step": 36970
    },
    {
      "epoch": 0.25512773634500885,
      "grad_norm": 0.0,
      "learning_rate": 7.450033460506254e-06,
      "loss": 0.0,
      "step": 36980
    },
    {
      "epoch": 0.2551967270795532,
      "grad_norm": 0.0,
      "learning_rate": 7.449343553160812e-06,
      "loss": 0.0,
      "step": 36990
    },
    {
      "epoch": 0.25526571781409757,
      "grad_norm": 0.0,
      "learning_rate": 7.448653645815368e-06,
      "loss": 0.7023,
      "step": 37000
    },
    {
      "epoch": 0.25533470854864193,
      "grad_norm": 0.0,
      "learning_rate": 7.447963738469925e-06,
      "loss": 0.0,
      "step": 37010
    },
    {
      "epoch": 0.2554036992831863,
      "grad_norm": 0.0,
      "learning_rate": 7.447273831124481e-06,
      "loss": 0.0,
      "step": 37020
    },
    {
      "epoch": 0.2554726900177306,
      "grad_norm": 0.0,
      "learning_rate": 7.446583923779037e-06,
      "loss": 0.0,
      "step": 37030
    },
    {
      "epoch": 0.25554168075227496,
      "grad_norm": 4.5429510153738306e-10,
      "learning_rate": 7.445894016433593e-06,
      "loss": 0.0,
      "step": 37040
    },
    {
      "epoch": 0.2556106714868193,
      "grad_norm": 0.0,
      "learning_rate": 7.44520410908815e-06,
      "loss": 0.0,
      "step": 37050
    },
    {
      "epoch": 0.2556796622213637,
      "grad_norm": 0.0,
      "learning_rate": 7.444514201742706e-06,
      "loss": 0.0,
      "step": 37060
    },
    {
      "epoch": 0.25574865295590804,
      "grad_norm": 3.456605099927401e-06,
      "learning_rate": 7.443824294397263e-06,
      "loss": 0.0,
      "step": 37070
    },
    {
      "epoch": 0.25581764369045235,
      "grad_norm": 5.273923761706101e-06,
      "learning_rate": 7.443134387051819e-06,
      "loss": 0.0,
      "step": 37080
    },
    {
      "epoch": 0.2558866344249967,
      "grad_norm": 0.0,
      "learning_rate": 7.442444479706376e-06,
      "loss": 0.0,
      "step": 37090
    },
    {
      "epoch": 0.25595562515954107,
      "grad_norm": 0.0,
      "learning_rate": 7.441754572360933e-06,
      "loss": 0.0,
      "step": 37100
    },
    {
      "epoch": 0.2560246158940854,
      "grad_norm": 0.0,
      "learning_rate": 7.4410646650154895e-06,
      "loss": 0.0,
      "step": 37110
    },
    {
      "epoch": 0.2560936066286298,
      "grad_norm": 1.1108824082839419e-07,
      "learning_rate": 7.440374757670046e-06,
      "loss": 0.0,
      "step": 37120
    },
    {
      "epoch": 0.25616259736317415,
      "grad_norm": 0.0,
      "learning_rate": 7.4396848503246025e-06,
      "loss": 0.0,
      "step": 37130
    },
    {
      "epoch": 0.25623158809771845,
      "grad_norm": 4.6528797481570905e-10,
      "learning_rate": 7.438994942979159e-06,
      "loss": 0.0,
      "step": 37140
    },
    {
      "epoch": 0.2563005788322628,
      "grad_norm": 5.158657945258938e-09,
      "learning_rate": 7.438305035633715e-06,
      "loss": 0.2709,
      "step": 37150
    },
    {
      "epoch": 0.2563695695668072,
      "grad_norm": 0.0,
      "learning_rate": 7.437615128288271e-06,
      "loss": 0.0,
      "step": 37160
    },
    {
      "epoch": 0.25643856030135154,
      "grad_norm": 0.016699325293302536,
      "learning_rate": 7.436925220942828e-06,
      "loss": 0.0,
      "step": 37170
    },
    {
      "epoch": 0.2565075510358959,
      "grad_norm": 0.0,
      "learning_rate": 7.436235313597384e-06,
      "loss": 0.0,
      "step": 37180
    },
    {
      "epoch": 0.2565765417704402,
      "grad_norm": 0.0,
      "learning_rate": 7.435545406251941e-06,
      "loss": 0.0,
      "step": 37190
    },
    {
      "epoch": 0.25664553250498456,
      "grad_norm": 0.0,
      "learning_rate": 7.434855498906497e-06,
      "loss": 0.0,
      "step": 37200
    },
    {
      "epoch": 0.2567145232395289,
      "grad_norm": 0.809956431388855,
      "learning_rate": 7.4341655915610544e-06,
      "loss": 0.0001,
      "step": 37210
    },
    {
      "epoch": 0.2567835139740733,
      "grad_norm": 0.0,
      "learning_rate": 7.433475684215611e-06,
      "loss": 0.0031,
      "step": 37220
    },
    {
      "epoch": 0.25685250470861765,
      "grad_norm": 0.00017121054406743497,
      "learning_rate": 7.432785776870167e-06,
      "loss": 0.0,
      "step": 37230
    },
    {
      "epoch": 0.256921495443162,
      "grad_norm": 0.0,
      "learning_rate": 7.432095869524724e-06,
      "loss": 0.0,
      "step": 37240
    },
    {
      "epoch": 0.2569904861777063,
      "grad_norm": 0.0,
      "learning_rate": 7.43140596217928e-06,
      "loss": 0.0001,
      "step": 37250
    },
    {
      "epoch": 0.2570594769122507,
      "grad_norm": 0.0,
      "learning_rate": 7.430716054833836e-06,
      "loss": 0.0,
      "step": 37260
    },
    {
      "epoch": 0.25712846764679503,
      "grad_norm": 0.0,
      "learning_rate": 7.4300261474883925e-06,
      "loss": 0.0,
      "step": 37270
    },
    {
      "epoch": 0.2571974583813394,
      "grad_norm": 0.0,
      "learning_rate": 7.429336240142949e-06,
      "loss": 0.0,
      "step": 37280
    },
    {
      "epoch": 0.25726644911588376,
      "grad_norm": 0.0,
      "learning_rate": 7.4286463327975055e-06,
      "loss": 0.0007,
      "step": 37290
    },
    {
      "epoch": 0.25733543985042806,
      "grad_norm": 4.518082019622227e-10,
      "learning_rate": 7.427956425452062e-06,
      "loss": 0.0,
      "step": 37300
    },
    {
      "epoch": 0.2574044305849724,
      "grad_norm": 1.021700143814087,
      "learning_rate": 7.4272665181066185e-06,
      "loss": 0.0002,
      "step": 37310
    },
    {
      "epoch": 0.2574734213195168,
      "grad_norm": 0.0,
      "learning_rate": 7.426576610761176e-06,
      "loss": 0.001,
      "step": 37320
    },
    {
      "epoch": 0.25754241205406114,
      "grad_norm": 0.0,
      "learning_rate": 7.425886703415732e-06,
      "loss": 0.0,
      "step": 37330
    },
    {
      "epoch": 0.2576114027886055,
      "grad_norm": 0.0,
      "learning_rate": 7.425196796070289e-06,
      "loss": 0.1441,
      "step": 37340
    },
    {
      "epoch": 0.25768039352314986,
      "grad_norm": 9.680126822786406e-05,
      "learning_rate": 7.424506888724845e-06,
      "loss": 0.0,
      "step": 37350
    },
    {
      "epoch": 0.25774938425769417,
      "grad_norm": 0.0,
      "learning_rate": 7.423816981379402e-06,
      "loss": 0.0,
      "step": 37360
    },
    {
      "epoch": 0.25781837499223853,
      "grad_norm": 0.0,
      "learning_rate": 7.4231270740339574e-06,
      "loss": 0.0,
      "step": 37370
    },
    {
      "epoch": 0.2578873657267829,
      "grad_norm": 0.0,
      "learning_rate": 7.422437166688514e-06,
      "loss": 0.003,
      "step": 37380
    },
    {
      "epoch": 0.25795635646132725,
      "grad_norm": 2.1739184064006878e-10,
      "learning_rate": 7.42174725934307e-06,
      "loss": 0.0036,
      "step": 37390
    },
    {
      "epoch": 0.2580253471958716,
      "grad_norm": 0.0,
      "learning_rate": 7.421057351997627e-06,
      "loss": 0.0,
      "step": 37400
    },
    {
      "epoch": 0.2580943379304159,
      "grad_norm": 0.0003934430715162307,
      "learning_rate": 7.420367444652183e-06,
      "loss": 0.0104,
      "step": 37410
    },
    {
      "epoch": 0.2581633286649603,
      "grad_norm": 0.0,
      "learning_rate": 7.41967753730674e-06,
      "loss": 0.0,
      "step": 37420
    },
    {
      "epoch": 0.25823231939950464,
      "grad_norm": 0.0,
      "learning_rate": 7.419056620695841e-06,
      "loss": 0.1169,
      "step": 37430
    },
    {
      "epoch": 0.258301310134049,
      "grad_norm": 0.0,
      "learning_rate": 7.418366713350398e-06,
      "loss": 0.0178,
      "step": 37440
    },
    {
      "epoch": 0.25837030086859336,
      "grad_norm": 0.0,
      "learning_rate": 7.417676806004954e-06,
      "loss": 0.0009,
      "step": 37450
    },
    {
      "epoch": 0.2584392916031377,
      "grad_norm": 0.0,
      "learning_rate": 7.416986898659511e-06,
      "loss": 0.0,
      "step": 37460
    },
    {
      "epoch": 0.258508282337682,
      "grad_norm": 915.4453735351562,
      "learning_rate": 7.416296991314067e-06,
      "loss": 0.267,
      "step": 37470
    },
    {
      "epoch": 0.2585772730722264,
      "grad_norm": 1.6182926856345148e-06,
      "learning_rate": 7.415607083968623e-06,
      "loss": 0.0,
      "step": 37480
    },
    {
      "epoch": 0.25864626380677075,
      "grad_norm": 0.0,
      "learning_rate": 7.4149171766231795e-06,
      "loss": 0.0002,
      "step": 37490
    },
    {
      "epoch": 0.2587152545413151,
      "grad_norm": 0.0,
      "learning_rate": 7.414227269277736e-06,
      "loss": 0.0,
      "step": 37500
    },
    {
      "epoch": 0.25878424527585947,
      "grad_norm": 0.0,
      "learning_rate": 7.4135373619322925e-06,
      "loss": 0.0,
      "step": 37510
    },
    {
      "epoch": 0.2588532360104038,
      "grad_norm": 1062.3094482421875,
      "learning_rate": 7.41284745458685e-06,
      "loss": 0.7446,
      "step": 37520
    },
    {
      "epoch": 0.25892222674494814,
      "grad_norm": 0.0,
      "learning_rate": 7.412157547241406e-06,
      "loss": 0.0,
      "step": 37530
    },
    {
      "epoch": 0.2589912174794925,
      "grad_norm": 0.0003129865217488259,
      "learning_rate": 7.411467639895963e-06,
      "loss": 0.0093,
      "step": 37540
    },
    {
      "epoch": 0.25906020821403686,
      "grad_norm": 5.7864535847329535e-06,
      "learning_rate": 7.410777732550519e-06,
      "loss": 0.0001,
      "step": 37550
    },
    {
      "epoch": 0.2591291989485812,
      "grad_norm": 0.0,
      "learning_rate": 7.410087825205076e-06,
      "loss": 0.0,
      "step": 37560
    },
    {
      "epoch": 0.2591981896831256,
      "grad_norm": 9.169200954062262e-10,
      "learning_rate": 7.409397917859632e-06,
      "loss": 0.0,
      "step": 37570
    },
    {
      "epoch": 0.2592671804176699,
      "grad_norm": 4.127901411266066e-05,
      "learning_rate": 7.408708010514189e-06,
      "loss": 0.0,
      "step": 37580
    },
    {
      "epoch": 0.25933617115221425,
      "grad_norm": 1.1614266633987427,
      "learning_rate": 7.408018103168744e-06,
      "loss": 0.0001,
      "step": 37590
    },
    {
      "epoch": 0.2594051618867586,
      "grad_norm": 7.775157428113744e-05,
      "learning_rate": 7.407328195823301e-06,
      "loss": 0.0671,
      "step": 37600
    },
    {
      "epoch": 0.25947415262130297,
      "grad_norm": 0.0,
      "learning_rate": 7.406638288477857e-06,
      "loss": 0.0,
      "step": 37610
    },
    {
      "epoch": 0.25954314335584733,
      "grad_norm": 0.0,
      "learning_rate": 7.405948381132414e-06,
      "loss": 0.0,
      "step": 37620
    },
    {
      "epoch": 0.25961213409039163,
      "grad_norm": 0.0,
      "learning_rate": 7.405258473786971e-06,
      "loss": 0.0,
      "step": 37630
    },
    {
      "epoch": 0.259681124824936,
      "grad_norm": 0.003942232578992844,
      "learning_rate": 7.404568566441528e-06,
      "loss": 0.0,
      "step": 37640
    },
    {
      "epoch": 0.25975011555948035,
      "grad_norm": 0.0,
      "learning_rate": 7.403878659096084e-06,
      "loss": 0.0,
      "step": 37650
    },
    {
      "epoch": 0.2598191062940247,
      "grad_norm": 0.0,
      "learning_rate": 7.403188751750641e-06,
      "loss": 0.0002,
      "step": 37660
    },
    {
      "epoch": 0.2598880970285691,
      "grad_norm": 4.735766778729555e-10,
      "learning_rate": 7.402498844405197e-06,
      "loss": 0.0,
      "step": 37670
    },
    {
      "epoch": 0.25995708776311344,
      "grad_norm": 5.6813117765841525e-08,
      "learning_rate": 7.401808937059754e-06,
      "loss": 0.0,
      "step": 37680
    },
    {
      "epoch": 0.26002607849765774,
      "grad_norm": 0.0026334025897085667,
      "learning_rate": 7.40111902971431e-06,
      "loss": 0.0,
      "step": 37690
    },
    {
      "epoch": 0.2600950692322021,
      "grad_norm": 0.0,
      "learning_rate": 7.400429122368867e-06,
      "loss": 0.0,
      "step": 37700
    },
    {
      "epoch": 0.26016405996674646,
      "grad_norm": 0.0027832696214318275,
      "learning_rate": 7.399739215023422e-06,
      "loss": 0.0,
      "step": 37710
    },
    {
      "epoch": 0.2602330507012908,
      "grad_norm": 0.0,
      "learning_rate": 7.399049307677979e-06,
      "loss": 0.0001,
      "step": 37720
    },
    {
      "epoch": 0.2603020414358352,
      "grad_norm": 0.0,
      "learning_rate": 7.398359400332535e-06,
      "loss": 0.0,
      "step": 37730
    },
    {
      "epoch": 0.2603710321703795,
      "grad_norm": 0.0,
      "learning_rate": 7.397669492987093e-06,
      "loss": 0.0,
      "step": 37740
    },
    {
      "epoch": 0.26044002290492385,
      "grad_norm": 1.7929034584085457e-05,
      "learning_rate": 7.396979585641649e-06,
      "loss": 0.0,
      "step": 37750
    },
    {
      "epoch": 0.2605090136394682,
      "grad_norm": 4.2693945601079974e-10,
      "learning_rate": 7.3962896782962056e-06,
      "loss": 0.0,
      "step": 37760
    },
    {
      "epoch": 0.2605780043740126,
      "grad_norm": 0.0006262511597014964,
      "learning_rate": 7.395599770950762e-06,
      "loss": 0.0013,
      "step": 37770
    },
    {
      "epoch": 0.26064699510855693,
      "grad_norm": 0.0,
      "learning_rate": 7.3949098636053186e-06,
      "loss": 0.0,
      "step": 37780
    },
    {
      "epoch": 0.2607159858431013,
      "grad_norm": 0.0,
      "learning_rate": 7.394219956259875e-06,
      "loss": 0.0,
      "step": 37790
    },
    {
      "epoch": 0.2607849765776456,
      "grad_norm": 9.403579026567854e-10,
      "learning_rate": 7.3935300489144315e-06,
      "loss": 0.0003,
      "step": 37800
    },
    {
      "epoch": 0.26085396731218996,
      "grad_norm": 0.0,
      "learning_rate": 7.392840141568988e-06,
      "loss": 0.0,
      "step": 37810
    },
    {
      "epoch": 0.2609229580467343,
      "grad_norm": 0.0,
      "learning_rate": 7.392150234223544e-06,
      "loss": 0.0037,
      "step": 37820
    },
    {
      "epoch": 0.2609919487812787,
      "grad_norm": 0.0,
      "learning_rate": 7.3914603268781e-06,
      "loss": 0.0004,
      "step": 37830
    },
    {
      "epoch": 0.26106093951582304,
      "grad_norm": 0.049956776201725006,
      "learning_rate": 7.390770419532657e-06,
      "loss": 0.0,
      "step": 37840
    },
    {
      "epoch": 0.26112993025036735,
      "grad_norm": 2.764890538742293e-09,
      "learning_rate": 7.390080512187214e-06,
      "loss": 0.0001,
      "step": 37850
    },
    {
      "epoch": 0.2611989209849117,
      "grad_norm": 0.0,
      "learning_rate": 7.3893906048417705e-06,
      "loss": 0.0,
      "step": 37860
    },
    {
      "epoch": 0.26126791171945607,
      "grad_norm": 4.611935764842201e-06,
      "learning_rate": 7.388700697496327e-06,
      "loss": 0.0,
      "step": 37870
    },
    {
      "epoch": 0.26133690245400043,
      "grad_norm": 0.0,
      "learning_rate": 7.3880107901508835e-06,
      "loss": 0.0,
      "step": 37880
    },
    {
      "epoch": 0.2614058931885448,
      "grad_norm": 0.0,
      "learning_rate": 7.38732088280544e-06,
      "loss": 0.0,
      "step": 37890
    },
    {
      "epoch": 0.26147488392308915,
      "grad_norm": 0.0,
      "learning_rate": 7.3866309754599964e-06,
      "loss": 0.0009,
      "step": 37900
    },
    {
      "epoch": 0.26154387465763346,
      "grad_norm": 0.0,
      "learning_rate": 7.385941068114553e-06,
      "loss": 0.616,
      "step": 37910
    },
    {
      "epoch": 0.2616128653921778,
      "grad_norm": 0.0,
      "learning_rate": 7.385251160769109e-06,
      "loss": 0.0,
      "step": 37920
    },
    {
      "epoch": 0.2616818561267222,
      "grad_norm": 0.0,
      "learning_rate": 7.384561253423665e-06,
      "loss": 0.0,
      "step": 37930
    },
    {
      "epoch": 0.26175084686126654,
      "grad_norm": 0.0,
      "learning_rate": 7.3838713460782216e-06,
      "loss": 0.0,
      "step": 37940
    },
    {
      "epoch": 0.2618198375958109,
      "grad_norm": 1.8060086759419391e-09,
      "learning_rate": 7.383250429467324e-06,
      "loss": 0.9157,
      "step": 37950
    },
    {
      "epoch": 0.2618888283303552,
      "grad_norm": 0.0,
      "learning_rate": 7.3825605221218796e-06,
      "loss": 0.0,
      "step": 37960
    },
    {
      "epoch": 0.26195781906489957,
      "grad_norm": 0.00021310083684511483,
      "learning_rate": 7.381870614776436e-06,
      "loss": 0.0,
      "step": 37970
    },
    {
      "epoch": 0.2620268097994439,
      "grad_norm": 0.0029894940089434385,
      "learning_rate": 7.3811807074309925e-06,
      "loss": 0.0,
      "step": 37980
    },
    {
      "epoch": 0.2620958005339883,
      "grad_norm": 0.0,
      "learning_rate": 7.380490800085549e-06,
      "loss": 0.0358,
      "step": 37990
    },
    {
      "epoch": 0.26216479126853265,
      "grad_norm": 0.0005743624060414732,
      "learning_rate": 7.3798008927401055e-06,
      "loss": 0.0,
      "step": 38000
    },
    {
      "epoch": 0.262233782003077,
      "grad_norm": 1.6207940234380658e-07,
      "learning_rate": 7.379110985394662e-06,
      "loss": 0.0,
      "step": 38010
    },
    {
      "epoch": 0.2623027727376213,
      "grad_norm": 0.0,
      "learning_rate": 7.3784210780492185e-06,
      "loss": 0.0,
      "step": 38020
    },
    {
      "epoch": 0.2623717634721657,
      "grad_norm": 8.575650215148926,
      "learning_rate": 7.377731170703775e-06,
      "loss": 0.0015,
      "step": 38030
    },
    {
      "epoch": 0.26244075420671004,
      "grad_norm": 0.0,
      "learning_rate": 7.377041263358331e-06,
      "loss": 0.0,
      "step": 38040
    },
    {
      "epoch": 0.2625097449412544,
      "grad_norm": 0.0,
      "learning_rate": 7.376351356012889e-06,
      "loss": 0.0,
      "step": 38050
    },
    {
      "epoch": 0.26257873567579876,
      "grad_norm": 0.0,
      "learning_rate": 7.375661448667445e-06,
      "loss": 0.0,
      "step": 38060
    },
    {
      "epoch": 0.2626477264103431,
      "grad_norm": 0.0,
      "learning_rate": 7.374971541322001e-06,
      "loss": 0.0001,
      "step": 38070
    },
    {
      "epoch": 0.2627167171448874,
      "grad_norm": 8.552069630241022e-08,
      "learning_rate": 7.3742816339765574e-06,
      "loss": 0.0001,
      "step": 38080
    },
    {
      "epoch": 0.2627857078794318,
      "grad_norm": 0.0,
      "learning_rate": 7.373591726631114e-06,
      "loss": 0.0,
      "step": 38090
    },
    {
      "epoch": 0.26285469861397615,
      "grad_norm": 0.0,
      "learning_rate": 7.3729018192856704e-06,
      "loss": 0.0,
      "step": 38100
    },
    {
      "epoch": 0.2629236893485205,
      "grad_norm": 0.12948361039161682,
      "learning_rate": 7.372211911940227e-06,
      "loss": 0.0001,
      "step": 38110
    },
    {
      "epoch": 0.26299268008306487,
      "grad_norm": 0.01798807829618454,
      "learning_rate": 7.371522004594783e-06,
      "loss": 0.0,
      "step": 38120
    },
    {
      "epoch": 0.2630616708176092,
      "grad_norm": 0.0,
      "learning_rate": 7.37083209724934e-06,
      "loss": 0.0,
      "step": 38130
    },
    {
      "epoch": 0.26313066155215353,
      "grad_norm": 0.0,
      "learning_rate": 7.370142189903896e-06,
      "loss": 0.0,
      "step": 38140
    },
    {
      "epoch": 0.2631996522866979,
      "grad_norm": 0.0,
      "learning_rate": 7.369452282558453e-06,
      "loss": 0.0,
      "step": 38150
    },
    {
      "epoch": 0.26326864302124225,
      "grad_norm": 0.0,
      "learning_rate": 7.36876237521301e-06,
      "loss": 0.0001,
      "step": 38160
    },
    {
      "epoch": 0.2633376337557866,
      "grad_norm": 0.0,
      "learning_rate": 7.368072467867567e-06,
      "loss": 0.0,
      "step": 38170
    },
    {
      "epoch": 0.263406624490331,
      "grad_norm": 0.0,
      "learning_rate": 7.367382560522122e-06,
      "loss": 0.0002,
      "step": 38180
    },
    {
      "epoch": 0.2634756152248753,
      "grad_norm": 0.0,
      "learning_rate": 7.366692653176679e-06,
      "loss": 0.0,
      "step": 38190
    },
    {
      "epoch": 0.26354460595941964,
      "grad_norm": 0.0,
      "learning_rate": 7.366002745831235e-06,
      "loss": 0.0,
      "step": 38200
    },
    {
      "epoch": 0.263613596693964,
      "grad_norm": 0.0,
      "learning_rate": 7.365312838485792e-06,
      "loss": 0.0,
      "step": 38210
    },
    {
      "epoch": 0.26368258742850836,
      "grad_norm": 0.0,
      "learning_rate": 7.364622931140348e-06,
      "loss": 0.0,
      "step": 38220
    },
    {
      "epoch": 0.2637515781630527,
      "grad_norm": 0.0,
      "learning_rate": 7.363933023794905e-06,
      "loss": 0.0006,
      "step": 38230
    },
    {
      "epoch": 0.26382056889759703,
      "grad_norm": 0.0,
      "learning_rate": 7.363243116449461e-06,
      "loss": 0.0,
      "step": 38240
    },
    {
      "epoch": 0.2638895596321414,
      "grad_norm": 0.0,
      "learning_rate": 7.362553209104018e-06,
      "loss": 0.0,
      "step": 38250
    },
    {
      "epoch": 0.26395855036668575,
      "grad_norm": 0.0,
      "learning_rate": 7.361863301758574e-06,
      "loss": 0.0001,
      "step": 38260
    },
    {
      "epoch": 0.2640275411012301,
      "grad_norm": 0.0,
      "learning_rate": 7.361173394413132e-06,
      "loss": 0.0563,
      "step": 38270
    },
    {
      "epoch": 0.2640965318357745,
      "grad_norm": 0.0,
      "learning_rate": 7.360483487067688e-06,
      "loss": 0.0002,
      "step": 38280
    },
    {
      "epoch": 0.26416552257031883,
      "grad_norm": 0.0,
      "learning_rate": 7.359793579722245e-06,
      "loss": 0.0,
      "step": 38290
    },
    {
      "epoch": 0.26423451330486314,
      "grad_norm": 0.0,
      "learning_rate": 7.3591036723768e-06,
      "loss": 0.0,
      "step": 38300
    },
    {
      "epoch": 0.2643035040394075,
      "grad_norm": 0.0,
      "learning_rate": 7.358413765031357e-06,
      "loss": 0.0,
      "step": 38310
    },
    {
      "epoch": 0.26437249477395186,
      "grad_norm": 0.0,
      "learning_rate": 7.357723857685913e-06,
      "loss": 0.0,
      "step": 38320
    },
    {
      "epoch": 0.2644414855084962,
      "grad_norm": 0.0,
      "learning_rate": 7.35703395034047e-06,
      "loss": 0.0,
      "step": 38330
    },
    {
      "epoch": 0.2645104762430406,
      "grad_norm": 0.0,
      "learning_rate": 7.356344042995026e-06,
      "loss": 0.0,
      "step": 38340
    },
    {
      "epoch": 0.2645794669775849,
      "grad_norm": 0.0,
      "learning_rate": 7.355654135649583e-06,
      "loss": 0.0,
      "step": 38350
    },
    {
      "epoch": 0.26464845771212925,
      "grad_norm": 1.5695299282469932e-07,
      "learning_rate": 7.354964228304139e-06,
      "loss": 0.0,
      "step": 38360
    },
    {
      "epoch": 0.2647174484466736,
      "grad_norm": 0.0,
      "learning_rate": 7.354274320958696e-06,
      "loss": 0.0,
      "step": 38370
    },
    {
      "epoch": 0.26478643918121797,
      "grad_norm": 0.0,
      "learning_rate": 7.353584413613253e-06,
      "loss": 0.0239,
      "step": 38380
    },
    {
      "epoch": 0.26485542991576233,
      "grad_norm": 0.0,
      "learning_rate": 7.3528945062678095e-06,
      "loss": 0.0,
      "step": 38390
    },
    {
      "epoch": 0.2649244206503067,
      "grad_norm": 2.2398413420887664e-05,
      "learning_rate": 7.352204598922366e-06,
      "loss": 0.2436,
      "step": 38400
    },
    {
      "epoch": 0.264993411384851,
      "grad_norm": 0.0,
      "learning_rate": 7.351514691576922e-06,
      "loss": 0.0,
      "step": 38410
    },
    {
      "epoch": 0.26506240211939536,
      "grad_norm": 0.0,
      "learning_rate": 7.350824784231478e-06,
      "loss": 0.0004,
      "step": 38420
    },
    {
      "epoch": 0.2651313928539397,
      "grad_norm": 0.0,
      "learning_rate": 7.350134876886035e-06,
      "loss": 0.0,
      "step": 38430
    },
    {
      "epoch": 0.2652003835884841,
      "grad_norm": 0.0,
      "learning_rate": 7.349444969540591e-06,
      "loss": 0.0,
      "step": 38440
    },
    {
      "epoch": 0.26526937432302844,
      "grad_norm": 0.0,
      "learning_rate": 7.348755062195148e-06,
      "loss": 0.0,
      "step": 38450
    },
    {
      "epoch": 0.26533836505757274,
      "grad_norm": 0.001017861533910036,
      "learning_rate": 7.348065154849704e-06,
      "loss": 0.0,
      "step": 38460
    },
    {
      "epoch": 0.2654073557921171,
      "grad_norm": 0.6606370210647583,
      "learning_rate": 7.3473752475042606e-06,
      "loss": 0.0001,
      "step": 38470
    },
    {
      "epoch": 0.26547634652666147,
      "grad_norm": 0.0009400597773492336,
      "learning_rate": 7.346685340158817e-06,
      "loss": 0.0,
      "step": 38480
    },
    {
      "epoch": 0.2655453372612058,
      "grad_norm": 0.0,
      "learning_rate": 7.345995432813374e-06,
      "loss": 0.0,
      "step": 38490
    },
    {
      "epoch": 0.2656143279957502,
      "grad_norm": 1.123922954704426e-09,
      "learning_rate": 7.345305525467931e-06,
      "loss": 0.0,
      "step": 38500
    },
    {
      "epoch": 0.26568331873029455,
      "grad_norm": 0.0,
      "learning_rate": 7.344615618122487e-06,
      "loss": 0.0,
      "step": 38510
    },
    {
      "epoch": 0.26575230946483885,
      "grad_norm": 0.0,
      "learning_rate": 7.343925710777043e-06,
      "loss": 0.0,
      "step": 38520
    },
    {
      "epoch": 0.2658213001993832,
      "grad_norm": 0.07188908755779266,
      "learning_rate": 7.3432358034315995e-06,
      "loss": 0.0028,
      "step": 38530
    },
    {
      "epoch": 0.2658902909339276,
      "grad_norm": 0.0,
      "learning_rate": 7.342545896086156e-06,
      "loss": 0.0014,
      "step": 38540
    },
    {
      "epoch": 0.26595928166847194,
      "grad_norm": 0.0,
      "learning_rate": 7.3418559887407125e-06,
      "loss": 0.0,
      "step": 38550
    },
    {
      "epoch": 0.2660282724030163,
      "grad_norm": 0.0,
      "learning_rate": 7.341166081395269e-06,
      "loss": 0.0,
      "step": 38560
    },
    {
      "epoch": 0.2660972631375606,
      "grad_norm": 0.0,
      "learning_rate": 7.3404761740498255e-06,
      "loss": 0.0,
      "step": 38570
    },
    {
      "epoch": 0.26616625387210496,
      "grad_norm": 0.0,
      "learning_rate": 7.339786266704382e-06,
      "loss": 0.0,
      "step": 38580
    },
    {
      "epoch": 0.2662352446066493,
      "grad_norm": 0.0,
      "learning_rate": 7.3390963593589385e-06,
      "loss": 0.1409,
      "step": 38590
    },
    {
      "epoch": 0.2663042353411937,
      "grad_norm": 0.0,
      "learning_rate": 7.338406452013496e-06,
      "loss": 0.0,
      "step": 38600
    },
    {
      "epoch": 0.26637322607573805,
      "grad_norm": 1.7232873688044492e-06,
      "learning_rate": 7.337716544668052e-06,
      "loss": 0.1604,
      "step": 38610
    },
    {
      "epoch": 0.2664422168102824,
      "grad_norm": 1.96116634469945e-05,
      "learning_rate": 7.337026637322609e-06,
      "loss": 0.0,
      "step": 38620
    },
    {
      "epoch": 0.2665112075448267,
      "grad_norm": 0.0,
      "learning_rate": 7.336336729977164e-06,
      "loss": 0.0,
      "step": 38630
    },
    {
      "epoch": 0.2665801982793711,
      "grad_norm": 0.0,
      "learning_rate": 7.335646822631721e-06,
      "loss": 0.0053,
      "step": 38640
    },
    {
      "epoch": 0.26664918901391543,
      "grad_norm": 0.0,
      "learning_rate": 7.334956915286277e-06,
      "loss": 0.0,
      "step": 38650
    },
    {
      "epoch": 0.2667181797484598,
      "grad_norm": 7.976258542541359e-10,
      "learning_rate": 7.334267007940834e-06,
      "loss": 0.0,
      "step": 38660
    },
    {
      "epoch": 0.26678717048300415,
      "grad_norm": 0.0,
      "learning_rate": 7.33357710059539e-06,
      "loss": 0.0,
      "step": 38670
    },
    {
      "epoch": 0.26685616121754846,
      "grad_norm": 0.0,
      "learning_rate": 7.332887193249947e-06,
      "loss": 0.0,
      "step": 38680
    },
    {
      "epoch": 0.2669251519520928,
      "grad_norm": 3.788072522326047e-09,
      "learning_rate": 7.332197285904503e-06,
      "loss": 0.0009,
      "step": 38690
    },
    {
      "epoch": 0.2669941426866372,
      "grad_norm": 0.0,
      "learning_rate": 7.33150737855906e-06,
      "loss": 0.0,
      "step": 38700
    },
    {
      "epoch": 0.26706313342118154,
      "grad_norm": 0.0,
      "learning_rate": 7.330817471213617e-06,
      "loss": 0.0,
      "step": 38710
    },
    {
      "epoch": 0.2671321241557259,
      "grad_norm": 8.979275101239637e-10,
      "learning_rate": 7.330127563868174e-06,
      "loss": 0.0,
      "step": 38720
    },
    {
      "epoch": 0.26720111489027026,
      "grad_norm": 0.0,
      "learning_rate": 7.32943765652273e-06,
      "loss": 0.0001,
      "step": 38730
    },
    {
      "epoch": 0.26727010562481457,
      "grad_norm": 2.423665250717022e-07,
      "learning_rate": 7.328747749177287e-06,
      "loss": 0.0,
      "step": 38740
    },
    {
      "epoch": 0.26733909635935893,
      "grad_norm": 0.0,
      "learning_rate": 7.328057841831842e-06,
      "loss": 0.0,
      "step": 38750
    },
    {
      "epoch": 0.2674080870939033,
      "grad_norm": 0.02464974857866764,
      "learning_rate": 7.327367934486399e-06,
      "loss": 0.0,
      "step": 38760
    },
    {
      "epoch": 0.26747707782844765,
      "grad_norm": 0.0,
      "learning_rate": 7.326678027140955e-06,
      "loss": 0.0,
      "step": 38770
    },
    {
      "epoch": 0.267546068562992,
      "grad_norm": 0.0,
      "learning_rate": 7.325988119795512e-06,
      "loss": 0.0,
      "step": 38780
    },
    {
      "epoch": 0.2676150592975363,
      "grad_norm": 0.0,
      "learning_rate": 7.325298212450068e-06,
      "loss": 0.0,
      "step": 38790
    },
    {
      "epoch": 0.2676840500320807,
      "grad_norm": 0.0,
      "learning_rate": 7.324608305104625e-06,
      "loss": 0.0006,
      "step": 38800
    },
    {
      "epoch": 0.26775304076662504,
      "grad_norm": 0.0,
      "learning_rate": 7.323918397759181e-06,
      "loss": 0.0,
      "step": 38810
    },
    {
      "epoch": 0.2678220315011694,
      "grad_norm": 0.0,
      "learning_rate": 7.323228490413739e-06,
      "loss": 0.0,
      "step": 38820
    },
    {
      "epoch": 0.26789102223571376,
      "grad_norm": 0.001431636861525476,
      "learning_rate": 7.322538583068295e-06,
      "loss": 0.0,
      "step": 38830
    },
    {
      "epoch": 0.2679600129702581,
      "grad_norm": 0.0,
      "learning_rate": 7.3218486757228516e-06,
      "loss": 0.0026,
      "step": 38840
    },
    {
      "epoch": 0.2680290037048024,
      "grad_norm": 0.9360167384147644,
      "learning_rate": 7.321158768377408e-06,
      "loss": 0.0001,
      "step": 38850
    },
    {
      "epoch": 0.2680979944393468,
      "grad_norm": 0.0,
      "learning_rate": 7.320468861031964e-06,
      "loss": 0.0,
      "step": 38860
    },
    {
      "epoch": 0.26816698517389115,
      "grad_norm": 0.004220404662191868,
      "learning_rate": 7.31977895368652e-06,
      "loss": 0.0,
      "step": 38870
    },
    {
      "epoch": 0.2682359759084355,
      "grad_norm": 0.0,
      "learning_rate": 7.319089046341077e-06,
      "loss": 0.0,
      "step": 38880
    },
    {
      "epoch": 0.26830496664297987,
      "grad_norm": 0.0,
      "learning_rate": 7.318399138995633e-06,
      "loss": 0.0,
      "step": 38890
    },
    {
      "epoch": 0.2683739573775242,
      "grad_norm": 0.027038592845201492,
      "learning_rate": 7.31770923165019e-06,
      "loss": 0.0149,
      "step": 38900
    },
    {
      "epoch": 0.26844294811206854,
      "grad_norm": 1.9568692266602739e-07,
      "learning_rate": 7.317019324304746e-06,
      "loss": 0.0001,
      "step": 38910
    },
    {
      "epoch": 0.2685119388466129,
      "grad_norm": 0.12096309661865234,
      "learning_rate": 7.316329416959303e-06,
      "loss": 0.0045,
      "step": 38920
    },
    {
      "epoch": 0.26858092958115726,
      "grad_norm": 0.0,
      "learning_rate": 7.31563950961386e-06,
      "loss": 0.0,
      "step": 38930
    },
    {
      "epoch": 0.2686499203157016,
      "grad_norm": 6.261028318022e-08,
      "learning_rate": 7.3149496022684165e-06,
      "loss": 0.0005,
      "step": 38940
    },
    {
      "epoch": 0.268718911050246,
      "grad_norm": 2.6456669729668647e-05,
      "learning_rate": 7.314259694922973e-06,
      "loss": 0.0121,
      "step": 38950
    },
    {
      "epoch": 0.2687879017847903,
      "grad_norm": 2.862464931752129e-09,
      "learning_rate": 7.3135697875775294e-06,
      "loss": 0.0534,
      "step": 38960
    },
    {
      "epoch": 0.26885689251933464,
      "grad_norm": 0.0,
      "learning_rate": 7.312879880232085e-06,
      "loss": 0.0,
      "step": 38970
    },
    {
      "epoch": 0.268925883253879,
      "grad_norm": 0.0,
      "learning_rate": 7.312189972886642e-06,
      "loss": 0.0,
      "step": 38980
    },
    {
      "epoch": 0.26899487398842337,
      "grad_norm": 0.0,
      "learning_rate": 7.311500065541198e-06,
      "loss": 0.0,
      "step": 38990
    },
    {
      "epoch": 0.2690638647229677,
      "grad_norm": 0.0,
      "learning_rate": 7.3108101581957546e-06,
      "loss": 0.0,
      "step": 39000
    },
    {
      "epoch": 0.26913285545751203,
      "grad_norm": 0.0,
      "learning_rate": 7.310120250850311e-06,
      "loss": 0.0001,
      "step": 39010
    },
    {
      "epoch": 0.2692018461920564,
      "grad_norm": 6.099893198552309e-06,
      "learning_rate": 7.3094303435048675e-06,
      "loss": 0.0,
      "step": 39020
    },
    {
      "epoch": 0.26927083692660075,
      "grad_norm": 1.1155313117683363e-08,
      "learning_rate": 7.308740436159424e-06,
      "loss": 0.0,
      "step": 39030
    },
    {
      "epoch": 0.2693398276611451,
      "grad_norm": 0.0,
      "learning_rate": 7.308050528813981e-06,
      "loss": 0.0,
      "step": 39040
    },
    {
      "epoch": 0.2694088183956895,
      "grad_norm": 0.0,
      "learning_rate": 7.307360621468538e-06,
      "loss": 0.0,
      "step": 39050
    },
    {
      "epoch": 0.26947780913023384,
      "grad_norm": 0.0,
      "learning_rate": 7.306670714123094e-06,
      "loss": 0.0391,
      "step": 39060
    },
    {
      "epoch": 0.26954679986477814,
      "grad_norm": 1.2339822319518134e-07,
      "learning_rate": 7.305980806777651e-06,
      "loss": 0.2168,
      "step": 39070
    },
    {
      "epoch": 0.2696157905993225,
      "grad_norm": 0.0,
      "learning_rate": 7.305290899432207e-06,
      "loss": 0.0,
      "step": 39080
    },
    {
      "epoch": 0.26968478133386686,
      "grad_norm": 0.0,
      "learning_rate": 7.304600992086763e-06,
      "loss": 0.0,
      "step": 39090
    },
    {
      "epoch": 0.2697537720684112,
      "grad_norm": 251.88563537597656,
      "learning_rate": 7.3039110847413195e-06,
      "loss": 0.0616,
      "step": 39100
    },
    {
      "epoch": 0.2698227628029556,
      "grad_norm": 0.0,
      "learning_rate": 7.303221177395876e-06,
      "loss": 0.0,
      "step": 39110
    },
    {
      "epoch": 0.2698917535374999,
      "grad_norm": 0.0,
      "learning_rate": 7.3025312700504324e-06,
      "loss": 0.0,
      "step": 39120
    },
    {
      "epoch": 0.26996074427204425,
      "grad_norm": 0.0,
      "learning_rate": 7.301841362704989e-06,
      "loss": 0.0,
      "step": 39130
    },
    {
      "epoch": 0.2700297350065886,
      "grad_norm": 0.0,
      "learning_rate": 7.3011514553595454e-06,
      "loss": 0.0,
      "step": 39140
    },
    {
      "epoch": 0.270098725741133,
      "grad_norm": 0.0,
      "learning_rate": 7.300461548014103e-06,
      "loss": 0.0028,
      "step": 39150
    },
    {
      "epoch": 0.27016771647567733,
      "grad_norm": 0.0,
      "learning_rate": 7.299771640668659e-06,
      "loss": 0.0,
      "step": 39160
    },
    {
      "epoch": 0.2702367072102217,
      "grad_norm": 0.0022596721537411213,
      "learning_rate": 7.299081733323216e-06,
      "loss": 0.0,
      "step": 39170
    },
    {
      "epoch": 0.270305697944766,
      "grad_norm": 0.0035123704001307487,
      "learning_rate": 7.298391825977772e-06,
      "loss": 0.2731,
      "step": 39180
    },
    {
      "epoch": 0.27037468867931036,
      "grad_norm": 0.0,
      "learning_rate": 7.297701918632329e-06,
      "loss": 0.0,
      "step": 39190
    },
    {
      "epoch": 0.2704436794138547,
      "grad_norm": 0.0,
      "learning_rate": 7.297012011286884e-06,
      "loss": 0.0,
      "step": 39200
    },
    {
      "epoch": 0.2705126701483991,
      "grad_norm": 6.752162153134122e-05,
      "learning_rate": 7.296322103941441e-06,
      "loss": 0.0014,
      "step": 39210
    },
    {
      "epoch": 0.27058166088294344,
      "grad_norm": 0.0,
      "learning_rate": 7.295632196595997e-06,
      "loss": 0.0,
      "step": 39220
    },
    {
      "epoch": 0.27065065161748775,
      "grad_norm": 0.0,
      "learning_rate": 7.294942289250554e-06,
      "loss": 0.0,
      "step": 39230
    },
    {
      "epoch": 0.2707196423520321,
      "grad_norm": 0.0,
      "learning_rate": 7.29425238190511e-06,
      "loss": 0.0,
      "step": 39240
    },
    {
      "epoch": 0.27078863308657647,
      "grad_norm": 0.0,
      "learning_rate": 7.293562474559667e-06,
      "loss": 0.0,
      "step": 39250
    },
    {
      "epoch": 0.27085762382112083,
      "grad_norm": 0.0,
      "learning_rate": 7.292872567214224e-06,
      "loss": 1.2336,
      "step": 39260
    },
    {
      "epoch": 0.2709266145556652,
      "grad_norm": 0.0,
      "learning_rate": 7.292182659868781e-06,
      "loss": 0.0006,
      "step": 39270
    },
    {
      "epoch": 0.27099560529020955,
      "grad_norm": 0.0,
      "learning_rate": 7.291492752523337e-06,
      "loss": 0.0001,
      "step": 39280
    },
    {
      "epoch": 0.27106459602475386,
      "grad_norm": 0.0,
      "learning_rate": 7.290802845177894e-06,
      "loss": 0.0001,
      "step": 39290
    },
    {
      "epoch": 0.2711335867592982,
      "grad_norm": 0.0,
      "learning_rate": 7.29011293783245e-06,
      "loss": 0.0,
      "step": 39300
    },
    {
      "epoch": 0.2712025774938426,
      "grad_norm": 0.0,
      "learning_rate": 7.289423030487006e-06,
      "loss": 0.0,
      "step": 39310
    },
    {
      "epoch": 0.27127156822838694,
      "grad_norm": 0.0,
      "learning_rate": 7.288733123141562e-06,
      "loss": 0.0,
      "step": 39320
    },
    {
      "epoch": 0.2713405589629313,
      "grad_norm": 2.115774577760021e-06,
      "learning_rate": 7.288043215796119e-06,
      "loss": 0.0001,
      "step": 39330
    },
    {
      "epoch": 0.2714095496974756,
      "grad_norm": 0.0,
      "learning_rate": 7.287353308450675e-06,
      "loss": 0.187,
      "step": 39340
    },
    {
      "epoch": 0.27147854043201997,
      "grad_norm": 0.0,
      "learning_rate": 7.286663401105232e-06,
      "loss": 0.0011,
      "step": 39350
    },
    {
      "epoch": 0.2715475311665643,
      "grad_norm": 0.0,
      "learning_rate": 7.285973493759788e-06,
      "loss": 0.0,
      "step": 39360
    },
    {
      "epoch": 0.2716165219011087,
      "grad_norm": 0.0,
      "learning_rate": 7.2852835864143456e-06,
      "loss": 0.0,
      "step": 39370
    },
    {
      "epoch": 0.27168551263565305,
      "grad_norm": 4.052931323883513e-09,
      "learning_rate": 7.284593679068902e-06,
      "loss": 0.0,
      "step": 39380
    },
    {
      "epoch": 0.2717545033701974,
      "grad_norm": 0.0,
      "learning_rate": 7.2839037717234585e-06,
      "loss": 0.0045,
      "step": 39390
    },
    {
      "epoch": 0.2718234941047417,
      "grad_norm": 2.0672324296810984e-08,
      "learning_rate": 7.283213864378015e-06,
      "loss": 0.0,
      "step": 39400
    },
    {
      "epoch": 0.2718924848392861,
      "grad_norm": 0.0,
      "learning_rate": 7.2825239570325715e-06,
      "loss": 0.0038,
      "step": 39410
    },
    {
      "epoch": 0.27196147557383044,
      "grad_norm": 0.0,
      "learning_rate": 7.281834049687128e-06,
      "loss": 0.0105,
      "step": 39420
    },
    {
      "epoch": 0.2720304663083748,
      "grad_norm": 0.0,
      "learning_rate": 7.281144142341684e-06,
      "loss": 0.0,
      "step": 39430
    },
    {
      "epoch": 0.27209945704291916,
      "grad_norm": 0.0,
      "learning_rate": 7.28045423499624e-06,
      "loss": 0.0522,
      "step": 39440
    },
    {
      "epoch": 0.27216844777746346,
      "grad_norm": 0.04534636810421944,
      "learning_rate": 7.279764327650797e-06,
      "loss": 0.0,
      "step": 39450
    },
    {
      "epoch": 0.2722374385120078,
      "grad_norm": 3.3380654258508002e-06,
      "learning_rate": 7.279074420305353e-06,
      "loss": 0.0,
      "step": 39460
    },
    {
      "epoch": 0.2723064292465522,
      "grad_norm": 0.0,
      "learning_rate": 7.27838451295991e-06,
      "loss": 0.0,
      "step": 39470
    },
    {
      "epoch": 0.27237541998109654,
      "grad_norm": 1.5880146975177922e-06,
      "learning_rate": 7.277694605614467e-06,
      "loss": 0.0,
      "step": 39480
    },
    {
      "epoch": 0.2724444107156409,
      "grad_norm": 0.0,
      "learning_rate": 7.2770046982690234e-06,
      "loss": 0.4078,
      "step": 39490
    },
    {
      "epoch": 0.27251340145018527,
      "grad_norm": 0.0,
      "learning_rate": 7.27631479092358e-06,
      "loss": 0.0,
      "step": 39500
    },
    {
      "epoch": 0.27258239218472957,
      "grad_norm": 0.0,
      "learning_rate": 7.275624883578136e-06,
      "loss": 0.0,
      "step": 39510
    },
    {
      "epoch": 0.27265138291927393,
      "grad_norm": 0.0,
      "learning_rate": 7.274934976232693e-06,
      "loss": 0.0,
      "step": 39520
    },
    {
      "epoch": 0.2727203736538183,
      "grad_norm": 1.6298672278480808e-07,
      "learning_rate": 7.274245068887249e-06,
      "loss": 0.0002,
      "step": 39530
    },
    {
      "epoch": 0.27278936438836265,
      "grad_norm": 0.0,
      "learning_rate": 7.273555161541805e-06,
      "loss": 0.0001,
      "step": 39540
    },
    {
      "epoch": 0.272858355122907,
      "grad_norm": 0.0,
      "learning_rate": 7.2728652541963615e-06,
      "loss": 0.0,
      "step": 39550
    },
    {
      "epoch": 0.2729273458574513,
      "grad_norm": 0.0,
      "learning_rate": 7.272175346850918e-06,
      "loss": 0.0,
      "step": 39560
    },
    {
      "epoch": 0.2729963365919957,
      "grad_norm": 0.0,
      "learning_rate": 7.2714854395054745e-06,
      "loss": 0.0,
      "step": 39570
    },
    {
      "epoch": 0.27306532732654004,
      "grad_norm": 1.044731902766216e-06,
      "learning_rate": 7.270795532160031e-06,
      "loss": 0.0,
      "step": 39580
    },
    {
      "epoch": 0.2731343180610844,
      "grad_norm": 0.0,
      "learning_rate": 7.270105624814588e-06,
      "loss": 0.0,
      "step": 39590
    },
    {
      "epoch": 0.27320330879562876,
      "grad_norm": 0.0,
      "learning_rate": 7.269415717469145e-06,
      "loss": 0.0,
      "step": 39600
    },
    {
      "epoch": 0.2732722995301731,
      "grad_norm": 0.0,
      "learning_rate": 7.268725810123701e-06,
      "loss": 0.0,
      "step": 39610
    },
    {
      "epoch": 0.27334129026471743,
      "grad_norm": 0.04081987589597702,
      "learning_rate": 7.268035902778258e-06,
      "loss": 0.0543,
      "step": 39620
    },
    {
      "epoch": 0.2734102809992618,
      "grad_norm": 0.0,
      "learning_rate": 7.267345995432814e-06,
      "loss": 0.0,
      "step": 39630
    },
    {
      "epoch": 0.27347927173380615,
      "grad_norm": 2.2865371704101562,
      "learning_rate": 7.266656088087371e-06,
      "loss": 0.0002,
      "step": 39640
    },
    {
      "epoch": 0.2735482624683505,
      "grad_norm": 0.0,
      "learning_rate": 7.2659661807419264e-06,
      "loss": 0.0,
      "step": 39650
    },
    {
      "epoch": 0.2736172532028949,
      "grad_norm": 0.0,
      "learning_rate": 7.265276273396483e-06,
      "loss": 0.0,
      "step": 39660
    },
    {
      "epoch": 0.2736862439374392,
      "grad_norm": 0.0,
      "learning_rate": 7.264586366051039e-06,
      "loss": 0.0,
      "step": 39670
    },
    {
      "epoch": 0.27375523467198354,
      "grad_norm": 0.0,
      "learning_rate": 7.263896458705596e-06,
      "loss": 0.0,
      "step": 39680
    },
    {
      "epoch": 0.2738242254065279,
      "grad_norm": 0.0,
      "learning_rate": 7.263206551360152e-06,
      "loss": 0.0,
      "step": 39690
    },
    {
      "epoch": 0.27389321614107226,
      "grad_norm": 0.0,
      "learning_rate": 7.26251664401471e-06,
      "loss": 0.0,
      "step": 39700
    },
    {
      "epoch": 0.2739622068756166,
      "grad_norm": 3.687272510433104e-06,
      "learning_rate": 7.261826736669266e-06,
      "loss": 0.0,
      "step": 39710
    },
    {
      "epoch": 0.274031197610161,
      "grad_norm": 0.0,
      "learning_rate": 7.261136829323823e-06,
      "loss": 0.0,
      "step": 39720
    },
    {
      "epoch": 0.2741001883447053,
      "grad_norm": 0.0,
      "learning_rate": 7.260446921978379e-06,
      "loss": 0.0,
      "step": 39730
    },
    {
      "epoch": 0.27416917907924965,
      "grad_norm": 0.0,
      "learning_rate": 7.259757014632936e-06,
      "loss": 0.0,
      "step": 39740
    },
    {
      "epoch": 0.274238169813794,
      "grad_norm": 0.0,
      "learning_rate": 7.259067107287492e-06,
      "loss": 0.0,
      "step": 39750
    },
    {
      "epoch": 0.27430716054833837,
      "grad_norm": 0.0046210926957428455,
      "learning_rate": 7.258377199942049e-06,
      "loss": 0.0,
      "step": 39760
    },
    {
      "epoch": 0.27437615128288273,
      "grad_norm": 0.0,
      "learning_rate": 7.257687292596604e-06,
      "loss": 0.0,
      "step": 39770
    },
    {
      "epoch": 0.27444514201742704,
      "grad_norm": 0.0,
      "learning_rate": 7.256997385251161e-06,
      "loss": 0.0,
      "step": 39780
    },
    {
      "epoch": 0.2745141327519714,
      "grad_norm": 0.0,
      "learning_rate": 7.256307477905717e-06,
      "loss": 0.0001,
      "step": 39790
    },
    {
      "epoch": 0.27458312348651576,
      "grad_norm": 0.0,
      "learning_rate": 7.255617570560274e-06,
      "loss": 0.0,
      "step": 39800
    },
    {
      "epoch": 0.2746521142210601,
      "grad_norm": 0.0,
      "learning_rate": 7.254927663214831e-06,
      "loss": 0.0,
      "step": 39810
    },
    {
      "epoch": 0.2747211049556045,
      "grad_norm": 5.98169071963639e-06,
      "learning_rate": 7.254237755869388e-06,
      "loss": 0.0001,
      "step": 39820
    },
    {
      "epoch": 0.27479009569014884,
      "grad_norm": 0.0,
      "learning_rate": 7.253547848523944e-06,
      "loss": 0.0,
      "step": 39830
    },
    {
      "epoch": 0.27485908642469314,
      "grad_norm": 0.0,
      "learning_rate": 7.252857941178501e-06,
      "loss": 0.0,
      "step": 39840
    },
    {
      "epoch": 0.2749280771592375,
      "grad_norm": 0.0,
      "learning_rate": 7.252168033833057e-06,
      "loss": 0.0,
      "step": 39850
    },
    {
      "epoch": 0.27499706789378187,
      "grad_norm": 0.0,
      "learning_rate": 7.251478126487614e-06,
      "loss": 0.0,
      "step": 39860
    },
    {
      "epoch": 0.2750660586283262,
      "grad_norm": 0.0,
      "learning_rate": 7.25078821914217e-06,
      "loss": 0.0,
      "step": 39870
    },
    {
      "epoch": 0.2751350493628706,
      "grad_norm": 0.38289105892181396,
      "learning_rate": 7.250098311796726e-06,
      "loss": 0.0001,
      "step": 39880
    },
    {
      "epoch": 0.2752040400974149,
      "grad_norm": 0.0,
      "learning_rate": 7.249408404451282e-06,
      "loss": 0.0,
      "step": 39890
    },
    {
      "epoch": 0.27527303083195925,
      "grad_norm": 0.0,
      "learning_rate": 7.248718497105839e-06,
      "loss": 0.0,
      "step": 39900
    },
    {
      "epoch": 0.2753420215665036,
      "grad_norm": 0.031553372740745544,
      "learning_rate": 7.248028589760395e-06,
      "loss": 0.0,
      "step": 39910
    },
    {
      "epoch": 0.275411012301048,
      "grad_norm": 0.0,
      "learning_rate": 7.2473386824149525e-06,
      "loss": 0.0,
      "step": 39920
    },
    {
      "epoch": 0.27548000303559234,
      "grad_norm": 0.0,
      "learning_rate": 7.246648775069509e-06,
      "loss": 0.0,
      "step": 39930
    },
    {
      "epoch": 0.2755489937701367,
      "grad_norm": 0.0,
      "learning_rate": 7.2459588677240655e-06,
      "loss": 0.0,
      "step": 39940
    },
    {
      "epoch": 0.275617984504681,
      "grad_norm": 0.0,
      "learning_rate": 7.245268960378622e-06,
      "loss": 0.0,
      "step": 39950
    },
    {
      "epoch": 0.27568697523922536,
      "grad_norm": 0.0,
      "learning_rate": 7.2445790530331785e-06,
      "loss": 0.0001,
      "step": 39960
    },
    {
      "epoch": 0.2757559659737697,
      "grad_norm": 0.0,
      "learning_rate": 7.243889145687735e-06,
      "loss": 0.0,
      "step": 39970
    },
    {
      "epoch": 0.2758249567083141,
      "grad_norm": 2.154805770260282e-06,
      "learning_rate": 7.2431992383422915e-06,
      "loss": 0.0,
      "step": 39980
    },
    {
      "epoch": 0.27589394744285844,
      "grad_norm": 2.7835710625367938e-06,
      "learning_rate": 7.242509330996847e-06,
      "loss": 0.0022,
      "step": 39990
    },
    {
      "epoch": 0.27596293817740275,
      "grad_norm": 0.0,
      "learning_rate": 7.241819423651404e-06,
      "loss": 0.0361,
      "step": 40000
    },
    {
      "epoch": 0.2760319289119471,
      "grad_norm": 9.956645286024468e-09,
      "learning_rate": 7.24112951630596e-06,
      "loss": 0.0733,
      "step": 40010
    },
    {
      "epoch": 0.27610091964649147,
      "grad_norm": 0.0,
      "learning_rate": 7.240439608960517e-06,
      "loss": 0.0031,
      "step": 40020
    },
    {
      "epoch": 0.27616991038103583,
      "grad_norm": 0.0,
      "learning_rate": 7.239749701615074e-06,
      "loss": 0.0,
      "step": 40030
    },
    {
      "epoch": 0.2762389011155802,
      "grad_norm": 0.0,
      "learning_rate": 7.23905979426963e-06,
      "loss": 0.0,
      "step": 40040
    },
    {
      "epoch": 0.27630789185012455,
      "grad_norm": 0.0,
      "learning_rate": 7.238369886924187e-06,
      "loss": 0.0,
      "step": 40050
    },
    {
      "epoch": 0.27637688258466886,
      "grad_norm": 5.2426734065136316e-09,
      "learning_rate": 7.237679979578743e-06,
      "loss": 0.0,
      "step": 40060
    },
    {
      "epoch": 0.2764458733192132,
      "grad_norm": 0.0,
      "learning_rate": 7.2369900722333e-06,
      "loss": 0.0001,
      "step": 40070
    },
    {
      "epoch": 0.2765148640537576,
      "grad_norm": 0.0,
      "learning_rate": 7.236300164887856e-06,
      "loss": 0.0001,
      "step": 40080
    },
    {
      "epoch": 0.27658385478830194,
      "grad_norm": 0.0,
      "learning_rate": 7.235610257542413e-06,
      "loss": 0.0,
      "step": 40090
    },
    {
      "epoch": 0.2766528455228463,
      "grad_norm": 0.00014590761566068977,
      "learning_rate": 7.2349203501969685e-06,
      "loss": 0.0,
      "step": 40100
    },
    {
      "epoch": 0.2767218362573906,
      "grad_norm": 0.0,
      "learning_rate": 7.234230442851525e-06,
      "loss": 0.0,
      "step": 40110
    },
    {
      "epoch": 0.27679082699193497,
      "grad_norm": 0.0,
      "learning_rate": 7.2335405355060815e-06,
      "loss": 0.0,
      "step": 40120
    },
    {
      "epoch": 0.27685981772647933,
      "grad_norm": 0.0,
      "learning_rate": 7.232850628160638e-06,
      "loss": 0.0,
      "step": 40130
    },
    {
      "epoch": 0.2769288084610237,
      "grad_norm": 0.0,
      "learning_rate": 7.232160720815195e-06,
      "loss": 0.0001,
      "step": 40140
    },
    {
      "epoch": 0.27699779919556805,
      "grad_norm": 0.0,
      "learning_rate": 7.231470813469752e-06,
      "loss": 0.0,
      "step": 40150
    },
    {
      "epoch": 0.2770667899301124,
      "grad_norm": 4.691009358737119e-09,
      "learning_rate": 7.230780906124308e-06,
      "loss": 0.0,
      "step": 40160
    },
    {
      "epoch": 0.2771357806646567,
      "grad_norm": 0.0,
      "learning_rate": 7.230090998778865e-06,
      "loss": 0.0,
      "step": 40170
    },
    {
      "epoch": 0.2772047713992011,
      "grad_norm": 0.0,
      "learning_rate": 7.229401091433421e-06,
      "loss": 0.0,
      "step": 40180
    },
    {
      "epoch": 0.27727376213374544,
      "grad_norm": 0.0,
      "learning_rate": 7.228711184087978e-06,
      "loss": 0.0,
      "step": 40190
    },
    {
      "epoch": 0.2773427528682898,
      "grad_norm": 0.0,
      "learning_rate": 7.228021276742534e-06,
      "loss": 0.0088,
      "step": 40200
    },
    {
      "epoch": 0.27741174360283416,
      "grad_norm": 0.0,
      "learning_rate": 7.227331369397091e-06,
      "loss": 0.0009,
      "step": 40210
    },
    {
      "epoch": 0.27748073433737847,
      "grad_norm": 0.0,
      "learning_rate": 7.226641462051646e-06,
      "loss": 0.0002,
      "step": 40220
    },
    {
      "epoch": 0.2775497250719228,
      "grad_norm": 0.0,
      "learning_rate": 7.225951554706203e-06,
      "loss": 0.0,
      "step": 40230
    },
    {
      "epoch": 0.2776187158064672,
      "grad_norm": 0.0,
      "learning_rate": 7.225261647360759e-06,
      "loss": 0.0,
      "step": 40240
    },
    {
      "epoch": 0.27768770654101155,
      "grad_norm": 1.2461619007808622e-06,
      "learning_rate": 7.224571740015317e-06,
      "loss": 0.0038,
      "step": 40250
    },
    {
      "epoch": 0.2777566972755559,
      "grad_norm": 0.0,
      "learning_rate": 7.223881832669873e-06,
      "loss": 0.0,
      "step": 40260
    },
    {
      "epoch": 0.27782568801010027,
      "grad_norm": 0.0,
      "learning_rate": 7.22319192532443e-06,
      "loss": 0.0,
      "step": 40270
    },
    {
      "epoch": 0.2778946787446446,
      "grad_norm": 0.17264869809150696,
      "learning_rate": 7.222502017978986e-06,
      "loss": 0.0,
      "step": 40280
    },
    {
      "epoch": 0.27796366947918894,
      "grad_norm": 0.0,
      "learning_rate": 7.221812110633543e-06,
      "loss": 0.0,
      "step": 40290
    },
    {
      "epoch": 0.2780326602137333,
      "grad_norm": 0.0,
      "learning_rate": 7.221122203288099e-06,
      "loss": 0.0,
      "step": 40300
    },
    {
      "epoch": 0.27810165094827766,
      "grad_norm": 0.0,
      "learning_rate": 7.220432295942656e-06,
      "loss": 0.0001,
      "step": 40310
    },
    {
      "epoch": 0.278170641682822,
      "grad_norm": 0.0,
      "learning_rate": 7.219742388597212e-06,
      "loss": 0.0,
      "step": 40320
    },
    {
      "epoch": 0.2782396324173663,
      "grad_norm": 0.0,
      "learning_rate": 7.219052481251768e-06,
      "loss": 0.0,
      "step": 40330
    },
    {
      "epoch": 0.2783086231519107,
      "grad_norm": 694.5888671875,
      "learning_rate": 7.218362573906324e-06,
      "loss": 0.4496,
      "step": 40340
    },
    {
      "epoch": 0.27837761388645504,
      "grad_norm": 0.0,
      "learning_rate": 7.217672666560881e-06,
      "loss": 0.0023,
      "step": 40350
    },
    {
      "epoch": 0.2784466046209994,
      "grad_norm": 0.0,
      "learning_rate": 7.216982759215438e-06,
      "loss": 0.0,
      "step": 40360
    },
    {
      "epoch": 0.27851559535554377,
      "grad_norm": 0.0,
      "learning_rate": 7.216292851869995e-06,
      "loss": 0.0,
      "step": 40370
    },
    {
      "epoch": 0.2785845860900881,
      "grad_norm": 0.0,
      "learning_rate": 7.215602944524551e-06,
      "loss": 0.0,
      "step": 40380
    },
    {
      "epoch": 0.27865357682463243,
      "grad_norm": 1.381496758767753e-06,
      "learning_rate": 7.2149130371791076e-06,
      "loss": 0.0,
      "step": 40390
    },
    {
      "epoch": 0.2787225675591768,
      "grad_norm": 2.1915736851951806e-06,
      "learning_rate": 7.214223129833664e-06,
      "loss": 0.001,
      "step": 40400
    },
    {
      "epoch": 0.27879155829372115,
      "grad_norm": 23.778778076171875,
      "learning_rate": 7.2135332224882205e-06,
      "loss": 0.0053,
      "step": 40410
    },
    {
      "epoch": 0.2788605490282655,
      "grad_norm": 0.0,
      "learning_rate": 7.212912305877321e-06,
      "loss": 0.316,
      "step": 40420
    },
    {
      "epoch": 0.2789295397628099,
      "grad_norm": 0.0,
      "learning_rate": 7.212222398531878e-06,
      "loss": 0.0,
      "step": 40430
    },
    {
      "epoch": 0.2789985304973542,
      "grad_norm": 0.0,
      "learning_rate": 7.211532491186433e-06,
      "loss": 0.2004,
      "step": 40440
    },
    {
      "epoch": 0.27906752123189854,
      "grad_norm": 0.5479373335838318,
      "learning_rate": 7.2108425838409915e-06,
      "loss": 0.0008,
      "step": 40450
    },
    {
      "epoch": 0.2791365119664429,
      "grad_norm": 9.191258865115515e-10,
      "learning_rate": 7.210152676495548e-06,
      "loss": 0.0,
      "step": 40460
    },
    {
      "epoch": 0.27920550270098726,
      "grad_norm": 0.0,
      "learning_rate": 7.209462769150104e-06,
      "loss": 0.0,
      "step": 40470
    },
    {
      "epoch": 0.2792744934355316,
      "grad_norm": 0.0,
      "learning_rate": 7.20877286180466e-06,
      "loss": 0.0,
      "step": 40480
    },
    {
      "epoch": 0.279343484170076,
      "grad_norm": 0.5799373388290405,
      "learning_rate": 7.208082954459217e-06,
      "loss": 0.0002,
      "step": 40490
    },
    {
      "epoch": 0.2794124749046203,
      "grad_norm": 0.0,
      "learning_rate": 7.207393047113773e-06,
      "loss": 0.0,
      "step": 40500
    },
    {
      "epoch": 0.27948146563916465,
      "grad_norm": 0.0,
      "learning_rate": 7.20670313976833e-06,
      "loss": 0.008,
      "step": 40510
    },
    {
      "epoch": 0.279550456373709,
      "grad_norm": 0.0,
      "learning_rate": 7.206013232422886e-06,
      "loss": 0.0,
      "step": 40520
    },
    {
      "epoch": 0.27961944710825337,
      "grad_norm": 0.0,
      "learning_rate": 7.205323325077443e-06,
      "loss": 0.0,
      "step": 40530
    },
    {
      "epoch": 0.27968843784279773,
      "grad_norm": 0.0,
      "learning_rate": 7.204633417731999e-06,
      "loss": 0.003,
      "step": 40540
    },
    {
      "epoch": 0.27975742857734204,
      "grad_norm": 0.0,
      "learning_rate": 7.203943510386555e-06,
      "loss": 0.0,
      "step": 40550
    },
    {
      "epoch": 0.2798264193118864,
      "grad_norm": 0.0,
      "learning_rate": 7.203253603041113e-06,
      "loss": 0.0001,
      "step": 40560
    },
    {
      "epoch": 0.27989541004643076,
      "grad_norm": 0.0,
      "learning_rate": 7.202563695695669e-06,
      "loss": 0.0001,
      "step": 40570
    },
    {
      "epoch": 0.2799644007809751,
      "grad_norm": 0.0,
      "learning_rate": 7.201873788350225e-06,
      "loss": 0.0,
      "step": 40580
    },
    {
      "epoch": 0.2800333915155195,
      "grad_norm": 0.0,
      "learning_rate": 7.2011838810047816e-06,
      "loss": 0.0079,
      "step": 40590
    },
    {
      "epoch": 0.28010238225006384,
      "grad_norm": 0.0,
      "learning_rate": 7.200493973659338e-06,
      "loss": 0.0305,
      "step": 40600
    },
    {
      "epoch": 0.28017137298460815,
      "grad_norm": 0.0,
      "learning_rate": 7.1998040663138945e-06,
      "loss": 0.0,
      "step": 40610
    },
    {
      "epoch": 0.2802403637191525,
      "grad_norm": 0.004756584297865629,
      "learning_rate": 7.199114158968451e-06,
      "loss": 0.0407,
      "step": 40620
    },
    {
      "epoch": 0.28030935445369687,
      "grad_norm": 0.0,
      "learning_rate": 7.1984242516230075e-06,
      "loss": 0.0,
      "step": 40630
    },
    {
      "epoch": 0.28037834518824123,
      "grad_norm": 0.00023076798242982477,
      "learning_rate": 7.197734344277564e-06,
      "loss": 0.0,
      "step": 40640
    },
    {
      "epoch": 0.2804473359227856,
      "grad_norm": 0.0,
      "learning_rate": 7.1970444369321205e-06,
      "loss": 0.0,
      "step": 40650
    },
    {
      "epoch": 0.2805163266573299,
      "grad_norm": 0.0,
      "learning_rate": 7.196354529586677e-06,
      "loss": 0.0,
      "step": 40660
    },
    {
      "epoch": 0.28058531739187426,
      "grad_norm": 0.0,
      "learning_rate": 7.195664622241234e-06,
      "loss": 0.0,
      "step": 40670
    },
    {
      "epoch": 0.2806543081264186,
      "grad_norm": 0.0,
      "learning_rate": 7.194974714895791e-06,
      "loss": 0.0,
      "step": 40680
    },
    {
      "epoch": 0.280723298860963,
      "grad_norm": 0.0,
      "learning_rate": 7.1942848075503465e-06,
      "loss": 0.0,
      "step": 40690
    },
    {
      "epoch": 0.28079228959550734,
      "grad_norm": 0.0,
      "learning_rate": 7.193594900204903e-06,
      "loss": 0.5321,
      "step": 40700
    },
    {
      "epoch": 0.2808612803300517,
      "grad_norm": 0.0,
      "learning_rate": 7.1929049928594594e-06,
      "loss": 0.0012,
      "step": 40710
    },
    {
      "epoch": 0.280930271064596,
      "grad_norm": 0.0,
      "learning_rate": 7.192215085514016e-06,
      "loss": 0.0,
      "step": 40720
    },
    {
      "epoch": 0.28099926179914037,
      "grad_norm": 0.0,
      "learning_rate": 7.191525178168572e-06,
      "loss": 0.0,
      "step": 40730
    },
    {
      "epoch": 0.2810682525336847,
      "grad_norm": 0.0,
      "learning_rate": 7.190835270823129e-06,
      "loss": 0.0001,
      "step": 40740
    },
    {
      "epoch": 0.2811372432682291,
      "grad_norm": 0.0,
      "learning_rate": 7.190145363477685e-06,
      "loss": 0.0007,
      "step": 40750
    },
    {
      "epoch": 0.28120623400277345,
      "grad_norm": 8.962988129468386e-10,
      "learning_rate": 7.189455456132242e-06,
      "loss": 0.0237,
      "step": 40760
    },
    {
      "epoch": 0.28127522473731775,
      "grad_norm": 0.00010635334183461964,
      "learning_rate": 7.188765548786798e-06,
      "loss": 0.0001,
      "step": 40770
    },
    {
      "epoch": 0.2813442154718621,
      "grad_norm": 0.0,
      "learning_rate": 7.188075641441356e-06,
      "loss": 0.0001,
      "step": 40780
    },
    {
      "epoch": 0.2814132062064065,
      "grad_norm": 0.0,
      "learning_rate": 7.187385734095912e-06,
      "loss": 0.001,
      "step": 40790
    },
    {
      "epoch": 0.28148219694095084,
      "grad_norm": 0.0,
      "learning_rate": 7.186695826750469e-06,
      "loss": 0.0,
      "step": 40800
    },
    {
      "epoch": 0.2815511876754952,
      "grad_norm": 1.777445080008988e-09,
      "learning_rate": 7.186005919405024e-06,
      "loss": 0.0001,
      "step": 40810
    },
    {
      "epoch": 0.28162017841003956,
      "grad_norm": 0.0,
      "learning_rate": 7.185316012059581e-06,
      "loss": 0.0007,
      "step": 40820
    },
    {
      "epoch": 0.28168916914458386,
      "grad_norm": 0.0,
      "learning_rate": 7.184626104714137e-06,
      "loss": 0.0,
      "step": 40830
    },
    {
      "epoch": 0.2817581598791282,
      "grad_norm": 0.0,
      "learning_rate": 7.183936197368694e-06,
      "loss": 0.0,
      "step": 40840
    },
    {
      "epoch": 0.2818271506136726,
      "grad_norm": 1.822227924108688e-09,
      "learning_rate": 7.18324629002325e-06,
      "loss": 0.0,
      "step": 40850
    },
    {
      "epoch": 0.28189614134821694,
      "grad_norm": 0.0,
      "learning_rate": 7.182556382677807e-06,
      "loss": 0.0901,
      "step": 40860
    },
    {
      "epoch": 0.2819651320827613,
      "grad_norm": 0.0,
      "learning_rate": 7.181866475332363e-06,
      "loss": 0.0,
      "step": 40870
    },
    {
      "epoch": 0.2820341228173056,
      "grad_norm": 0.0,
      "learning_rate": 7.18117656798692e-06,
      "loss": 0.0,
      "step": 40880
    },
    {
      "epoch": 0.28210311355184997,
      "grad_norm": 0.0,
      "learning_rate": 7.180486660641477e-06,
      "loss": 0.0,
      "step": 40890
    },
    {
      "epoch": 0.28217210428639433,
      "grad_norm": 0.0,
      "learning_rate": 7.179796753296034e-06,
      "loss": 0.0001,
      "step": 40900
    },
    {
      "epoch": 0.2822410950209387,
      "grad_norm": 0.0,
      "learning_rate": 7.17910684595059e-06,
      "loss": 0.0,
      "step": 40910
    },
    {
      "epoch": 0.28231008575548305,
      "grad_norm": 0.0,
      "learning_rate": 7.178416938605146e-06,
      "loss": 0.0001,
      "step": 40920
    },
    {
      "epoch": 0.2823790764900274,
      "grad_norm": 0.0,
      "learning_rate": 7.177727031259702e-06,
      "loss": 0.0,
      "step": 40930
    },
    {
      "epoch": 0.2824480672245717,
      "grad_norm": 5.843711505804094e-07,
      "learning_rate": 7.177037123914259e-06,
      "loss": 0.0,
      "step": 40940
    },
    {
      "epoch": 0.2825170579591161,
      "grad_norm": 0.00039787410059943795,
      "learning_rate": 7.176347216568815e-06,
      "loss": 0.0281,
      "step": 40950
    },
    {
      "epoch": 0.28258604869366044,
      "grad_norm": 0.0,
      "learning_rate": 7.175657309223372e-06,
      "loss": 0.0,
      "step": 40960
    },
    {
      "epoch": 0.2826550394282048,
      "grad_norm": 0.0,
      "learning_rate": 7.174967401877928e-06,
      "loss": 0.0,
      "step": 40970
    },
    {
      "epoch": 0.28272403016274916,
      "grad_norm": 0.0,
      "learning_rate": 7.174277494532485e-06,
      "loss": 0.0,
      "step": 40980
    },
    {
      "epoch": 0.28279302089729347,
      "grad_norm": 0.0,
      "learning_rate": 7.173587587187041e-06,
      "loss": 0.0,
      "step": 40990
    },
    {
      "epoch": 0.28286201163183783,
      "grad_norm": 0.0,
      "learning_rate": 7.1728976798415985e-06,
      "loss": 0.0,
      "step": 41000
    },
    {
      "epoch": 0.2829310023663822,
      "grad_norm": 1.244142936229764e-06,
      "learning_rate": 7.172207772496155e-06,
      "loss": 0.0,
      "step": 41010
    },
    {
      "epoch": 0.28299999310092655,
      "grad_norm": 0.0,
      "learning_rate": 7.1715178651507115e-06,
      "loss": 0.1564,
      "step": 41020
    },
    {
      "epoch": 0.2830689838354709,
      "grad_norm": 0.0,
      "learning_rate": 7.170827957805267e-06,
      "loss": 0.0,
      "step": 41030
    },
    {
      "epoch": 0.28313797457001527,
      "grad_norm": 0.0,
      "learning_rate": 7.170138050459824e-06,
      "loss": 0.0,
      "step": 41040
    },
    {
      "epoch": 0.2832069653045596,
      "grad_norm": 0.0,
      "learning_rate": 7.16944814311438e-06,
      "loss": 0.0008,
      "step": 41050
    },
    {
      "epoch": 0.28327595603910394,
      "grad_norm": 0.0,
      "learning_rate": 7.168758235768937e-06,
      "loss": 0.239,
      "step": 41060
    },
    {
      "epoch": 0.2833449467736483,
      "grad_norm": 0.0,
      "learning_rate": 7.168068328423493e-06,
      "loss": 0.0,
      "step": 41070
    },
    {
      "epoch": 0.28341393750819266,
      "grad_norm": 0.0,
      "learning_rate": 7.16737842107805e-06,
      "loss": 0.2239,
      "step": 41080
    },
    {
      "epoch": 0.283482928242737,
      "grad_norm": 0.02193543314933777,
      "learning_rate": 7.166688513732606e-06,
      "loss": 0.0,
      "step": 41090
    },
    {
      "epoch": 0.2835519189772813,
      "grad_norm": 0.0,
      "learning_rate": 7.1659986063871626e-06,
      "loss": 0.0,
      "step": 41100
    },
    {
      "epoch": 0.2836209097118257,
      "grad_norm": 0.0,
      "learning_rate": 7.16530869904172e-06,
      "loss": 0.0,
      "step": 41110
    },
    {
      "epoch": 0.28368990044637005,
      "grad_norm": 0.0,
      "learning_rate": 7.164618791696276e-06,
      "loss": 0.025,
      "step": 41120
    },
    {
      "epoch": 0.2837588911809144,
      "grad_norm": 0.16067703068256378,
      "learning_rate": 7.163928884350833e-06,
      "loss": 0.1651,
      "step": 41130
    },
    {
      "epoch": 0.28382788191545877,
      "grad_norm": 0.0,
      "learning_rate": 7.163238977005389e-06,
      "loss": 0.019,
      "step": 41140
    },
    {
      "epoch": 0.28389687265000313,
      "grad_norm": 0.0,
      "learning_rate": 7.162549069659945e-06,
      "loss": 0.0,
      "step": 41150
    },
    {
      "epoch": 0.28396586338454743,
      "grad_norm": 0.0,
      "learning_rate": 7.1618591623145015e-06,
      "loss": 0.0,
      "step": 41160
    },
    {
      "epoch": 0.2840348541190918,
      "grad_norm": 1.7417595472579706e-06,
      "learning_rate": 7.161169254969058e-06,
      "loss": 0.0,
      "step": 41170
    },
    {
      "epoch": 0.28410384485363616,
      "grad_norm": 0.0,
      "learning_rate": 7.1604793476236145e-06,
      "loss": 0.0,
      "step": 41180
    },
    {
      "epoch": 0.2841728355881805,
      "grad_norm": 0.0,
      "learning_rate": 7.159789440278171e-06,
      "loss": 0.0,
      "step": 41190
    },
    {
      "epoch": 0.2842418263227249,
      "grad_norm": 0.0,
      "learning_rate": 7.1590995329327275e-06,
      "loss": 0.0,
      "step": 41200
    },
    {
      "epoch": 0.2843108170572692,
      "grad_norm": 1.790538476598158e-06,
      "learning_rate": 7.158409625587284e-06,
      "loss": 0.0,
      "step": 41210
    },
    {
      "epoch": 0.28437980779181354,
      "grad_norm": 0.0,
      "learning_rate": 7.157719718241841e-06,
      "loss": 0.0,
      "step": 41220
    },
    {
      "epoch": 0.2844487985263579,
      "grad_norm": 0.0,
      "learning_rate": 7.157029810896398e-06,
      "loss": 0.0,
      "step": 41230
    },
    {
      "epoch": 0.28451778926090227,
      "grad_norm": 5.097764699257823e-08,
      "learning_rate": 7.156339903550954e-06,
      "loss": 0.0,
      "step": 41240
    },
    {
      "epoch": 0.2845867799954466,
      "grad_norm": 0.0,
      "learning_rate": 7.155649996205511e-06,
      "loss": 0.0,
      "step": 41250
    },
    {
      "epoch": 0.284655770729991,
      "grad_norm": 0.0,
      "learning_rate": 7.154960088860066e-06,
      "loss": 0.0,
      "step": 41260
    },
    {
      "epoch": 0.2847247614645353,
      "grad_norm": 0.0,
      "learning_rate": 7.154270181514623e-06,
      "loss": 0.0,
      "step": 41270
    },
    {
      "epoch": 0.28479375219907965,
      "grad_norm": 0.002611007774248719,
      "learning_rate": 7.153580274169179e-06,
      "loss": 0.0,
      "step": 41280
    },
    {
      "epoch": 0.284862742933624,
      "grad_norm": 0.01389256864786148,
      "learning_rate": 7.152890366823736e-06,
      "loss": 0.0,
      "step": 41290
    },
    {
      "epoch": 0.2849317336681684,
      "grad_norm": 0.0,
      "learning_rate": 7.152200459478292e-06,
      "loss": 0.0,
      "step": 41300
    },
    {
      "epoch": 0.28500072440271274,
      "grad_norm": 0.0,
      "learning_rate": 7.151510552132849e-06,
      "loss": 0.0,
      "step": 41310
    },
    {
      "epoch": 0.28506971513725704,
      "grad_norm": 0.6857854723930359,
      "learning_rate": 7.150820644787405e-06,
      "loss": 0.0001,
      "step": 41320
    },
    {
      "epoch": 0.2851387058718014,
      "grad_norm": 4.0679592530068476e-06,
      "learning_rate": 7.150130737441963e-06,
      "loss": 0.0,
      "step": 41330
    },
    {
      "epoch": 0.28520769660634576,
      "grad_norm": 0.0,
      "learning_rate": 7.149440830096519e-06,
      "loss": 0.0,
      "step": 41340
    },
    {
      "epoch": 0.2852766873408901,
      "grad_norm": 0.0,
      "learning_rate": 7.148750922751076e-06,
      "loss": 0.0,
      "step": 41350
    },
    {
      "epoch": 0.2853456780754345,
      "grad_norm": 0.0,
      "learning_rate": 7.148061015405632e-06,
      "loss": 0.0,
      "step": 41360
    },
    {
      "epoch": 0.28541466880997884,
      "grad_norm": 0.0,
      "learning_rate": 7.147371108060188e-06,
      "loss": 0.0012,
      "step": 41370
    },
    {
      "epoch": 0.28548365954452315,
      "grad_norm": 0.0,
      "learning_rate": 7.146681200714744e-06,
      "loss": 0.0,
      "step": 41380
    },
    {
      "epoch": 0.2855526502790675,
      "grad_norm": 0.0,
      "learning_rate": 7.145991293369301e-06,
      "loss": 0.0,
      "step": 41390
    },
    {
      "epoch": 0.28562164101361187,
      "grad_norm": 0.0,
      "learning_rate": 7.145301386023857e-06,
      "loss": 0.0,
      "step": 41400
    },
    {
      "epoch": 0.28569063174815623,
      "grad_norm": 0.0,
      "learning_rate": 7.144611478678414e-06,
      "loss": 0.0,
      "step": 41410
    },
    {
      "epoch": 0.2857596224827006,
      "grad_norm": 0.0,
      "learning_rate": 7.14392157133297e-06,
      "loss": 0.0,
      "step": 41420
    },
    {
      "epoch": 0.2858286132172449,
      "grad_norm": 8.529092432318919e-10,
      "learning_rate": 7.143231663987527e-06,
      "loss": 0.1011,
      "step": 41430
    },
    {
      "epoch": 0.28589760395178926,
      "grad_norm": 0.0,
      "learning_rate": 7.142541756642084e-06,
      "loss": 0.0001,
      "step": 41440
    },
    {
      "epoch": 0.2859665946863336,
      "grad_norm": 0.0,
      "learning_rate": 7.1418518492966406e-06,
      "loss": 0.0,
      "step": 41450
    },
    {
      "epoch": 0.286035585420878,
      "grad_norm": 0.0,
      "learning_rate": 7.141161941951197e-06,
      "loss": 0.0,
      "step": 41460
    },
    {
      "epoch": 0.28610457615542234,
      "grad_norm": 0.0,
      "learning_rate": 7.1404720346057536e-06,
      "loss": 0.0013,
      "step": 41470
    },
    {
      "epoch": 0.2861735668899667,
      "grad_norm": 0.0,
      "learning_rate": 7.139782127260309e-06,
      "loss": 0.0,
      "step": 41480
    },
    {
      "epoch": 0.286242557624511,
      "grad_norm": 4.25914326029897e-09,
      "learning_rate": 7.139092219914866e-06,
      "loss": 1.3399,
      "step": 41490
    },
    {
      "epoch": 0.28631154835905537,
      "grad_norm": 0.0,
      "learning_rate": 7.138402312569422e-06,
      "loss": 0.0,
      "step": 41500
    },
    {
      "epoch": 0.28638053909359973,
      "grad_norm": 0.0,
      "learning_rate": 7.137712405223979e-06,
      "loss": 0.0,
      "step": 41510
    },
    {
      "epoch": 0.2864495298281441,
      "grad_norm": 0.0,
      "learning_rate": 7.137022497878535e-06,
      "loss": 0.0,
      "step": 41520
    },
    {
      "epoch": 0.28651852056268845,
      "grad_norm": 0.0,
      "learning_rate": 7.136332590533092e-06,
      "loss": 0.0,
      "step": 41530
    },
    {
      "epoch": 0.28658751129723276,
      "grad_norm": 0.0,
      "learning_rate": 7.135642683187648e-06,
      "loss": 0.0,
      "step": 41540
    },
    {
      "epoch": 0.2866565020317771,
      "grad_norm": 6.631841529269877e-07,
      "learning_rate": 7.1349527758422055e-06,
      "loss": 0.0,
      "step": 41550
    },
    {
      "epoch": 0.2867254927663215,
      "grad_norm": 0.0,
      "learning_rate": 7.134262868496762e-06,
      "loss": 0.0,
      "step": 41560
    },
    {
      "epoch": 0.28679448350086584,
      "grad_norm": 0.0,
      "learning_rate": 7.1335729611513185e-06,
      "loss": 0.0011,
      "step": 41570
    },
    {
      "epoch": 0.2868634742354102,
      "grad_norm": 0.0,
      "learning_rate": 7.132883053805875e-06,
      "loss": 0.0,
      "step": 41580
    },
    {
      "epoch": 0.28693246496995456,
      "grad_norm": 0.0,
      "learning_rate": 7.1321931464604314e-06,
      "loss": 0.0,
      "step": 41590
    },
    {
      "epoch": 0.28700145570449886,
      "grad_norm": 0.07003507018089294,
      "learning_rate": 7.131503239114987e-06,
      "loss": 0.0,
      "step": 41600
    },
    {
      "epoch": 0.2870704464390432,
      "grad_norm": 0.0,
      "learning_rate": 7.1308133317695436e-06,
      "loss": 0.0,
      "step": 41610
    },
    {
      "epoch": 0.2871394371735876,
      "grad_norm": 2.5664125132607296e-05,
      "learning_rate": 7.1301234244241e-06,
      "loss": 0.0,
      "step": 41620
    },
    {
      "epoch": 0.28720842790813195,
      "grad_norm": 0.0,
      "learning_rate": 7.1294335170786566e-06,
      "loss": 0.0,
      "step": 41630
    },
    {
      "epoch": 0.2872774186426763,
      "grad_norm": 0.0,
      "learning_rate": 7.128743609733213e-06,
      "loss": 0.0,
      "step": 41640
    },
    {
      "epoch": 0.2873464093772206,
      "grad_norm": 0.0,
      "learning_rate": 7.1280537023877695e-06,
      "loss": 0.1031,
      "step": 41650
    },
    {
      "epoch": 0.287415400111765,
      "grad_norm": 0.0,
      "learning_rate": 7.127363795042327e-06,
      "loss": 0.0,
      "step": 41660
    },
    {
      "epoch": 0.28748439084630933,
      "grad_norm": 0.0,
      "learning_rate": 7.126673887696883e-06,
      "loss": 0.0,
      "step": 41670
    },
    {
      "epoch": 0.2875533815808537,
      "grad_norm": 0.0,
      "learning_rate": 7.12598398035144e-06,
      "loss": 0.0,
      "step": 41680
    },
    {
      "epoch": 0.28762237231539806,
      "grad_norm": 0.0,
      "learning_rate": 7.125294073005996e-06,
      "loss": 0.0,
      "step": 41690
    },
    {
      "epoch": 0.2876913630499424,
      "grad_norm": 0.0,
      "learning_rate": 7.124604165660553e-06,
      "loss": 0.0,
      "step": 41700
    },
    {
      "epoch": 0.2877603537844867,
      "grad_norm": 0.007582913618534803,
      "learning_rate": 7.1239142583151085e-06,
      "loss": 0.0,
      "step": 41710
    },
    {
      "epoch": 0.2878293445190311,
      "grad_norm": 0.0,
      "learning_rate": 7.123224350969665e-06,
      "loss": 0.0,
      "step": 41720
    },
    {
      "epoch": 0.28789833525357544,
      "grad_norm": 0.0,
      "learning_rate": 7.1225344436242215e-06,
      "loss": 0.0,
      "step": 41730
    },
    {
      "epoch": 0.2879673259881198,
      "grad_norm": 2.593980207166169e-05,
      "learning_rate": 7.121844536278778e-06,
      "loss": 0.0,
      "step": 41740
    },
    {
      "epoch": 0.28803631672266417,
      "grad_norm": 0.0,
      "learning_rate": 7.1211546289333344e-06,
      "loss": 0.0,
      "step": 41750
    },
    {
      "epoch": 0.28810530745720847,
      "grad_norm": 1.7740355730056763,
      "learning_rate": 7.120464721587891e-06,
      "loss": 0.0003,
      "step": 41760
    },
    {
      "epoch": 0.28817429819175283,
      "grad_norm": 0.0,
      "learning_rate": 7.119774814242448e-06,
      "loss": 0.0429,
      "step": 41770
    },
    {
      "epoch": 0.2882432889262972,
      "grad_norm": 0.788506805896759,
      "learning_rate": 7.119084906897005e-06,
      "loss": 0.0097,
      "step": 41780
    },
    {
      "epoch": 0.28831227966084155,
      "grad_norm": 0.0,
      "learning_rate": 7.118394999551561e-06,
      "loss": 0.0,
      "step": 41790
    },
    {
      "epoch": 0.2883812703953859,
      "grad_norm": 0.0,
      "learning_rate": 7.117705092206118e-06,
      "loss": 0.0,
      "step": 41800
    },
    {
      "epoch": 0.2884502611299303,
      "grad_norm": 0.0,
      "learning_rate": 7.117015184860674e-06,
      "loss": 0.0,
      "step": 41810
    },
    {
      "epoch": 0.2885192518644746,
      "grad_norm": 0.0,
      "learning_rate": 7.11632527751523e-06,
      "loss": 0.0597,
      "step": 41820
    },
    {
      "epoch": 0.28858824259901894,
      "grad_norm": 0.0,
      "learning_rate": 7.115635370169786e-06,
      "loss": 0.0,
      "step": 41830
    },
    {
      "epoch": 0.2886572333335633,
      "grad_norm": 0.0,
      "learning_rate": 7.114945462824343e-06,
      "loss": 0.0,
      "step": 41840
    },
    {
      "epoch": 0.28872622406810766,
      "grad_norm": 0.0,
      "learning_rate": 7.114255555478899e-06,
      "loss": 0.0001,
      "step": 41850
    },
    {
      "epoch": 0.288795214802652,
      "grad_norm": 0.00859209056943655,
      "learning_rate": 7.113565648133456e-06,
      "loss": 0.0,
      "step": 41860
    },
    {
      "epoch": 0.28886420553719633,
      "grad_norm": 0.0,
      "learning_rate": 7.112875740788012e-06,
      "loss": 0.6328,
      "step": 41870
    },
    {
      "epoch": 0.2889331962717407,
      "grad_norm": 0.0,
      "learning_rate": 7.11218583344257e-06,
      "loss": 0.0,
      "step": 41880
    },
    {
      "epoch": 0.28900218700628505,
      "grad_norm": 0.0,
      "learning_rate": 7.111495926097126e-06,
      "loss": 0.0,
      "step": 41890
    },
    {
      "epoch": 0.2890711777408294,
      "grad_norm": 3.907179291218199e-07,
      "learning_rate": 7.110806018751683e-06,
      "loss": 0.0,
      "step": 41900
    },
    {
      "epoch": 0.28914016847537377,
      "grad_norm": 0.0,
      "learning_rate": 7.110116111406239e-06,
      "loss": 0.1059,
      "step": 41910
    },
    {
      "epoch": 0.28920915920991813,
      "grad_norm": 0.0,
      "learning_rate": 7.109426204060796e-06,
      "loss": 0.0,
      "step": 41920
    },
    {
      "epoch": 0.28927814994446244,
      "grad_norm": 0.0,
      "learning_rate": 7.108736296715352e-06,
      "loss": 0.0,
      "step": 41930
    },
    {
      "epoch": 0.2893471406790068,
      "grad_norm": 0.0,
      "learning_rate": 7.108046389369908e-06,
      "loss": 0.0,
      "step": 41940
    },
    {
      "epoch": 0.28941613141355116,
      "grad_norm": 4.362367462817929e-07,
      "learning_rate": 7.107356482024464e-06,
      "loss": 0.0,
      "step": 41950
    },
    {
      "epoch": 0.2894851221480955,
      "grad_norm": 1.1224799436604371e-06,
      "learning_rate": 7.106666574679021e-06,
      "loss": 0.0006,
      "step": 41960
    },
    {
      "epoch": 0.2895541128826399,
      "grad_norm": 0.0,
      "learning_rate": 7.105976667333577e-06,
      "loss": 0.0,
      "step": 41970
    },
    {
      "epoch": 0.2896231036171842,
      "grad_norm": 0.0,
      "learning_rate": 7.105286759988134e-06,
      "loss": 0.0,
      "step": 41980
    },
    {
      "epoch": 0.28969209435172855,
      "grad_norm": 3.174858420607052e-07,
      "learning_rate": 7.104596852642691e-06,
      "loss": 0.0,
      "step": 41990
    },
    {
      "epoch": 0.2897610850862729,
      "grad_norm": 0.0,
      "learning_rate": 7.1039069452972475e-06,
      "loss": 0.0,
      "step": 42000
    },
    {
      "epoch": 0.28983007582081727,
      "grad_norm": 0.0,
      "learning_rate": 7.103217037951804e-06,
      "loss": 0.0,
      "step": 42010
    },
    {
      "epoch": 0.28989906655536163,
      "grad_norm": 0.0,
      "learning_rate": 7.1025271306063605e-06,
      "loss": 0.174,
      "step": 42020
    },
    {
      "epoch": 0.289968057289906,
      "grad_norm": 0.0,
      "learning_rate": 7.101837223260917e-06,
      "loss": 0.0,
      "step": 42030
    },
    {
      "epoch": 0.2900370480244503,
      "grad_norm": 0.0,
      "learning_rate": 7.1011473159154735e-06,
      "loss": 0.0,
      "step": 42040
    },
    {
      "epoch": 0.29010603875899466,
      "grad_norm": 0.0,
      "learning_rate": 7.100457408570029e-06,
      "loss": 0.0,
      "step": 42050
    },
    {
      "epoch": 0.290175029493539,
      "grad_norm": 1.8941272994510427e-09,
      "learning_rate": 7.099767501224586e-06,
      "loss": 0.0,
      "step": 42060
    },
    {
      "epoch": 0.2902440202280834,
      "grad_norm": 1.6150670489878394e-05,
      "learning_rate": 7.099077593879142e-06,
      "loss": 0.0,
      "step": 42070
    },
    {
      "epoch": 0.29031301096262774,
      "grad_norm": 2.188157850468997e-05,
      "learning_rate": 7.098387686533699e-06,
      "loss": 0.0,
      "step": 42080
    },
    {
      "epoch": 0.29038200169717204,
      "grad_norm": 0.0,
      "learning_rate": 7.097697779188255e-06,
      "loss": 0.0,
      "step": 42090
    },
    {
      "epoch": 0.2904509924317164,
      "grad_norm": 0.0,
      "learning_rate": 7.0970078718428124e-06,
      "loss": 0.0,
      "step": 42100
    },
    {
      "epoch": 0.29051998316626076,
      "grad_norm": 1.5726680430816486e-05,
      "learning_rate": 7.096317964497369e-06,
      "loss": 0.0,
      "step": 42110
    },
    {
      "epoch": 0.2905889739008051,
      "grad_norm": 0.002558254636824131,
      "learning_rate": 7.0956280571519254e-06,
      "loss": 0.0,
      "step": 42120
    },
    {
      "epoch": 0.2906579646353495,
      "grad_norm": 0.0,
      "learning_rate": 7.094938149806482e-06,
      "loss": 0.0001,
      "step": 42130
    },
    {
      "epoch": 0.29072695536989385,
      "grad_norm": 0.0,
      "learning_rate": 7.094248242461038e-06,
      "loss": 0.001,
      "step": 42140
    },
    {
      "epoch": 0.29079594610443815,
      "grad_norm": 0.14574797451496124,
      "learning_rate": 7.093558335115595e-06,
      "loss": 0.2092,
      "step": 42150
    },
    {
      "epoch": 0.2908649368389825,
      "grad_norm": 7.462674034286465e-07,
      "learning_rate": 7.0928684277701505e-06,
      "loss": 0.0,
      "step": 42160
    },
    {
      "epoch": 0.2909339275735269,
      "grad_norm": 0.0,
      "learning_rate": 7.092178520424707e-06,
      "loss": 0.0011,
      "step": 42170
    },
    {
      "epoch": 0.29100291830807123,
      "grad_norm": 0.0,
      "learning_rate": 7.0914886130792635e-06,
      "loss": 0.0,
      "step": 42180
    },
    {
      "epoch": 0.2910719090426156,
      "grad_norm": 0.0,
      "learning_rate": 7.09079870573382e-06,
      "loss": 0.0,
      "step": 42190
    },
    {
      "epoch": 0.2911408997771599,
      "grad_norm": 0.0,
      "learning_rate": 7.0901087983883765e-06,
      "loss": 0.0,
      "step": 42200
    },
    {
      "epoch": 0.29120989051170426,
      "grad_norm": 1.726275877445005e-05,
      "learning_rate": 7.089418891042934e-06,
      "loss": 0.3881,
      "step": 42210
    },
    {
      "epoch": 0.2912788812462486,
      "grad_norm": 0.0,
      "learning_rate": 7.08872898369749e-06,
      "loss": 0.0,
      "step": 42220
    },
    {
      "epoch": 0.291347871980793,
      "grad_norm": 0.0,
      "learning_rate": 7.088039076352047e-06,
      "loss": 0.0,
      "step": 42230
    },
    {
      "epoch": 0.29141686271533734,
      "grad_norm": 0.0,
      "learning_rate": 7.087349169006603e-06,
      "loss": 0.0,
      "step": 42240
    },
    {
      "epoch": 0.2914858534498817,
      "grad_norm": 7.967135502440215e-07,
      "learning_rate": 7.08665926166116e-06,
      "loss": 0.0,
      "step": 42250
    },
    {
      "epoch": 0.291554844184426,
      "grad_norm": 0.0,
      "learning_rate": 7.085969354315716e-06,
      "loss": 0.0002,
      "step": 42260
    },
    {
      "epoch": 0.29162383491897037,
      "grad_norm": 1.3892648098590143e-07,
      "learning_rate": 7.085279446970273e-06,
      "loss": 0.0,
      "step": 42270
    },
    {
      "epoch": 0.29169282565351473,
      "grad_norm": 0.0,
      "learning_rate": 7.0845895396248284e-06,
      "loss": 0.0,
      "step": 42280
    },
    {
      "epoch": 0.2917618163880591,
      "grad_norm": 0.0,
      "learning_rate": 7.083899632279385e-06,
      "loss": 0.0,
      "step": 42290
    },
    {
      "epoch": 0.29183080712260345,
      "grad_norm": 0.0,
      "learning_rate": 7.083209724933941e-06,
      "loss": 0.0001,
      "step": 42300
    },
    {
      "epoch": 0.29189979785714776,
      "grad_norm": 0.0,
      "learning_rate": 7.082519817588498e-06,
      "loss": 0.0,
      "step": 42310
    },
    {
      "epoch": 0.2919687885916921,
      "grad_norm": 0.0,
      "learning_rate": 7.081829910243055e-06,
      "loss": 0.0,
      "step": 42320
    },
    {
      "epoch": 0.2920377793262365,
      "grad_norm": 0.0,
      "learning_rate": 7.081140002897612e-06,
      "loss": 0.0,
      "step": 42330
    },
    {
      "epoch": 0.29210677006078084,
      "grad_norm": 0.0,
      "learning_rate": 7.080450095552168e-06,
      "loss": 0.6539,
      "step": 42340
    },
    {
      "epoch": 0.2921757607953252,
      "grad_norm": 0.0,
      "learning_rate": 7.079760188206725e-06,
      "loss": 0.0,
      "step": 42350
    },
    {
      "epoch": 0.29224475152986956,
      "grad_norm": 0.0,
      "learning_rate": 7.079070280861281e-06,
      "loss": 0.0,
      "step": 42360
    },
    {
      "epoch": 0.29231374226441387,
      "grad_norm": 0.0,
      "learning_rate": 7.078380373515838e-06,
      "loss": 0.0045,
      "step": 42370
    },
    {
      "epoch": 0.29238273299895823,
      "grad_norm": 0.0,
      "learning_rate": 7.077690466170394e-06,
      "loss": 0.0,
      "step": 42380
    },
    {
      "epoch": 0.2924517237335026,
      "grad_norm": 2.979498503918876e-06,
      "learning_rate": 7.07700055882495e-06,
      "loss": 0.0,
      "step": 42390
    },
    {
      "epoch": 0.29252071446804695,
      "grad_norm": 0.0,
      "learning_rate": 7.076310651479506e-06,
      "loss": 0.0,
      "step": 42400
    },
    {
      "epoch": 0.2925897052025913,
      "grad_norm": 0.0,
      "learning_rate": 7.075620744134063e-06,
      "loss": 0.0001,
      "step": 42410
    },
    {
      "epoch": 0.2926586959371356,
      "grad_norm": 0.0,
      "learning_rate": 7.074930836788619e-06,
      "loss": 0.0,
      "step": 42420
    },
    {
      "epoch": 0.29272768667168,
      "grad_norm": 0.0,
      "learning_rate": 7.074240929443177e-06,
      "loss": 0.0,
      "step": 42430
    },
    {
      "epoch": 0.29279667740622434,
      "grad_norm": 0.0,
      "learning_rate": 7.073551022097733e-06,
      "loss": 0.0,
      "step": 42440
    },
    {
      "epoch": 0.2928656681407687,
      "grad_norm": 0.00047385727521032095,
      "learning_rate": 7.07286111475229e-06,
      "loss": 0.878,
      "step": 42450
    },
    {
      "epoch": 0.29293465887531306,
      "grad_norm": 0.0,
      "learning_rate": 7.072171207406846e-06,
      "loss": 0.0013,
      "step": 42460
    },
    {
      "epoch": 0.2930036496098574,
      "grad_norm": 0.0,
      "learning_rate": 7.071481300061403e-06,
      "loss": 0.0001,
      "step": 42470
    },
    {
      "epoch": 0.2930726403444017,
      "grad_norm": 8.396614074707031,
      "learning_rate": 7.070791392715959e-06,
      "loss": 1.1338,
      "step": 42480
    },
    {
      "epoch": 0.2931416310789461,
      "grad_norm": 0.0,
      "learning_rate": 7.0701014853705156e-06,
      "loss": 0.0,
      "step": 42490
    },
    {
      "epoch": 0.29321062181349045,
      "grad_norm": 0.0,
      "learning_rate": 7.069411578025071e-06,
      "loss": 0.0,
      "step": 42500
    },
    {
      "epoch": 0.2932796125480348,
      "grad_norm": 0.0,
      "learning_rate": 7.068721670679628e-06,
      "loss": 0.2036,
      "step": 42510
    },
    {
      "epoch": 0.29334860328257917,
      "grad_norm": 0.0,
      "learning_rate": 7.068031763334184e-06,
      "loss": 0.0,
      "step": 42520
    },
    {
      "epoch": 0.2934175940171235,
      "grad_norm": 0.0,
      "learning_rate": 7.067341855988741e-06,
      "loss": 0.0693,
      "step": 42530
    },
    {
      "epoch": 0.29348658475166783,
      "grad_norm": 3.40055771630432e-06,
      "learning_rate": 7.066651948643298e-06,
      "loss": 0.0,
      "step": 42540
    },
    {
      "epoch": 0.2935555754862122,
      "grad_norm": 0.0,
      "learning_rate": 7.0659620412978545e-06,
      "loss": 0.0,
      "step": 42550
    },
    {
      "epoch": 0.29362456622075656,
      "grad_norm": 0.0,
      "learning_rate": 7.065272133952411e-06,
      "loss": 0.0,
      "step": 42560
    },
    {
      "epoch": 0.2936935569553009,
      "grad_norm": 4.543967702108631e-10,
      "learning_rate": 7.0645822266069675e-06,
      "loss": 0.0,
      "step": 42570
    },
    {
      "epoch": 0.2937625476898453,
      "grad_norm": 9.44025635440937e-10,
      "learning_rate": 7.063892319261524e-06,
      "loss": 0.0,
      "step": 42580
    },
    {
      "epoch": 0.2938315384243896,
      "grad_norm": 0.0013908512191846967,
      "learning_rate": 7.0632024119160805e-06,
      "loss": 0.191,
      "step": 42590
    },
    {
      "epoch": 0.29390052915893394,
      "grad_norm": 0.0,
      "learning_rate": 7.062512504570637e-06,
      "loss": 0.0,
      "step": 42600
    },
    {
      "epoch": 0.2939695198934783,
      "grad_norm": 0.003961845766752958,
      "learning_rate": 7.0618225972251935e-06,
      "loss": 0.0003,
      "step": 42610
    },
    {
      "epoch": 0.29403851062802266,
      "grad_norm": 0.0,
      "learning_rate": 7.061132689879749e-06,
      "loss": 0.0,
      "step": 42620
    },
    {
      "epoch": 0.294107501362567,
      "grad_norm": 0.0,
      "learning_rate": 7.060442782534306e-06,
      "loss": 0.0036,
      "step": 42630
    },
    {
      "epoch": 0.29417649209711133,
      "grad_norm": 0.0,
      "learning_rate": 7.059752875188862e-06,
      "loss": 0.0,
      "step": 42640
    },
    {
      "epoch": 0.2942454828316557,
      "grad_norm": 0.0,
      "learning_rate": 7.059062967843419e-06,
      "loss": 0.0019,
      "step": 42650
    },
    {
      "epoch": 0.29431447356620005,
      "grad_norm": 0.0,
      "learning_rate": 7.058373060497976e-06,
      "loss": 0.0,
      "step": 42660
    },
    {
      "epoch": 0.2943834643007444,
      "grad_norm": 9.014798365569732e-07,
      "learning_rate": 7.057683153152532e-06,
      "loss": 0.0,
      "step": 42670
    },
    {
      "epoch": 0.2944524550352888,
      "grad_norm": 1.2542999982833862,
      "learning_rate": 7.056993245807089e-06,
      "loss": 0.0005,
      "step": 42680
    },
    {
      "epoch": 0.29452144576983313,
      "grad_norm": 0.0,
      "learning_rate": 7.056303338461645e-06,
      "loss": 0.1514,
      "step": 42690
    },
    {
      "epoch": 0.29459043650437744,
      "grad_norm": 0.004014682490378618,
      "learning_rate": 7.055613431116202e-06,
      "loss": 0.0,
      "step": 42700
    },
    {
      "epoch": 0.2946594272389218,
      "grad_norm": 0.0,
      "learning_rate": 7.054923523770758e-06,
      "loss": 0.0,
      "step": 42710
    },
    {
      "epoch": 0.29472841797346616,
      "grad_norm": 0.0,
      "learning_rate": 7.054233616425315e-06,
      "loss": 0.0,
      "step": 42720
    },
    {
      "epoch": 0.2947974087080105,
      "grad_norm": 0.0,
      "learning_rate": 7.0535437090798705e-06,
      "loss": 0.0,
      "step": 42730
    },
    {
      "epoch": 0.2948663994425549,
      "grad_norm": 0.0,
      "learning_rate": 7.052853801734427e-06,
      "loss": 0.0,
      "step": 42740
    },
    {
      "epoch": 0.2949353901770992,
      "grad_norm": 7.616115181008354e-05,
      "learning_rate": 7.0521638943889835e-06,
      "loss": 0.0,
      "step": 42750
    },
    {
      "epoch": 0.29500438091164355,
      "grad_norm": 0.00018584623467177153,
      "learning_rate": 7.051473987043541e-06,
      "loss": 0.0,
      "step": 42760
    },
    {
      "epoch": 0.2950733716461879,
      "grad_norm": 2.0667194178258796e-08,
      "learning_rate": 7.050784079698097e-06,
      "loss": 0.0,
      "step": 42770
    },
    {
      "epoch": 0.29514236238073227,
      "grad_norm": 0.0,
      "learning_rate": 7.050094172352654e-06,
      "loss": 0.0,
      "step": 42780
    },
    {
      "epoch": 0.29521135311527663,
      "grad_norm": 0.0001325804478256032,
      "learning_rate": 7.04940426500721e-06,
      "loss": 0.0,
      "step": 42790
    },
    {
      "epoch": 0.295280343849821,
      "grad_norm": 0.0,
      "learning_rate": 7.048714357661767e-06,
      "loss": 0.0,
      "step": 42800
    },
    {
      "epoch": 0.2953493345843653,
      "grad_norm": 0.0,
      "learning_rate": 7.048024450316323e-06,
      "loss": 0.0,
      "step": 42810
    },
    {
      "epoch": 0.29541832531890966,
      "grad_norm": 0.00033801706740632653,
      "learning_rate": 7.04733454297088e-06,
      "loss": 0.6336,
      "step": 42820
    },
    {
      "epoch": 0.295487316053454,
      "grad_norm": 0.0,
      "learning_rate": 7.046644635625436e-06,
      "loss": 0.0,
      "step": 42830
    },
    {
      "epoch": 0.2955563067879984,
      "grad_norm": 0.0,
      "learning_rate": 7.045954728279992e-06,
      "loss": 0.0,
      "step": 42840
    },
    {
      "epoch": 0.29562529752254274,
      "grad_norm": 0.0,
      "learning_rate": 7.045264820934548e-06,
      "loss": 0.0,
      "step": 42850
    },
    {
      "epoch": 0.29569428825708705,
      "grad_norm": 0.0,
      "learning_rate": 7.044574913589105e-06,
      "loss": 0.0002,
      "step": 42860
    },
    {
      "epoch": 0.2957632789916314,
      "grad_norm": 0.00027899316046386957,
      "learning_rate": 7.043885006243662e-06,
      "loss": 0.0,
      "step": 42870
    },
    {
      "epoch": 0.29583226972617577,
      "grad_norm": 4.4702400114893237e-10,
      "learning_rate": 7.043195098898219e-06,
      "loss": 0.0,
      "step": 42880
    },
    {
      "epoch": 0.29590126046072013,
      "grad_norm": 0.0,
      "learning_rate": 7.042505191552775e-06,
      "loss": 0.0,
      "step": 42890
    },
    {
      "epoch": 0.2959702511952645,
      "grad_norm": 0.0,
      "learning_rate": 7.041815284207332e-06,
      "loss": 0.0,
      "step": 42900
    },
    {
      "epoch": 0.29603924192980885,
      "grad_norm": 0.0,
      "learning_rate": 7.041125376861888e-06,
      "loss": 0.0,
      "step": 42910
    },
    {
      "epoch": 0.29610823266435315,
      "grad_norm": 2.475000870916233e-09,
      "learning_rate": 7.040435469516445e-06,
      "loss": 0.0,
      "step": 42920
    },
    {
      "epoch": 0.2961772233988975,
      "grad_norm": 0.08551587164402008,
      "learning_rate": 7.039745562171001e-06,
      "loss": 0.0,
      "step": 42930
    },
    {
      "epoch": 0.2962462141334419,
      "grad_norm": 0.0,
      "learning_rate": 7.039055654825558e-06,
      "loss": 0.0001,
      "step": 42940
    },
    {
      "epoch": 0.29631520486798624,
      "grad_norm": 4.688519017470583e-10,
      "learning_rate": 7.038365747480113e-06,
      "loss": 0.0,
      "step": 42950
    },
    {
      "epoch": 0.2963841956025306,
      "grad_norm": 2.0624385797418654e-05,
      "learning_rate": 7.03767584013467e-06,
      "loss": 0.0,
      "step": 42960
    },
    {
      "epoch": 0.2964531863370749,
      "grad_norm": 0.0,
      "learning_rate": 7.036985932789226e-06,
      "loss": 0.0,
      "step": 42970
    },
    {
      "epoch": 0.29652217707161926,
      "grad_norm": 0.0,
      "learning_rate": 7.036296025443784e-06,
      "loss": 0.0,
      "step": 42980
    },
    {
      "epoch": 0.2965911678061636,
      "grad_norm": 0.0,
      "learning_rate": 7.03560611809834e-06,
      "loss": 0.0,
      "step": 42990
    },
    {
      "epoch": 0.296660158540708,
      "grad_norm": 1.1841663763334509e-05,
      "learning_rate": 7.034916210752897e-06,
      "loss": 0.0,
      "step": 43000
    },
    {
      "epoch": 0.29672914927525235,
      "grad_norm": 0.05389532074332237,
      "learning_rate": 7.034226303407453e-06,
      "loss": 0.5758,
      "step": 43010
    },
    {
      "epoch": 0.2967981400097967,
      "grad_norm": 0.0,
      "learning_rate": 7.0335363960620096e-06,
      "loss": 0.0003,
      "step": 43020
    },
    {
      "epoch": 0.296867130744341,
      "grad_norm": 0.0,
      "learning_rate": 7.032846488716566e-06,
      "loss": 0.0041,
      "step": 43030
    },
    {
      "epoch": 0.2969361214788854,
      "grad_norm": 1.9829495556678012e-07,
      "learning_rate": 7.0321565813711225e-06,
      "loss": 0.0,
      "step": 43040
    },
    {
      "epoch": 0.29700511221342973,
      "grad_norm": 0.0,
      "learning_rate": 7.031466674025679e-06,
      "loss": 0.0,
      "step": 43050
    },
    {
      "epoch": 0.2970741029479741,
      "grad_norm": 0.0,
      "learning_rate": 7.0307767666802355e-06,
      "loss": 0.0,
      "step": 43060
    },
    {
      "epoch": 0.29714309368251846,
      "grad_norm": 0.00023221809533424675,
      "learning_rate": 7.030086859334791e-06,
      "loss": 0.5473,
      "step": 43070
    },
    {
      "epoch": 0.2972120844170628,
      "grad_norm": 0.274690181016922,
      "learning_rate": 7.029396951989348e-06,
      "loss": 0.0001,
      "step": 43080
    },
    {
      "epoch": 0.2972810751516071,
      "grad_norm": 0.0,
      "learning_rate": 7.028707044643905e-06,
      "loss": 0.0,
      "step": 43090
    },
    {
      "epoch": 0.2973500658861515,
      "grad_norm": 0.0,
      "learning_rate": 7.0280171372984615e-06,
      "loss": 0.0,
      "step": 43100
    },
    {
      "epoch": 0.29741905662069584,
      "grad_norm": 0.0,
      "learning_rate": 7.027327229953018e-06,
      "loss": 0.0,
      "step": 43110
    },
    {
      "epoch": 0.2974880473552402,
      "grad_norm": 0.0,
      "learning_rate": 7.0266373226075745e-06,
      "loss": 0.0,
      "step": 43120
    },
    {
      "epoch": 0.29755703808978456,
      "grad_norm": 0.0,
      "learning_rate": 7.025947415262131e-06,
      "loss": 0.0,
      "step": 43130
    },
    {
      "epoch": 0.29762602882432887,
      "grad_norm": 0.0,
      "learning_rate": 7.0252575079166874e-06,
      "loss": 0.0,
      "step": 43140
    },
    {
      "epoch": 0.29769501955887323,
      "grad_norm": 0.0,
      "learning_rate": 7.024567600571244e-06,
      "loss": 0.0,
      "step": 43150
    },
    {
      "epoch": 0.2977640102934176,
      "grad_norm": 0.0,
      "learning_rate": 7.0238776932258004e-06,
      "loss": 0.0,
      "step": 43160
    },
    {
      "epoch": 0.29783300102796195,
      "grad_norm": 9.832288014877122e-06,
      "learning_rate": 7.023187785880357e-06,
      "loss": 0.0974,
      "step": 43170
    },
    {
      "epoch": 0.2979019917625063,
      "grad_norm": 0.0,
      "learning_rate": 7.0224978785349126e-06,
      "loss": 0.0,
      "step": 43180
    },
    {
      "epoch": 0.2979709824970507,
      "grad_norm": 1.982948916179339e-09,
      "learning_rate": 7.021807971189469e-06,
      "loss": 0.0094,
      "step": 43190
    },
    {
      "epoch": 0.298039973231595,
      "grad_norm": 0.0,
      "learning_rate": 7.021118063844027e-06,
      "loss": 0.0,
      "step": 43200
    },
    {
      "epoch": 0.29810896396613934,
      "grad_norm": 0.0,
      "learning_rate": 7.020428156498583e-06,
      "loss": 0.0,
      "step": 43210
    },
    {
      "epoch": 0.2981779547006837,
      "grad_norm": 0.0,
      "learning_rate": 7.019738249153139e-06,
      "loss": 0.0,
      "step": 43220
    },
    {
      "epoch": 0.29824694543522806,
      "grad_norm": 0.018246019259095192,
      "learning_rate": 7.019048341807696e-06,
      "loss": 0.0,
      "step": 43230
    },
    {
      "epoch": 0.2983159361697724,
      "grad_norm": 0.0,
      "learning_rate": 7.018358434462252e-06,
      "loss": 0.0,
      "step": 43240
    },
    {
      "epoch": 0.2983849269043167,
      "grad_norm": 0.0,
      "learning_rate": 7.017668527116809e-06,
      "loss": 0.0003,
      "step": 43250
    },
    {
      "epoch": 0.2984539176388611,
      "grad_norm": 0.0,
      "learning_rate": 7.016978619771365e-06,
      "loss": 0.0,
      "step": 43260
    },
    {
      "epoch": 0.29852290837340545,
      "grad_norm": 0.0,
      "learning_rate": 7.016288712425922e-06,
      "loss": 0.0,
      "step": 43270
    },
    {
      "epoch": 0.2985918991079498,
      "grad_norm": 0.02237069420516491,
      "learning_rate": 7.015598805080478e-06,
      "loss": 0.0,
      "step": 43280
    },
    {
      "epoch": 0.29866088984249417,
      "grad_norm": 0.0,
      "learning_rate": 7.014908897735034e-06,
      "loss": 0.0002,
      "step": 43290
    },
    {
      "epoch": 0.29872988057703853,
      "grad_norm": 0.0,
      "learning_rate": 7.0142189903895904e-06,
      "loss": 0.0,
      "step": 43300
    },
    {
      "epoch": 0.29879887131158284,
      "grad_norm": 0.0421195812523365,
      "learning_rate": 7.013529083044149e-06,
      "loss": 0.0,
      "step": 43310
    },
    {
      "epoch": 0.2988678620461272,
      "grad_norm": 8.994599509648538e-10,
      "learning_rate": 7.012839175698704e-06,
      "loss": 0.0,
      "step": 43320
    },
    {
      "epoch": 0.29893685278067156,
      "grad_norm": 0.0,
      "learning_rate": 7.012149268353261e-06,
      "loss": 0.0,
      "step": 43330
    },
    {
      "epoch": 0.2990058435152159,
      "grad_norm": 0.0,
      "learning_rate": 7.011459361007817e-06,
      "loss": 0.0,
      "step": 43340
    },
    {
      "epoch": 0.2990748342497603,
      "grad_norm": 0.0,
      "learning_rate": 7.010769453662374e-06,
      "loss": 0.0,
      "step": 43350
    },
    {
      "epoch": 0.2991438249843046,
      "grad_norm": 0.0,
      "learning_rate": 7.01007954631693e-06,
      "loss": 0.0,
      "step": 43360
    },
    {
      "epoch": 0.29921281571884895,
      "grad_norm": 0.001235152711160481,
      "learning_rate": 7.009389638971487e-06,
      "loss": 0.0,
      "step": 43370
    },
    {
      "epoch": 0.2992818064533933,
      "grad_norm": 4.251310969927147e-10,
      "learning_rate": 7.008699731626043e-06,
      "loss": 0.0,
      "step": 43380
    },
    {
      "epoch": 0.29935079718793767,
      "grad_norm": 0.0,
      "learning_rate": 7.0080098242806e-06,
      "loss": 0.0,
      "step": 43390
    },
    {
      "epoch": 0.29941978792248203,
      "grad_norm": 0.0,
      "learning_rate": 7.007319916935156e-06,
      "loss": 0.0,
      "step": 43400
    },
    {
      "epoch": 0.2994887786570264,
      "grad_norm": 0.5563706755638123,
      "learning_rate": 7.006630009589712e-06,
      "loss": 0.0002,
      "step": 43410
    },
    {
      "epoch": 0.2995577693915707,
      "grad_norm": 3.5235859741078457e-07,
      "learning_rate": 7.00594010224427e-06,
      "loss": 0.0001,
      "step": 43420
    },
    {
      "epoch": 0.29962676012611505,
      "grad_norm": 0.0,
      "learning_rate": 7.005250194898826e-06,
      "loss": 0.0001,
      "step": 43430
    },
    {
      "epoch": 0.2996957508606594,
      "grad_norm": 0.0,
      "learning_rate": 7.004560287553382e-06,
      "loss": 0.0,
      "step": 43440
    },
    {
      "epoch": 0.2997647415952038,
      "grad_norm": 0.0,
      "learning_rate": 7.003870380207939e-06,
      "loss": 0.0,
      "step": 43450
    },
    {
      "epoch": 0.29983373232974814,
      "grad_norm": 0.0,
      "learning_rate": 7.003180472862495e-06,
      "loss": 0.0,
      "step": 43460
    },
    {
      "epoch": 0.29990272306429244,
      "grad_norm": 0.0,
      "learning_rate": 7.002490565517052e-06,
      "loss": 0.0,
      "step": 43470
    },
    {
      "epoch": 0.2999717137988368,
      "grad_norm": 0.0,
      "learning_rate": 7.001800658171608e-06,
      "loss": 0.0,
      "step": 43480
    },
    {
      "epoch": 0.30004070453338116,
      "grad_norm": 0.0,
      "learning_rate": 7.001110750826165e-06,
      "loss": 0.0,
      "step": 43490
    },
    {
      "epoch": 0.3001096952679255,
      "grad_norm": 6.271287045223062e-09,
      "learning_rate": 7.000420843480721e-06,
      "loss": 0.0,
      "step": 43500
    },
    {
      "epoch": 0.3001786860024699,
      "grad_norm": 0.0,
      "learning_rate": 6.999730936135278e-06,
      "loss": 0.0098,
      "step": 43510
    },
    {
      "epoch": 0.30024767673701425,
      "grad_norm": 0.0,
      "learning_rate": 6.999041028789833e-06,
      "loss": 0.0,
      "step": 43520
    },
    {
      "epoch": 0.30031666747155855,
      "grad_norm": 0.0,
      "learning_rate": 6.998351121444391e-06,
      "loss": 0.0,
      "step": 43530
    },
    {
      "epoch": 0.3003856582061029,
      "grad_norm": 0.00022184659610502422,
      "learning_rate": 6.997661214098948e-06,
      "loss": 0.0,
      "step": 43540
    },
    {
      "epoch": 0.3004546489406473,
      "grad_norm": 4.742188308703987e-10,
      "learning_rate": 6.9969713067535035e-06,
      "loss": 0.0,
      "step": 43550
    },
    {
      "epoch": 0.30052363967519163,
      "grad_norm": 0.0,
      "learning_rate": 6.99628139940806e-06,
      "loss": 0.0,
      "step": 43560
    },
    {
      "epoch": 0.300592630409736,
      "grad_norm": 0.0,
      "learning_rate": 6.9955914920626165e-06,
      "loss": 0.0,
      "step": 43570
    },
    {
      "epoch": 0.3006616211442803,
      "grad_norm": 0.0,
      "learning_rate": 6.994901584717173e-06,
      "loss": 0.0001,
      "step": 43580
    },
    {
      "epoch": 0.30073061187882466,
      "grad_norm": 0.0,
      "learning_rate": 6.9942116773717295e-06,
      "loss": 0.0003,
      "step": 43590
    },
    {
      "epoch": 0.300799602613369,
      "grad_norm": 0.0,
      "learning_rate": 6.993521770026286e-06,
      "loss": 0.0001,
      "step": 43600
    },
    {
      "epoch": 0.3008685933479134,
      "grad_norm": 0.0,
      "learning_rate": 6.9928318626808425e-06,
      "loss": 0.0001,
      "step": 43610
    },
    {
      "epoch": 0.30093758408245774,
      "grad_norm": 69.81422424316406,
      "learning_rate": 6.992141955335399e-06,
      "loss": 0.0079,
      "step": 43620
    },
    {
      "epoch": 0.3010065748170021,
      "grad_norm": 0.0,
      "learning_rate": 6.991452047989955e-06,
      "loss": 0.0,
      "step": 43630
    },
    {
      "epoch": 0.3010755655515464,
      "grad_norm": 0.09207481145858765,
      "learning_rate": 6.990762140644513e-06,
      "loss": 0.0001,
      "step": 43640
    },
    {
      "epoch": 0.30114455628609077,
      "grad_norm": 8.705681102583185e-05,
      "learning_rate": 6.990072233299069e-06,
      "loss": 0.0,
      "step": 43650
    },
    {
      "epoch": 0.30121354702063513,
      "grad_norm": 0.0,
      "learning_rate": 6.989382325953625e-06,
      "loss": 0.0004,
      "step": 43660
    },
    {
      "epoch": 0.3012825377551795,
      "grad_norm": 6.277215902628086e-08,
      "learning_rate": 6.9886924186081814e-06,
      "loss": 0.0,
      "step": 43670
    },
    {
      "epoch": 0.30135152848972385,
      "grad_norm": 0.47472986578941345,
      "learning_rate": 6.988002511262738e-06,
      "loss": 0.2183,
      "step": 43680
    },
    {
      "epoch": 0.30142051922426816,
      "grad_norm": 0.0,
      "learning_rate": 6.987312603917294e-06,
      "loss": 0.0,
      "step": 43690
    },
    {
      "epoch": 0.3014895099588125,
      "grad_norm": 0.0,
      "learning_rate": 6.986622696571851e-06,
      "loss": 0.0,
      "step": 43700
    },
    {
      "epoch": 0.3015585006933569,
      "grad_norm": 0.255240797996521,
      "learning_rate": 6.985932789226407e-06,
      "loss": 0.0,
      "step": 43710
    },
    {
      "epoch": 0.30162749142790124,
      "grad_norm": 0.0,
      "learning_rate": 6.985242881880964e-06,
      "loss": 0.0,
      "step": 43720
    },
    {
      "epoch": 0.3016964821624456,
      "grad_norm": 0.0,
      "learning_rate": 6.98455297453552e-06,
      "loss": 0.0008,
      "step": 43730
    },
    {
      "epoch": 0.30176547289698996,
      "grad_norm": 3.417747575440444e-05,
      "learning_rate": 6.983863067190077e-06,
      "loss": 0.0001,
      "step": 43740
    },
    {
      "epoch": 0.30183446363153427,
      "grad_norm": 0.0,
      "learning_rate": 6.983173159844634e-06,
      "loss": 0.0013,
      "step": 43750
    },
    {
      "epoch": 0.3019034543660786,
      "grad_norm": 0.0,
      "learning_rate": 6.982483252499191e-06,
      "loss": 0.0,
      "step": 43760
    },
    {
      "epoch": 0.301972445100623,
      "grad_norm": 102.90641021728516,
      "learning_rate": 6.981793345153746e-06,
      "loss": 0.0115,
      "step": 43770
    },
    {
      "epoch": 0.30204143583516735,
      "grad_norm": 0.0,
      "learning_rate": 6.981103437808303e-06,
      "loss": 0.0,
      "step": 43780
    },
    {
      "epoch": 0.3021104265697117,
      "grad_norm": 0.0,
      "learning_rate": 6.980413530462859e-06,
      "loss": 0.0,
      "step": 43790
    },
    {
      "epoch": 0.302179417304256,
      "grad_norm": 0.0,
      "learning_rate": 6.979723623117416e-06,
      "loss": 0.0,
      "step": 43800
    },
    {
      "epoch": 0.3022484080388004,
      "grad_norm": 0.0,
      "learning_rate": 6.979033715771972e-06,
      "loss": 0.0,
      "step": 43810
    },
    {
      "epoch": 0.30231739877334474,
      "grad_norm": 0.0,
      "learning_rate": 6.978343808426529e-06,
      "loss": 0.0,
      "step": 43820
    },
    {
      "epoch": 0.3023863895078891,
      "grad_norm": 4.636920292178104e-10,
      "learning_rate": 6.977653901081085e-06,
      "loss": 0.0,
      "step": 43830
    },
    {
      "epoch": 0.30245538024243346,
      "grad_norm": 0.0,
      "learning_rate": 6.976963993735642e-06,
      "loss": 0.0,
      "step": 43840
    },
    {
      "epoch": 0.3025243709769778,
      "grad_norm": 0.0,
      "learning_rate": 6.976274086390198e-06,
      "loss": 0.0,
      "step": 43850
    },
    {
      "epoch": 0.3025933617115221,
      "grad_norm": 1.4008020343680982e-06,
      "learning_rate": 6.975584179044756e-06,
      "loss": 0.0007,
      "step": 43860
    },
    {
      "epoch": 0.3026623524460665,
      "grad_norm": 0.0,
      "learning_rate": 6.974894271699312e-06,
      "loss": 0.0,
      "step": 43870
    },
    {
      "epoch": 0.30273134318061085,
      "grad_norm": 0.0,
      "learning_rate": 6.974204364353869e-06,
      "loss": 0.0419,
      "step": 43880
    },
    {
      "epoch": 0.3028003339151552,
      "grad_norm": 2.9344546419451945e-05,
      "learning_rate": 6.973514457008424e-06,
      "loss": 0.0,
      "step": 43890
    },
    {
      "epoch": 0.30286932464969957,
      "grad_norm": 0.0,
      "learning_rate": 6.972824549662981e-06,
      "loss": 0.0,
      "step": 43900
    },
    {
      "epoch": 0.3029383153842439,
      "grad_norm": 0.0,
      "learning_rate": 6.972134642317537e-06,
      "loss": 0.0,
      "step": 43910
    },
    {
      "epoch": 0.30300730611878823,
      "grad_norm": 0.0,
      "learning_rate": 6.971444734972094e-06,
      "loss": 0.0,
      "step": 43920
    },
    {
      "epoch": 0.3030762968533326,
      "grad_norm": 0.0,
      "learning_rate": 6.97075482762665e-06,
      "loss": 0.0001,
      "step": 43930
    },
    {
      "epoch": 0.30314528758787695,
      "grad_norm": 0.0,
      "learning_rate": 6.970064920281207e-06,
      "loss": 0.0,
      "step": 43940
    },
    {
      "epoch": 0.3032142783224213,
      "grad_norm": 0.02213207818567753,
      "learning_rate": 6.969375012935763e-06,
      "loss": 0.4531,
      "step": 43950
    },
    {
      "epoch": 0.3032832690569657,
      "grad_norm": 0.0,
      "learning_rate": 6.96868510559032e-06,
      "loss": 0.0,
      "step": 43960
    },
    {
      "epoch": 0.30335225979151,
      "grad_norm": 0.0002772458828985691,
      "learning_rate": 6.967995198244877e-06,
      "loss": 0.0113,
      "step": 43970
    },
    {
      "epoch": 0.30342125052605434,
      "grad_norm": 0.0,
      "learning_rate": 6.9673052908994335e-06,
      "loss": 0.0011,
      "step": 43980
    },
    {
      "epoch": 0.3034902412605987,
      "grad_norm": 0.0,
      "learning_rate": 6.96661538355399e-06,
      "loss": 0.0,
      "step": 43990
    },
    {
      "epoch": 0.30355923199514306,
      "grad_norm": 0.0,
      "learning_rate": 6.965925476208546e-06,
      "loss": 0.0,
      "step": 44000
    },
    {
      "epoch": 0.3036282227296874,
      "grad_norm": 1.1430282592773438,
      "learning_rate": 6.965235568863102e-06,
      "loss": 0.0011,
      "step": 44010
    },
    {
      "epoch": 0.30369721346423173,
      "grad_norm": 0.0,
      "learning_rate": 6.964545661517659e-06,
      "loss": 0.0,
      "step": 44020
    },
    {
      "epoch": 0.3037662041987761,
      "grad_norm": 0.0,
      "learning_rate": 6.963855754172215e-06,
      "loss": 0.0,
      "step": 44030
    },
    {
      "epoch": 0.30383519493332045,
      "grad_norm": 0.0,
      "learning_rate": 6.9631658468267716e-06,
      "loss": 0.0001,
      "step": 44040
    },
    {
      "epoch": 0.3039041856678648,
      "grad_norm": 0.0,
      "learning_rate": 6.962475939481328e-06,
      "loss": 0.0,
      "step": 44050
    },
    {
      "epoch": 0.3039731764024092,
      "grad_norm": 0.00100987171754241,
      "learning_rate": 6.9617860321358846e-06,
      "loss": 0.0,
      "step": 44060
    },
    {
      "epoch": 0.30404216713695353,
      "grad_norm": 0.0,
      "learning_rate": 6.961096124790441e-06,
      "loss": 0.0,
      "step": 44070
    },
    {
      "epoch": 0.30411115787149784,
      "grad_norm": 0.0,
      "learning_rate": 6.960406217444998e-06,
      "loss": 0.0,
      "step": 44080
    },
    {
      "epoch": 0.3041801486060422,
      "grad_norm": 0.0,
      "learning_rate": 6.959716310099555e-06,
      "loss": 0.0,
      "step": 44090
    },
    {
      "epoch": 0.30424913934058656,
      "grad_norm": 0.245906800031662,
      "learning_rate": 6.959026402754111e-06,
      "loss": 0.0,
      "step": 44100
    },
    {
      "epoch": 0.3043181300751309,
      "grad_norm": 2.147875832747559e-08,
      "learning_rate": 6.958336495408667e-06,
      "loss": 0.0,
      "step": 44110
    },
    {
      "epoch": 0.3043871208096753,
      "grad_norm": 0.0,
      "learning_rate": 6.9576465880632235e-06,
      "loss": 0.0,
      "step": 44120
    },
    {
      "epoch": 0.3044561115442196,
      "grad_norm": 1.2877981703240948e-07,
      "learning_rate": 6.95695668071778e-06,
      "loss": 0.0003,
      "step": 44130
    },
    {
      "epoch": 0.30452510227876395,
      "grad_norm": 0.0,
      "learning_rate": 6.9562667733723365e-06,
      "loss": 0.001,
      "step": 44140
    },
    {
      "epoch": 0.3045940930133083,
      "grad_norm": 0.0,
      "learning_rate": 6.955576866026893e-06,
      "loss": 0.0,
      "step": 44150
    },
    {
      "epoch": 0.30466308374785267,
      "grad_norm": 0.0,
      "learning_rate": 6.9548869586814495e-06,
      "loss": 0.0,
      "step": 44160
    },
    {
      "epoch": 0.30473207448239703,
      "grad_norm": 0.0,
      "learning_rate": 6.954197051336006e-06,
      "loss": 0.0,
      "step": 44170
    },
    {
      "epoch": 0.3048010652169414,
      "grad_norm": 0.0006027265335433185,
      "learning_rate": 6.9535071439905624e-06,
      "loss": 0.0,
      "step": 44180
    },
    {
      "epoch": 0.3048700559514857,
      "grad_norm": 7.980028021847829e-05,
      "learning_rate": 6.95281723664512e-06,
      "loss": 0.1831,
      "step": 44190
    },
    {
      "epoch": 0.30493904668603006,
      "grad_norm": 0.0,
      "learning_rate": 6.952127329299676e-06,
      "loss": 0.1854,
      "step": 44200
    },
    {
      "epoch": 0.3050080374205744,
      "grad_norm": 0.0,
      "learning_rate": 6.951437421954233e-06,
      "loss": 0.0005,
      "step": 44210
    },
    {
      "epoch": 0.3050770281551188,
      "grad_norm": 0.0,
      "learning_rate": 6.950747514608789e-06,
      "loss": 0.0,
      "step": 44220
    },
    {
      "epoch": 0.30514601888966314,
      "grad_norm": 0.0,
      "learning_rate": 6.950057607263345e-06,
      "loss": 0.0,
      "step": 44230
    },
    {
      "epoch": 0.30521500962420745,
      "grad_norm": 4.230862032272853e-06,
      "learning_rate": 6.949367699917901e-06,
      "loss": 0.0,
      "step": 44240
    },
    {
      "epoch": 0.3052840003587518,
      "grad_norm": 4.910173379002458e-10,
      "learning_rate": 6.948677792572458e-06,
      "loss": 0.0046,
      "step": 44250
    },
    {
      "epoch": 0.30535299109329617,
      "grad_norm": 0.0001485193643020466,
      "learning_rate": 6.947987885227014e-06,
      "loss": 0.0,
      "step": 44260
    },
    {
      "epoch": 0.3054219818278405,
      "grad_norm": 0.0,
      "learning_rate": 6.947297977881571e-06,
      "loss": 0.0,
      "step": 44270
    },
    {
      "epoch": 0.3054909725623849,
      "grad_norm": 0.0,
      "learning_rate": 6.946677061270672e-06,
      "loss": 1.3853,
      "step": 44280
    },
    {
      "epoch": 0.30555996329692925,
      "grad_norm": 0.0,
      "learning_rate": 6.945987153925229e-06,
      "loss": 0.0,
      "step": 44290
    },
    {
      "epoch": 0.30562895403147355,
      "grad_norm": 0.0,
      "learning_rate": 6.945297246579785e-06,
      "loss": 0.0,
      "step": 44300
    },
    {
      "epoch": 0.3056979447660179,
      "grad_norm": 0.0,
      "learning_rate": 6.944607339234342e-06,
      "loss": 0.0001,
      "step": 44310
    },
    {
      "epoch": 0.3057669355005623,
      "grad_norm": 0.0,
      "learning_rate": 6.943917431888898e-06,
      "loss": 0.0,
      "step": 44320
    },
    {
      "epoch": 0.30583592623510664,
      "grad_norm": 0.0,
      "learning_rate": 6.943227524543454e-06,
      "loss": 0.0,
      "step": 44330
    },
    {
      "epoch": 0.305904916969651,
      "grad_norm": 0.00016434176359325647,
      "learning_rate": 6.9425376171980105e-06,
      "loss": 0.0,
      "step": 44340
    },
    {
      "epoch": 0.3059739077041953,
      "grad_norm": 0.0,
      "learning_rate": 6.941847709852567e-06,
      "loss": 0.0,
      "step": 44350
    },
    {
      "epoch": 0.30604289843873966,
      "grad_norm": 0.0,
      "learning_rate": 6.9411578025071234e-06,
      "loss": 0.0,
      "step": 44360
    },
    {
      "epoch": 0.306111889173284,
      "grad_norm": 0.0,
      "learning_rate": 6.94046789516168e-06,
      "loss": 0.0,
      "step": 44370
    },
    {
      "epoch": 0.3061808799078284,
      "grad_norm": 0.0,
      "learning_rate": 6.9397779878162364e-06,
      "loss": 0.0,
      "step": 44380
    },
    {
      "epoch": 0.30624987064237275,
      "grad_norm": 0.0,
      "learning_rate": 6.939088080470794e-06,
      "loss": 0.0259,
      "step": 44390
    },
    {
      "epoch": 0.3063188613769171,
      "grad_norm": 0.0,
      "learning_rate": 6.93839817312535e-06,
      "loss": 0.3742,
      "step": 44400
    },
    {
      "epoch": 0.3063878521114614,
      "grad_norm": 0.06201404333114624,
      "learning_rate": 6.937708265779907e-06,
      "loss": 0.0,
      "step": 44410
    },
    {
      "epoch": 0.3064568428460058,
      "grad_norm": 6.698346169287106e-06,
      "learning_rate": 6.937018358434463e-06,
      "loss": 0.005,
      "step": 44420
    },
    {
      "epoch": 0.30652583358055013,
      "grad_norm": 0.0,
      "learning_rate": 6.93632845108902e-06,
      "loss": 0.0,
      "step": 44430
    },
    {
      "epoch": 0.3065948243150945,
      "grad_norm": 0.0,
      "learning_rate": 6.935638543743576e-06,
      "loss": 0.0,
      "step": 44440
    },
    {
      "epoch": 0.30666381504963885,
      "grad_norm": 659.7449340820312,
      "learning_rate": 6.934948636398132e-06,
      "loss": 0.5816,
      "step": 44450
    },
    {
      "epoch": 0.30673280578418316,
      "grad_norm": 0.0,
      "learning_rate": 6.934258729052688e-06,
      "loss": 0.0,
      "step": 44460
    },
    {
      "epoch": 0.3068017965187275,
      "grad_norm": 0.0,
      "learning_rate": 6.933568821707245e-06,
      "loss": 0.0,
      "step": 44470
    },
    {
      "epoch": 0.3068707872532719,
      "grad_norm": 0.0,
      "learning_rate": 6.932878914361801e-06,
      "loss": 0.0001,
      "step": 44480
    },
    {
      "epoch": 0.30693977798781624,
      "grad_norm": 0.0,
      "learning_rate": 6.932189007016358e-06,
      "loss": 0.0,
      "step": 44490
    },
    {
      "epoch": 0.3070087687223606,
      "grad_norm": 1.300905267953567e-07,
      "learning_rate": 6.931499099670915e-06,
      "loss": 0.0,
      "step": 44500
    },
    {
      "epoch": 0.30707775945690496,
      "grad_norm": 0.0,
      "learning_rate": 6.930809192325472e-06,
      "loss": 0.0,
      "step": 44510
    },
    {
      "epoch": 0.30714675019144927,
      "grad_norm": 0.0,
      "learning_rate": 6.930119284980028e-06,
      "loss": 0.0,
      "step": 44520
    },
    {
      "epoch": 0.30721574092599363,
      "grad_norm": 0.0,
      "learning_rate": 6.929429377634585e-06,
      "loss": 0.0,
      "step": 44530
    },
    {
      "epoch": 0.307284731660538,
      "grad_norm": 0.0,
      "learning_rate": 6.928739470289141e-06,
      "loss": 0.0,
      "step": 44540
    },
    {
      "epoch": 0.30735372239508235,
      "grad_norm": 0.0,
      "learning_rate": 6.928049562943698e-06,
      "loss": 0.0,
      "step": 44550
    },
    {
      "epoch": 0.3074227131296267,
      "grad_norm": 0.0,
      "learning_rate": 6.927359655598253e-06,
      "loss": 0.0001,
      "step": 44560
    },
    {
      "epoch": 0.307491703864171,
      "grad_norm": 0.00045866615255363286,
      "learning_rate": 6.92666974825281e-06,
      "loss": 0.0,
      "step": 44570
    },
    {
      "epoch": 0.3075606945987154,
      "grad_norm": 0.0,
      "learning_rate": 6.925979840907366e-06,
      "loss": 0.0,
      "step": 44580
    },
    {
      "epoch": 0.30762968533325974,
      "grad_norm": 0.0,
      "learning_rate": 6.925289933561923e-06,
      "loss": 0.0,
      "step": 44590
    },
    {
      "epoch": 0.3076986760678041,
      "grad_norm": 0.0,
      "learning_rate": 6.924600026216479e-06,
      "loss": 0.0,
      "step": 44600
    },
    {
      "epoch": 0.30776766680234846,
      "grad_norm": 8.274710125988349e-05,
      "learning_rate": 6.9239101188710366e-06,
      "loss": 0.0,
      "step": 44610
    },
    {
      "epoch": 0.3078366575368928,
      "grad_norm": 2.4043238511239906e-08,
      "learning_rate": 6.923220211525593e-06,
      "loss": 0.0,
      "step": 44620
    },
    {
      "epoch": 0.3079056482714371,
      "grad_norm": 0.0,
      "learning_rate": 6.9225303041801495e-06,
      "loss": 0.0,
      "step": 44630
    },
    {
      "epoch": 0.3079746390059815,
      "grad_norm": 0.0,
      "learning_rate": 6.921840396834706e-06,
      "loss": 0.0,
      "step": 44640
    },
    {
      "epoch": 0.30804362974052585,
      "grad_norm": 0.0,
      "learning_rate": 6.9211504894892625e-06,
      "loss": 0.0008,
      "step": 44650
    },
    {
      "epoch": 0.3081126204750702,
      "grad_norm": 0.0,
      "learning_rate": 6.920460582143819e-06,
      "loss": 0.0,
      "step": 44660
    },
    {
      "epoch": 0.30818161120961457,
      "grad_norm": 1.9058694622486883e-09,
      "learning_rate": 6.919770674798375e-06,
      "loss": 0.0,
      "step": 44670
    },
    {
      "epoch": 0.3082506019441589,
      "grad_norm": 0.021299777552485466,
      "learning_rate": 6.919080767452931e-06,
      "loss": 0.0,
      "step": 44680
    },
    {
      "epoch": 0.30831959267870324,
      "grad_norm": 0.0,
      "learning_rate": 6.918390860107488e-06,
      "loss": 0.0204,
      "step": 44690
    },
    {
      "epoch": 0.3083885834132476,
      "grad_norm": 4.867079042014666e-07,
      "learning_rate": 6.917700952762044e-06,
      "loss": 0.5073,
      "step": 44700
    },
    {
      "epoch": 0.30845757414779196,
      "grad_norm": 0.0,
      "learning_rate": 6.917011045416601e-06,
      "loss": 0.0,
      "step": 44710
    },
    {
      "epoch": 0.3085265648823363,
      "grad_norm": 0.0,
      "learning_rate": 6.916321138071158e-06,
      "loss": 0.0,
      "step": 44720
    },
    {
      "epoch": 0.3085955556168807,
      "grad_norm": 0.0,
      "learning_rate": 6.9156312307257144e-06,
      "loss": 0.0034,
      "step": 44730
    },
    {
      "epoch": 0.308664546351425,
      "grad_norm": 0.0,
      "learning_rate": 6.914941323380271e-06,
      "loss": 0.0199,
      "step": 44740
    },
    {
      "epoch": 0.30873353708596935,
      "grad_norm": 0.041417695581912994,
      "learning_rate": 6.914251416034827e-06,
      "loss": 0.0849,
      "step": 44750
    },
    {
      "epoch": 0.3088025278205137,
      "grad_norm": 0.0,
      "learning_rate": 6.913561508689384e-06,
      "loss": 0.0,
      "step": 44760
    },
    {
      "epoch": 0.30887151855505807,
      "grad_norm": 5.211788811720908e-05,
      "learning_rate": 6.91287160134394e-06,
      "loss": 0.0,
      "step": 44770
    },
    {
      "epoch": 0.3089405092896024,
      "grad_norm": 7.899868488311768,
      "learning_rate": 6.912181693998497e-06,
      "loss": 0.0009,
      "step": 44780
    },
    {
      "epoch": 0.30900950002414673,
      "grad_norm": 1.8002893620305827e-09,
      "learning_rate": 6.9114917866530525e-06,
      "loss": 0.0,
      "step": 44790
    },
    {
      "epoch": 0.3090784907586911,
      "grad_norm": 0.0,
      "learning_rate": 6.910801879307609e-06,
      "loss": 0.0,
      "step": 44800
    },
    {
      "epoch": 0.30914748149323545,
      "grad_norm": 0.0,
      "learning_rate": 6.9101119719621655e-06,
      "loss": 0.0,
      "step": 44810
    },
    {
      "epoch": 0.3092164722277798,
      "grad_norm": 0.0,
      "learning_rate": 6.909422064616722e-06,
      "loss": 0.0,
      "step": 44820
    },
    {
      "epoch": 0.3092854629623242,
      "grad_norm": 0.0,
      "learning_rate": 6.908732157271279e-06,
      "loss": 0.0,
      "step": 44830
    },
    {
      "epoch": 0.30935445369686854,
      "grad_norm": 0.0,
      "learning_rate": 6.908042249925836e-06,
      "loss": 0.0,
      "step": 44840
    },
    {
      "epoch": 0.30942344443141284,
      "grad_norm": 8.181193500433892e-10,
      "learning_rate": 6.907352342580392e-06,
      "loss": 0.0,
      "step": 44850
    },
    {
      "epoch": 0.3094924351659572,
      "grad_norm": 1.98007299445635e-09,
      "learning_rate": 6.906662435234949e-06,
      "loss": 0.0994,
      "step": 44860
    },
    {
      "epoch": 0.30956142590050156,
      "grad_norm": 9.021978719658819e-10,
      "learning_rate": 6.905972527889505e-06,
      "loss": 0.0,
      "step": 44870
    },
    {
      "epoch": 0.3096304166350459,
      "grad_norm": 0.0,
      "learning_rate": 6.905282620544062e-06,
      "loss": 0.0,
      "step": 44880
    },
    {
      "epoch": 0.3096994073695903,
      "grad_norm": 15.982380867004395,
      "learning_rate": 6.904592713198618e-06,
      "loss": 0.0038,
      "step": 44890
    },
    {
      "epoch": 0.3097683981041346,
      "grad_norm": 0.0,
      "learning_rate": 6.903902805853174e-06,
      "loss": 0.0,
      "step": 44900
    },
    {
      "epoch": 0.30983738883867895,
      "grad_norm": 0.00024265515094157308,
      "learning_rate": 6.90321289850773e-06,
      "loss": 0.0,
      "step": 44910
    },
    {
      "epoch": 0.3099063795732233,
      "grad_norm": 0.0,
      "learning_rate": 6.902522991162287e-06,
      "loss": 0.0,
      "step": 44920
    },
    {
      "epoch": 0.3099753703077677,
      "grad_norm": 0.0,
      "learning_rate": 6.901833083816843e-06,
      "loss": 0.0,
      "step": 44930
    },
    {
      "epoch": 0.31004436104231203,
      "grad_norm": 0.0,
      "learning_rate": 6.901143176471401e-06,
      "loss": 0.0,
      "step": 44940
    },
    {
      "epoch": 0.3101133517768564,
      "grad_norm": 0.0,
      "learning_rate": 6.900453269125957e-06,
      "loss": 0.0,
      "step": 44950
    },
    {
      "epoch": 0.3101823425114007,
      "grad_norm": 0.0,
      "learning_rate": 6.899763361780514e-06,
      "loss": 0.0,
      "step": 44960
    },
    {
      "epoch": 0.31025133324594506,
      "grad_norm": 0.0,
      "learning_rate": 6.89907345443507e-06,
      "loss": 0.0,
      "step": 44970
    },
    {
      "epoch": 0.3103203239804894,
      "grad_norm": 0.0,
      "learning_rate": 6.898383547089627e-06,
      "loss": 0.0,
      "step": 44980
    },
    {
      "epoch": 0.3103893147150338,
      "grad_norm": 0.0,
      "learning_rate": 6.897693639744183e-06,
      "loss": 0.0001,
      "step": 44990
    },
    {
      "epoch": 0.31045830544957814,
      "grad_norm": 3.703691390910535e-06,
      "learning_rate": 6.89700373239874e-06,
      "loss": 0.1872,
      "step": 45000
    },
    {
      "epoch": 0.31052729618412245,
      "grad_norm": 0.0036246581003069878,
      "learning_rate": 6.896313825053295e-06,
      "loss": 0.0,
      "step": 45010
    },
    {
      "epoch": 0.3105962869186668,
      "grad_norm": 6.165812374092638e-05,
      "learning_rate": 6.895623917707852e-06,
      "loss": 0.0,
      "step": 45020
    },
    {
      "epoch": 0.31066527765321117,
      "grad_norm": 0.0,
      "learning_rate": 6.894934010362408e-06,
      "loss": 0.0004,
      "step": 45030
    },
    {
      "epoch": 0.31073426838775553,
      "grad_norm": 0.0,
      "learning_rate": 6.894244103016965e-06,
      "loss": 0.0,
      "step": 45040
    },
    {
      "epoch": 0.3108032591222999,
      "grad_norm": 5.5043828979250975e-06,
      "learning_rate": 6.893554195671522e-06,
      "loss": 0.0,
      "step": 45050
    },
    {
      "epoch": 0.31087224985684425,
      "grad_norm": 0.0,
      "learning_rate": 6.892864288326079e-06,
      "loss": 0.0,
      "step": 45060
    },
    {
      "epoch": 0.31094124059138856,
      "grad_norm": 0.0,
      "learning_rate": 6.892174380980635e-06,
      "loss": 0.0,
      "step": 45070
    },
    {
      "epoch": 0.3110102313259329,
      "grad_norm": 2.5875824576360174e-06,
      "learning_rate": 6.891484473635192e-06,
      "loss": 0.0003,
      "step": 45080
    },
    {
      "epoch": 0.3110792220604773,
      "grad_norm": 0.0,
      "learning_rate": 6.890794566289748e-06,
      "loss": 0.0,
      "step": 45090
    },
    {
      "epoch": 0.31114821279502164,
      "grad_norm": 0.0,
      "learning_rate": 6.890104658944305e-06,
      "loss": 0.0013,
      "step": 45100
    },
    {
      "epoch": 0.311217203529566,
      "grad_norm": 0.0,
      "learning_rate": 6.889414751598861e-06,
      "loss": 0.0,
      "step": 45110
    },
    {
      "epoch": 0.3112861942641103,
      "grad_norm": 0.0,
      "learning_rate": 6.8887248442534176e-06,
      "loss": 0.0,
      "step": 45120
    },
    {
      "epoch": 0.31135518499865467,
      "grad_norm": 0.0,
      "learning_rate": 6.888034936907973e-06,
      "loss": 0.0026,
      "step": 45130
    },
    {
      "epoch": 0.311424175733199,
      "grad_norm": 0.00012032729864586145,
      "learning_rate": 6.88734502956253e-06,
      "loss": 0.0004,
      "step": 45140
    },
    {
      "epoch": 0.3114931664677434,
      "grad_norm": 2.408187583569088e-06,
      "learning_rate": 6.886655122217086e-06,
      "loss": 0.0014,
      "step": 45150
    },
    {
      "epoch": 0.31156215720228775,
      "grad_norm": 5.01133046171276e-09,
      "learning_rate": 6.8859652148716435e-06,
      "loss": 0.0,
      "step": 45160
    },
    {
      "epoch": 0.3116311479368321,
      "grad_norm": 0.0,
      "learning_rate": 6.8852753075262e-06,
      "loss": 0.0,
      "step": 45170
    },
    {
      "epoch": 0.3117001386713764,
      "grad_norm": 0.0,
      "learning_rate": 6.8845854001807565e-06,
      "loss": 0.0078,
      "step": 45180
    },
    {
      "epoch": 0.3117691294059208,
      "grad_norm": 0.008029049262404442,
      "learning_rate": 6.883895492835313e-06,
      "loss": 0.0001,
      "step": 45190
    },
    {
      "epoch": 0.31183812014046514,
      "grad_norm": 4.350970357336337e-06,
      "learning_rate": 6.8832055854898695e-06,
      "loss": 0.0004,
      "step": 45200
    },
    {
      "epoch": 0.3119071108750095,
      "grad_norm": 0.0,
      "learning_rate": 6.882515678144426e-06,
      "loss": 0.2322,
      "step": 45210
    },
    {
      "epoch": 0.31197610160955386,
      "grad_norm": 8.942540596912352e-10,
      "learning_rate": 6.8818257707989825e-06,
      "loss": 0.0,
      "step": 45220
    },
    {
      "epoch": 0.31204509234409816,
      "grad_norm": 2.541708226999617e-06,
      "learning_rate": 6.881135863453539e-06,
      "loss": 0.0069,
      "step": 45230
    },
    {
      "epoch": 0.3121140830786425,
      "grad_norm": 4.609160900115967,
      "learning_rate": 6.880445956108095e-06,
      "loss": 0.0008,
      "step": 45240
    },
    {
      "epoch": 0.3121830738131869,
      "grad_norm": 0.022396419197320938,
      "learning_rate": 6.879756048762651e-06,
      "loss": 0.0,
      "step": 45250
    },
    {
      "epoch": 0.31225206454773125,
      "grad_norm": 0.0,
      "learning_rate": 6.879066141417208e-06,
      "loss": 0.0,
      "step": 45260
    },
    {
      "epoch": 0.3123210552822756,
      "grad_norm": 0.645513653755188,
      "learning_rate": 6.878376234071765e-06,
      "loss": 0.0001,
      "step": 45270
    },
    {
      "epoch": 0.31239004601681997,
      "grad_norm": 0.05424048751592636,
      "learning_rate": 6.877686326726321e-06,
      "loss": 0.0,
      "step": 45280
    },
    {
      "epoch": 0.31245903675136427,
      "grad_norm": 0.0,
      "learning_rate": 6.876996419380878e-06,
      "loss": 0.0023,
      "step": 45290
    },
    {
      "epoch": 0.31252802748590863,
      "grad_norm": 0.23057273030281067,
      "learning_rate": 6.876306512035434e-06,
      "loss": 0.0001,
      "step": 45300
    },
    {
      "epoch": 0.312597018220453,
      "grad_norm": 0.0,
      "learning_rate": 6.875616604689991e-06,
      "loss": 0.118,
      "step": 45310
    },
    {
      "epoch": 0.31266600895499735,
      "grad_norm": 0.0,
      "learning_rate": 6.874926697344547e-06,
      "loss": 0.0523,
      "step": 45320
    },
    {
      "epoch": 0.3127349996895417,
      "grad_norm": 0.0,
      "learning_rate": 6.874236789999104e-06,
      "loss": 0.0,
      "step": 45330
    },
    {
      "epoch": 0.312803990424086,
      "grad_norm": 0.0,
      "learning_rate": 6.87354688265366e-06,
      "loss": 0.0,
      "step": 45340
    },
    {
      "epoch": 0.3128729811586304,
      "grad_norm": 0.0,
      "learning_rate": 6.872856975308216e-06,
      "loss": 0.0,
      "step": 45350
    },
    {
      "epoch": 0.31294197189317474,
      "grad_norm": 0.0010468738619238138,
      "learning_rate": 6.8721670679627725e-06,
      "loss": 0.0113,
      "step": 45360
    },
    {
      "epoch": 0.3130109626277191,
      "grad_norm": 0.014644600450992584,
      "learning_rate": 6.871477160617329e-06,
      "loss": 0.0051,
      "step": 45370
    },
    {
      "epoch": 0.31307995336226346,
      "grad_norm": 0.0,
      "learning_rate": 6.870787253271886e-06,
      "loss": 0.0,
      "step": 45380
    },
    {
      "epoch": 0.3131489440968078,
      "grad_norm": 0.0,
      "learning_rate": 6.870097345926443e-06,
      "loss": 0.0,
      "step": 45390
    },
    {
      "epoch": 0.31321793483135213,
      "grad_norm": 0.0,
      "learning_rate": 6.869407438580999e-06,
      "loss": 0.0,
      "step": 45400
    },
    {
      "epoch": 0.3132869255658965,
      "grad_norm": 0.0,
      "learning_rate": 6.868717531235556e-06,
      "loss": 0.0,
      "step": 45410
    },
    {
      "epoch": 0.31335591630044085,
      "grad_norm": 0.0,
      "learning_rate": 6.868027623890112e-06,
      "loss": 0.0,
      "step": 45420
    },
    {
      "epoch": 0.3134249070349852,
      "grad_norm": 0.0,
      "learning_rate": 6.867337716544669e-06,
      "loss": 0.0,
      "step": 45430
    },
    {
      "epoch": 0.3134938977695296,
      "grad_norm": 272.8149719238281,
      "learning_rate": 6.866647809199225e-06,
      "loss": 0.1084,
      "step": 45440
    },
    {
      "epoch": 0.3135628885040739,
      "grad_norm": 0.0,
      "learning_rate": 6.865957901853782e-06,
      "loss": 0.0001,
      "step": 45450
    },
    {
      "epoch": 0.31363187923861824,
      "grad_norm": 1.0238304648524377e-09,
      "learning_rate": 6.865267994508337e-06,
      "loss": 0.0,
      "step": 45460
    },
    {
      "epoch": 0.3137008699731626,
      "grad_norm": 0.0,
      "learning_rate": 6.864578087162894e-06,
      "loss": 0.0,
      "step": 45470
    },
    {
      "epoch": 0.31376986070770696,
      "grad_norm": 0.0,
      "learning_rate": 6.86388817981745e-06,
      "loss": 0.0001,
      "step": 45480
    },
    {
      "epoch": 0.3138388514422513,
      "grad_norm": 0.0,
      "learning_rate": 6.863198272472008e-06,
      "loss": 0.0,
      "step": 45490
    },
    {
      "epoch": 0.3139078421767957,
      "grad_norm": 0.0,
      "learning_rate": 6.862508365126564e-06,
      "loss": 0.0,
      "step": 45500
    },
    {
      "epoch": 0.31397683291134,
      "grad_norm": 0.0,
      "learning_rate": 6.861818457781121e-06,
      "loss": 0.0,
      "step": 45510
    },
    {
      "epoch": 0.31404582364588435,
      "grad_norm": 0.0,
      "learning_rate": 6.861128550435677e-06,
      "loss": 0.0018,
      "step": 45520
    },
    {
      "epoch": 0.3141148143804287,
      "grad_norm": 0.0,
      "learning_rate": 6.860438643090234e-06,
      "loss": 0.0,
      "step": 45530
    },
    {
      "epoch": 0.31418380511497307,
      "grad_norm": 0.0,
      "learning_rate": 6.85974873574479e-06,
      "loss": 0.0,
      "step": 45540
    },
    {
      "epoch": 0.31425279584951743,
      "grad_norm": 0.0005821464583277702,
      "learning_rate": 6.859058828399347e-06,
      "loss": 0.0038,
      "step": 45550
    },
    {
      "epoch": 0.31432178658406174,
      "grad_norm": 0.8175028562545776,
      "learning_rate": 6.858368921053903e-06,
      "loss": 0.0002,
      "step": 45560
    },
    {
      "epoch": 0.3143907773186061,
      "grad_norm": 0.0,
      "learning_rate": 6.85767901370846e-06,
      "loss": 0.0003,
      "step": 45570
    },
    {
      "epoch": 0.31445976805315046,
      "grad_norm": 0.0,
      "learning_rate": 6.856989106363015e-06,
      "loss": 0.0,
      "step": 45580
    },
    {
      "epoch": 0.3145287587876948,
      "grad_norm": 0.0,
      "learning_rate": 6.856299199017572e-06,
      "loss": 0.0,
      "step": 45590
    },
    {
      "epoch": 0.3145977495222392,
      "grad_norm": 0.0,
      "learning_rate": 6.855609291672129e-06,
      "loss": 0.0,
      "step": 45600
    },
    {
      "epoch": 0.31466674025678354,
      "grad_norm": 0.0,
      "learning_rate": 6.854919384326686e-06,
      "loss": 0.0,
      "step": 45610
    },
    {
      "epoch": 0.31473573099132784,
      "grad_norm": 1.785837810963642e-09,
      "learning_rate": 6.854229476981242e-06,
      "loss": 0.0006,
      "step": 45620
    },
    {
      "epoch": 0.3148047217258722,
      "grad_norm": 0.0,
      "learning_rate": 6.8535395696357986e-06,
      "loss": 0.0,
      "step": 45630
    },
    {
      "epoch": 0.31487371246041657,
      "grad_norm": 0.0,
      "learning_rate": 6.852849662290355e-06,
      "loss": 0.0,
      "step": 45640
    },
    {
      "epoch": 0.3149427031949609,
      "grad_norm": 0.002674132352694869,
      "learning_rate": 6.8521597549449116e-06,
      "loss": 0.0003,
      "step": 45650
    },
    {
      "epoch": 0.3150116939295053,
      "grad_norm": 0.0,
      "learning_rate": 6.851469847599468e-06,
      "loss": 0.0001,
      "step": 45660
    },
    {
      "epoch": 0.3150806846640496,
      "grad_norm": 0.0,
      "learning_rate": 6.8507799402540245e-06,
      "loss": 0.0001,
      "step": 45670
    },
    {
      "epoch": 0.31514967539859395,
      "grad_norm": 0.0,
      "learning_rate": 6.850090032908581e-06,
      "loss": 0.0671,
      "step": 45680
    },
    {
      "epoch": 0.3152186661331383,
      "grad_norm": 0.00021236490283627063,
      "learning_rate": 6.849400125563137e-06,
      "loss": 0.0,
      "step": 45690
    },
    {
      "epoch": 0.3152876568676827,
      "grad_norm": 0.0,
      "learning_rate": 6.848710218217693e-06,
      "loss": 0.0,
      "step": 45700
    },
    {
      "epoch": 0.31535664760222704,
      "grad_norm": 0.0,
      "learning_rate": 6.848020310872251e-06,
      "loss": 0.0,
      "step": 45710
    },
    {
      "epoch": 0.3154256383367714,
      "grad_norm": 0.022920195013284683,
      "learning_rate": 6.847330403526807e-06,
      "loss": 0.0,
      "step": 45720
    },
    {
      "epoch": 0.3154946290713157,
      "grad_norm": 3.2675626277923584,
      "learning_rate": 6.8466404961813635e-06,
      "loss": 0.0362,
      "step": 45730
    },
    {
      "epoch": 0.31556361980586006,
      "grad_norm": 0.0,
      "learning_rate": 6.84595058883592e-06,
      "loss": 0.0003,
      "step": 45740
    },
    {
      "epoch": 0.3156326105404044,
      "grad_norm": 12.057875633239746,
      "learning_rate": 6.8452606814904765e-06,
      "loss": 0.004,
      "step": 45750
    },
    {
      "epoch": 0.3157016012749488,
      "grad_norm": 0.0,
      "learning_rate": 6.844570774145033e-06,
      "loss": 0.0,
      "step": 45760
    },
    {
      "epoch": 0.31577059200949315,
      "grad_norm": 2.0129700715187937e-06,
      "learning_rate": 6.8438808667995894e-06,
      "loss": 0.0,
      "step": 45770
    },
    {
      "epoch": 0.31583958274403745,
      "grad_norm": 0.0,
      "learning_rate": 6.843190959454146e-06,
      "loss": 0.013,
      "step": 45780
    },
    {
      "epoch": 0.3159085734785818,
      "grad_norm": 8.719306788407266e-05,
      "learning_rate": 6.842501052108702e-06,
      "loss": 0.0,
      "step": 45790
    },
    {
      "epoch": 0.31597756421312617,
      "grad_norm": 0.021643323823809624,
      "learning_rate": 6.841811144763258e-06,
      "loss": 0.0001,
      "step": 45800
    },
    {
      "epoch": 0.31604655494767053,
      "grad_norm": 6.259943620534614e-05,
      "learning_rate": 6.8411212374178145e-06,
      "loss": 0.6844,
      "step": 45810
    },
    {
      "epoch": 0.3161155456822149,
      "grad_norm": 7.235644261527341e-08,
      "learning_rate": 6.840431330072373e-06,
      "loss": 0.0005,
      "step": 45820
    },
    {
      "epoch": 0.31618453641675925,
      "grad_norm": 0.0,
      "learning_rate": 6.839741422726928e-06,
      "loss": 0.0,
      "step": 45830
    },
    {
      "epoch": 0.31625352715130356,
      "grad_norm": 8.794885375529304e-10,
      "learning_rate": 6.839051515381485e-06,
      "loss": 0.0,
      "step": 45840
    },
    {
      "epoch": 0.3163225178858479,
      "grad_norm": 0.0,
      "learning_rate": 6.838361608036041e-06,
      "loss": 0.0387,
      "step": 45850
    },
    {
      "epoch": 0.3163915086203923,
      "grad_norm": 0.0,
      "learning_rate": 6.837671700690598e-06,
      "loss": 0.0005,
      "step": 45860
    },
    {
      "epoch": 0.31646049935493664,
      "grad_norm": 0.0,
      "learning_rate": 6.836981793345154e-06,
      "loss": 0.0,
      "step": 45870
    },
    {
      "epoch": 0.316529490089481,
      "grad_norm": 0.0,
      "learning_rate": 6.836291885999711e-06,
      "loss": 0.0,
      "step": 45880
    },
    {
      "epoch": 0.3165984808240253,
      "grad_norm": 0.0,
      "learning_rate": 6.835601978654267e-06,
      "loss": 0.0,
      "step": 45890
    },
    {
      "epoch": 0.31666747155856967,
      "grad_norm": 0.0,
      "learning_rate": 6.834912071308824e-06,
      "loss": 0.0001,
      "step": 45900
    },
    {
      "epoch": 0.31673646229311403,
      "grad_norm": 9.855404270453505e-10,
      "learning_rate": 6.83422216396338e-06,
      "loss": 0.0,
      "step": 45910
    },
    {
      "epoch": 0.3168054530276584,
      "grad_norm": 0.0,
      "learning_rate": 6.833532256617936e-06,
      "loss": 0.0001,
      "step": 45920
    },
    {
      "epoch": 0.31687444376220275,
      "grad_norm": 0.0,
      "learning_rate": 6.832842349272494e-06,
      "loss": 0.0,
      "step": 45930
    },
    {
      "epoch": 0.3169434344967471,
      "grad_norm": 0.0,
      "learning_rate": 6.83215244192705e-06,
      "loss": 0.0001,
      "step": 45940
    },
    {
      "epoch": 0.3170124252312914,
      "grad_norm": 0.0,
      "learning_rate": 6.831462534581606e-06,
      "loss": 0.0,
      "step": 45950
    },
    {
      "epoch": 0.3170814159658358,
      "grad_norm": 0.0,
      "learning_rate": 6.830772627236163e-06,
      "loss": 0.0,
      "step": 45960
    },
    {
      "epoch": 0.31715040670038014,
      "grad_norm": 0.0,
      "learning_rate": 6.830082719890719e-06,
      "loss": 0.0,
      "step": 45970
    },
    {
      "epoch": 0.3172193974349245,
      "grad_norm": 0.0,
      "learning_rate": 6.829392812545276e-06,
      "loss": 0.0,
      "step": 45980
    },
    {
      "epoch": 0.31728838816946886,
      "grad_norm": 0.0,
      "learning_rate": 6.828702905199832e-06,
      "loss": 0.0048,
      "step": 45990
    },
    {
      "epoch": 0.31735737890401317,
      "grad_norm": 0.0,
      "learning_rate": 6.828012997854389e-06,
      "loss": 0.1422,
      "step": 46000
    },
    {
      "epoch": 0.3174263696385575,
      "grad_norm": 8.094303560746141e-10,
      "learning_rate": 6.827323090508945e-06,
      "loss": 0.0,
      "step": 46010
    },
    {
      "epoch": 0.3174953603731019,
      "grad_norm": 0.0,
      "learning_rate": 6.826633183163502e-06,
      "loss": 0.0,
      "step": 46020
    },
    {
      "epoch": 0.31756435110764625,
      "grad_norm": 0.0,
      "learning_rate": 6.825943275818057e-06,
      "loss": 0.0189,
      "step": 46030
    },
    {
      "epoch": 0.3176333418421906,
      "grad_norm": 4.439801273292687e-07,
      "learning_rate": 6.8252533684726155e-06,
      "loss": 0.0,
      "step": 46040
    },
    {
      "epoch": 0.31770233257673497,
      "grad_norm": 0.0668044239282608,
      "learning_rate": 6.824563461127172e-06,
      "loss": 0.0,
      "step": 46050
    },
    {
      "epoch": 0.3177713233112793,
      "grad_norm": 2.518414248697809e-06,
      "learning_rate": 6.823873553781728e-06,
      "loss": 0.0,
      "step": 46060
    },
    {
      "epoch": 0.31784031404582364,
      "grad_norm": 0.0,
      "learning_rate": 6.823183646436284e-06,
      "loss": 0.0172,
      "step": 46070
    },
    {
      "epoch": 0.317909304780368,
      "grad_norm": 0.0,
      "learning_rate": 6.822493739090841e-06,
      "loss": 0.0,
      "step": 46080
    },
    {
      "epoch": 0.31797829551491236,
      "grad_norm": 0.0,
      "learning_rate": 6.821803831745397e-06,
      "loss": 0.0017,
      "step": 46090
    },
    {
      "epoch": 0.3180472862494567,
      "grad_norm": 0.0,
      "learning_rate": 6.821113924399954e-06,
      "loss": 0.0,
      "step": 46100
    },
    {
      "epoch": 0.318116276984001,
      "grad_norm": 0.0,
      "learning_rate": 6.82042401705451e-06,
      "loss": 0.0,
      "step": 46110
    },
    {
      "epoch": 0.3181852677185454,
      "grad_norm": 5.813530515297316e-05,
      "learning_rate": 6.819734109709067e-06,
      "loss": 0.0,
      "step": 46120
    },
    {
      "epoch": 0.31825425845308974,
      "grad_norm": 0.0,
      "learning_rate": 6.819044202363623e-06,
      "loss": 0.0,
      "step": 46130
    },
    {
      "epoch": 0.3183232491876341,
      "grad_norm": 89.85318756103516,
      "learning_rate": 6.818354295018179e-06,
      "loss": 0.0151,
      "step": 46140
    },
    {
      "epoch": 0.31839223992217847,
      "grad_norm": 0.0,
      "learning_rate": 6.817664387672737e-06,
      "loss": 0.0,
      "step": 46150
    },
    {
      "epoch": 0.3184612306567228,
      "grad_norm": 8.862480194160582e-10,
      "learning_rate": 6.816974480327293e-06,
      "loss": 0.0024,
      "step": 46160
    },
    {
      "epoch": 0.31853022139126713,
      "grad_norm": 5.272399334899092e-07,
      "learning_rate": 6.816284572981849e-06,
      "loss": 0.0,
      "step": 46170
    },
    {
      "epoch": 0.3185992121258115,
      "grad_norm": 0.0,
      "learning_rate": 6.8155946656364055e-06,
      "loss": 0.0,
      "step": 46180
    },
    {
      "epoch": 0.31866820286035585,
      "grad_norm": 0.0,
      "learning_rate": 6.814904758290962e-06,
      "loss": 0.5758,
      "step": 46190
    },
    {
      "epoch": 0.3187371935949002,
      "grad_norm": 0.0,
      "learning_rate": 6.8142148509455185e-06,
      "loss": 0.0,
      "step": 46200
    },
    {
      "epoch": 0.3188061843294446,
      "grad_norm": 0.0,
      "learning_rate": 6.813524943600075e-06,
      "loss": 0.3789,
      "step": 46210
    },
    {
      "epoch": 0.3188751750639889,
      "grad_norm": 0.0,
      "learning_rate": 6.8128350362546315e-06,
      "loss": 0.0,
      "step": 46220
    },
    {
      "epoch": 0.31894416579853324,
      "grad_norm": 0.0,
      "learning_rate": 6.812145128909188e-06,
      "loss": 0.2004,
      "step": 46230
    },
    {
      "epoch": 0.3190131565330776,
      "grad_norm": 0.0,
      "learning_rate": 6.8114552215637445e-06,
      "loss": 0.0,
      "step": 46240
    },
    {
      "epoch": 0.31908214726762196,
      "grad_norm": 0.0,
      "learning_rate": 6.810765314218301e-06,
      "loss": 0.0,
      "step": 46250
    },
    {
      "epoch": 0.3191511380021663,
      "grad_norm": 1.7817486375193425e-09,
      "learning_rate": 6.810075406872858e-06,
      "loss": 0.0,
      "step": 46260
    },
    {
      "epoch": 0.3192201287367107,
      "grad_norm": 0.0,
      "learning_rate": 6.809385499527415e-06,
      "loss": 0.0,
      "step": 46270
    },
    {
      "epoch": 0.319289119471255,
      "grad_norm": 4.777670481459495e-10,
      "learning_rate": 6.8086955921819704e-06,
      "loss": 0.0019,
      "step": 46280
    },
    {
      "epoch": 0.31935811020579935,
      "grad_norm": 0.0,
      "learning_rate": 6.808005684836527e-06,
      "loss": 0.0,
      "step": 46290
    },
    {
      "epoch": 0.3194271009403437,
      "grad_norm": 0.0,
      "learning_rate": 6.8073157774910834e-06,
      "loss": 0.0001,
      "step": 46300
    },
    {
      "epoch": 0.31949609167488807,
      "grad_norm": 0.0,
      "learning_rate": 6.80662587014564e-06,
      "loss": 0.0,
      "step": 46310
    },
    {
      "epoch": 0.31956508240943243,
      "grad_norm": 2.0159306757250306e-07,
      "learning_rate": 6.805935962800196e-06,
      "loss": 0.0116,
      "step": 46320
    },
    {
      "epoch": 0.31963407314397674,
      "grad_norm": 0.0,
      "learning_rate": 6.805246055454753e-06,
      "loss": 0.0,
      "step": 46330
    },
    {
      "epoch": 0.3197030638785211,
      "grad_norm": 0.0,
      "learning_rate": 6.804556148109309e-06,
      "loss": 0.0033,
      "step": 46340
    },
    {
      "epoch": 0.31977205461306546,
      "grad_norm": 3.4986820220947266,
      "learning_rate": 6.803866240763866e-06,
      "loss": 0.0007,
      "step": 46350
    },
    {
      "epoch": 0.3198410453476098,
      "grad_norm": 0.0,
      "learning_rate": 6.803176333418422e-06,
      "loss": 0.0,
      "step": 46360
    },
    {
      "epoch": 0.3199100360821542,
      "grad_norm": 0.0,
      "learning_rate": 6.80248642607298e-06,
      "loss": 0.0001,
      "step": 46370
    },
    {
      "epoch": 0.31997902681669854,
      "grad_norm": 0.03837305307388306,
      "learning_rate": 6.801796518727536e-06,
      "loss": 0.0,
      "step": 46380
    },
    {
      "epoch": 0.32004801755124285,
      "grad_norm": 0.0,
      "learning_rate": 6.801106611382093e-06,
      "loss": 0.0003,
      "step": 46390
    },
    {
      "epoch": 0.3201170082857872,
      "grad_norm": 0.0,
      "learning_rate": 6.800416704036648e-06,
      "loss": 0.0,
      "step": 46400
    },
    {
      "epoch": 0.32018599902033157,
      "grad_norm": 0.0,
      "learning_rate": 6.799726796691205e-06,
      "loss": 0.0,
      "step": 46410
    },
    {
      "epoch": 0.32025498975487593,
      "grad_norm": 0.0,
      "learning_rate": 6.799036889345761e-06,
      "loss": 0.0,
      "step": 46420
    },
    {
      "epoch": 0.3203239804894203,
      "grad_norm": 0.0,
      "learning_rate": 6.798346982000318e-06,
      "loss": 0.0,
      "step": 46430
    },
    {
      "epoch": 0.3203929712239646,
      "grad_norm": 0.0028375277761369944,
      "learning_rate": 6.797657074654874e-06,
      "loss": 0.0,
      "step": 46440
    },
    {
      "epoch": 0.32046196195850896,
      "grad_norm": 0.0,
      "learning_rate": 6.796967167309431e-06,
      "loss": 0.0,
      "step": 46450
    },
    {
      "epoch": 0.3205309526930533,
      "grad_norm": 0.0,
      "learning_rate": 6.796277259963987e-06,
      "loss": 0.0,
      "step": 46460
    },
    {
      "epoch": 0.3205999434275977,
      "grad_norm": 0.0,
      "learning_rate": 6.795587352618544e-06,
      "loss": 0.0,
      "step": 46470
    },
    {
      "epoch": 0.32066893416214204,
      "grad_norm": 9.338423367921678e-10,
      "learning_rate": 6.794897445273101e-06,
      "loss": 0.0,
      "step": 46480
    },
    {
      "epoch": 0.3207379248966864,
      "grad_norm": 0.0,
      "learning_rate": 6.794207537927658e-06,
      "loss": 0.0,
      "step": 46490
    },
    {
      "epoch": 0.3208069156312307,
      "grad_norm": 0.0,
      "learning_rate": 6.793517630582214e-06,
      "loss": 0.0,
      "step": 46500
    },
    {
      "epoch": 0.32087590636577507,
      "grad_norm": 0.0,
      "learning_rate": 6.79282772323677e-06,
      "loss": 0.0,
      "step": 46510
    },
    {
      "epoch": 0.3209448971003194,
      "grad_norm": 0.0,
      "learning_rate": 6.792137815891326e-06,
      "loss": 0.0,
      "step": 46520
    },
    {
      "epoch": 0.3210138878348638,
      "grad_norm": 7.458052095898893e-06,
      "learning_rate": 6.791447908545883e-06,
      "loss": 0.0,
      "step": 46530
    },
    {
      "epoch": 0.32108287856940815,
      "grad_norm": 0.0,
      "learning_rate": 6.790758001200439e-06,
      "loss": 0.0,
      "step": 46540
    },
    {
      "epoch": 0.32115186930395245,
      "grad_norm": 2.636823737134364e-08,
      "learning_rate": 6.790068093854996e-06,
      "loss": 0.0,
      "step": 46550
    },
    {
      "epoch": 0.3212208600384968,
      "grad_norm": 0.0,
      "learning_rate": 6.789378186509552e-06,
      "loss": 0.0,
      "step": 46560
    },
    {
      "epoch": 0.3212898507730412,
      "grad_norm": 0.0,
      "learning_rate": 6.788688279164109e-06,
      "loss": 0.0,
      "step": 46570
    },
    {
      "epoch": 0.32135884150758554,
      "grad_norm": 0.0,
      "learning_rate": 6.787998371818665e-06,
      "loss": 0.0814,
      "step": 46580
    },
    {
      "epoch": 0.3214278322421299,
      "grad_norm": 9.081887464290617e-10,
      "learning_rate": 6.7873084644732225e-06,
      "loss": 0.0,
      "step": 46590
    },
    {
      "epoch": 0.32149682297667426,
      "grad_norm": 0.0,
      "learning_rate": 6.786618557127779e-06,
      "loss": 0.0,
      "step": 46600
    },
    {
      "epoch": 0.32156581371121856,
      "grad_norm": 0.0,
      "learning_rate": 6.7859286497823355e-06,
      "loss": 0.0001,
      "step": 46610
    },
    {
      "epoch": 0.3216348044457629,
      "grad_norm": 1.1349088424594811e-07,
      "learning_rate": 6.785238742436891e-06,
      "loss": 0.0,
      "step": 46620
    },
    {
      "epoch": 0.3217037951803073,
      "grad_norm": 0.001222744700498879,
      "learning_rate": 6.784548835091448e-06,
      "loss": 0.0,
      "step": 46630
    },
    {
      "epoch": 0.32177278591485164,
      "grad_norm": 0.0,
      "learning_rate": 6.783858927746004e-06,
      "loss": 0.0,
      "step": 46640
    },
    {
      "epoch": 0.321841776649396,
      "grad_norm": 0.03581872209906578,
      "learning_rate": 6.783169020400561e-06,
      "loss": 0.0,
      "step": 46650
    },
    {
      "epoch": 0.3219107673839403,
      "grad_norm": 0.0,
      "learning_rate": 6.782479113055117e-06,
      "loss": 0.0009,
      "step": 46660
    },
    {
      "epoch": 0.32197975811848467,
      "grad_norm": 0.0,
      "learning_rate": 6.7817892057096736e-06,
      "loss": 0.0004,
      "step": 46670
    },
    {
      "epoch": 0.32204874885302903,
      "grad_norm": 0.0,
      "learning_rate": 6.78109929836423e-06,
      "loss": 0.0,
      "step": 46680
    },
    {
      "epoch": 0.3221177395875734,
      "grad_norm": 0.0,
      "learning_rate": 6.7804093910187865e-06,
      "loss": 0.0,
      "step": 46690
    },
    {
      "epoch": 0.32218673032211775,
      "grad_norm": 0.0,
      "learning_rate": 6.779719483673344e-06,
      "loss": 0.0,
      "step": 46700
    },
    {
      "epoch": 0.3222557210566621,
      "grad_norm": 0.0,
      "learning_rate": 6.7790295763279e-06,
      "loss": 0.0,
      "step": 46710
    },
    {
      "epoch": 0.3223247117912064,
      "grad_norm": 0.0,
      "learning_rate": 6.778339668982457e-06,
      "loss": 0.0002,
      "step": 46720
    },
    {
      "epoch": 0.3223937025257508,
      "grad_norm": 0.0,
      "learning_rate": 6.777649761637013e-06,
      "loss": 0.0,
      "step": 46730
    },
    {
      "epoch": 0.32246269326029514,
      "grad_norm": 0.0,
      "learning_rate": 6.776959854291569e-06,
      "loss": 0.0,
      "step": 46740
    },
    {
      "epoch": 0.3225316839948395,
      "grad_norm": 0.0,
      "learning_rate": 6.7762699469461255e-06,
      "loss": 0.0,
      "step": 46750
    },
    {
      "epoch": 0.32260067472938386,
      "grad_norm": 35.30790710449219,
      "learning_rate": 6.775580039600682e-06,
      "loss": 0.01,
      "step": 46760
    },
    {
      "epoch": 0.32266966546392817,
      "grad_norm": 0.0,
      "learning_rate": 6.7748901322552385e-06,
      "loss": 0.0,
      "step": 46770
    },
    {
      "epoch": 0.32273865619847253,
      "grad_norm": 0.0,
      "learning_rate": 6.774200224909795e-06,
      "loss": 0.0,
      "step": 46780
    },
    {
      "epoch": 0.3228076469330169,
      "grad_norm": 0.0,
      "learning_rate": 6.7735103175643515e-06,
      "loss": 0.0,
      "step": 46790
    },
    {
      "epoch": 0.32287663766756125,
      "grad_norm": 0.0,
      "learning_rate": 6.772820410218908e-06,
      "loss": 0.0,
      "step": 46800
    },
    {
      "epoch": 0.3229456284021056,
      "grad_norm": 0.0,
      "learning_rate": 6.772130502873465e-06,
      "loss": 0.0,
      "step": 46810
    },
    {
      "epoch": 0.32301461913664997,
      "grad_norm": 0.0,
      "learning_rate": 6.771440595528022e-06,
      "loss": 0.0,
      "step": 46820
    },
    {
      "epoch": 0.3230836098711943,
      "grad_norm": 0.0,
      "learning_rate": 6.770750688182578e-06,
      "loss": 0.0004,
      "step": 46830
    },
    {
      "epoch": 0.32315260060573864,
      "grad_norm": 0.0,
      "learning_rate": 6.770060780837135e-06,
      "loss": 0.0,
      "step": 46840
    },
    {
      "epoch": 0.323221591340283,
      "grad_norm": 0.0,
      "learning_rate": 6.76937087349169e-06,
      "loss": 0.0003,
      "step": 46850
    },
    {
      "epoch": 0.32329058207482736,
      "grad_norm": 0.0,
      "learning_rate": 6.768680966146247e-06,
      "loss": 0.0,
      "step": 46860
    },
    {
      "epoch": 0.3233595728093717,
      "grad_norm": 0.0,
      "learning_rate": 6.767991058800803e-06,
      "loss": 0.0,
      "step": 46870
    },
    {
      "epoch": 0.323428563543916,
      "grad_norm": 0.0,
      "learning_rate": 6.76730115145536e-06,
      "loss": 0.0,
      "step": 46880
    },
    {
      "epoch": 0.3234975542784604,
      "grad_norm": 0.0,
      "learning_rate": 6.766611244109916e-06,
      "loss": 0.0,
      "step": 46890
    },
    {
      "epoch": 0.32356654501300475,
      "grad_norm": 0.0,
      "learning_rate": 6.765921336764473e-06,
      "loss": 0.0001,
      "step": 46900
    },
    {
      "epoch": 0.3236355357475491,
      "grad_norm": 0.0,
      "learning_rate": 6.765231429419029e-06,
      "loss": 0.0006,
      "step": 46910
    },
    {
      "epoch": 0.32370452648209347,
      "grad_norm": 0.0,
      "learning_rate": 6.764541522073587e-06,
      "loss": 0.0,
      "step": 46920
    },
    {
      "epoch": 0.32377351721663783,
      "grad_norm": 0.0,
      "learning_rate": 6.763851614728143e-06,
      "loss": 0.0,
      "step": 46930
    },
    {
      "epoch": 0.32384250795118213,
      "grad_norm": 0.0,
      "learning_rate": 6.7631617073827e-06,
      "loss": 0.0,
      "step": 46940
    },
    {
      "epoch": 0.3239114986857265,
      "grad_norm": 4.902747097190741e-10,
      "learning_rate": 6.762471800037256e-06,
      "loss": 0.0,
      "step": 46950
    },
    {
      "epoch": 0.32398048942027086,
      "grad_norm": 1.2466879297790001e-06,
      "learning_rate": 6.761781892691812e-06,
      "loss": 0.0,
      "step": 46960
    },
    {
      "epoch": 0.3240494801548152,
      "grad_norm": 7.000086732489308e-09,
      "learning_rate": 6.761091985346368e-06,
      "loss": 0.0,
      "step": 46970
    },
    {
      "epoch": 0.3241184708893596,
      "grad_norm": 0.026929859071969986,
      "learning_rate": 6.760402078000925e-06,
      "loss": 0.0,
      "step": 46980
    },
    {
      "epoch": 0.3241874616239039,
      "grad_norm": 0.0,
      "learning_rate": 6.759712170655481e-06,
      "loss": 0.0,
      "step": 46990
    },
    {
      "epoch": 0.32425645235844824,
      "grad_norm": 0.0,
      "learning_rate": 6.759022263310038e-06,
      "loss": 0.0,
      "step": 47000
    },
    {
      "epoch": 0.3243254430929926,
      "grad_norm": 0.0,
      "learning_rate": 6.758332355964594e-06,
      "loss": 0.0,
      "step": 47010
    },
    {
      "epoch": 0.32439443382753697,
      "grad_norm": 0.0,
      "learning_rate": 6.757642448619151e-06,
      "loss": 0.0,
      "step": 47020
    },
    {
      "epoch": 0.3244634245620813,
      "grad_norm": 0.0,
      "learning_rate": 6.756952541273708e-06,
      "loss": 0.0015,
      "step": 47030
    },
    {
      "epoch": 0.3245324152966257,
      "grad_norm": 0.0,
      "learning_rate": 6.7562626339282646e-06,
      "loss": 0.0,
      "step": 47040
    },
    {
      "epoch": 0.32460140603117,
      "grad_norm": 0.16211307048797607,
      "learning_rate": 6.755572726582821e-06,
      "loss": 0.0002,
      "step": 47050
    },
    {
      "epoch": 0.32467039676571435,
      "grad_norm": 0.0,
      "learning_rate": 6.7548828192373775e-06,
      "loss": 0.0013,
      "step": 47060
    },
    {
      "epoch": 0.3247393875002587,
      "grad_norm": 8.754976079217158e-06,
      "learning_rate": 6.754192911891933e-06,
      "loss": 0.0,
      "step": 47070
    },
    {
      "epoch": 0.3248083782348031,
      "grad_norm": 0.0,
      "learning_rate": 6.75350300454649e-06,
      "loss": 0.0,
      "step": 47080
    },
    {
      "epoch": 0.32487736896934744,
      "grad_norm": 0.0,
      "learning_rate": 6.752813097201046e-06,
      "loss": 0.0569,
      "step": 47090
    },
    {
      "epoch": 0.32494635970389174,
      "grad_norm": 0.0,
      "learning_rate": 6.752123189855603e-06,
      "loss": 0.0,
      "step": 47100
    },
    {
      "epoch": 0.3250153504384361,
      "grad_norm": 0.0001604845019755885,
      "learning_rate": 6.751433282510159e-06,
      "loss": 0.0,
      "step": 47110
    },
    {
      "epoch": 0.32508434117298046,
      "grad_norm": 0.0,
      "learning_rate": 6.750743375164716e-06,
      "loss": 0.0,
      "step": 47120
    },
    {
      "epoch": 0.3251533319075248,
      "grad_norm": 0.0,
      "learning_rate": 6.750053467819272e-06,
      "loss": 0.0055,
      "step": 47130
    },
    {
      "epoch": 0.3252223226420692,
      "grad_norm": 0.0,
      "learning_rate": 6.7493635604738295e-06,
      "loss": 0.0,
      "step": 47140
    },
    {
      "epoch": 0.32529131337661354,
      "grad_norm": 0.0,
      "learning_rate": 6.748673653128386e-06,
      "loss": 0.048,
      "step": 47150
    },
    {
      "epoch": 0.32536030411115785,
      "grad_norm": 0.0,
      "learning_rate": 6.7479837457829424e-06,
      "loss": 0.0,
      "step": 47160
    },
    {
      "epoch": 0.3254292948457022,
      "grad_norm": 0.0,
      "learning_rate": 6.747293838437499e-06,
      "loss": 0.0001,
      "step": 47170
    },
    {
      "epoch": 0.32549828558024657,
      "grad_norm": 0.0,
      "learning_rate": 6.746672921826599e-06,
      "loss": 0.6029,
      "step": 47180
    },
    {
      "epoch": 0.32556727631479093,
      "grad_norm": 0.0,
      "learning_rate": 6.745983014481155e-06,
      "loss": 0.1036,
      "step": 47190
    },
    {
      "epoch": 0.3256362670493353,
      "grad_norm": 1.7777416706085205,
      "learning_rate": 6.745293107135712e-06,
      "loss": 0.0005,
      "step": 47200
    },
    {
      "epoch": 0.3257052577838796,
      "grad_norm": 0.0,
      "learning_rate": 6.744603199790268e-06,
      "loss": 0.0,
      "step": 47210
    },
    {
      "epoch": 0.32577424851842396,
      "grad_norm": 0.0,
      "learning_rate": 6.743913292444825e-06,
      "loss": 0.0,
      "step": 47220
    },
    {
      "epoch": 0.3258432392529683,
      "grad_norm": 0.0,
      "learning_rate": 6.743223385099382e-06,
      "loss": 0.0064,
      "step": 47230
    },
    {
      "epoch": 0.3259122299875127,
      "grad_norm": 0.0006976527511142194,
      "learning_rate": 6.7425334777539385e-06,
      "loss": 0.0,
      "step": 47240
    },
    {
      "epoch": 0.32598122072205704,
      "grad_norm": 0.00018832985369954258,
      "learning_rate": 6.741843570408495e-06,
      "loss": 0.0001,
      "step": 47250
    },
    {
      "epoch": 0.3260502114566014,
      "grad_norm": 0.0,
      "learning_rate": 6.7411536630630515e-06,
      "loss": 0.0003,
      "step": 47260
    },
    {
      "epoch": 0.3261192021911457,
      "grad_norm": 0.0,
      "learning_rate": 6.740463755717608e-06,
      "loss": 0.0,
      "step": 47270
    },
    {
      "epoch": 0.32618819292569007,
      "grad_norm": 8.414518561039586e-06,
      "learning_rate": 6.7397738483721645e-06,
      "loss": 0.0001,
      "step": 47280
    },
    {
      "epoch": 0.32625718366023443,
      "grad_norm": 0.012728461995720863,
      "learning_rate": 6.739083941026721e-06,
      "loss": 0.0,
      "step": 47290
    },
    {
      "epoch": 0.3263261743947788,
      "grad_norm": 0.0,
      "learning_rate": 6.738394033681277e-06,
      "loss": 0.0,
      "step": 47300
    },
    {
      "epoch": 0.32639516512932315,
      "grad_norm": 0.01205992791801691,
      "learning_rate": 6.737704126335833e-06,
      "loss": 0.0,
      "step": 47310
    },
    {
      "epoch": 0.32646415586386746,
      "grad_norm": 0.0,
      "learning_rate": 6.73701421899039e-06,
      "loss": 0.0,
      "step": 47320
    },
    {
      "epoch": 0.3265331465984118,
      "grad_norm": 0.0,
      "learning_rate": 6.736324311644946e-06,
      "loss": 0.0,
      "step": 47330
    },
    {
      "epoch": 0.3266021373329562,
      "grad_norm": 0.0,
      "learning_rate": 6.7356344042995034e-06,
      "loss": 0.0,
      "step": 47340
    },
    {
      "epoch": 0.32667112806750054,
      "grad_norm": 3.6986063101096534e-09,
      "learning_rate": 6.73494449695406e-06,
      "loss": 0.0,
      "step": 47350
    },
    {
      "epoch": 0.3267401188020449,
      "grad_norm": 0.0,
      "learning_rate": 6.7342545896086164e-06,
      "loss": 0.0,
      "step": 47360
    },
    {
      "epoch": 0.32680910953658926,
      "grad_norm": 0.0,
      "learning_rate": 6.733564682263173e-06,
      "loss": 0.0,
      "step": 47370
    },
    {
      "epoch": 0.32687810027113356,
      "grad_norm": 0.0,
      "learning_rate": 6.732874774917729e-06,
      "loss": 0.0009,
      "step": 47380
    },
    {
      "epoch": 0.3269470910056779,
      "grad_norm": 9.114627630424366e-08,
      "learning_rate": 6.732184867572286e-06,
      "loss": 0.0003,
      "step": 47390
    },
    {
      "epoch": 0.3270160817402223,
      "grad_norm": 0.0,
      "learning_rate": 6.731494960226842e-06,
      "loss": 0.0,
      "step": 47400
    },
    {
      "epoch": 0.32708507247476665,
      "grad_norm": 0.0,
      "learning_rate": 6.730805052881398e-06,
      "loss": 0.0007,
      "step": 47410
    },
    {
      "epoch": 0.327154063209311,
      "grad_norm": 0.0,
      "learning_rate": 6.7301151455359545e-06,
      "loss": 0.0291,
      "step": 47420
    },
    {
      "epoch": 0.3272230539438553,
      "grad_norm": 5.850496563652996e-06,
      "learning_rate": 6.729425238190511e-06,
      "loss": 0.0,
      "step": 47430
    },
    {
      "epoch": 0.3272920446783997,
      "grad_norm": 0.0,
      "learning_rate": 6.7287353308450675e-06,
      "loss": 0.0,
      "step": 47440
    },
    {
      "epoch": 0.32736103541294403,
      "grad_norm": 0.0,
      "learning_rate": 6.728045423499625e-06,
      "loss": 0.0,
      "step": 47450
    },
    {
      "epoch": 0.3274300261474884,
      "grad_norm": 1.930912540970553e-09,
      "learning_rate": 6.727355516154181e-06,
      "loss": 0.0,
      "step": 47460
    },
    {
      "epoch": 0.32749901688203276,
      "grad_norm": 0.0,
      "learning_rate": 6.726665608808738e-06,
      "loss": 0.0,
      "step": 47470
    },
    {
      "epoch": 0.3275680076165771,
      "grad_norm": 0.0059599317610263824,
      "learning_rate": 6.725975701463294e-06,
      "loss": 0.0,
      "step": 47480
    },
    {
      "epoch": 0.3276369983511214,
      "grad_norm": 0.0,
      "learning_rate": 6.725285794117851e-06,
      "loss": 0.0,
      "step": 47490
    },
    {
      "epoch": 0.3277059890856658,
      "grad_norm": 0.0,
      "learning_rate": 6.724595886772407e-06,
      "loss": 0.0,
      "step": 47500
    },
    {
      "epoch": 0.32777497982021014,
      "grad_norm": 0.0,
      "learning_rate": 6.723905979426964e-06,
      "loss": 0.0,
      "step": 47510
    },
    {
      "epoch": 0.3278439705547545,
      "grad_norm": 0.0,
      "learning_rate": 6.7232160720815194e-06,
      "loss": 0.0,
      "step": 47520
    },
    {
      "epoch": 0.32791296128929887,
      "grad_norm": 0.0,
      "learning_rate": 6.722526164736076e-06,
      "loss": 0.0,
      "step": 47530
    },
    {
      "epoch": 0.32798195202384317,
      "grad_norm": 0.0,
      "learning_rate": 6.721836257390632e-06,
      "loss": 0.0,
      "step": 47540
    },
    {
      "epoch": 0.32805094275838753,
      "grad_norm": 0.0,
      "learning_rate": 6.721146350045189e-06,
      "loss": 0.0,
      "step": 47550
    },
    {
      "epoch": 0.3281199334929319,
      "grad_norm": 0.019669951871037483,
      "learning_rate": 6.720456442699746e-06,
      "loss": 0.0023,
      "step": 47560
    },
    {
      "epoch": 0.32818892422747625,
      "grad_norm": 0.0,
      "learning_rate": 6.719766535354303e-06,
      "loss": 0.0018,
      "step": 47570
    },
    {
      "epoch": 0.3282579149620206,
      "grad_norm": 0.0,
      "learning_rate": 6.719076628008859e-06,
      "loss": 0.0,
      "step": 47580
    },
    {
      "epoch": 0.328326905696565,
      "grad_norm": 0.0,
      "learning_rate": 6.718386720663416e-06,
      "loss": 0.0,
      "step": 47590
    },
    {
      "epoch": 0.3283958964311093,
      "grad_norm": 0.0,
      "learning_rate": 6.717696813317972e-06,
      "loss": 0.0,
      "step": 47600
    },
    {
      "epoch": 0.32846488716565364,
      "grad_norm": 0.0,
      "learning_rate": 6.717006905972529e-06,
      "loss": 0.0,
      "step": 47610
    },
    {
      "epoch": 0.328533877900198,
      "grad_norm": 0.0,
      "learning_rate": 6.716316998627085e-06,
      "loss": 0.0,
      "step": 47620
    },
    {
      "epoch": 0.32860286863474236,
      "grad_norm": 0.0,
      "learning_rate": 6.715627091281642e-06,
      "loss": 0.0,
      "step": 47630
    },
    {
      "epoch": 0.3286718593692867,
      "grad_norm": 0.0,
      "learning_rate": 6.714937183936197e-06,
      "loss": 0.2436,
      "step": 47640
    },
    {
      "epoch": 0.32874085010383103,
      "grad_norm": 9.838677406311035,
      "learning_rate": 6.714247276590754e-06,
      "loss": 0.0021,
      "step": 47650
    },
    {
      "epoch": 0.3288098408383754,
      "grad_norm": 2.3980476271390216e-08,
      "learning_rate": 6.71355736924531e-06,
      "loss": 0.0,
      "step": 47660
    },
    {
      "epoch": 0.32887883157291975,
      "grad_norm": 0.000427614722866565,
      "learning_rate": 6.712867461899868e-06,
      "loss": 0.0,
      "step": 47670
    },
    {
      "epoch": 0.3289478223074641,
      "grad_norm": 0.0,
      "learning_rate": 6.712177554554424e-06,
      "loss": 0.0,
      "step": 47680
    },
    {
      "epoch": 0.32901681304200847,
      "grad_norm": 0.0,
      "learning_rate": 6.711487647208981e-06,
      "loss": 0.0,
      "step": 47690
    },
    {
      "epoch": 0.32908580377655283,
      "grad_norm": 0.0,
      "learning_rate": 6.710797739863537e-06,
      "loss": 0.0,
      "step": 47700
    },
    {
      "epoch": 0.32915479451109714,
      "grad_norm": 0.0,
      "learning_rate": 6.710107832518094e-06,
      "loss": 0.0,
      "step": 47710
    },
    {
      "epoch": 0.3292237852456415,
      "grad_norm": 3.1523313737125136e-06,
      "learning_rate": 6.70941792517265e-06,
      "loss": 0.0,
      "step": 47720
    },
    {
      "epoch": 0.32929277598018586,
      "grad_norm": 0.0,
      "learning_rate": 6.7087280178272066e-06,
      "loss": 0.0,
      "step": 47730
    },
    {
      "epoch": 0.3293617667147302,
      "grad_norm": 0.0,
      "learning_rate": 6.708038110481763e-06,
      "loss": 0.0,
      "step": 47740
    },
    {
      "epoch": 0.3294307574492746,
      "grad_norm": 485.5085754394531,
      "learning_rate": 6.707348203136319e-06,
      "loss": 0.1348,
      "step": 47750
    },
    {
      "epoch": 0.3294997481838189,
      "grad_norm": 0.0,
      "learning_rate": 6.706658295790875e-06,
      "loss": 0.0,
      "step": 47760
    },
    {
      "epoch": 0.32956873891836325,
      "grad_norm": 0.0,
      "learning_rate": 6.705968388445432e-06,
      "loss": 0.0,
      "step": 47770
    },
    {
      "epoch": 0.3296377296529076,
      "grad_norm": 0.0,
      "learning_rate": 6.705278481099989e-06,
      "loss": 0.0,
      "step": 47780
    },
    {
      "epoch": 0.32970672038745197,
      "grad_norm": 0.0,
      "learning_rate": 6.7045885737545455e-06,
      "loss": 0.0,
      "step": 47790
    },
    {
      "epoch": 0.32977571112199633,
      "grad_norm": 0.0,
      "learning_rate": 6.703898666409102e-06,
      "loss": 0.0092,
      "step": 47800
    },
    {
      "epoch": 0.3298447018565407,
      "grad_norm": 0.0,
      "learning_rate": 6.7032087590636585e-06,
      "loss": 0.0021,
      "step": 47810
    },
    {
      "epoch": 0.329913692591085,
      "grad_norm": 0.0,
      "learning_rate": 6.702518851718215e-06,
      "loss": 0.0001,
      "step": 47820
    },
    {
      "epoch": 0.32998268332562936,
      "grad_norm": 3.451175345503543e-08,
      "learning_rate": 6.7018289443727715e-06,
      "loss": 0.0,
      "step": 47830
    },
    {
      "epoch": 0.3300516740601737,
      "grad_norm": 0.0,
      "learning_rate": 6.701139037027328e-06,
      "loss": 0.0,
      "step": 47840
    },
    {
      "epoch": 0.3301206647947181,
      "grad_norm": 0.015162219293415546,
      "learning_rate": 6.7004491296818845e-06,
      "loss": 0.8594,
      "step": 47850
    },
    {
      "epoch": 0.33018965552926244,
      "grad_norm": 0.0,
      "learning_rate": 6.69975922233644e-06,
      "loss": 0.0,
      "step": 47860
    },
    {
      "epoch": 0.33025864626380674,
      "grad_norm": 3.937416477128863e-05,
      "learning_rate": 6.699069314990997e-06,
      "loss": 0.0,
      "step": 47870
    },
    {
      "epoch": 0.3303276369983511,
      "grad_norm": 0.0,
      "learning_rate": 6.698379407645553e-06,
      "loss": 0.0,
      "step": 47880
    },
    {
      "epoch": 0.33039662773289546,
      "grad_norm": 0.00018171903502661735,
      "learning_rate": 6.69768950030011e-06,
      "loss": 0.0,
      "step": 47890
    },
    {
      "epoch": 0.3304656184674398,
      "grad_norm": 0.0,
      "learning_rate": 6.696999592954667e-06,
      "loss": 0.0007,
      "step": 47900
    },
    {
      "epoch": 0.3305346092019842,
      "grad_norm": 1.445470047656272e-06,
      "learning_rate": 6.696309685609223e-06,
      "loss": 0.0,
      "step": 47910
    },
    {
      "epoch": 0.33060359993652855,
      "grad_norm": 1.4895603044351446e-06,
      "learning_rate": 6.69561977826378e-06,
      "loss": 0.0,
      "step": 47920
    },
    {
      "epoch": 0.33067259067107285,
      "grad_norm": 0.0,
      "learning_rate": 6.694929870918336e-06,
      "loss": 0.0,
      "step": 47930
    },
    {
      "epoch": 0.3307415814056172,
      "grad_norm": 0.0,
      "learning_rate": 6.694239963572893e-06,
      "loss": 0.0,
      "step": 47940
    },
    {
      "epoch": 0.3308105721401616,
      "grad_norm": 0.0,
      "learning_rate": 6.693550056227449e-06,
      "loss": 0.0,
      "step": 47950
    },
    {
      "epoch": 0.33087956287470593,
      "grad_norm": 0.0,
      "learning_rate": 6.692860148882006e-06,
      "loss": 0.0,
      "step": 47960
    },
    {
      "epoch": 0.3309485536092503,
      "grad_norm": 0.0,
      "learning_rate": 6.6921702415365615e-06,
      "loss": 0.0,
      "step": 47970
    },
    {
      "epoch": 0.3310175443437946,
      "grad_norm": 0.0,
      "learning_rate": 6.691480334191118e-06,
      "loss": 0.0001,
      "step": 47980
    },
    {
      "epoch": 0.33108653507833896,
      "grad_norm": 0.0012809181353077292,
      "learning_rate": 6.6907904268456745e-06,
      "loss": 0.0,
      "step": 47990
    },
    {
      "epoch": 0.3311555258128833,
      "grad_norm": 0.0001356671709800139,
      "learning_rate": 6.690100519500232e-06,
      "loss": 0.482,
      "step": 48000
    },
    {
      "epoch": 0.3312245165474277,
      "grad_norm": 0.0,
      "learning_rate": 6.689410612154788e-06,
      "loss": 0.0,
      "step": 48010
    },
    {
      "epoch": 0.33129350728197204,
      "grad_norm": 0.0,
      "learning_rate": 6.688720704809345e-06,
      "loss": 0.2182,
      "step": 48020
    },
    {
      "epoch": 0.3313624980165164,
      "grad_norm": 0.0,
      "learning_rate": 6.688030797463901e-06,
      "loss": 0.0011,
      "step": 48030
    },
    {
      "epoch": 0.3314314887510607,
      "grad_norm": 0.0,
      "learning_rate": 6.687340890118458e-06,
      "loss": 0.0,
      "step": 48040
    },
    {
      "epoch": 0.33150047948560507,
      "grad_norm": 0.0,
      "learning_rate": 6.686650982773014e-06,
      "loss": 0.0,
      "step": 48050
    },
    {
      "epoch": 0.33156947022014943,
      "grad_norm": 0.0,
      "learning_rate": 6.685961075427571e-06,
      "loss": 0.0,
      "step": 48060
    },
    {
      "epoch": 0.3316384609546938,
      "grad_norm": 0.0,
      "learning_rate": 6.685271168082127e-06,
      "loss": 0.0,
      "step": 48070
    },
    {
      "epoch": 0.33170745168923815,
      "grad_norm": 1.834815634538245e-06,
      "learning_rate": 6.684581260736684e-06,
      "loss": 0.0003,
      "step": 48080
    },
    {
      "epoch": 0.33177644242378246,
      "grad_norm": 5.670102382282494e-06,
      "learning_rate": 6.683891353391239e-06,
      "loss": 0.0024,
      "step": 48090
    },
    {
      "epoch": 0.3318454331583268,
      "grad_norm": 0.0,
      "learning_rate": 6.683201446045796e-06,
      "loss": 0.0633,
      "step": 48100
    },
    {
      "epoch": 0.3319144238928712,
      "grad_norm": 8.239067761373065e-10,
      "learning_rate": 6.682511538700353e-06,
      "loss": 0.0002,
      "step": 48110
    },
    {
      "epoch": 0.33198341462741554,
      "grad_norm": 0.0,
      "learning_rate": 6.68182163135491e-06,
      "loss": 0.0,
      "step": 48120
    },
    {
      "epoch": 0.3320524053619599,
      "grad_norm": 2.2167314650545222e-09,
      "learning_rate": 6.681131724009466e-06,
      "loss": 0.0,
      "step": 48130
    },
    {
      "epoch": 0.33212139609650426,
      "grad_norm": 0.00016086299729067832,
      "learning_rate": 6.680441816664023e-06,
      "loss": 0.0,
      "step": 48140
    },
    {
      "epoch": 0.33219038683104857,
      "grad_norm": 5.424868376735503e-08,
      "learning_rate": 6.679751909318579e-06,
      "loss": 0.0,
      "step": 48150
    },
    {
      "epoch": 0.33225937756559293,
      "grad_norm": 0.0,
      "learning_rate": 6.679062001973136e-06,
      "loss": 0.0,
      "step": 48160
    },
    {
      "epoch": 0.3323283683001373,
      "grad_norm": 0.0,
      "learning_rate": 6.678372094627692e-06,
      "loss": 0.0,
      "step": 48170
    },
    {
      "epoch": 0.33239735903468165,
      "grad_norm": 0.0,
      "learning_rate": 6.677682187282249e-06,
      "loss": 0.0,
      "step": 48180
    },
    {
      "epoch": 0.332466349769226,
      "grad_norm": 0.054791051894426346,
      "learning_rate": 6.676992279936805e-06,
      "loss": 0.0,
      "step": 48190
    },
    {
      "epoch": 0.33253534050377037,
      "grad_norm": 0.0,
      "learning_rate": 6.676302372591361e-06,
      "loss": 0.0151,
      "step": 48200
    },
    {
      "epoch": 0.3326043312383147,
      "grad_norm": 0.0009703313116915524,
      "learning_rate": 6.675612465245917e-06,
      "loss": 0.005,
      "step": 48210
    },
    {
      "epoch": 0.33267332197285904,
      "grad_norm": 0.0,
      "learning_rate": 6.6749225579004754e-06,
      "loss": 0.0,
      "step": 48220
    },
    {
      "epoch": 0.3327423127074034,
      "grad_norm": 0.0,
      "learning_rate": 6.674232650555031e-06,
      "loss": 0.0,
      "step": 48230
    },
    {
      "epoch": 0.33281130344194776,
      "grad_norm": 0.0,
      "learning_rate": 6.673542743209588e-06,
      "loss": 1.0576,
      "step": 48240
    },
    {
      "epoch": 0.3328802941764921,
      "grad_norm": 0.0,
      "learning_rate": 6.672852835864144e-06,
      "loss": 0.0,
      "step": 48250
    },
    {
      "epoch": 0.3329492849110364,
      "grad_norm": 0.0,
      "learning_rate": 6.6721629285187006e-06,
      "loss": 0.0,
      "step": 48260
    },
    {
      "epoch": 0.3330182756455808,
      "grad_norm": 0.0,
      "learning_rate": 6.671473021173257e-06,
      "loss": 0.0029,
      "step": 48270
    },
    {
      "epoch": 0.33308726638012515,
      "grad_norm": 0.0,
      "learning_rate": 6.6707831138278135e-06,
      "loss": 0.0,
      "step": 48280
    },
    {
      "epoch": 0.3331562571146695,
      "grad_norm": 0.0,
      "learning_rate": 6.67009320648237e-06,
      "loss": 0.0,
      "step": 48290
    },
    {
      "epoch": 0.33322524784921387,
      "grad_norm": 0.0,
      "learning_rate": 6.6694032991369265e-06,
      "loss": 0.0,
      "step": 48300
    },
    {
      "epoch": 0.33329423858375823,
      "grad_norm": 0.0,
      "learning_rate": 6.668713391791482e-06,
      "loss": 0.0,
      "step": 48310
    },
    {
      "epoch": 0.33336322931830253,
      "grad_norm": 0.0,
      "learning_rate": 6.668023484446039e-06,
      "loss": 0.0,
      "step": 48320
    },
    {
      "epoch": 0.3334322200528469,
      "grad_norm": 1.0825621510335282e-09,
      "learning_rate": 6.667333577100597e-06,
      "loss": 0.0,
      "step": 48330
    },
    {
      "epoch": 0.33350121078739126,
      "grad_norm": 0.0,
      "learning_rate": 6.6666436697551525e-06,
      "loss": 0.0,
      "step": 48340
    },
    {
      "epoch": 0.3335702015219356,
      "grad_norm": 0.0,
      "learning_rate": 6.665953762409709e-06,
      "loss": 0.0011,
      "step": 48350
    },
    {
      "epoch": 0.33363919225648,
      "grad_norm": 0.0,
      "learning_rate": 6.6652638550642655e-06,
      "loss": 0.0,
      "step": 48360
    },
    {
      "epoch": 0.3337081829910243,
      "grad_norm": 0.002793417079374194,
      "learning_rate": 6.664573947718822e-06,
      "loss": 0.0006,
      "step": 48370
    },
    {
      "epoch": 0.33377717372556864,
      "grad_norm": 0.0016971139702945948,
      "learning_rate": 6.6638840403733784e-06,
      "loss": 0.0,
      "step": 48380
    },
    {
      "epoch": 0.333846164460113,
      "grad_norm": 0.0,
      "learning_rate": 6.663194133027935e-06,
      "loss": 0.0,
      "step": 48390
    },
    {
      "epoch": 0.33391515519465736,
      "grad_norm": 0.038245365023612976,
      "learning_rate": 6.6625042256824914e-06,
      "loss": 0.0,
      "step": 48400
    },
    {
      "epoch": 0.3339841459292017,
      "grad_norm": 0.0,
      "learning_rate": 6.661814318337048e-06,
      "loss": 0.0,
      "step": 48410
    },
    {
      "epoch": 0.3340531366637461,
      "grad_norm": 6.902739713154915e-09,
      "learning_rate": 6.661124410991604e-06,
      "loss": 0.0,
      "step": 48420
    },
    {
      "epoch": 0.3341221273982904,
      "grad_norm": 0.0,
      "learning_rate": 6.66043450364616e-06,
      "loss": 0.0,
      "step": 48430
    },
    {
      "epoch": 0.33419111813283475,
      "grad_norm": 0.0,
      "learning_rate": 6.659744596300718e-06,
      "loss": 0.0,
      "step": 48440
    },
    {
      "epoch": 0.3342601088673791,
      "grad_norm": 0.0,
      "learning_rate": 6.659054688955274e-06,
      "loss": 0.0001,
      "step": 48450
    },
    {
      "epoch": 0.3343290996019235,
      "grad_norm": 0.0,
      "learning_rate": 6.65836478160983e-06,
      "loss": 0.0,
      "step": 48460
    },
    {
      "epoch": 0.33439809033646783,
      "grad_norm": 0.0,
      "learning_rate": 6.657674874264387e-06,
      "loss": 0.0,
      "step": 48470
    },
    {
      "epoch": 0.33446708107101214,
      "grad_norm": 0.0,
      "learning_rate": 6.656984966918943e-06,
      "loss": 0.6441,
      "step": 48480
    },
    {
      "epoch": 0.3345360718055565,
      "grad_norm": 0.0,
      "learning_rate": 6.6562950595735e-06,
      "loss": 0.001,
      "step": 48490
    },
    {
      "epoch": 0.33460506254010086,
      "grad_norm": 0.0,
      "learning_rate": 6.655605152228056e-06,
      "loss": 0.0043,
      "step": 48500
    },
    {
      "epoch": 0.3346740532746452,
      "grad_norm": 0.15096665918827057,
      "learning_rate": 6.654915244882613e-06,
      "loss": 0.0,
      "step": 48510
    },
    {
      "epoch": 0.3347430440091896,
      "grad_norm": 0.0,
      "learning_rate": 6.654225337537169e-06,
      "loss": 0.0,
      "step": 48520
    },
    {
      "epoch": 0.33481203474373394,
      "grad_norm": 0.0,
      "learning_rate": 6.653535430191726e-06,
      "loss": 0.0177,
      "step": 48530
    },
    {
      "epoch": 0.33488102547827825,
      "grad_norm": 0.0,
      "learning_rate": 6.6528455228462814e-06,
      "loss": 0.0,
      "step": 48540
    },
    {
      "epoch": 0.3349500162128226,
      "grad_norm": 0.11346650123596191,
      "learning_rate": 6.65215561550084e-06,
      "loss": 0.0,
      "step": 48550
    },
    {
      "epoch": 0.33501900694736697,
      "grad_norm": 0.0,
      "learning_rate": 6.651465708155396e-06,
      "loss": 0.0,
      "step": 48560
    },
    {
      "epoch": 0.33508799768191133,
      "grad_norm": 0.0,
      "learning_rate": 6.650775800809952e-06,
      "loss": 0.0,
      "step": 48570
    },
    {
      "epoch": 0.3351569884164557,
      "grad_norm": 0.0,
      "learning_rate": 6.650085893464508e-06,
      "loss": 0.0,
      "step": 48580
    },
    {
      "epoch": 0.335225979151,
      "grad_norm": 0.0,
      "learning_rate": 6.649395986119065e-06,
      "loss": 0.0,
      "step": 48590
    },
    {
      "epoch": 0.33529496988554436,
      "grad_norm": 0.0,
      "learning_rate": 6.648706078773621e-06,
      "loss": 0.0002,
      "step": 48600
    },
    {
      "epoch": 0.3353639606200887,
      "grad_norm": 0.0,
      "learning_rate": 6.648016171428178e-06,
      "loss": 0.0,
      "step": 48610
    },
    {
      "epoch": 0.3354329513546331,
      "grad_norm": 0.0,
      "learning_rate": 6.647326264082734e-06,
      "loss": 0.0,
      "step": 48620
    },
    {
      "epoch": 0.33550194208917744,
      "grad_norm": 0.0,
      "learning_rate": 6.646636356737291e-06,
      "loss": 0.0,
      "step": 48630
    },
    {
      "epoch": 0.3355709328237218,
      "grad_norm": 0.0,
      "learning_rate": 6.645946449391847e-06,
      "loss": 0.0,
      "step": 48640
    },
    {
      "epoch": 0.3356399235582661,
      "grad_norm": 5.627112287243108e-08,
      "learning_rate": 6.645256542046403e-06,
      "loss": 0.0,
      "step": 48650
    },
    {
      "epoch": 0.33570891429281047,
      "grad_norm": 0.0,
      "learning_rate": 6.644566634700961e-06,
      "loss": 0.0,
      "step": 48660
    },
    {
      "epoch": 0.33577790502735483,
      "grad_norm": 0.0,
      "learning_rate": 6.6438767273555175e-06,
      "loss": 0.0,
      "step": 48670
    },
    {
      "epoch": 0.3358468957618992,
      "grad_norm": 0.002178758382797241,
      "learning_rate": 6.643186820010073e-06,
      "loss": 0.0,
      "step": 48680
    },
    {
      "epoch": 0.33591588649644355,
      "grad_norm": 0.0,
      "learning_rate": 6.64249691266463e-06,
      "loss": 0.0279,
      "step": 48690
    },
    {
      "epoch": 0.33598487723098786,
      "grad_norm": 0.0,
      "learning_rate": 6.641807005319186e-06,
      "loss": 0.0,
      "step": 48700
    },
    {
      "epoch": 0.3360538679655322,
      "grad_norm": 0.0014523827703669667,
      "learning_rate": 6.641117097973743e-06,
      "loss": 0.002,
      "step": 48710
    },
    {
      "epoch": 0.3361228587000766,
      "grad_norm": 0.0,
      "learning_rate": 6.640427190628299e-06,
      "loss": 0.0,
      "step": 48720
    },
    {
      "epoch": 0.33619184943462094,
      "grad_norm": 0.00023559408145956695,
      "learning_rate": 6.639737283282856e-06,
      "loss": 0.0,
      "step": 48730
    },
    {
      "epoch": 0.3362608401691653,
      "grad_norm": 0.0,
      "learning_rate": 6.639047375937412e-06,
      "loss": 0.0026,
      "step": 48740
    },
    {
      "epoch": 0.33632983090370966,
      "grad_norm": 9.214473073448914e-10,
      "learning_rate": 6.638357468591969e-06,
      "loss": 0.0653,
      "step": 48750
    },
    {
      "epoch": 0.33639882163825396,
      "grad_norm": 0.0,
      "learning_rate": 6.637667561246525e-06,
      "loss": 0.0,
      "step": 48760
    },
    {
      "epoch": 0.3364678123727983,
      "grad_norm": 0.0,
      "learning_rate": 6.636977653901082e-06,
      "loss": 0.0,
      "step": 48770
    },
    {
      "epoch": 0.3365368031073427,
      "grad_norm": 0.0,
      "learning_rate": 6.636287746555639e-06,
      "loss": 0.0,
      "step": 48780
    },
    {
      "epoch": 0.33660579384188705,
      "grad_norm": 0.0012921111192554235,
      "learning_rate": 6.6355978392101946e-06,
      "loss": 0.0,
      "step": 48790
    },
    {
      "epoch": 0.3366747845764314,
      "grad_norm": 0.0,
      "learning_rate": 6.634907931864751e-06,
      "loss": 0.0,
      "step": 48800
    },
    {
      "epoch": 0.3367437753109757,
      "grad_norm": 7.016439429463617e-09,
      "learning_rate": 6.6342180245193075e-06,
      "loss": 0.0,
      "step": 48810
    },
    {
      "epoch": 0.3368127660455201,
      "grad_norm": 322.56787109375,
      "learning_rate": 6.633528117173864e-06,
      "loss": 0.129,
      "step": 48820
    },
    {
      "epoch": 0.33688175678006443,
      "grad_norm": 1.9350633806425321e-07,
      "learning_rate": 6.6328382098284205e-06,
      "loss": 0.0042,
      "step": 48830
    },
    {
      "epoch": 0.3369507475146088,
      "grad_norm": 0.0039228470996022224,
      "learning_rate": 6.632148302482977e-06,
      "loss": 0.0,
      "step": 48840
    },
    {
      "epoch": 0.33701973824915316,
      "grad_norm": 0.0,
      "learning_rate": 6.6314583951375335e-06,
      "loss": 0.0,
      "step": 48850
    },
    {
      "epoch": 0.3370887289836975,
      "grad_norm": 8.374750177608803e-05,
      "learning_rate": 6.63076848779209e-06,
      "loss": 0.0,
      "step": 48860
    },
    {
      "epoch": 0.3371577197182418,
      "grad_norm": 9.157400948645034e-10,
      "learning_rate": 6.6300785804466465e-06,
      "loss": 0.0,
      "step": 48870
    },
    {
      "epoch": 0.3372267104527862,
      "grad_norm": 0.0,
      "learning_rate": 6.629388673101204e-06,
      "loss": 0.0,
      "step": 48880
    },
    {
      "epoch": 0.33729570118733054,
      "grad_norm": 0.04398076981306076,
      "learning_rate": 6.62869876575576e-06,
      "loss": 0.5645,
      "step": 48890
    },
    {
      "epoch": 0.3373646919218749,
      "grad_norm": 0.0,
      "learning_rate": 6.628008858410317e-06,
      "loss": 0.0402,
      "step": 48900
    },
    {
      "epoch": 0.33743368265641926,
      "grad_norm": 1.1633978829195257e-06,
      "learning_rate": 6.6273189510648724e-06,
      "loss": 0.0,
      "step": 48910
    },
    {
      "epoch": 0.33750267339096357,
      "grad_norm": 5.3339068273317025e-08,
      "learning_rate": 6.626629043719429e-06,
      "loss": 0.0,
      "step": 48920
    },
    {
      "epoch": 0.33757166412550793,
      "grad_norm": 0.0,
      "learning_rate": 6.625939136373985e-06,
      "loss": 0.0,
      "step": 48930
    },
    {
      "epoch": 0.3376406548600523,
      "grad_norm": 0.0,
      "learning_rate": 6.625249229028542e-06,
      "loss": 0.0007,
      "step": 48940
    },
    {
      "epoch": 0.33770964559459665,
      "grad_norm": 0.0,
      "learning_rate": 6.624559321683098e-06,
      "loss": 0.0,
      "step": 48950
    },
    {
      "epoch": 0.337778636329141,
      "grad_norm": 0.0,
      "learning_rate": 6.623869414337655e-06,
      "loss": 0.0002,
      "step": 48960
    },
    {
      "epoch": 0.3378476270636854,
      "grad_norm": 0.0,
      "learning_rate": 6.623179506992211e-06,
      "loss": 0.0,
      "step": 48970
    },
    {
      "epoch": 0.3379166177982297,
      "grad_norm": 4.9775690236231185e-09,
      "learning_rate": 6.622489599646768e-06,
      "loss": 0.0,
      "step": 48980
    },
    {
      "epoch": 0.33798560853277404,
      "grad_norm": 0.0041794306598603725,
      "learning_rate": 6.621799692301325e-06,
      "loss": 0.187,
      "step": 48990
    },
    {
      "epoch": 0.3380545992673184,
      "grad_norm": 8.096119330502916e-08,
      "learning_rate": 6.621109784955882e-06,
      "loss": 0.0001,
      "step": 49000
    },
    {
      "epoch": 0.33812359000186276,
      "grad_norm": 0.0027439144905656576,
      "learning_rate": 6.620419877610438e-06,
      "loss": 0.0001,
      "step": 49010
    },
    {
      "epoch": 0.3381925807364071,
      "grad_norm": 0.0,
      "learning_rate": 6.619729970264994e-06,
      "loss": 0.0,
      "step": 49020
    },
    {
      "epoch": 0.3382615714709514,
      "grad_norm": 1.600804171175696e-05,
      "learning_rate": 6.61904006291955e-06,
      "loss": 0.0,
      "step": 49030
    },
    {
      "epoch": 0.3383305622054958,
      "grad_norm": 4.895523488812614e-07,
      "learning_rate": 6.618350155574107e-06,
      "loss": 0.0,
      "step": 49040
    },
    {
      "epoch": 0.33839955294004015,
      "grad_norm": 0.005733453668653965,
      "learning_rate": 6.617660248228663e-06,
      "loss": 0.0,
      "step": 49050
    },
    {
      "epoch": 0.3384685436745845,
      "grad_norm": 1.7886375713871416e-09,
      "learning_rate": 6.61697034088322e-06,
      "loss": 0.0,
      "step": 49060
    },
    {
      "epoch": 0.33853753440912887,
      "grad_norm": 9.078182483790442e-05,
      "learning_rate": 6.616280433537776e-06,
      "loss": 0.0,
      "step": 49070
    },
    {
      "epoch": 0.33860652514367323,
      "grad_norm": 0.0,
      "learning_rate": 6.615590526192333e-06,
      "loss": 0.0,
      "step": 49080
    },
    {
      "epoch": 0.33867551587821754,
      "grad_norm": 0.06932124495506287,
      "learning_rate": 6.614900618846889e-06,
      "loss": 0.0,
      "step": 49090
    },
    {
      "epoch": 0.3387445066127619,
      "grad_norm": 9.345848894781739e-08,
      "learning_rate": 6.614210711501447e-06,
      "loss": 0.0055,
      "step": 49100
    },
    {
      "epoch": 0.33881349734730626,
      "grad_norm": 4.323508262634277,
      "learning_rate": 6.613520804156003e-06,
      "loss": 0.0205,
      "step": 49110
    },
    {
      "epoch": 0.3388824880818506,
      "grad_norm": 0.0,
      "learning_rate": 6.61283089681056e-06,
      "loss": 0.0031,
      "step": 49120
    },
    {
      "epoch": 0.338951478816395,
      "grad_norm": 0.0,
      "learning_rate": 6.612140989465115e-06,
      "loss": 0.0001,
      "step": 49130
    },
    {
      "epoch": 0.3390204695509393,
      "grad_norm": 0.0,
      "learning_rate": 6.611451082119672e-06,
      "loss": 0.0,
      "step": 49140
    },
    {
      "epoch": 0.33908946028548365,
      "grad_norm": 0.0,
      "learning_rate": 6.610761174774228e-06,
      "loss": 0.2012,
      "step": 49150
    },
    {
      "epoch": 0.339158451020028,
      "grad_norm": 0.0,
      "learning_rate": 6.610071267428785e-06,
      "loss": 0.0,
      "step": 49160
    },
    {
      "epoch": 0.33922744175457237,
      "grad_norm": 0.0,
      "learning_rate": 6.609381360083341e-06,
      "loss": 0.0,
      "step": 49170
    },
    {
      "epoch": 0.33929643248911673,
      "grad_norm": 0.0,
      "learning_rate": 6.608691452737898e-06,
      "loss": 0.0,
      "step": 49180
    },
    {
      "epoch": 0.3393654232236611,
      "grad_norm": 0.0,
      "learning_rate": 6.608070536126999e-06,
      "loss": 0.5285,
      "step": 49190
    },
    {
      "epoch": 0.3394344139582054,
      "grad_norm": 0.0,
      "learning_rate": 6.607380628781556e-06,
      "loss": 0.0,
      "step": 49200
    },
    {
      "epoch": 0.33950340469274976,
      "grad_norm": 0.0,
      "learning_rate": 6.606690721436112e-06,
      "loss": 0.0,
      "step": 49210
    },
    {
      "epoch": 0.3395723954272941,
      "grad_norm": 0.00011352659930707887,
      "learning_rate": 6.606000814090669e-06,
      "loss": 0.0,
      "step": 49220
    },
    {
      "epoch": 0.3396413861618385,
      "grad_norm": 0.0,
      "learning_rate": 6.605310906745225e-06,
      "loss": 0.0523,
      "step": 49230
    },
    {
      "epoch": 0.33971037689638284,
      "grad_norm": 0.3528650104999542,
      "learning_rate": 6.604620999399781e-06,
      "loss": 0.0,
      "step": 49240
    },
    {
      "epoch": 0.33977936763092714,
      "grad_norm": 0.0,
      "learning_rate": 6.603931092054337e-06,
      "loss": 0.0001,
      "step": 49250
    },
    {
      "epoch": 0.3398483583654715,
      "grad_norm": 0.0,
      "learning_rate": 6.603241184708894e-06,
      "loss": 0.0,
      "step": 49260
    },
    {
      "epoch": 0.33991734910001586,
      "grad_norm": 0.0,
      "learning_rate": 6.60255127736345e-06,
      "loss": 0.0,
      "step": 49270
    },
    {
      "epoch": 0.3399863398345602,
      "grad_norm": 0.0,
      "learning_rate": 6.601861370018007e-06,
      "loss": 0.0,
      "step": 49280
    },
    {
      "epoch": 0.3400553305691046,
      "grad_norm": 0.0,
      "learning_rate": 6.601171462672563e-06,
      "loss": 0.0,
      "step": 49290
    },
    {
      "epoch": 0.34012432130364895,
      "grad_norm": 0.0,
      "learning_rate": 6.600481555327121e-06,
      "loss": 0.0,
      "step": 49300
    },
    {
      "epoch": 0.34019331203819325,
      "grad_norm": 0.0,
      "learning_rate": 6.599791647981677e-06,
      "loss": 0.0003,
      "step": 49310
    },
    {
      "epoch": 0.3402623027727376,
      "grad_norm": 0.0,
      "learning_rate": 6.5991017406362336e-06,
      "loss": 0.0,
      "step": 49320
    },
    {
      "epoch": 0.340331293507282,
      "grad_norm": 0.0,
      "learning_rate": 6.59841183329079e-06,
      "loss": 0.0,
      "step": 49330
    },
    {
      "epoch": 0.34040028424182633,
      "grad_norm": 0.0023016713093966246,
      "learning_rate": 6.5977219259453465e-06,
      "loss": 0.0,
      "step": 49340
    },
    {
      "epoch": 0.3404692749763707,
      "grad_norm": 0.0,
      "learning_rate": 6.597032018599902e-06,
      "loss": 0.0,
      "step": 49350
    },
    {
      "epoch": 0.340538265710915,
      "grad_norm": 1.1255200149662414e-07,
      "learning_rate": 6.596342111254459e-06,
      "loss": 0.0,
      "step": 49360
    },
    {
      "epoch": 0.34060725644545936,
      "grad_norm": 0.0,
      "learning_rate": 6.595652203909015e-06,
      "loss": 0.0,
      "step": 49370
    },
    {
      "epoch": 0.3406762471800037,
      "grad_norm": 0.0,
      "learning_rate": 6.594962296563572e-06,
      "loss": 0.0,
      "step": 49380
    },
    {
      "epoch": 0.3407452379145481,
      "grad_norm": 0.0,
      "learning_rate": 6.594272389218128e-06,
      "loss": 0.0047,
      "step": 49390
    },
    {
      "epoch": 0.34081422864909244,
      "grad_norm": 0.0,
      "learning_rate": 6.593582481872685e-06,
      "loss": 0.0001,
      "step": 49400
    },
    {
      "epoch": 0.3408832193836368,
      "grad_norm": 0.0,
      "learning_rate": 6.592892574527242e-06,
      "loss": 0.0,
      "step": 49410
    },
    {
      "epoch": 0.3409522101181811,
      "grad_norm": 0.0,
      "learning_rate": 6.5922026671817985e-06,
      "loss": 0.0004,
      "step": 49420
    },
    {
      "epoch": 0.34102120085272547,
      "grad_norm": 0.0,
      "learning_rate": 6.591512759836355e-06,
      "loss": 0.0,
      "step": 49430
    },
    {
      "epoch": 0.34109019158726983,
      "grad_norm": 0.0,
      "learning_rate": 6.5908228524909115e-06,
      "loss": 0.1472,
      "step": 49440
    },
    {
      "epoch": 0.3411591823218142,
      "grad_norm": 0.0,
      "learning_rate": 6.590132945145468e-06,
      "loss": 0.0,
      "step": 49450
    },
    {
      "epoch": 0.34122817305635855,
      "grad_norm": 0.0011365123791620135,
      "learning_rate": 6.5894430378000244e-06,
      "loss": 0.0,
      "step": 49460
    },
    {
      "epoch": 0.34129716379090286,
      "grad_norm": 0.0,
      "learning_rate": 6.58875313045458e-06,
      "loss": 0.0,
      "step": 49470
    },
    {
      "epoch": 0.3413661545254472,
      "grad_norm": 0.00041695666732266545,
      "learning_rate": 6.5880632231091366e-06,
      "loss": 0.0,
      "step": 49480
    },
    {
      "epoch": 0.3414351452599916,
      "grad_norm": 0.0,
      "learning_rate": 6.587373315763693e-06,
      "loss": 0.0003,
      "step": 49490
    },
    {
      "epoch": 0.34150413599453594,
      "grad_norm": 0.0,
      "learning_rate": 6.5866834084182495e-06,
      "loss": 0.0,
      "step": 49500
    },
    {
      "epoch": 0.3415731267290803,
      "grad_norm": 0.0,
      "learning_rate": 6.585993501072806e-06,
      "loss": 0.0,
      "step": 49510
    },
    {
      "epoch": 0.34164211746362466,
      "grad_norm": 0.0,
      "learning_rate": 6.585303593727363e-06,
      "loss": 0.0402,
      "step": 49520
    },
    {
      "epoch": 0.34171110819816897,
      "grad_norm": 0.0,
      "learning_rate": 6.58461368638192e-06,
      "loss": 0.0,
      "step": 49530
    },
    {
      "epoch": 0.3417800989327133,
      "grad_norm": 1.6668776652295492e-06,
      "learning_rate": 6.583923779036476e-06,
      "loss": 0.0,
      "step": 49540
    },
    {
      "epoch": 0.3418490896672577,
      "grad_norm": 0.003756061429157853,
      "learning_rate": 6.583233871691033e-06,
      "loss": 0.0065,
      "step": 49550
    },
    {
      "epoch": 0.34191808040180205,
      "grad_norm": 9.396153854979161e-10,
      "learning_rate": 6.582543964345589e-06,
      "loss": 0.0,
      "step": 49560
    },
    {
      "epoch": 0.3419870711363464,
      "grad_norm": 1.390722843552794e-07,
      "learning_rate": 6.581854057000146e-06,
      "loss": 0.0,
      "step": 49570
    },
    {
      "epoch": 0.3420560618708907,
      "grad_norm": 0.0,
      "learning_rate": 6.5811641496547015e-06,
      "loss": 0.0,
      "step": 49580
    },
    {
      "epoch": 0.3421250526054351,
      "grad_norm": 0.0,
      "learning_rate": 6.580474242309258e-06,
      "loss": 0.0,
      "step": 49590
    },
    {
      "epoch": 0.34219404333997944,
      "grad_norm": 0.0,
      "learning_rate": 6.5797843349638145e-06,
      "loss": 0.0,
      "step": 49600
    },
    {
      "epoch": 0.3422630340745238,
      "grad_norm": 0.0,
      "learning_rate": 6.579094427618371e-06,
      "loss": 0.0,
      "step": 49610
    },
    {
      "epoch": 0.34233202480906816,
      "grad_norm": 0.0,
      "learning_rate": 6.5784045202729274e-06,
      "loss": 0.0,
      "step": 49620
    },
    {
      "epoch": 0.3424010155436125,
      "grad_norm": 0.0,
      "learning_rate": 6.577714612927485e-06,
      "loss": 0.0,
      "step": 49630
    },
    {
      "epoch": 0.3424700062781568,
      "grad_norm": 0.0,
      "learning_rate": 6.577024705582041e-06,
      "loss": 0.0,
      "step": 49640
    },
    {
      "epoch": 0.3425389970127012,
      "grad_norm": 0.0,
      "learning_rate": 6.576334798236598e-06,
      "loss": 0.0,
      "step": 49650
    },
    {
      "epoch": 0.34260798774724555,
      "grad_norm": 0.0022532902657985687,
      "learning_rate": 6.575644890891154e-06,
      "loss": 0.0,
      "step": 49660
    },
    {
      "epoch": 0.3426769784817899,
      "grad_norm": 0.0,
      "learning_rate": 6.574954983545711e-06,
      "loss": 0.4844,
      "step": 49670
    },
    {
      "epoch": 0.34274596921633427,
      "grad_norm": 0.0,
      "learning_rate": 6.574265076200267e-06,
      "loss": 0.0,
      "step": 49680
    },
    {
      "epoch": 0.3428149599508786,
      "grad_norm": 0.0,
      "learning_rate": 6.573575168854823e-06,
      "loss": 0.0,
      "step": 49690
    },
    {
      "epoch": 0.34288395068542293,
      "grad_norm": 1.9226769065738836e-09,
      "learning_rate": 6.572885261509379e-06,
      "loss": 0.0,
      "step": 49700
    },
    {
      "epoch": 0.3429529414199673,
      "grad_norm": 0.0,
      "learning_rate": 6.572195354163936e-06,
      "loss": 0.0,
      "step": 49710
    },
    {
      "epoch": 0.34302193215451166,
      "grad_norm": 0.0,
      "learning_rate": 6.571505446818492e-06,
      "loss": 0.0344,
      "step": 49720
    },
    {
      "epoch": 0.343090922889056,
      "grad_norm": 0.0,
      "learning_rate": 6.570815539473049e-06,
      "loss": 0.0,
      "step": 49730
    },
    {
      "epoch": 0.3431599136236004,
      "grad_norm": 0.0,
      "learning_rate": 6.570125632127606e-06,
      "loss": 0.0007,
      "step": 49740
    },
    {
      "epoch": 0.3432289043581447,
      "grad_norm": 0.0,
      "learning_rate": 6.569435724782163e-06,
      "loss": 0.0,
      "step": 49750
    },
    {
      "epoch": 0.34329789509268904,
      "grad_norm": 0.0030286326073110104,
      "learning_rate": 6.568745817436719e-06,
      "loss": 0.0,
      "step": 49760
    },
    {
      "epoch": 0.3433668858272334,
      "grad_norm": 0.0,
      "learning_rate": 6.568055910091276e-06,
      "loss": 0.0,
      "step": 49770
    },
    {
      "epoch": 0.34343587656177776,
      "grad_norm": 0.0,
      "learning_rate": 6.567366002745832e-06,
      "loss": 0.003,
      "step": 49780
    },
    {
      "epoch": 0.3435048672963221,
      "grad_norm": 0.0,
      "learning_rate": 6.566676095400389e-06,
      "loss": 0.0,
      "step": 49790
    },
    {
      "epoch": 0.34357385803086643,
      "grad_norm": 0.0,
      "learning_rate": 6.565986188054945e-06,
      "loss": 0.0001,
      "step": 49800
    },
    {
      "epoch": 0.3436428487654108,
      "grad_norm": 0.0,
      "learning_rate": 6.565296280709501e-06,
      "loss": 0.0,
      "step": 49810
    },
    {
      "epoch": 0.34371183949995515,
      "grad_norm": 0.0,
      "learning_rate": 6.564606373364057e-06,
      "loss": 0.0,
      "step": 49820
    },
    {
      "epoch": 0.3437808302344995,
      "grad_norm": 0.0,
      "learning_rate": 6.563916466018614e-06,
      "loss": 0.0,
      "step": 49830
    },
    {
      "epoch": 0.3438498209690439,
      "grad_norm": 0.0,
      "learning_rate": 6.56322655867317e-06,
      "loss": 0.0,
      "step": 49840
    },
    {
      "epoch": 0.34391881170358823,
      "grad_norm": 2.9462868042173795e-05,
      "learning_rate": 6.5625366513277276e-06,
      "loss": 0.0,
      "step": 49850
    },
    {
      "epoch": 0.34398780243813254,
      "grad_norm": 0.0,
      "learning_rate": 6.561846743982284e-06,
      "loss": 0.0,
      "step": 49860
    },
    {
      "epoch": 0.3440567931726769,
      "grad_norm": 0.0,
      "learning_rate": 6.5611568366368405e-06,
      "loss": 0.0,
      "step": 49870
    },
    {
      "epoch": 0.34412578390722126,
      "grad_norm": 0.0,
      "learning_rate": 6.560466929291397e-06,
      "loss": 0.0,
      "step": 49880
    },
    {
      "epoch": 0.3441947746417656,
      "grad_norm": 0.0,
      "learning_rate": 6.5597770219459535e-06,
      "loss": 0.0,
      "step": 49890
    },
    {
      "epoch": 0.34426376537631,
      "grad_norm": 0.0,
      "learning_rate": 6.55908711460051e-06,
      "loss": 0.0,
      "step": 49900
    },
    {
      "epoch": 0.3443327561108543,
      "grad_norm": 0.0,
      "learning_rate": 6.5583972072550665e-06,
      "loss": 0.0,
      "step": 49910
    },
    {
      "epoch": 0.34440174684539865,
      "grad_norm": 1.3277878679218702e-05,
      "learning_rate": 6.557707299909622e-06,
      "loss": 0.0,
      "step": 49920
    },
    {
      "epoch": 0.344470737579943,
      "grad_norm": 0.0,
      "learning_rate": 6.557017392564179e-06,
      "loss": 0.0,
      "step": 49930
    },
    {
      "epoch": 0.34453972831448737,
      "grad_norm": 0.0,
      "learning_rate": 6.556327485218735e-06,
      "loss": 0.0,
      "step": 49940
    },
    {
      "epoch": 0.34460871904903173,
      "grad_norm": 0.0,
      "learning_rate": 6.555637577873292e-06,
      "loss": 0.0,
      "step": 49950
    },
    {
      "epoch": 0.3446777097835761,
      "grad_norm": 1.1702499389648438,
      "learning_rate": 6.554947670527849e-06,
      "loss": 0.0001,
      "step": 49960
    },
    {
      "epoch": 0.3447467005181204,
      "grad_norm": 0.0,
      "learning_rate": 6.5542577631824054e-06,
      "loss": 0.0,
      "step": 49970
    },
    {
      "epoch": 0.34481569125266476,
      "grad_norm": 0.0001836222072597593,
      "learning_rate": 6.553567855836962e-06,
      "loss": 0.0613,
      "step": 49980
    },
    {
      "epoch": 0.3448846819872091,
      "grad_norm": 9.513488885559696e-10,
      "learning_rate": 6.552877948491518e-06,
      "loss": 0.0,
      "step": 49990
    },
    {
      "epoch": 0.3449536727217535,
      "grad_norm": 0.0,
      "learning_rate": 6.552188041146075e-06,
      "loss": 0.0001,
      "step": 50000
    },
    {
      "epoch": 0.34502266345629784,
      "grad_norm": 0.0,
      "learning_rate": 6.551498133800631e-06,
      "loss": 0.0,
      "step": 50010
    },
    {
      "epoch": 0.34509165419084215,
      "grad_norm": 0.0,
      "learning_rate": 6.550808226455188e-06,
      "loss": 0.0,
      "step": 50020
    },
    {
      "epoch": 0.3451606449253865,
      "grad_norm": 0.0,
      "learning_rate": 6.5501183191097435e-06,
      "loss": 0.0,
      "step": 50030
    },
    {
      "epoch": 0.34522963565993087,
      "grad_norm": 8.063936718372133e-08,
      "learning_rate": 6.5494284117643e-06,
      "loss": 0.0089,
      "step": 50040
    },
    {
      "epoch": 0.3452986263944752,
      "grad_norm": 0.0,
      "learning_rate": 6.5487385044188565e-06,
      "loss": 0.0,
      "step": 50050
    },
    {
      "epoch": 0.3453676171290196,
      "grad_norm": 1.1236663794989e-06,
      "learning_rate": 6.548048597073413e-06,
      "loss": 0.0,
      "step": 50060
    },
    {
      "epoch": 0.34543660786356395,
      "grad_norm": 9.154676461342603e-10,
      "learning_rate": 6.54735868972797e-06,
      "loss": 0.002,
      "step": 50070
    },
    {
      "epoch": 0.34550559859810825,
      "grad_norm": 0.0,
      "learning_rate": 6.546668782382527e-06,
      "loss": 0.0,
      "step": 50080
    },
    {
      "epoch": 0.3455745893326526,
      "grad_norm": 0.00041500889346934855,
      "learning_rate": 6.545978875037083e-06,
      "loss": 0.0,
      "step": 50090
    },
    {
      "epoch": 0.345643580067197,
      "grad_norm": 0.0,
      "learning_rate": 6.54528896769164e-06,
      "loss": 0.0003,
      "step": 50100
    },
    {
      "epoch": 0.34571257080174134,
      "grad_norm": 0.0,
      "learning_rate": 6.544599060346196e-06,
      "loss": 0.0,
      "step": 50110
    },
    {
      "epoch": 0.3457815615362857,
      "grad_norm": 0.0,
      "learning_rate": 6.543909153000753e-06,
      "loss": 0.0,
      "step": 50120
    },
    {
      "epoch": 0.34585055227083,
      "grad_norm": 0.0009828883921727538,
      "learning_rate": 6.543219245655309e-06,
      "loss": 0.0,
      "step": 50130
    },
    {
      "epoch": 0.34591954300537436,
      "grad_norm": 0.0,
      "learning_rate": 6.542529338309866e-06,
      "loss": 0.0001,
      "step": 50140
    },
    {
      "epoch": 0.3459885337399187,
      "grad_norm": 0.0,
      "learning_rate": 6.541839430964421e-06,
      "loss": 0.0014,
      "step": 50150
    },
    {
      "epoch": 0.3460575244744631,
      "grad_norm": 0.0,
      "learning_rate": 6.541149523618978e-06,
      "loss": 0.0,
      "step": 50160
    },
    {
      "epoch": 0.34612651520900745,
      "grad_norm": 0.0,
      "learning_rate": 6.540459616273534e-06,
      "loss": 0.0,
      "step": 50170
    },
    {
      "epoch": 0.3461955059435518,
      "grad_norm": 0.0,
      "learning_rate": 6.539769708928092e-06,
      "loss": 0.0021,
      "step": 50180
    },
    {
      "epoch": 0.3462644966780961,
      "grad_norm": 8.57240763707523e-07,
      "learning_rate": 6.539079801582648e-06,
      "loss": 0.0,
      "step": 50190
    },
    {
      "epoch": 0.3463334874126405,
      "grad_norm": 0.0,
      "learning_rate": 6.538389894237205e-06,
      "loss": 0.0,
      "step": 50200
    },
    {
      "epoch": 0.34640247814718483,
      "grad_norm": 0.0,
      "learning_rate": 6.537699986891761e-06,
      "loss": 0.0,
      "step": 50210
    },
    {
      "epoch": 0.3464714688817292,
      "grad_norm": 0.0,
      "learning_rate": 6.537010079546318e-06,
      "loss": 0.0,
      "step": 50220
    },
    {
      "epoch": 0.34654045961627356,
      "grad_norm": 0.0,
      "learning_rate": 6.536320172200874e-06,
      "loss": 0.0,
      "step": 50230
    },
    {
      "epoch": 0.34660945035081786,
      "grad_norm": 0.0,
      "learning_rate": 6.535630264855431e-06,
      "loss": 0.0,
      "step": 50240
    },
    {
      "epoch": 0.3466784410853622,
      "grad_norm": 0.0,
      "learning_rate": 6.534940357509987e-06,
      "loss": 0.0,
      "step": 50250
    },
    {
      "epoch": 0.3467474318199066,
      "grad_norm": 0.0,
      "learning_rate": 6.534250450164543e-06,
      "loss": 0.0,
      "step": 50260
    },
    {
      "epoch": 0.34681642255445094,
      "grad_norm": 9.376717567443848,
      "learning_rate": 6.533560542819099e-06,
      "loss": 0.0013,
      "step": 50270
    },
    {
      "epoch": 0.3468854132889953,
      "grad_norm": 0.03148139640688896,
      "learning_rate": 6.532870635473656e-06,
      "loss": 0.0,
      "step": 50280
    },
    {
      "epoch": 0.34695440402353966,
      "grad_norm": 0.0,
      "learning_rate": 6.532180728128213e-06,
      "loss": 0.0002,
      "step": 50290
    },
    {
      "epoch": 0.34702339475808397,
      "grad_norm": 0.612483024597168,
      "learning_rate": 6.53149082078277e-06,
      "loss": 0.0001,
      "step": 50300
    },
    {
      "epoch": 0.34709238549262833,
      "grad_norm": 0.0,
      "learning_rate": 6.530800913437326e-06,
      "loss": 0.0,
      "step": 50310
    },
    {
      "epoch": 0.3471613762271727,
      "grad_norm": 0.0,
      "learning_rate": 6.530111006091883e-06,
      "loss": 0.0,
      "step": 50320
    },
    {
      "epoch": 0.34723036696171705,
      "grad_norm": 0.0,
      "learning_rate": 6.529421098746439e-06,
      "loss": 0.0,
      "step": 50330
    },
    {
      "epoch": 0.3472993576962614,
      "grad_norm": 4.289747721486492e-06,
      "learning_rate": 6.528731191400996e-06,
      "loss": 0.0001,
      "step": 50340
    },
    {
      "epoch": 0.3473683484308057,
      "grad_norm": 0.0,
      "learning_rate": 6.528041284055552e-06,
      "loss": 0.0005,
      "step": 50350
    },
    {
      "epoch": 0.3474373391653501,
      "grad_norm": 0.0,
      "learning_rate": 6.5273513767101086e-06,
      "loss": 0.0127,
      "step": 50360
    },
    {
      "epoch": 0.34750632989989444,
      "grad_norm": 0.008071240969002247,
      "learning_rate": 6.526661469364664e-06,
      "loss": 0.0,
      "step": 50370
    },
    {
      "epoch": 0.3475753206344388,
      "grad_norm": 68.7182846069336,
      "learning_rate": 6.525971562019221e-06,
      "loss": 0.0135,
      "step": 50380
    },
    {
      "epoch": 0.34764431136898316,
      "grad_norm": 0.0,
      "learning_rate": 6.525281654673777e-06,
      "loss": 0.0002,
      "step": 50390
    },
    {
      "epoch": 0.3477133021035275,
      "grad_norm": 0.6752731800079346,
      "learning_rate": 6.5245917473283345e-06,
      "loss": 0.0001,
      "step": 50400
    },
    {
      "epoch": 0.3477822928380718,
      "grad_norm": 0.0,
      "learning_rate": 6.523901839982891e-06,
      "loss": 0.0,
      "step": 50410
    },
    {
      "epoch": 0.3478512835726162,
      "grad_norm": 0.0,
      "learning_rate": 6.5232119326374475e-06,
      "loss": 0.0007,
      "step": 50420
    },
    {
      "epoch": 0.34792027430716055,
      "grad_norm": 0.0,
      "learning_rate": 6.522522025292004e-06,
      "loss": 0.0,
      "step": 50430
    },
    {
      "epoch": 0.3479892650417049,
      "grad_norm": 0.0,
      "learning_rate": 6.5218321179465605e-06,
      "loss": 0.0,
      "step": 50440
    },
    {
      "epoch": 0.34805825577624927,
      "grad_norm": 0.0,
      "learning_rate": 6.521142210601117e-06,
      "loss": 0.0,
      "step": 50450
    },
    {
      "epoch": 0.3481272465107936,
      "grad_norm": 0.015025528147816658,
      "learning_rate": 6.5204523032556735e-06,
      "loss": 0.0,
      "step": 50460
    },
    {
      "epoch": 0.34819623724533794,
      "grad_norm": 0.0,
      "learning_rate": 6.51976239591023e-06,
      "loss": 0.0,
      "step": 50470
    },
    {
      "epoch": 0.3482652279798823,
      "grad_norm": 0.0,
      "learning_rate": 6.5190724885647864e-06,
      "loss": 0.0,
      "step": 50480
    },
    {
      "epoch": 0.34833421871442666,
      "grad_norm": 0.0,
      "learning_rate": 6.518382581219342e-06,
      "loss": 0.0,
      "step": 50490
    },
    {
      "epoch": 0.348403209448971,
      "grad_norm": 0.0,
      "learning_rate": 6.517692673873899e-06,
      "loss": 0.0,
      "step": 50500
    },
    {
      "epoch": 0.3484722001835154,
      "grad_norm": 0.29596391320228577,
      "learning_rate": 6.517002766528456e-06,
      "loss": 0.0,
      "step": 50510
    },
    {
      "epoch": 0.3485411909180597,
      "grad_norm": 0.0,
      "learning_rate": 6.516312859183012e-06,
      "loss": 0.0,
      "step": 50520
    },
    {
      "epoch": 0.34861018165260405,
      "grad_norm": 0.0,
      "learning_rate": 6.515622951837569e-06,
      "loss": 0.0,
      "step": 50530
    },
    {
      "epoch": 0.3486791723871484,
      "grad_norm": 0.0,
      "learning_rate": 6.514933044492125e-06,
      "loss": 0.0,
      "step": 50540
    },
    {
      "epoch": 0.34874816312169277,
      "grad_norm": 0.0016665668226778507,
      "learning_rate": 6.514243137146682e-06,
      "loss": 0.0,
      "step": 50550
    },
    {
      "epoch": 0.3488171538562371,
      "grad_norm": 0.0,
      "learning_rate": 6.513553229801238e-06,
      "loss": 0.0043,
      "step": 50560
    },
    {
      "epoch": 0.34888614459078143,
      "grad_norm": 0.0,
      "learning_rate": 6.512863322455795e-06,
      "loss": 0.0103,
      "step": 50570
    },
    {
      "epoch": 0.3489551353253258,
      "grad_norm": 0.0,
      "learning_rate": 6.512173415110351e-06,
      "loss": 0.0,
      "step": 50580
    },
    {
      "epoch": 0.34902412605987015,
      "grad_norm": 0.0,
      "learning_rate": 6.511483507764908e-06,
      "loss": 0.0,
      "step": 50590
    },
    {
      "epoch": 0.3490931167944145,
      "grad_norm": 0.0,
      "learning_rate": 6.5107936004194635e-06,
      "loss": 0.0,
      "step": 50600
    },
    {
      "epoch": 0.3491621075289589,
      "grad_norm": 0.0,
      "learning_rate": 6.51010369307402e-06,
      "loss": 0.0,
      "step": 50610
    },
    {
      "epoch": 0.34923109826350324,
      "grad_norm": 0.0,
      "learning_rate": 6.509413785728578e-06,
      "loss": 0.0,
      "step": 50620
    },
    {
      "epoch": 0.34930008899804754,
      "grad_norm": 9.754700380781856e-10,
      "learning_rate": 6.508723878383134e-06,
      "loss": 0.0,
      "step": 50630
    },
    {
      "epoch": 0.3493690797325919,
      "grad_norm": 0.0,
      "learning_rate": 6.50803397103769e-06,
      "loss": 0.0,
      "step": 50640
    },
    {
      "epoch": 0.34943807046713626,
      "grad_norm": 9.152853475136169e-10,
      "learning_rate": 6.507344063692247e-06,
      "loss": 0.0,
      "step": 50650
    },
    {
      "epoch": 0.3495070612016806,
      "grad_norm": 0.0,
      "learning_rate": 6.506654156346803e-06,
      "loss": 0.0004,
      "step": 50660
    },
    {
      "epoch": 0.349576051936225,
      "grad_norm": 0.0,
      "learning_rate": 6.50596424900136e-06,
      "loss": 0.0,
      "step": 50670
    },
    {
      "epoch": 0.3496450426707693,
      "grad_norm": 0.0,
      "learning_rate": 6.505274341655916e-06,
      "loss": 0.0,
      "step": 50680
    },
    {
      "epoch": 0.34971403340531365,
      "grad_norm": 0.0,
      "learning_rate": 6.504584434310473e-06,
      "loss": 0.3583,
      "step": 50690
    },
    {
      "epoch": 0.349783024139858,
      "grad_norm": 0.0,
      "learning_rate": 6.503894526965029e-06,
      "loss": 0.0,
      "step": 50700
    },
    {
      "epoch": 0.3498520148744024,
      "grad_norm": 0.0,
      "learning_rate": 6.503204619619585e-06,
      "loss": 0.0001,
      "step": 50710
    },
    {
      "epoch": 0.34992100560894673,
      "grad_norm": 0.0,
      "learning_rate": 6.502514712274141e-06,
      "loss": 0.0,
      "step": 50720
    },
    {
      "epoch": 0.3499899963434911,
      "grad_norm": 0.0,
      "learning_rate": 6.501824804928698e-06,
      "loss": 0.0,
      "step": 50730
    },
    {
      "epoch": 0.3500589870780354,
      "grad_norm": 0.0,
      "learning_rate": 6.501134897583255e-06,
      "loss": 0.0,
      "step": 50740
    },
    {
      "epoch": 0.35012797781257976,
      "grad_norm": 0.0,
      "learning_rate": 6.500444990237812e-06,
      "loss": 0.0,
      "step": 50750
    },
    {
      "epoch": 0.3501969685471241,
      "grad_norm": 0.0,
      "learning_rate": 6.499755082892368e-06,
      "loss": 0.0,
      "step": 50760
    },
    {
      "epoch": 0.3502659592816685,
      "grad_norm": 8.00547277322039e-05,
      "learning_rate": 6.499065175546925e-06,
      "loss": 0.0001,
      "step": 50770
    },
    {
      "epoch": 0.35033495001621284,
      "grad_norm": 0.0025550287682563066,
      "learning_rate": 6.498375268201481e-06,
      "loss": 0.0,
      "step": 50780
    },
    {
      "epoch": 0.35040394075075715,
      "grad_norm": 0.0010697155958041549,
      "learning_rate": 6.497685360856038e-06,
      "loss": 0.0,
      "step": 50790
    },
    {
      "epoch": 0.3504729314853015,
      "grad_norm": 0.0,
      "learning_rate": 6.496995453510594e-06,
      "loss": 0.0,
      "step": 50800
    },
    {
      "epoch": 0.35054192221984587,
      "grad_norm": 0.0,
      "learning_rate": 6.496305546165151e-06,
      "loss": 0.0,
      "step": 50810
    },
    {
      "epoch": 0.35061091295439023,
      "grad_norm": 3.235627900721738e-06,
      "learning_rate": 6.495615638819706e-06,
      "loss": 0.0,
      "step": 50820
    },
    {
      "epoch": 0.3506799036889346,
      "grad_norm": 0.0,
      "learning_rate": 6.494925731474263e-06,
      "loss": 0.0,
      "step": 50830
    },
    {
      "epoch": 0.35074889442347895,
      "grad_norm": 0.0,
      "learning_rate": 6.494235824128819e-06,
      "loss": 0.0,
      "step": 50840
    },
    {
      "epoch": 0.35081788515802326,
      "grad_norm": 0.0,
      "learning_rate": 6.493545916783377e-06,
      "loss": 0.0,
      "step": 50850
    },
    {
      "epoch": 0.3508868758925676,
      "grad_norm": 0.0,
      "learning_rate": 6.492856009437933e-06,
      "loss": 0.0044,
      "step": 50860
    },
    {
      "epoch": 0.350955866627112,
      "grad_norm": 0.0,
      "learning_rate": 6.4921661020924896e-06,
      "loss": 0.0,
      "step": 50870
    },
    {
      "epoch": 0.35102485736165634,
      "grad_norm": 0.0,
      "learning_rate": 6.491476194747046e-06,
      "loss": 0.0,
      "step": 50880
    },
    {
      "epoch": 0.3510938480962007,
      "grad_norm": 0.0,
      "learning_rate": 6.4907862874016026e-06,
      "loss": 0.0002,
      "step": 50890
    },
    {
      "epoch": 0.351162838830745,
      "grad_norm": 0.035069920122623444,
      "learning_rate": 6.490096380056159e-06,
      "loss": 0.0,
      "step": 50900
    },
    {
      "epoch": 0.35123182956528937,
      "grad_norm": 0.0,
      "learning_rate": 6.4894064727107155e-06,
      "loss": 0.0,
      "step": 50910
    },
    {
      "epoch": 0.3513008202998337,
      "grad_norm": 0.0,
      "learning_rate": 6.488716565365272e-06,
      "loss": 0.0,
      "step": 50920
    },
    {
      "epoch": 0.3513698110343781,
      "grad_norm": 0.0,
      "learning_rate": 6.4880266580198285e-06,
      "loss": 0.0564,
      "step": 50930
    },
    {
      "epoch": 0.35143880176892245,
      "grad_norm": 0.0,
      "learning_rate": 6.487336750674384e-06,
      "loss": 0.0,
      "step": 50940
    },
    {
      "epoch": 0.3515077925034668,
      "grad_norm": 0.0,
      "learning_rate": 6.486646843328941e-06,
      "loss": 0.0339,
      "step": 50950
    },
    {
      "epoch": 0.3515767832380111,
      "grad_norm": 1.8043915250842701e-09,
      "learning_rate": 6.485956935983498e-06,
      "loss": 0.0,
      "step": 50960
    },
    {
      "epoch": 0.3516457739725555,
      "grad_norm": 0.0,
      "learning_rate": 6.4852670286380545e-06,
      "loss": 0.0,
      "step": 50970
    },
    {
      "epoch": 0.35171476470709984,
      "grad_norm": 0.5766947269439697,
      "learning_rate": 6.484577121292611e-06,
      "loss": 0.0001,
      "step": 50980
    },
    {
      "epoch": 0.3517837554416442,
      "grad_norm": 0.0,
      "learning_rate": 6.4838872139471675e-06,
      "loss": 0.0,
      "step": 50990
    },
    {
      "epoch": 0.35185274617618856,
      "grad_norm": 1.9761574776566704e-07,
      "learning_rate": 6.483197306601724e-06,
      "loss": 0.0,
      "step": 51000
    },
    {
      "epoch": 0.35192173691073286,
      "grad_norm": 0.0,
      "learning_rate": 6.4825073992562804e-06,
      "loss": 0.0,
      "step": 51010
    },
    {
      "epoch": 0.3519907276452772,
      "grad_norm": 0.0,
      "learning_rate": 6.481817491910837e-06,
      "loss": 0.0,
      "step": 51020
    },
    {
      "epoch": 0.3520597183798216,
      "grad_norm": 0.0,
      "learning_rate": 6.481127584565393e-06,
      "loss": 0.0,
      "step": 51030
    },
    {
      "epoch": 0.35212870911436595,
      "grad_norm": 836.3893432617188,
      "learning_rate": 6.48043767721995e-06,
      "loss": 0.4836,
      "step": 51040
    },
    {
      "epoch": 0.3521976998489103,
      "grad_norm": 0.0,
      "learning_rate": 6.4797477698745056e-06,
      "loss": 0.0,
      "step": 51050
    },
    {
      "epoch": 0.35226669058345467,
      "grad_norm": 0.0009629593114368618,
      "learning_rate": 6.479057862529062e-06,
      "loss": 0.0002,
      "step": 51060
    },
    {
      "epoch": 0.35233568131799897,
      "grad_norm": 0.0,
      "learning_rate": 6.47836795518362e-06,
      "loss": 0.0,
      "step": 51070
    },
    {
      "epoch": 0.35240467205254333,
      "grad_norm": 0.0,
      "learning_rate": 6.477678047838176e-06,
      "loss": 0.0,
      "step": 51080
    },
    {
      "epoch": 0.3524736627870877,
      "grad_norm": 0.0,
      "learning_rate": 6.476988140492732e-06,
      "loss": 0.0,
      "step": 51090
    },
    {
      "epoch": 0.35254265352163205,
      "grad_norm": 0.0,
      "learning_rate": 6.476298233147289e-06,
      "loss": 0.7922,
      "step": 51100
    },
    {
      "epoch": 0.3526116442561764,
      "grad_norm": 6.40731886960566e-05,
      "learning_rate": 6.475608325801845e-06,
      "loss": 0.0032,
      "step": 51110
    },
    {
      "epoch": 0.3526806349907207,
      "grad_norm": 9.686899829830509e-06,
      "learning_rate": 6.474918418456402e-06,
      "loss": 0.0,
      "step": 51120
    },
    {
      "epoch": 0.3527496257252651,
      "grad_norm": 0.0,
      "learning_rate": 6.474228511110958e-06,
      "loss": 0.0,
      "step": 51130
    },
    {
      "epoch": 0.35281861645980944,
      "grad_norm": 0.0,
      "learning_rate": 6.473538603765515e-06,
      "loss": 0.0,
      "step": 51140
    },
    {
      "epoch": 0.3528876071943538,
      "grad_norm": 0.0006227701669558883,
      "learning_rate": 6.472848696420071e-06,
      "loss": 0.0,
      "step": 51150
    },
    {
      "epoch": 0.35295659792889816,
      "grad_norm": 0.0,
      "learning_rate": 6.472158789074627e-06,
      "loss": 0.0002,
      "step": 51160
    },
    {
      "epoch": 0.3530255886634425,
      "grad_norm": 0.0,
      "learning_rate": 6.4714688817291834e-06,
      "loss": 0.0003,
      "step": 51170
    },
    {
      "epoch": 0.35309457939798683,
      "grad_norm": 2.170688458136283e-05,
      "learning_rate": 6.470778974383742e-06,
      "loss": 0.0,
      "step": 51180
    },
    {
      "epoch": 0.3531635701325312,
      "grad_norm": 6.692272958019885e-09,
      "learning_rate": 6.470089067038297e-06,
      "loss": 0.0,
      "step": 51190
    },
    {
      "epoch": 0.35323256086707555,
      "grad_norm": 0.0,
      "learning_rate": 6.469399159692854e-06,
      "loss": 0.0,
      "step": 51200
    },
    {
      "epoch": 0.3533015516016199,
      "grad_norm": 3.0576808285331936e-07,
      "learning_rate": 6.46870925234741e-06,
      "loss": 0.0001,
      "step": 51210
    },
    {
      "epoch": 0.3533705423361643,
      "grad_norm": 0.0,
      "learning_rate": 6.468019345001967e-06,
      "loss": 0.0003,
      "step": 51220
    },
    {
      "epoch": 0.3534395330707086,
      "grad_norm": 4.628592265021325e-08,
      "learning_rate": 6.467329437656523e-06,
      "loss": 0.0,
      "step": 51230
    },
    {
      "epoch": 0.35350852380525294,
      "grad_norm": 0.0,
      "learning_rate": 6.46663953031108e-06,
      "loss": 0.0,
      "step": 51240
    },
    {
      "epoch": 0.3535775145397973,
      "grad_norm": 0.0,
      "learning_rate": 6.465949622965636e-06,
      "loss": 0.0647,
      "step": 51250
    },
    {
      "epoch": 0.35364650527434166,
      "grad_norm": 868.2990112304688,
      "learning_rate": 6.465259715620193e-06,
      "loss": 0.2975,
      "step": 51260
    },
    {
      "epoch": 0.353715496008886,
      "grad_norm": 0.0,
      "learning_rate": 6.464638799009294e-06,
      "loss": 0.7109,
      "step": 51270
    },
    {
      "epoch": 0.3537844867434304,
      "grad_norm": 0.0,
      "learning_rate": 6.463948891663851e-06,
      "loss": 0.0,
      "step": 51280
    },
    {
      "epoch": 0.3538534774779747,
      "grad_norm": 0.0,
      "learning_rate": 6.463258984318407e-06,
      "loss": 0.0,
      "step": 51290
    },
    {
      "epoch": 0.35392246821251905,
      "grad_norm": 0.0,
      "learning_rate": 6.462569076972963e-06,
      "loss": 0.0001,
      "step": 51300
    },
    {
      "epoch": 0.3539914589470634,
      "grad_norm": 5.896670245419955e-07,
      "learning_rate": 6.461879169627519e-06,
      "loss": 0.2535,
      "step": 51310
    },
    {
      "epoch": 0.35406044968160777,
      "grad_norm": 0.0,
      "learning_rate": 6.461189262282076e-06,
      "loss": 0.0,
      "step": 51320
    },
    {
      "epoch": 0.35412944041615213,
      "grad_norm": 0.0,
      "learning_rate": 6.460499354936632e-06,
      "loss": 0.0,
      "step": 51330
    },
    {
      "epoch": 0.35419843115069644,
      "grad_norm": 0.0,
      "learning_rate": 6.459809447591189e-06,
      "loss": 0.0009,
      "step": 51340
    },
    {
      "epoch": 0.3542674218852408,
      "grad_norm": 0.004679547622799873,
      "learning_rate": 6.459119540245745e-06,
      "loss": 0.1567,
      "step": 51350
    },
    {
      "epoch": 0.35433641261978516,
      "grad_norm": 0.0,
      "learning_rate": 6.458429632900302e-06,
      "loss": 0.0,
      "step": 51360
    },
    {
      "epoch": 0.3544054033543295,
      "grad_norm": 0.0,
      "learning_rate": 6.457739725554859e-06,
      "loss": 0.0028,
      "step": 51370
    },
    {
      "epoch": 0.3544743940888739,
      "grad_norm": 0.0,
      "learning_rate": 6.457049818209416e-06,
      "loss": 0.0,
      "step": 51380
    },
    {
      "epoch": 0.35454338482341824,
      "grad_norm": 0.0,
      "learning_rate": 6.456359910863972e-06,
      "loss": 0.0,
      "step": 51390
    },
    {
      "epoch": 0.35461237555796254,
      "grad_norm": 0.0,
      "learning_rate": 6.455670003518529e-06,
      "loss": 0.0,
      "step": 51400
    },
    {
      "epoch": 0.3546813662925069,
      "grad_norm": 0.04218251258134842,
      "learning_rate": 6.454980096173084e-06,
      "loss": 0.0,
      "step": 51410
    },
    {
      "epoch": 0.35475035702705127,
      "grad_norm": 0.004686353728175163,
      "learning_rate": 6.454290188827641e-06,
      "loss": 0.0,
      "step": 51420
    },
    {
      "epoch": 0.3548193477615956,
      "grad_norm": 0.0,
      "learning_rate": 6.453600281482197e-06,
      "loss": 0.0011,
      "step": 51430
    },
    {
      "epoch": 0.35488833849614,
      "grad_norm": 0.0,
      "learning_rate": 6.452910374136754e-06,
      "loss": 0.0001,
      "step": 51440
    },
    {
      "epoch": 0.3549573292306843,
      "grad_norm": 0.0,
      "learning_rate": 6.45222046679131e-06,
      "loss": 0.0,
      "step": 51450
    },
    {
      "epoch": 0.35502631996522865,
      "grad_norm": 0.0,
      "learning_rate": 6.451530559445867e-06,
      "loss": 0.0,
      "step": 51460
    },
    {
      "epoch": 0.355095310699773,
      "grad_norm": 0.0,
      "learning_rate": 6.450840652100423e-06,
      "loss": 0.0,
      "step": 51470
    },
    {
      "epoch": 0.3551643014343174,
      "grad_norm": 0.0002492847852408886,
      "learning_rate": 6.4501507447549805e-06,
      "loss": 0.0,
      "step": 51480
    },
    {
      "epoch": 0.35523329216886174,
      "grad_norm": 0.0,
      "learning_rate": 6.449460837409537e-06,
      "loss": 0.0,
      "step": 51490
    },
    {
      "epoch": 0.3553022829034061,
      "grad_norm": 3.526331738612498e-06,
      "learning_rate": 6.4487709300640935e-06,
      "loss": 0.0,
      "step": 51500
    },
    {
      "epoch": 0.3553712736379504,
      "grad_norm": 0.0,
      "learning_rate": 6.44808102271865e-06,
      "loss": 0.0006,
      "step": 51510
    },
    {
      "epoch": 0.35544026437249476,
      "grad_norm": 0.0,
      "learning_rate": 6.4473911153732065e-06,
      "loss": 0.0,
      "step": 51520
    },
    {
      "epoch": 0.3555092551070391,
      "grad_norm": 0.0,
      "learning_rate": 6.446701208027762e-06,
      "loss": 0.0,
      "step": 51530
    },
    {
      "epoch": 0.3555782458415835,
      "grad_norm": 0.0,
      "learning_rate": 6.446011300682319e-06,
      "loss": 0.0,
      "step": 51540
    },
    {
      "epoch": 0.35564723657612785,
      "grad_norm": 0.0,
      "learning_rate": 6.445321393336875e-06,
      "loss": 0.0,
      "step": 51550
    },
    {
      "epoch": 0.35571622731067215,
      "grad_norm": 1.6291174631533067e-07,
      "learning_rate": 6.444631485991432e-06,
      "loss": 0.0,
      "step": 51560
    },
    {
      "epoch": 0.3557852180452165,
      "grad_norm": 0.0,
      "learning_rate": 6.443941578645988e-06,
      "loss": 0.0,
      "step": 51570
    },
    {
      "epoch": 0.35585420877976087,
      "grad_norm": 0.0,
      "learning_rate": 6.4432516713005446e-06,
      "loss": 0.0,
      "step": 51580
    },
    {
      "epoch": 0.35592319951430523,
      "grad_norm": 0.0,
      "learning_rate": 6.442561763955102e-06,
      "loss": 0.0,
      "step": 51590
    },
    {
      "epoch": 0.3559921902488496,
      "grad_norm": 0.0,
      "learning_rate": 6.441871856609658e-06,
      "loss": 0.0111,
      "step": 51600
    },
    {
      "epoch": 0.35606118098339395,
      "grad_norm": 0.0,
      "learning_rate": 6.441181949264215e-06,
      "loss": 0.0,
      "step": 51610
    },
    {
      "epoch": 0.35613017171793826,
      "grad_norm": 0.0,
      "learning_rate": 6.440492041918771e-06,
      "loss": 0.0,
      "step": 51620
    },
    {
      "epoch": 0.3561991624524826,
      "grad_norm": 0.0,
      "learning_rate": 6.439802134573328e-06,
      "loss": 0.0,
      "step": 51630
    },
    {
      "epoch": 0.356268153187027,
      "grad_norm": 0.007423896808177233,
      "learning_rate": 6.4391122272278835e-06,
      "loss": 0.0,
      "step": 51640
    },
    {
      "epoch": 0.35633714392157134,
      "grad_norm": 4.474450143732156e-09,
      "learning_rate": 6.43842231988244e-06,
      "loss": 0.0,
      "step": 51650
    },
    {
      "epoch": 0.3564061346561157,
      "grad_norm": 0.0,
      "learning_rate": 6.4377324125369965e-06,
      "loss": 0.0,
      "step": 51660
    },
    {
      "epoch": 0.35647512539066,
      "grad_norm": 0.0,
      "learning_rate": 6.437042505191553e-06,
      "loss": 0.0,
      "step": 51670
    },
    {
      "epoch": 0.35654411612520437,
      "grad_norm": 0.28651612997055054,
      "learning_rate": 6.4363525978461095e-06,
      "loss": 0.0001,
      "step": 51680
    },
    {
      "epoch": 0.35661310685974873,
      "grad_norm": 0.0,
      "learning_rate": 6.435662690500666e-06,
      "loss": 0.0,
      "step": 51690
    },
    {
      "epoch": 0.3566820975942931,
      "grad_norm": 0.0,
      "learning_rate": 6.434972783155223e-06,
      "loss": 0.0,
      "step": 51700
    },
    {
      "epoch": 0.35675108832883745,
      "grad_norm": 0.0,
      "learning_rate": 6.43428287580978e-06,
      "loss": 0.0,
      "step": 51710
    },
    {
      "epoch": 0.3568200790633818,
      "grad_norm": 0.0,
      "learning_rate": 6.433592968464336e-06,
      "loss": 0.0,
      "step": 51720
    },
    {
      "epoch": 0.3568890697979261,
      "grad_norm": 0.32183560729026794,
      "learning_rate": 6.432903061118893e-06,
      "loss": 0.0001,
      "step": 51730
    },
    {
      "epoch": 0.3569580605324705,
      "grad_norm": 0.0,
      "learning_rate": 6.432213153773449e-06,
      "loss": 0.0,
      "step": 51740
    },
    {
      "epoch": 0.35702705126701484,
      "grad_norm": 8.300338749656078e-10,
      "learning_rate": 6.431523246428005e-06,
      "loss": 0.0,
      "step": 51750
    },
    {
      "epoch": 0.3570960420015592,
      "grad_norm": 0.0,
      "learning_rate": 6.430833339082561e-06,
      "loss": 0.0,
      "step": 51760
    },
    {
      "epoch": 0.35716503273610356,
      "grad_norm": 0.0,
      "learning_rate": 6.430143431737118e-06,
      "loss": 0.0,
      "step": 51770
    },
    {
      "epoch": 0.35723402347064787,
      "grad_norm": 0.000443589553469792,
      "learning_rate": 6.429453524391674e-06,
      "loss": 0.0,
      "step": 51780
    },
    {
      "epoch": 0.3573030142051922,
      "grad_norm": 0.00010486302926437929,
      "learning_rate": 6.428763617046231e-06,
      "loss": 0.0001,
      "step": 51790
    },
    {
      "epoch": 0.3573720049397366,
      "grad_norm": 0.0,
      "learning_rate": 6.428073709700787e-06,
      "loss": 0.0,
      "step": 51800
    },
    {
      "epoch": 0.35744099567428095,
      "grad_norm": 0.0,
      "learning_rate": 6.427383802355344e-06,
      "loss": 0.4902,
      "step": 51810
    },
    {
      "epoch": 0.3575099864088253,
      "grad_norm": 0.0,
      "learning_rate": 6.426693895009901e-06,
      "loss": 0.0,
      "step": 51820
    },
    {
      "epoch": 0.35757897714336967,
      "grad_norm": 0.0,
      "learning_rate": 6.426003987664458e-06,
      "loss": 0.0,
      "step": 51830
    },
    {
      "epoch": 0.357647967877914,
      "grad_norm": 0.004734083544462919,
      "learning_rate": 6.425314080319014e-06,
      "loss": 0.0,
      "step": 51840
    },
    {
      "epoch": 0.35771695861245834,
      "grad_norm": 0.0,
      "learning_rate": 6.424624172973571e-06,
      "loss": 0.0,
      "step": 51850
    },
    {
      "epoch": 0.3577859493470027,
      "grad_norm": 0.0,
      "learning_rate": 6.423934265628126e-06,
      "loss": 0.0,
      "step": 51860
    },
    {
      "epoch": 0.35785494008154706,
      "grad_norm": 0.0,
      "learning_rate": 6.423244358282683e-06,
      "loss": 0.0,
      "step": 51870
    },
    {
      "epoch": 0.3579239308160914,
      "grad_norm": 0.0,
      "learning_rate": 6.422554450937239e-06,
      "loss": 0.0,
      "step": 51880
    },
    {
      "epoch": 0.3579929215506357,
      "grad_norm": 0.0,
      "learning_rate": 6.421864543591796e-06,
      "loss": 0.0,
      "step": 51890
    },
    {
      "epoch": 0.3580619122851801,
      "grad_norm": 0.0,
      "learning_rate": 6.421174636246352e-06,
      "loss": 0.0,
      "step": 51900
    },
    {
      "epoch": 0.35813090301972444,
      "grad_norm": 0.0,
      "learning_rate": 6.420484728900909e-06,
      "loss": 0.0,
      "step": 51910
    },
    {
      "epoch": 0.3581998937542688,
      "grad_norm": 0.0,
      "learning_rate": 6.419794821555465e-06,
      "loss": 0.0,
      "step": 51920
    },
    {
      "epoch": 0.35826888448881317,
      "grad_norm": 9.824264735058819e-10,
      "learning_rate": 6.419104914210023e-06,
      "loss": 0.0001,
      "step": 51930
    },
    {
      "epoch": 0.3583378752233575,
      "grad_norm": 1.946612684378124e-07,
      "learning_rate": 6.418415006864579e-06,
      "loss": 0.0,
      "step": 51940
    },
    {
      "epoch": 0.35840686595790183,
      "grad_norm": 0.0,
      "learning_rate": 6.4177250995191356e-06,
      "loss": 0.0,
      "step": 51950
    },
    {
      "epoch": 0.3584758566924462,
      "grad_norm": 0.0,
      "learning_rate": 6.417035192173692e-06,
      "loss": 0.0,
      "step": 51960
    },
    {
      "epoch": 0.35854484742699055,
      "grad_norm": 0.0,
      "learning_rate": 6.4163452848282485e-06,
      "loss": 0.0,
      "step": 51970
    },
    {
      "epoch": 0.3586138381615349,
      "grad_norm": 0.0,
      "learning_rate": 6.415655377482804e-06,
      "loss": 0.0,
      "step": 51980
    },
    {
      "epoch": 0.3586828288960793,
      "grad_norm": 0.0,
      "learning_rate": 6.414965470137361e-06,
      "loss": 0.0,
      "step": 51990
    },
    {
      "epoch": 0.3587518196306236,
      "grad_norm": 3.0257751859608106e-05,
      "learning_rate": 6.414275562791917e-06,
      "loss": 0.0,
      "step": 52000
    },
    {
      "epoch": 0.35882081036516794,
      "grad_norm": 0.0,
      "learning_rate": 6.413585655446474e-06,
      "loss": 0.0,
      "step": 52010
    },
    {
      "epoch": 0.3588898010997123,
      "grad_norm": 0.0,
      "learning_rate": 6.41289574810103e-06,
      "loss": 0.0,
      "step": 52020
    },
    {
      "epoch": 0.35895879183425666,
      "grad_norm": 2.1959505829727277e-05,
      "learning_rate": 6.412205840755587e-06,
      "loss": 0.0,
      "step": 52030
    },
    {
      "epoch": 0.359027782568801,
      "grad_norm": 0.0,
      "learning_rate": 6.411515933410144e-06,
      "loss": 0.0,
      "step": 52040
    },
    {
      "epoch": 0.3590967733033454,
      "grad_norm": 0.0,
      "learning_rate": 6.4108260260647005e-06,
      "loss": 0.0,
      "step": 52050
    },
    {
      "epoch": 0.3591657640378897,
      "grad_norm": 0.0,
      "learning_rate": 6.410136118719257e-06,
      "loss": 0.7734,
      "step": 52060
    },
    {
      "epoch": 0.35923475477243405,
      "grad_norm": 0.0,
      "learning_rate": 6.4094462113738134e-06,
      "loss": 0.1903,
      "step": 52070
    },
    {
      "epoch": 0.3593037455069784,
      "grad_norm": 0.0008258084417320788,
      "learning_rate": 6.40875630402837e-06,
      "loss": 0.0,
      "step": 52080
    },
    {
      "epoch": 0.35937273624152277,
      "grad_norm": 0.0,
      "learning_rate": 6.408066396682926e-06,
      "loss": 0.0,
      "step": 52090
    },
    {
      "epoch": 0.35944172697606713,
      "grad_norm": 0.0,
      "learning_rate": 6.407376489337482e-06,
      "loss": 0.0,
      "step": 52100
    },
    {
      "epoch": 0.35951071771061144,
      "grad_norm": 0.0,
      "learning_rate": 6.4066865819920386e-06,
      "loss": 0.0,
      "step": 52110
    },
    {
      "epoch": 0.3595797084451558,
      "grad_norm": 9.867168193622433e-10,
      "learning_rate": 6.405996674646595e-06,
      "loss": 0.0,
      "step": 52120
    },
    {
      "epoch": 0.35964869917970016,
      "grad_norm": 0.0,
      "learning_rate": 6.4053067673011515e-06,
      "loss": 0.0001,
      "step": 52130
    },
    {
      "epoch": 0.3597176899142445,
      "grad_norm": 0.0,
      "learning_rate": 6.404616859955708e-06,
      "loss": 0.0003,
      "step": 52140
    },
    {
      "epoch": 0.3597866806487889,
      "grad_norm": 0.0,
      "learning_rate": 6.403926952610265e-06,
      "loss": 0.0,
      "step": 52150
    },
    {
      "epoch": 0.35985567138333324,
      "grad_norm": 0.0,
      "learning_rate": 6.403237045264822e-06,
      "loss": 0.0,
      "step": 52160
    },
    {
      "epoch": 0.35992466211787755,
      "grad_norm": 6.122380000306293e-06,
      "learning_rate": 6.402547137919378e-06,
      "loss": 0.0,
      "step": 52170
    },
    {
      "epoch": 0.3599936528524219,
      "grad_norm": 0.0,
      "learning_rate": 6.401857230573935e-06,
      "loss": 0.0,
      "step": 52180
    },
    {
      "epoch": 0.36006264358696627,
      "grad_norm": 0.0,
      "learning_rate": 6.401167323228491e-06,
      "loss": 0.0,
      "step": 52190
    },
    {
      "epoch": 0.36013163432151063,
      "grad_norm": 0.2074791043996811,
      "learning_rate": 6.400477415883047e-06,
      "loss": 0.0,
      "step": 52200
    },
    {
      "epoch": 0.360200625056055,
      "grad_norm": 0.0,
      "learning_rate": 6.3997875085376035e-06,
      "loss": 0.0,
      "step": 52210
    },
    {
      "epoch": 0.3602696157905993,
      "grad_norm": 2.339615292612507e-07,
      "learning_rate": 6.39909760119216e-06,
      "loss": 0.0,
      "step": 52220
    },
    {
      "epoch": 0.36033860652514366,
      "grad_norm": 1.2783141301042633e-07,
      "learning_rate": 6.3984076938467164e-06,
      "loss": 0.0,
      "step": 52230
    },
    {
      "epoch": 0.360407597259688,
      "grad_norm": 0.0,
      "learning_rate": 6.397717786501273e-06,
      "loss": 0.0001,
      "step": 52240
    },
    {
      "epoch": 0.3604765879942324,
      "grad_norm": 0.0,
      "learning_rate": 6.397027879155829e-06,
      "loss": 0.0,
      "step": 52250
    },
    {
      "epoch": 0.36054557872877674,
      "grad_norm": 0.0,
      "learning_rate": 6.396337971810387e-06,
      "loss": 0.0,
      "step": 52260
    },
    {
      "epoch": 0.3606145694633211,
      "grad_norm": 0.0,
      "learning_rate": 6.395648064464943e-06,
      "loss": 0.0,
      "step": 52270
    },
    {
      "epoch": 0.3606835601978654,
      "grad_norm": 0.0,
      "learning_rate": 6.3949581571195e-06,
      "loss": 0.0001,
      "step": 52280
    },
    {
      "epoch": 0.36075255093240977,
      "grad_norm": 0.0,
      "learning_rate": 6.394268249774056e-06,
      "loss": 0.0,
      "step": 52290
    },
    {
      "epoch": 0.3608215416669541,
      "grad_norm": 0.0,
      "learning_rate": 6.393578342428613e-06,
      "loss": 0.0,
      "step": 52300
    },
    {
      "epoch": 0.3608905324014985,
      "grad_norm": 0.0,
      "learning_rate": 6.392888435083169e-06,
      "loss": 0.0,
      "step": 52310
    },
    {
      "epoch": 0.36095952313604285,
      "grad_norm": 0.0,
      "learning_rate": 6.392198527737725e-06,
      "loss": 0.0,
      "step": 52320
    },
    {
      "epoch": 0.36102851387058715,
      "grad_norm": 0.0,
      "learning_rate": 6.391508620392281e-06,
      "loss": 0.0,
      "step": 52330
    },
    {
      "epoch": 0.3610975046051315,
      "grad_norm": 0.3596288859844208,
      "learning_rate": 6.390818713046838e-06,
      "loss": 0.0022,
      "step": 52340
    },
    {
      "epoch": 0.3611664953396759,
      "grad_norm": 0.0,
      "learning_rate": 6.390128805701394e-06,
      "loss": 0.0,
      "step": 52350
    },
    {
      "epoch": 0.36123548607422024,
      "grad_norm": 0.0,
      "learning_rate": 6.389438898355951e-06,
      "loss": 0.0014,
      "step": 52360
    },
    {
      "epoch": 0.3613044768087646,
      "grad_norm": 0.0,
      "learning_rate": 6.388748991010508e-06,
      "loss": 0.0,
      "step": 52370
    },
    {
      "epoch": 0.36137346754330896,
      "grad_norm": 0.0,
      "learning_rate": 6.388059083665065e-06,
      "loss": 0.0,
      "step": 52380
    },
    {
      "epoch": 0.36144245827785326,
      "grad_norm": 0.0,
      "learning_rate": 6.387369176319621e-06,
      "loss": 0.0001,
      "step": 52390
    },
    {
      "epoch": 0.3615114490123976,
      "grad_norm": 7.475036589710271e-09,
      "learning_rate": 6.386679268974178e-06,
      "loss": 0.0,
      "step": 52400
    },
    {
      "epoch": 0.361580439746942,
      "grad_norm": 0.36978551745414734,
      "learning_rate": 6.385989361628734e-06,
      "loss": 0.0001,
      "step": 52410
    },
    {
      "epoch": 0.36164943048148634,
      "grad_norm": 0.0,
      "learning_rate": 6.385299454283291e-06,
      "loss": 0.0,
      "step": 52420
    },
    {
      "epoch": 0.3617184212160307,
      "grad_norm": 0.0,
      "learning_rate": 6.384609546937846e-06,
      "loss": 0.0,
      "step": 52430
    },
    {
      "epoch": 0.361787411950575,
      "grad_norm": 435.36041259765625,
      "learning_rate": 6.383919639592403e-06,
      "loss": 0.1095,
      "step": 52440
    },
    {
      "epoch": 0.36185640268511937,
      "grad_norm": 0.0,
      "learning_rate": 6.383229732246959e-06,
      "loss": 0.0,
      "step": 52450
    },
    {
      "epoch": 0.36192539341966373,
      "grad_norm": 0.0,
      "learning_rate": 6.382539824901516e-06,
      "loss": 0.0009,
      "step": 52460
    },
    {
      "epoch": 0.3619943841542081,
      "grad_norm": 0.0,
      "learning_rate": 6.381849917556072e-06,
      "loss": 0.0001,
      "step": 52470
    },
    {
      "epoch": 0.36206337488875245,
      "grad_norm": 4.5022838435215817e-07,
      "learning_rate": 6.3811600102106295e-06,
      "loss": 0.0001,
      "step": 52480
    },
    {
      "epoch": 0.3621323656232968,
      "grad_norm": 2.1329526134650223e-05,
      "learning_rate": 6.380470102865186e-06,
      "loss": 0.0,
      "step": 52490
    },
    {
      "epoch": 0.3622013563578411,
      "grad_norm": 2.401577035016089e-07,
      "learning_rate": 6.3797801955197425e-06,
      "loss": 0.0,
      "step": 52500
    },
    {
      "epoch": 0.3622703470923855,
      "grad_norm": 1.8900733493865118e-06,
      "learning_rate": 6.379090288174299e-06,
      "loss": 0.0,
      "step": 52510
    },
    {
      "epoch": 0.36233933782692984,
      "grad_norm": 0.0,
      "learning_rate": 6.3784003808288555e-06,
      "loss": 0.0,
      "step": 52520
    },
    {
      "epoch": 0.3624083285614742,
      "grad_norm": 5.384059136304131e-07,
      "learning_rate": 6.377710473483412e-06,
      "loss": 0.0009,
      "step": 52530
    },
    {
      "epoch": 0.36247731929601856,
      "grad_norm": 0.0,
      "learning_rate": 6.377020566137968e-06,
      "loss": 0.015,
      "step": 52540
    },
    {
      "epoch": 0.36254631003056287,
      "grad_norm": 0.0,
      "learning_rate": 6.376330658792524e-06,
      "loss": 0.0001,
      "step": 52550
    },
    {
      "epoch": 0.36261530076510723,
      "grad_norm": 4.186577129416946e-08,
      "learning_rate": 6.375640751447081e-06,
      "loss": 0.0005,
      "step": 52560
    },
    {
      "epoch": 0.3626842914996516,
      "grad_norm": 0.0,
      "learning_rate": 6.374950844101637e-06,
      "loss": 0.0,
      "step": 52570
    },
    {
      "epoch": 0.36275328223419595,
      "grad_norm": 0.0,
      "learning_rate": 6.374260936756194e-06,
      "loss": 0.0005,
      "step": 52580
    },
    {
      "epoch": 0.3628222729687403,
      "grad_norm": 0.0,
      "learning_rate": 6.373571029410751e-06,
      "loss": 0.0,
      "step": 52590
    },
    {
      "epoch": 0.36289126370328467,
      "grad_norm": 0.0,
      "learning_rate": 6.3728811220653074e-06,
      "loss": 0.0,
      "step": 52600
    },
    {
      "epoch": 0.362960254437829,
      "grad_norm": 0.0,
      "learning_rate": 6.372191214719864e-06,
      "loss": 0.0,
      "step": 52610
    },
    {
      "epoch": 0.36302924517237334,
      "grad_norm": 0.0,
      "learning_rate": 6.37150130737442e-06,
      "loss": 0.0,
      "step": 52620
    },
    {
      "epoch": 0.3630982359069177,
      "grad_norm": 0.0,
      "learning_rate": 6.370811400028977e-06,
      "loss": 0.0,
      "step": 52630
    },
    {
      "epoch": 0.36316722664146206,
      "grad_norm": 1.384516716003418,
      "learning_rate": 6.370121492683533e-06,
      "loss": 0.5147,
      "step": 52640
    },
    {
      "epoch": 0.3632362173760064,
      "grad_norm": 0.0,
      "learning_rate": 6.36943158533809e-06,
      "loss": 0.0,
      "step": 52650
    },
    {
      "epoch": 0.3633052081105507,
      "grad_norm": 0.0,
      "learning_rate": 6.3687416779926455e-06,
      "loss": 0.0,
      "step": 52660
    },
    {
      "epoch": 0.3633741988450951,
      "grad_norm": 0.0,
      "learning_rate": 6.368051770647202e-06,
      "loss": 0.0,
      "step": 52670
    },
    {
      "epoch": 0.36344318957963945,
      "grad_norm": 5.1653405535034835e-06,
      "learning_rate": 6.3673618633017585e-06,
      "loss": 0.0,
      "step": 52680
    },
    {
      "epoch": 0.3635121803141838,
      "grad_norm": 0.0,
      "learning_rate": 6.366671955956315e-06,
      "loss": 0.0,
      "step": 52690
    },
    {
      "epoch": 0.36358117104872817,
      "grad_norm": 0.0,
      "learning_rate": 6.365982048610872e-06,
      "loss": 0.0,
      "step": 52700
    },
    {
      "epoch": 0.36365016178327253,
      "grad_norm": 0.0,
      "learning_rate": 6.365292141265429e-06,
      "loss": 0.0,
      "step": 52710
    },
    {
      "epoch": 0.36371915251781683,
      "grad_norm": 0.0,
      "learning_rate": 6.364602233919985e-06,
      "loss": 0.0,
      "step": 52720
    },
    {
      "epoch": 0.3637881432523612,
      "grad_norm": 0.0,
      "learning_rate": 6.363912326574542e-06,
      "loss": 0.0,
      "step": 52730
    },
    {
      "epoch": 0.36385713398690556,
      "grad_norm": 0.0,
      "learning_rate": 6.363222419229098e-06,
      "loss": 0.0001,
      "step": 52740
    },
    {
      "epoch": 0.3639261247214499,
      "grad_norm": 0.0,
      "learning_rate": 6.362532511883655e-06,
      "loss": 0.0,
      "step": 52750
    },
    {
      "epoch": 0.3639951154559943,
      "grad_norm": 0.0,
      "learning_rate": 6.361842604538211e-06,
      "loss": 0.0,
      "step": 52760
    },
    {
      "epoch": 0.3640641061905386,
      "grad_norm": 0.0,
      "learning_rate": 6.361152697192767e-06,
      "loss": 0.0,
      "step": 52770
    },
    {
      "epoch": 0.36413309692508294,
      "grad_norm": 0.0,
      "learning_rate": 6.360462789847323e-06,
      "loss": 0.0429,
      "step": 52780
    },
    {
      "epoch": 0.3642020876596273,
      "grad_norm": 0.0,
      "learning_rate": 6.35977288250188e-06,
      "loss": 0.0001,
      "step": 52790
    },
    {
      "epoch": 0.36427107839417167,
      "grad_norm": 0.0,
      "learning_rate": 6.359082975156436e-06,
      "loss": 0.0,
      "step": 52800
    },
    {
      "epoch": 0.364340069128716,
      "grad_norm": 1.2245102709584899e-07,
      "learning_rate": 6.358393067810994e-06,
      "loss": 0.0001,
      "step": 52810
    },
    {
      "epoch": 0.3644090598632604,
      "grad_norm": 0.0,
      "learning_rate": 6.35770316046555e-06,
      "loss": 0.0,
      "step": 52820
    },
    {
      "epoch": 0.3644780505978047,
      "grad_norm": 0.0,
      "learning_rate": 6.357013253120107e-06,
      "loss": 0.0,
      "step": 52830
    },
    {
      "epoch": 0.36454704133234905,
      "grad_norm": 0.0,
      "learning_rate": 6.356323345774663e-06,
      "loss": 0.1137,
      "step": 52840
    },
    {
      "epoch": 0.3646160320668934,
      "grad_norm": 0.0,
      "learning_rate": 6.35563343842922e-06,
      "loss": 0.0,
      "step": 52850
    },
    {
      "epoch": 0.3646850228014378,
      "grad_norm": 0.0,
      "learning_rate": 6.354943531083776e-06,
      "loss": 0.0,
      "step": 52860
    },
    {
      "epoch": 0.36475401353598214,
      "grad_norm": 0.0,
      "learning_rate": 6.354253623738333e-06,
      "loss": 0.0,
      "step": 52870
    },
    {
      "epoch": 0.36482300427052644,
      "grad_norm": 0.0,
      "learning_rate": 6.353563716392888e-06,
      "loss": 0.0,
      "step": 52880
    },
    {
      "epoch": 0.3648919950050708,
      "grad_norm": 0.0,
      "learning_rate": 6.352873809047445e-06,
      "loss": 0.0,
      "step": 52890
    },
    {
      "epoch": 0.36496098573961516,
      "grad_norm": 0.0,
      "learning_rate": 6.352183901702001e-06,
      "loss": 0.0,
      "step": 52900
    },
    {
      "epoch": 0.3650299764741595,
      "grad_norm": 0.0004221514973323792,
      "learning_rate": 6.351493994356558e-06,
      "loss": 0.0,
      "step": 52910
    },
    {
      "epoch": 0.3650989672087039,
      "grad_norm": 0.0,
      "learning_rate": 6.350804087011115e-06,
      "loss": 0.0,
      "step": 52920
    },
    {
      "epoch": 0.36516795794324824,
      "grad_norm": 0.0,
      "learning_rate": 6.350114179665672e-06,
      "loss": 0.35,
      "step": 52930
    },
    {
      "epoch": 0.36523694867779255,
      "grad_norm": 0.0,
      "learning_rate": 6.349424272320228e-06,
      "loss": 0.0,
      "step": 52940
    },
    {
      "epoch": 0.3653059394123369,
      "grad_norm": 0.0,
      "learning_rate": 6.348734364974785e-06,
      "loss": 0.2987,
      "step": 52950
    },
    {
      "epoch": 0.36537493014688127,
      "grad_norm": 0.0,
      "learning_rate": 6.348044457629341e-06,
      "loss": 0.0,
      "step": 52960
    },
    {
      "epoch": 0.36544392088142563,
      "grad_norm": 0.0,
      "learning_rate": 6.3473545502838976e-06,
      "loss": 0.0,
      "step": 52970
    },
    {
      "epoch": 0.36551291161597,
      "grad_norm": 0.0,
      "learning_rate": 6.346664642938454e-06,
      "loss": 0.0,
      "step": 52980
    },
    {
      "epoch": 0.3655819023505143,
      "grad_norm": 0.0,
      "learning_rate": 6.3459747355930106e-06,
      "loss": 0.0021,
      "step": 52990
    },
    {
      "epoch": 0.36565089308505866,
      "grad_norm": 0.0,
      "learning_rate": 6.345284828247566e-06,
      "loss": 0.0,
      "step": 53000
    },
    {
      "epoch": 0.365719883819603,
      "grad_norm": 0.0,
      "learning_rate": 6.344594920902123e-06,
      "loss": 0.0,
      "step": 53010
    },
    {
      "epoch": 0.3657888745541474,
      "grad_norm": 1.1275965334789362e-05,
      "learning_rate": 6.343905013556679e-06,
      "loss": 0.0,
      "step": 53020
    },
    {
      "epoch": 0.36585786528869174,
      "grad_norm": 0.0,
      "learning_rate": 6.3432151062112365e-06,
      "loss": 0.0,
      "step": 53030
    },
    {
      "epoch": 0.3659268560232361,
      "grad_norm": 2.156047562706931e-09,
      "learning_rate": 6.342525198865793e-06,
      "loss": 0.0377,
      "step": 53040
    },
    {
      "epoch": 0.3659958467577804,
      "grad_norm": 0.0,
      "learning_rate": 6.3418352915203495e-06,
      "loss": 0.1613,
      "step": 53050
    },
    {
      "epoch": 0.36606483749232477,
      "grad_norm": 0.0,
      "learning_rate": 6.341145384174906e-06,
      "loss": 0.0429,
      "step": 53060
    },
    {
      "epoch": 0.36613382822686913,
      "grad_norm": 0.0,
      "learning_rate": 6.3404554768294625e-06,
      "loss": 0.0,
      "step": 53070
    },
    {
      "epoch": 0.3662028189614135,
      "grad_norm": 0.0,
      "learning_rate": 6.339765569484019e-06,
      "loss": 0.0,
      "step": 53080
    },
    {
      "epoch": 0.36627180969595785,
      "grad_norm": 0.0024852491915225983,
      "learning_rate": 6.3390756621385755e-06,
      "loss": 0.3215,
      "step": 53090
    },
    {
      "epoch": 0.36634080043050216,
      "grad_norm": 0.0,
      "learning_rate": 6.338385754793132e-06,
      "loss": 0.0,
      "step": 53100
    },
    {
      "epoch": 0.3664097911650465,
      "grad_norm": 0.0,
      "learning_rate": 6.337695847447688e-06,
      "loss": 2.1703,
      "step": 53110
    },
    {
      "epoch": 0.3664787818995909,
      "grad_norm": 0.0,
      "learning_rate": 6.337005940102244e-06,
      "loss": 0.0,
      "step": 53120
    },
    {
      "epoch": 0.36654777263413524,
      "grad_norm": 0.0,
      "learning_rate": 6.3363160327568006e-06,
      "loss": 0.0033,
      "step": 53130
    },
    {
      "epoch": 0.3666167633686796,
      "grad_norm": 0.0,
      "learning_rate": 6.335626125411358e-06,
      "loss": 0.0,
      "step": 53140
    },
    {
      "epoch": 0.36668575410322396,
      "grad_norm": 0.0,
      "learning_rate": 6.334936218065914e-06,
      "loss": 0.0,
      "step": 53150
    },
    {
      "epoch": 0.36675474483776827,
      "grad_norm": 1.0293479135725647e-05,
      "learning_rate": 6.334246310720471e-06,
      "loss": 0.0,
      "step": 53160
    },
    {
      "epoch": 0.3668237355723126,
      "grad_norm": 262.7918701171875,
      "learning_rate": 6.333556403375027e-06,
      "loss": 0.0224,
      "step": 53170
    },
    {
      "epoch": 0.366892726306857,
      "grad_norm": 0.0,
      "learning_rate": 6.332866496029584e-06,
      "loss": 0.0,
      "step": 53180
    },
    {
      "epoch": 0.36696171704140135,
      "grad_norm": 0.0,
      "learning_rate": 6.33217658868414e-06,
      "loss": 0.0,
      "step": 53190
    },
    {
      "epoch": 0.3670307077759457,
      "grad_norm": 0.0,
      "learning_rate": 6.331486681338697e-06,
      "loss": 0.8742,
      "step": 53200
    },
    {
      "epoch": 0.36709969851049,
      "grad_norm": 0.10914280265569687,
      "learning_rate": 6.330796773993253e-06,
      "loss": 0.0,
      "step": 53210
    },
    {
      "epoch": 0.3671686892450344,
      "grad_norm": 2.665992155925778e-07,
      "learning_rate": 6.330106866647809e-06,
      "loss": 0.0,
      "step": 53220
    },
    {
      "epoch": 0.36723767997957873,
      "grad_norm": 0.0,
      "learning_rate": 6.3294169593023655e-06,
      "loss": 0.0,
      "step": 53230
    },
    {
      "epoch": 0.3673066707141231,
      "grad_norm": 0.0,
      "learning_rate": 6.328727051956922e-06,
      "loss": 0.0,
      "step": 53240
    },
    {
      "epoch": 0.36737566144866746,
      "grad_norm": 0.0021569826640188694,
      "learning_rate": 6.328037144611479e-06,
      "loss": 0.0,
      "step": 53250
    },
    {
      "epoch": 0.3674446521832118,
      "grad_norm": 0.0,
      "learning_rate": 6.327347237266036e-06,
      "loss": 0.0,
      "step": 53260
    },
    {
      "epoch": 0.3675136429177561,
      "grad_norm": 0.002401307225227356,
      "learning_rate": 6.326657329920592e-06,
      "loss": 0.0,
      "step": 53270
    },
    {
      "epoch": 0.3675826336523005,
      "grad_norm": 0.0,
      "learning_rate": 6.325967422575149e-06,
      "loss": 0.0,
      "step": 53280
    },
    {
      "epoch": 0.36765162438684484,
      "grad_norm": 0.0,
      "learning_rate": 6.325277515229705e-06,
      "loss": 0.0,
      "step": 53290
    },
    {
      "epoch": 0.3677206151213892,
      "grad_norm": 0.0,
      "learning_rate": 6.324587607884262e-06,
      "loss": 0.0,
      "step": 53300
    },
    {
      "epoch": 0.36778960585593357,
      "grad_norm": 0.0,
      "learning_rate": 6.323897700538818e-06,
      "loss": 0.0001,
      "step": 53310
    },
    {
      "epoch": 0.3678585965904779,
      "grad_norm": 0.0,
      "learning_rate": 6.323207793193375e-06,
      "loss": 0.0,
      "step": 53320
    },
    {
      "epoch": 0.36792758732502223,
      "grad_norm": 0.0,
      "learning_rate": 6.32251788584793e-06,
      "loss": 0.0,
      "step": 53330
    },
    {
      "epoch": 0.3679965780595666,
      "grad_norm": 0.0,
      "learning_rate": 6.321827978502487e-06,
      "loss": 0.0002,
      "step": 53340
    },
    {
      "epoch": 0.36806556879411095,
      "grad_norm": 0.0,
      "learning_rate": 6.321138071157043e-06,
      "loss": 0.0,
      "step": 53350
    },
    {
      "epoch": 0.3681345595286553,
      "grad_norm": 0.0,
      "learning_rate": 6.320448163811601e-06,
      "loss": 0.0,
      "step": 53360
    },
    {
      "epoch": 0.3682035502631997,
      "grad_norm": 0.0,
      "learning_rate": 6.319758256466157e-06,
      "loss": 0.0,
      "step": 53370
    },
    {
      "epoch": 0.368272540997744,
      "grad_norm": 0.0,
      "learning_rate": 6.319068349120714e-06,
      "loss": 0.0,
      "step": 53380
    },
    {
      "epoch": 0.36834153173228834,
      "grad_norm": 0.0,
      "learning_rate": 6.31837844177527e-06,
      "loss": 0.0918,
      "step": 53390
    },
    {
      "epoch": 0.3684105224668327,
      "grad_norm": 0.0,
      "learning_rate": 6.317688534429827e-06,
      "loss": 0.0493,
      "step": 53400
    },
    {
      "epoch": 0.36847951320137706,
      "grad_norm": 0.0,
      "learning_rate": 6.316998627084383e-06,
      "loss": 0.0,
      "step": 53410
    },
    {
      "epoch": 0.3685485039359214,
      "grad_norm": 0.0,
      "learning_rate": 6.31630871973894e-06,
      "loss": 0.0,
      "step": 53420
    },
    {
      "epoch": 0.3686174946704658,
      "grad_norm": 2.9677866564270516e-07,
      "learning_rate": 6.315618812393496e-06,
      "loss": 0.0222,
      "step": 53430
    },
    {
      "epoch": 0.3686864854050101,
      "grad_norm": 0.0,
      "learning_rate": 6.314928905048053e-06,
      "loss": 0.0,
      "step": 53440
    },
    {
      "epoch": 0.36875547613955445,
      "grad_norm": 1.6090236902236938,
      "learning_rate": 6.314238997702608e-06,
      "loss": 0.0017,
      "step": 53450
    },
    {
      "epoch": 0.3688244668740988,
      "grad_norm": 0.0,
      "learning_rate": 6.313549090357165e-06,
      "loss": 0.0006,
      "step": 53460
    },
    {
      "epoch": 0.36889345760864317,
      "grad_norm": 0.0,
      "learning_rate": 6.312859183011722e-06,
      "loss": 0.0009,
      "step": 53470
    },
    {
      "epoch": 0.36896244834318753,
      "grad_norm": 257.950439453125,
      "learning_rate": 6.312169275666279e-06,
      "loss": 0.044,
      "step": 53480
    },
    {
      "epoch": 0.36903143907773184,
      "grad_norm": 1421.5469970703125,
      "learning_rate": 6.311479368320835e-06,
      "loss": 0.3107,
      "step": 53490
    },
    {
      "epoch": 0.3691004298122762,
      "grad_norm": 0.0,
      "learning_rate": 6.3107894609753916e-06,
      "loss": 0.0,
      "step": 53500
    },
    {
      "epoch": 0.36916942054682056,
      "grad_norm": 0.0,
      "learning_rate": 6.310099553629948e-06,
      "loss": 0.0,
      "step": 53510
    },
    {
      "epoch": 0.3692384112813649,
      "grad_norm": 0.6843840479850769,
      "learning_rate": 6.3094096462845045e-06,
      "loss": 0.0004,
      "step": 53520
    },
    {
      "epoch": 0.3693074020159093,
      "grad_norm": 0.0,
      "learning_rate": 6.308719738939061e-06,
      "loss": 0.0,
      "step": 53530
    },
    {
      "epoch": 0.36937639275045364,
      "grad_norm": 4.340824375503871e-07,
      "learning_rate": 6.3080298315936175e-06,
      "loss": 0.0,
      "step": 53540
    },
    {
      "epoch": 0.36944538348499795,
      "grad_norm": 0.0,
      "learning_rate": 6.307339924248174e-06,
      "loss": 0.0,
      "step": 53550
    },
    {
      "epoch": 0.3695143742195423,
      "grad_norm": 0.0,
      "learning_rate": 6.30665001690273e-06,
      "loss": 0.1663,
      "step": 53560
    },
    {
      "epoch": 0.36958336495408667,
      "grad_norm": 8.709826215635985e-05,
      "learning_rate": 6.305960109557286e-06,
      "loss": 0.0,
      "step": 53570
    },
    {
      "epoch": 0.36965235568863103,
      "grad_norm": 0.0,
      "learning_rate": 6.305270202211844e-06,
      "loss": 0.0006,
      "step": 53580
    },
    {
      "epoch": 0.3697213464231754,
      "grad_norm": 1.2828556350541476e-07,
      "learning_rate": 6.3045802948664e-06,
      "loss": 0.0,
      "step": 53590
    },
    {
      "epoch": 0.3697903371577197,
      "grad_norm": 0.008203966543078423,
      "learning_rate": 6.3038903875209565e-06,
      "loss": 0.0,
      "step": 53600
    },
    {
      "epoch": 0.36985932789226406,
      "grad_norm": 0.0,
      "learning_rate": 6.303200480175513e-06,
      "loss": 0.0,
      "step": 53610
    },
    {
      "epoch": 0.3699283186268084,
      "grad_norm": 0.0,
      "learning_rate": 6.3025105728300694e-06,
      "loss": 0.0,
      "step": 53620
    },
    {
      "epoch": 0.3699973093613528,
      "grad_norm": 0.0,
      "learning_rate": 6.301820665484626e-06,
      "loss": 0.0377,
      "step": 53630
    },
    {
      "epoch": 0.37006630009589714,
      "grad_norm": 0.0,
      "learning_rate": 6.3011307581391824e-06,
      "loss": 0.0,
      "step": 53640
    },
    {
      "epoch": 0.3701352908304415,
      "grad_norm": 0.0,
      "learning_rate": 6.300440850793739e-06,
      "loss": 0.0,
      "step": 53650
    },
    {
      "epoch": 0.3702042815649858,
      "grad_norm": 0.0,
      "learning_rate": 6.299750943448295e-06,
      "loss": 0.0,
      "step": 53660
    },
    {
      "epoch": 0.37027327229953017,
      "grad_norm": 0.0,
      "learning_rate": 6.299061036102851e-06,
      "loss": 0.0001,
      "step": 53670
    },
    {
      "epoch": 0.3703422630340745,
      "grad_norm": 0.0,
      "learning_rate": 6.2983711287574075e-06,
      "loss": 0.0,
      "step": 53680
    },
    {
      "epoch": 0.3704112537686189,
      "grad_norm": 4.501118766597756e-08,
      "learning_rate": 6.297681221411966e-06,
      "loss": 0.0,
      "step": 53690
    },
    {
      "epoch": 0.37048024450316325,
      "grad_norm": 0.0,
      "learning_rate": 6.296991314066521e-06,
      "loss": 0.0,
      "step": 53700
    },
    {
      "epoch": 0.37054923523770755,
      "grad_norm": 0.0,
      "learning_rate": 6.296301406721078e-06,
      "loss": 0.0,
      "step": 53710
    },
    {
      "epoch": 0.3706182259722519,
      "grad_norm": 2.6968313804331956e-08,
      "learning_rate": 6.295611499375634e-06,
      "loss": 0.0002,
      "step": 53720
    },
    {
      "epoch": 0.3706872167067963,
      "grad_norm": 0.0,
      "learning_rate": 6.294921592030191e-06,
      "loss": 0.0,
      "step": 53730
    },
    {
      "epoch": 0.37075620744134064,
      "grad_norm": 0.0,
      "learning_rate": 6.294231684684747e-06,
      "loss": 0.0,
      "step": 53740
    },
    {
      "epoch": 0.370825198175885,
      "grad_norm": 7.549827500952233e-07,
      "learning_rate": 6.293541777339304e-06,
      "loss": 0.0002,
      "step": 53750
    },
    {
      "epoch": 0.37089418891042936,
      "grad_norm": 0.0,
      "learning_rate": 6.29285186999386e-06,
      "loss": 0.0,
      "step": 53760
    },
    {
      "epoch": 0.37096317964497366,
      "grad_norm": 2.8684544304269366e-06,
      "learning_rate": 6.292161962648417e-06,
      "loss": 0.0,
      "step": 53770
    },
    {
      "epoch": 0.371032170379518,
      "grad_norm": 0.17768219113349915,
      "learning_rate": 6.291472055302973e-06,
      "loss": 0.0008,
      "step": 53780
    },
    {
      "epoch": 0.3711011611140624,
      "grad_norm": 0.0,
      "learning_rate": 6.290782147957529e-06,
      "loss": 0.0,
      "step": 53790
    },
    {
      "epoch": 0.37117015184860674,
      "grad_norm": 0.0,
      "learning_rate": 6.290092240612087e-06,
      "loss": 0.0,
      "step": 53800
    },
    {
      "epoch": 0.3712391425831511,
      "grad_norm": 0.010918626561760902,
      "learning_rate": 6.289402333266643e-06,
      "loss": 0.0091,
      "step": 53810
    },
    {
      "epoch": 0.3713081333176954,
      "grad_norm": 0.0,
      "learning_rate": 6.288712425921199e-06,
      "loss": 0.0,
      "step": 53820
    },
    {
      "epoch": 0.37137712405223977,
      "grad_norm": 0.0005918614915572107,
      "learning_rate": 6.288022518575756e-06,
      "loss": 0.0005,
      "step": 53830
    },
    {
      "epoch": 0.37144611478678413,
      "grad_norm": 0.0,
      "learning_rate": 6.287332611230312e-06,
      "loss": 0.0,
      "step": 53840
    },
    {
      "epoch": 0.3715151055213285,
      "grad_norm": 0.0,
      "learning_rate": 6.286642703884869e-06,
      "loss": 0.0,
      "step": 53850
    },
    {
      "epoch": 0.37158409625587285,
      "grad_norm": 0.0,
      "learning_rate": 6.285952796539425e-06,
      "loss": 0.1163,
      "step": 53860
    },
    {
      "epoch": 0.3716530869904172,
      "grad_norm": 0.0,
      "learning_rate": 6.285262889193982e-06,
      "loss": 0.0,
      "step": 53870
    },
    {
      "epoch": 0.3717220777249615,
      "grad_norm": 0.0,
      "learning_rate": 6.284572981848538e-06,
      "loss": 0.0,
      "step": 53880
    },
    {
      "epoch": 0.3717910684595059,
      "grad_norm": 0.0,
      "learning_rate": 6.283883074503095e-06,
      "loss": 0.0,
      "step": 53890
    },
    {
      "epoch": 0.37186005919405024,
      "grad_norm": 0.0,
      "learning_rate": 6.28319316715765e-06,
      "loss": 0.0003,
      "step": 53900
    },
    {
      "epoch": 0.3719290499285946,
      "grad_norm": 0.0,
      "learning_rate": 6.2825032598122085e-06,
      "loss": 0.0001,
      "step": 53910
    },
    {
      "epoch": 0.37199804066313896,
      "grad_norm": 0.0,
      "learning_rate": 6.281813352466765e-06,
      "loss": 0.0,
      "step": 53920
    },
    {
      "epoch": 0.37206703139768327,
      "grad_norm": 0.0,
      "learning_rate": 6.281123445121321e-06,
      "loss": 0.0,
      "step": 53930
    },
    {
      "epoch": 0.37213602213222763,
      "grad_norm": 0.0,
      "learning_rate": 6.280433537775877e-06,
      "loss": 0.0,
      "step": 53940
    },
    {
      "epoch": 0.372205012866772,
      "grad_norm": 0.0012977931182831526,
      "learning_rate": 6.279743630430434e-06,
      "loss": 0.0005,
      "step": 53950
    },
    {
      "epoch": 0.37227400360131635,
      "grad_norm": 0.0,
      "learning_rate": 6.27905372308499e-06,
      "loss": 0.0,
      "step": 53960
    },
    {
      "epoch": 0.3723429943358607,
      "grad_norm": 0.0,
      "learning_rate": 6.278363815739547e-06,
      "loss": 0.0,
      "step": 53970
    },
    {
      "epoch": 0.37241198507040507,
      "grad_norm": 0.18349036574363708,
      "learning_rate": 6.277673908394103e-06,
      "loss": 0.0,
      "step": 53980
    },
    {
      "epoch": 0.3724809758049494,
      "grad_norm": 0.0,
      "learning_rate": 6.27698400104866e-06,
      "loss": 0.0,
      "step": 53990
    },
    {
      "epoch": 0.37254996653949374,
      "grad_norm": 0.0,
      "learning_rate": 6.276294093703216e-06,
      "loss": 0.0,
      "step": 54000
    },
    {
      "epoch": 0.3726189572740381,
      "grad_norm": 0.0,
      "learning_rate": 6.275604186357772e-06,
      "loss": 0.0,
      "step": 54010
    },
    {
      "epoch": 0.37268794800858246,
      "grad_norm": 2.8242279626056188e-08,
      "learning_rate": 6.27491427901233e-06,
      "loss": 0.0,
      "step": 54020
    },
    {
      "epoch": 0.3727569387431268,
      "grad_norm": 0.0,
      "learning_rate": 6.274224371666886e-06,
      "loss": 0.0,
      "step": 54030
    },
    {
      "epoch": 0.3728259294776711,
      "grad_norm": 0.0,
      "learning_rate": 6.273534464321442e-06,
      "loss": 0.0,
      "step": 54040
    },
    {
      "epoch": 0.3728949202122155,
      "grad_norm": 0.0,
      "learning_rate": 6.2728445569759985e-06,
      "loss": 0.0004,
      "step": 54050
    },
    {
      "epoch": 0.37296391094675985,
      "grad_norm": 0.0,
      "learning_rate": 6.272154649630555e-06,
      "loss": 0.0,
      "step": 54060
    },
    {
      "epoch": 0.3730329016813042,
      "grad_norm": 2.5350940902058028e-08,
      "learning_rate": 6.2714647422851115e-06,
      "loss": 0.0001,
      "step": 54070
    },
    {
      "epoch": 0.37310189241584857,
      "grad_norm": 4.99712826673715e-10,
      "learning_rate": 6.270774834939668e-06,
      "loss": 0.0,
      "step": 54080
    },
    {
      "epoch": 0.37317088315039293,
      "grad_norm": 2.5723872809635395e-08,
      "learning_rate": 6.2700849275942245e-06,
      "loss": 0.0,
      "step": 54090
    },
    {
      "epoch": 0.37323987388493723,
      "grad_norm": 2.6685615495125603e-08,
      "learning_rate": 6.269395020248781e-06,
      "loss": 0.0,
      "step": 54100
    },
    {
      "epoch": 0.3733088646194816,
      "grad_norm": 0.0,
      "learning_rate": 6.2687051129033375e-06,
      "loss": 0.0,
      "step": 54110
    },
    {
      "epoch": 0.37337785535402596,
      "grad_norm": 0.0,
      "learning_rate": 6.268015205557894e-06,
      "loss": 0.0,
      "step": 54120
    },
    {
      "epoch": 0.3734468460885703,
      "grad_norm": 0.0,
      "learning_rate": 6.267325298212451e-06,
      "loss": 0.0,
      "step": 54130
    },
    {
      "epoch": 0.3735158368231147,
      "grad_norm": 2.799781322479248,
      "learning_rate": 6.266635390867008e-06,
      "loss": 0.0005,
      "step": 54140
    },
    {
      "epoch": 0.373584827557659,
      "grad_norm": 3.888081209879601e-06,
      "learning_rate": 6.2659454835215634e-06,
      "loss": 0.0,
      "step": 54150
    },
    {
      "epoch": 0.37365381829220334,
      "grad_norm": 0.0,
      "learning_rate": 6.26525557617612e-06,
      "loss": 0.0,
      "step": 54160
    },
    {
      "epoch": 0.3737228090267477,
      "grad_norm": 0.0,
      "learning_rate": 6.264565668830676e-06,
      "loss": 0.0,
      "step": 54170
    },
    {
      "epoch": 0.37379179976129207,
      "grad_norm": 9.067513929039706e-06,
      "learning_rate": 6.263875761485233e-06,
      "loss": 0.0006,
      "step": 54180
    },
    {
      "epoch": 0.3738607904958364,
      "grad_norm": 0.0,
      "learning_rate": 6.263185854139789e-06,
      "loss": 0.0,
      "step": 54190
    },
    {
      "epoch": 0.3739297812303808,
      "grad_norm": 1.236219532074756e-06,
      "learning_rate": 6.262495946794346e-06,
      "loss": 0.0,
      "step": 54200
    },
    {
      "epoch": 0.3739987719649251,
      "grad_norm": 0.0,
      "learning_rate": 6.261806039448902e-06,
      "loss": 0.0041,
      "step": 54210
    },
    {
      "epoch": 0.37406776269946945,
      "grad_norm": 0.0,
      "learning_rate": 6.261116132103459e-06,
      "loss": 0.3545,
      "step": 54220
    },
    {
      "epoch": 0.3741367534340138,
      "grad_norm": 0.0,
      "learning_rate": 6.260426224758015e-06,
      "loss": 0.1688,
      "step": 54230
    },
    {
      "epoch": 0.3742057441685582,
      "grad_norm": 0.0,
      "learning_rate": 6.259736317412573e-06,
      "loss": 0.0001,
      "step": 54240
    },
    {
      "epoch": 0.37427473490310254,
      "grad_norm": 0.0,
      "learning_rate": 6.259046410067129e-06,
      "loss": 0.0726,
      "step": 54250
    },
    {
      "epoch": 0.37434372563764684,
      "grad_norm": 0.0,
      "learning_rate": 6.258356502721686e-06,
      "loss": 0.0,
      "step": 54260
    },
    {
      "epoch": 0.3744127163721912,
      "grad_norm": 0.0,
      "learning_rate": 6.257666595376241e-06,
      "loss": 0.0009,
      "step": 54270
    },
    {
      "epoch": 0.37448170710673556,
      "grad_norm": 0.0,
      "learning_rate": 6.256976688030798e-06,
      "loss": 0.0,
      "step": 54280
    },
    {
      "epoch": 0.3745506978412799,
      "grad_norm": 1.5801893823663704e-05,
      "learning_rate": 6.256286780685354e-06,
      "loss": 0.0,
      "step": 54290
    },
    {
      "epoch": 0.3746196885758243,
      "grad_norm": 0.0009169537806883454,
      "learning_rate": 6.255596873339911e-06,
      "loss": 0.0,
      "step": 54300
    },
    {
      "epoch": 0.37468867931036864,
      "grad_norm": 0.0,
      "learning_rate": 6.254906965994467e-06,
      "loss": 0.0,
      "step": 54310
    },
    {
      "epoch": 0.37475767004491295,
      "grad_norm": 0.0,
      "learning_rate": 6.254217058649024e-06,
      "loss": 0.0,
      "step": 54320
    },
    {
      "epoch": 0.3748266607794573,
      "grad_norm": 0.0,
      "learning_rate": 6.25352715130358e-06,
      "loss": 0.0,
      "step": 54330
    },
    {
      "epoch": 0.37489565151400167,
      "grad_norm": 0.0,
      "learning_rate": 6.252837243958137e-06,
      "loss": 0.0004,
      "step": 54340
    },
    {
      "epoch": 0.37496464224854603,
      "grad_norm": 0.0035362797789275646,
      "learning_rate": 6.252147336612694e-06,
      "loss": 0.0,
      "step": 54350
    },
    {
      "epoch": 0.3750336329830904,
      "grad_norm": 0.0,
      "learning_rate": 6.251457429267251e-06,
      "loss": 0.0,
      "step": 54360
    },
    {
      "epoch": 0.3751026237176347,
      "grad_norm": 0.004642609041184187,
      "learning_rate": 6.250767521921807e-06,
      "loss": 0.0,
      "step": 54370
    },
    {
      "epoch": 0.37517161445217906,
      "grad_norm": 1.1241195352340583e-05,
      "learning_rate": 6.250077614576363e-06,
      "loss": 0.0,
      "step": 54380
    },
    {
      "epoch": 0.3752406051867234,
      "grad_norm": 0.07726994901895523,
      "learning_rate": 6.249387707230919e-06,
      "loss": 0.0,
      "step": 54390
    },
    {
      "epoch": 0.3753095959212678,
      "grad_norm": 0.0,
      "learning_rate": 6.248697799885476e-06,
      "loss": 0.0008,
      "step": 54400
    },
    {
      "epoch": 0.37537858665581214,
      "grad_norm": 0.053300295025110245,
      "learning_rate": 6.248007892540032e-06,
      "loss": 0.0,
      "step": 54410
    },
    {
      "epoch": 0.3754475773903565,
      "grad_norm": 0.0,
      "learning_rate": 6.247317985194589e-06,
      "loss": 0.0,
      "step": 54420
    },
    {
      "epoch": 0.3755165681249008,
      "grad_norm": 0.0,
      "learning_rate": 6.246628077849145e-06,
      "loss": 0.0001,
      "step": 54430
    },
    {
      "epoch": 0.37558555885944517,
      "grad_norm": 0.0,
      "learning_rate": 6.245938170503702e-06,
      "loss": 0.0633,
      "step": 54440
    },
    {
      "epoch": 0.37565454959398953,
      "grad_norm": 0.0,
      "learning_rate": 6.245248263158258e-06,
      "loss": 0.0002,
      "step": 54450
    },
    {
      "epoch": 0.3757235403285339,
      "grad_norm": 0.0,
      "learning_rate": 6.2445583558128155e-06,
      "loss": 0.0,
      "step": 54460
    },
    {
      "epoch": 0.37579253106307825,
      "grad_norm": 0.09672283381223679,
      "learning_rate": 6.243868448467372e-06,
      "loss": 0.0008,
      "step": 54470
    },
    {
      "epoch": 0.37586152179762256,
      "grad_norm": 0.007992210797965527,
      "learning_rate": 6.2431785411219285e-06,
      "loss": 0.0,
      "step": 54480
    },
    {
      "epoch": 0.3759305125321669,
      "grad_norm": 0.0,
      "learning_rate": 6.242488633776484e-06,
      "loss": 0.0,
      "step": 54490
    },
    {
      "epoch": 0.3759995032667113,
      "grad_norm": 0.0,
      "learning_rate": 6.241867717165585e-06,
      "loss": 0.2797,
      "step": 54500
    },
    {
      "epoch": 0.37606849400125564,
      "grad_norm": 0.0,
      "learning_rate": 6.241177809820141e-06,
      "loss": 0.0,
      "step": 54510
    },
    {
      "epoch": 0.3761374847358,
      "grad_norm": 0.0,
      "learning_rate": 6.240487902474698e-06,
      "loss": 0.0,
      "step": 54520
    },
    {
      "epoch": 0.37620647547034436,
      "grad_norm": 0.0,
      "learning_rate": 6.239797995129254e-06,
      "loss": 0.0,
      "step": 54530
    },
    {
      "epoch": 0.37627546620488866,
      "grad_norm": 0.0,
      "learning_rate": 6.239108087783811e-06,
      "loss": 0.0,
      "step": 54540
    },
    {
      "epoch": 0.376344456939433,
      "grad_norm": 0.0,
      "learning_rate": 6.238418180438368e-06,
      "loss": 0.0,
      "step": 54550
    },
    {
      "epoch": 0.3764134476739774,
      "grad_norm": 0.0,
      "learning_rate": 6.2377282730929246e-06,
      "loss": 0.0,
      "step": 54560
    },
    {
      "epoch": 0.37648243840852175,
      "grad_norm": 0.0,
      "learning_rate": 6.237038365747481e-06,
      "loss": 0.0011,
      "step": 54570
    },
    {
      "epoch": 0.3765514291430661,
      "grad_norm": 3.5893006611331657e-07,
      "learning_rate": 6.2363484584020376e-06,
      "loss": 0.0,
      "step": 54580
    },
    {
      "epoch": 0.3766204198776104,
      "grad_norm": 5.55206991492696e-09,
      "learning_rate": 6.235658551056594e-06,
      "loss": 0.0,
      "step": 54590
    },
    {
      "epoch": 0.3766894106121548,
      "grad_norm": 4.77972412109375,
      "learning_rate": 6.23496864371115e-06,
      "loss": 0.001,
      "step": 54600
    },
    {
      "epoch": 0.37675840134669913,
      "grad_norm": 0.0,
      "learning_rate": 6.234278736365706e-06,
      "loss": 0.0,
      "step": 54610
    },
    {
      "epoch": 0.3768273920812435,
      "grad_norm": 0.17097824811935425,
      "learning_rate": 6.233588829020263e-06,
      "loss": 0.0,
      "step": 54620
    },
    {
      "epoch": 0.37689638281578786,
      "grad_norm": 0.0,
      "learning_rate": 6.232898921674819e-06,
      "loss": 0.0,
      "step": 54630
    },
    {
      "epoch": 0.3769653735503322,
      "grad_norm": 2.2251473637879826e-05,
      "learning_rate": 6.232209014329376e-06,
      "loss": 0.0002,
      "step": 54640
    },
    {
      "epoch": 0.3770343642848765,
      "grad_norm": 0.0,
      "learning_rate": 6.231519106983932e-06,
      "loss": 0.0,
      "step": 54650
    },
    {
      "epoch": 0.3771033550194209,
      "grad_norm": 0.0,
      "learning_rate": 6.2308291996384895e-06,
      "loss": 0.0,
      "step": 54660
    },
    {
      "epoch": 0.37717234575396524,
      "grad_norm": 0.0,
      "learning_rate": 6.230139292293046e-06,
      "loss": 0.0136,
      "step": 54670
    },
    {
      "epoch": 0.3772413364885096,
      "grad_norm": 0.0,
      "learning_rate": 6.2294493849476025e-06,
      "loss": 0.2169,
      "step": 54680
    },
    {
      "epoch": 0.37731032722305397,
      "grad_norm": 0.0,
      "learning_rate": 6.228759477602159e-06,
      "loss": 0.0,
      "step": 54690
    },
    {
      "epoch": 0.37737931795759827,
      "grad_norm": 0.2226845622062683,
      "learning_rate": 6.2280695702567154e-06,
      "loss": 0.0008,
      "step": 54700
    },
    {
      "epoch": 0.37744830869214263,
      "grad_norm": 6.639002094743773e-07,
      "learning_rate": 6.227379662911271e-06,
      "loss": 0.8344,
      "step": 54710
    },
    {
      "epoch": 0.377517299426687,
      "grad_norm": 0.0,
      "learning_rate": 6.2266897555658276e-06,
      "loss": 0.0,
      "step": 54720
    },
    {
      "epoch": 0.37758629016123135,
      "grad_norm": 2.3116963348002173e-05,
      "learning_rate": 6.225999848220384e-06,
      "loss": 0.0,
      "step": 54730
    },
    {
      "epoch": 0.3776552808957757,
      "grad_norm": 0.0,
      "learning_rate": 6.2253099408749405e-06,
      "loss": 0.0,
      "step": 54740
    },
    {
      "epoch": 0.3777242716303201,
      "grad_norm": 0.0,
      "learning_rate": 6.224620033529497e-06,
      "loss": 0.0,
      "step": 54750
    },
    {
      "epoch": 0.3777932623648644,
      "grad_norm": 0.0,
      "learning_rate": 6.2239301261840535e-06,
      "loss": 0.0,
      "step": 54760
    },
    {
      "epoch": 0.37786225309940874,
      "grad_norm": 0.0,
      "learning_rate": 6.223240218838611e-06,
      "loss": 0.0,
      "step": 54770
    },
    {
      "epoch": 0.3779312438339531,
      "grad_norm": 0.0,
      "learning_rate": 6.222550311493167e-06,
      "loss": 0.0,
      "step": 54780
    },
    {
      "epoch": 0.37800023456849746,
      "grad_norm": 1.8271359205245972,
      "learning_rate": 6.221860404147724e-06,
      "loss": 0.0006,
      "step": 54790
    },
    {
      "epoch": 0.3780692253030418,
      "grad_norm": 0.0,
      "learning_rate": 6.22117049680228e-06,
      "loss": 0.0,
      "step": 54800
    },
    {
      "epoch": 0.37813821603758613,
      "grad_norm": 0.0,
      "learning_rate": 6.220480589456837e-06,
      "loss": 0.0,
      "step": 54810
    },
    {
      "epoch": 0.3782072067721305,
      "grad_norm": 0.0,
      "learning_rate": 6.219790682111393e-06,
      "loss": 0.0727,
      "step": 54820
    },
    {
      "epoch": 0.37827619750667485,
      "grad_norm": 0.0,
      "learning_rate": 6.219100774765949e-06,
      "loss": 0.0013,
      "step": 54830
    },
    {
      "epoch": 0.3783451882412192,
      "grad_norm": 0.4376320540904999,
      "learning_rate": 6.2184108674205055e-06,
      "loss": 0.0,
      "step": 54840
    },
    {
      "epoch": 0.37841417897576357,
      "grad_norm": 0.004327361937612295,
      "learning_rate": 6.217720960075062e-06,
      "loss": 0.0,
      "step": 54850
    },
    {
      "epoch": 0.37848316971030793,
      "grad_norm": 0.0,
      "learning_rate": 6.2170310527296184e-06,
      "loss": 0.0725,
      "step": 54860
    },
    {
      "epoch": 0.37855216044485224,
      "grad_norm": 0.0,
      "learning_rate": 6.216341145384175e-06,
      "loss": 0.0,
      "step": 54870
    },
    {
      "epoch": 0.3786211511793966,
      "grad_norm": 0.0,
      "learning_rate": 6.215651238038732e-06,
      "loss": 0.0113,
      "step": 54880
    },
    {
      "epoch": 0.37869014191394096,
      "grad_norm": 0.0,
      "learning_rate": 6.214961330693289e-06,
      "loss": 0.0,
      "step": 54890
    },
    {
      "epoch": 0.3787591326484853,
      "grad_norm": 0.0,
      "learning_rate": 6.214271423347845e-06,
      "loss": 0.0002,
      "step": 54900
    },
    {
      "epoch": 0.3788281233830297,
      "grad_norm": 0.0,
      "learning_rate": 6.213581516002402e-06,
      "loss": 0.0002,
      "step": 54910
    },
    {
      "epoch": 0.378897114117574,
      "grad_norm": 0.0,
      "learning_rate": 6.212891608656958e-06,
      "loss": 0.0053,
      "step": 54920
    },
    {
      "epoch": 0.37896610485211835,
      "grad_norm": 0.0,
      "learning_rate": 6.212201701311515e-06,
      "loss": 0.0,
      "step": 54930
    },
    {
      "epoch": 0.3790350955866627,
      "grad_norm": 0.0,
      "learning_rate": 6.21151179396607e-06,
      "loss": 0.0,
      "step": 54940
    },
    {
      "epoch": 0.37910408632120707,
      "grad_norm": 0.0,
      "learning_rate": 6.210821886620627e-06,
      "loss": 0.0,
      "step": 54950
    },
    {
      "epoch": 0.37917307705575143,
      "grad_norm": 0.0,
      "learning_rate": 6.210131979275183e-06,
      "loss": 0.0,
      "step": 54960
    },
    {
      "epoch": 0.3792420677902958,
      "grad_norm": 0.0,
      "learning_rate": 6.20944207192974e-06,
      "loss": 0.0,
      "step": 54970
    },
    {
      "epoch": 0.3793110585248401,
      "grad_norm": 0.0,
      "learning_rate": 6.208752164584296e-06,
      "loss": 0.0,
      "step": 54980
    },
    {
      "epoch": 0.37938004925938446,
      "grad_norm": 0.0,
      "learning_rate": 6.208062257238854e-06,
      "loss": 0.0,
      "step": 54990
    },
    {
      "epoch": 0.3794490399939288,
      "grad_norm": 0.0,
      "learning_rate": 6.20737234989341e-06,
      "loss": 0.0,
      "step": 55000
    },
    {
      "epoch": 0.3795180307284732,
      "grad_norm": 0.0,
      "learning_rate": 6.206682442547967e-06,
      "loss": 0.0001,
      "step": 55010
    },
    {
      "epoch": 0.37958702146301754,
      "grad_norm": 0.00025303696747869253,
      "learning_rate": 6.205992535202523e-06,
      "loss": 0.0,
      "step": 55020
    },
    {
      "epoch": 0.37965601219756184,
      "grad_norm": 0.0,
      "learning_rate": 6.20530262785708e-06,
      "loss": 0.0,
      "step": 55030
    },
    {
      "epoch": 0.3797250029321062,
      "grad_norm": 0.0,
      "learning_rate": 6.204612720511636e-06,
      "loss": 0.0028,
      "step": 55040
    },
    {
      "epoch": 0.37979399366665056,
      "grad_norm": 0.0,
      "learning_rate": 6.203922813166192e-06,
      "loss": 1.2266,
      "step": 55050
    },
    {
      "epoch": 0.3798629844011949,
      "grad_norm": 0.0,
      "learning_rate": 6.203232905820748e-06,
      "loss": 0.0,
      "step": 55060
    },
    {
      "epoch": 0.3799319751357393,
      "grad_norm": 0.0,
      "learning_rate": 6.202542998475305e-06,
      "loss": 0.0005,
      "step": 55070
    },
    {
      "epoch": 0.38000096587028365,
      "grad_norm": 0.0,
      "learning_rate": 6.201853091129861e-06,
      "loss": 0.0,
      "step": 55080
    },
    {
      "epoch": 0.38006995660482795,
      "grad_norm": 0.0,
      "learning_rate": 6.201163183784418e-06,
      "loss": 0.0,
      "step": 55090
    },
    {
      "epoch": 0.3801389473393723,
      "grad_norm": 0.0,
      "learning_rate": 6.200473276438975e-06,
      "loss": 0.0006,
      "step": 55100
    },
    {
      "epoch": 0.3802079380739167,
      "grad_norm": 0.0003746901056729257,
      "learning_rate": 6.1997833690935315e-06,
      "loss": 0.0,
      "step": 55110
    },
    {
      "epoch": 0.38027692880846103,
      "grad_norm": 0.0,
      "learning_rate": 6.199093461748088e-06,
      "loss": 0.0001,
      "step": 55120
    },
    {
      "epoch": 0.3803459195430054,
      "grad_norm": 0.0,
      "learning_rate": 6.1984035544026445e-06,
      "loss": 0.0,
      "step": 55130
    },
    {
      "epoch": 0.3804149102775497,
      "grad_norm": 0.0,
      "learning_rate": 6.197713647057201e-06,
      "loss": 0.0,
      "step": 55140
    },
    {
      "epoch": 0.38048390101209406,
      "grad_norm": 0.0,
      "learning_rate": 6.1970237397117575e-06,
      "loss": 0.0,
      "step": 55150
    },
    {
      "epoch": 0.3805528917466384,
      "grad_norm": 0.0,
      "learning_rate": 6.196333832366314e-06,
      "loss": 0.4266,
      "step": 55160
    },
    {
      "epoch": 0.3806218824811828,
      "grad_norm": 0.0004689408524427563,
      "learning_rate": 6.19564392502087e-06,
      "loss": 0.0,
      "step": 55170
    },
    {
      "epoch": 0.38069087321572714,
      "grad_norm": 0.0,
      "learning_rate": 6.194954017675426e-06,
      "loss": 0.0,
      "step": 55180
    },
    {
      "epoch": 0.3807598639502715,
      "grad_norm": 0.0,
      "learning_rate": 6.194264110329983e-06,
      "loss": 0.1701,
      "step": 55190
    },
    {
      "epoch": 0.3808288546848158,
      "grad_norm": 0.0,
      "learning_rate": 6.193574202984539e-06,
      "loss": 0.0,
      "step": 55200
    },
    {
      "epoch": 0.38089784541936017,
      "grad_norm": 0.0,
      "learning_rate": 6.1928842956390964e-06,
      "loss": 0.0,
      "step": 55210
    },
    {
      "epoch": 0.38096683615390453,
      "grad_norm": 0.00835868064314127,
      "learning_rate": 6.192194388293653e-06,
      "loss": 0.1136,
      "step": 55220
    },
    {
      "epoch": 0.3810358268884489,
      "grad_norm": 4.671475437589834e-07,
      "learning_rate": 6.1915044809482094e-06,
      "loss": 0.0137,
      "step": 55230
    },
    {
      "epoch": 0.38110481762299325,
      "grad_norm": 0.0,
      "learning_rate": 6.190814573602766e-06,
      "loss": 0.0,
      "step": 55240
    },
    {
      "epoch": 0.38117380835753756,
      "grad_norm": 0.0,
      "learning_rate": 6.190124666257322e-06,
      "loss": 0.0029,
      "step": 55250
    },
    {
      "epoch": 0.3812427990920819,
      "grad_norm": 0.0,
      "learning_rate": 6.189434758911879e-06,
      "loss": 0.0,
      "step": 55260
    },
    {
      "epoch": 0.3813117898266263,
      "grad_norm": 0.0,
      "learning_rate": 6.188744851566435e-06,
      "loss": 0.0,
      "step": 55270
    },
    {
      "epoch": 0.38138078056117064,
      "grad_norm": 3.6447808327011444e-08,
      "learning_rate": 6.188054944220991e-06,
      "loss": 0.0,
      "step": 55280
    },
    {
      "epoch": 0.381449771295715,
      "grad_norm": 0.11424826085567474,
      "learning_rate": 6.1873650368755475e-06,
      "loss": 0.0,
      "step": 55290
    },
    {
      "epoch": 0.38151876203025936,
      "grad_norm": 0.0,
      "learning_rate": 6.186675129530104e-06,
      "loss": 0.0,
      "step": 55300
    },
    {
      "epoch": 0.38158775276480367,
      "grad_norm": 0.0,
      "learning_rate": 6.1859852221846605e-06,
      "loss": 0.0,
      "step": 55310
    },
    {
      "epoch": 0.38165674349934803,
      "grad_norm": 0.0017013035248965025,
      "learning_rate": 6.185295314839218e-06,
      "loss": 0.0,
      "step": 55320
    },
    {
      "epoch": 0.3817257342338924,
      "grad_norm": 9.270196699162625e-08,
      "learning_rate": 6.184605407493774e-06,
      "loss": 0.0,
      "step": 55330
    },
    {
      "epoch": 0.38179472496843675,
      "grad_norm": 0.0,
      "learning_rate": 6.183915500148331e-06,
      "loss": 0.0001,
      "step": 55340
    },
    {
      "epoch": 0.3818637157029811,
      "grad_norm": 0.0,
      "learning_rate": 6.183225592802887e-06,
      "loss": 0.0043,
      "step": 55350
    },
    {
      "epoch": 0.3819327064375254,
      "grad_norm": 0.006450190674513578,
      "learning_rate": 6.182535685457444e-06,
      "loss": 0.0,
      "step": 55360
    },
    {
      "epoch": 0.3820016971720698,
      "grad_norm": 0.0,
      "learning_rate": 6.181845778112e-06,
      "loss": 0.0,
      "step": 55370
    },
    {
      "epoch": 0.38207068790661414,
      "grad_norm": 0.0,
      "learning_rate": 6.181155870766557e-06,
      "loss": 0.0047,
      "step": 55380
    },
    {
      "epoch": 0.3821396786411585,
      "grad_norm": 0.0,
      "learning_rate": 6.180465963421112e-06,
      "loss": 0.0,
      "step": 55390
    },
    {
      "epoch": 0.38220866937570286,
      "grad_norm": 9.17267373168329e-10,
      "learning_rate": 6.179776056075669e-06,
      "loss": 0.0,
      "step": 55400
    },
    {
      "epoch": 0.3822776601102472,
      "grad_norm": 0.0,
      "learning_rate": 6.179086148730225e-06,
      "loss": 0.0,
      "step": 55410
    },
    {
      "epoch": 0.3823466508447915,
      "grad_norm": 0.0,
      "learning_rate": 6.178396241384782e-06,
      "loss": 0.0003,
      "step": 55420
    },
    {
      "epoch": 0.3824156415793359,
      "grad_norm": 0.0,
      "learning_rate": 6.177706334039339e-06,
      "loss": 0.0,
      "step": 55430
    },
    {
      "epoch": 0.38248463231388025,
      "grad_norm": 1.2796429871286819e-07,
      "learning_rate": 6.177016426693896e-06,
      "loss": 0.0,
      "step": 55440
    },
    {
      "epoch": 0.3825536230484246,
      "grad_norm": 0.0,
      "learning_rate": 6.176326519348452e-06,
      "loss": 0.0,
      "step": 55450
    },
    {
      "epoch": 0.38262261378296897,
      "grad_norm": 1.7372332461462747e-09,
      "learning_rate": 6.175636612003009e-06,
      "loss": 0.0,
      "step": 55460
    },
    {
      "epoch": 0.3826916045175133,
      "grad_norm": 0.22413499653339386,
      "learning_rate": 6.174946704657565e-06,
      "loss": 0.0,
      "step": 55470
    },
    {
      "epoch": 0.38276059525205763,
      "grad_norm": 0.0,
      "learning_rate": 6.174256797312122e-06,
      "loss": 0.0,
      "step": 55480
    },
    {
      "epoch": 0.382829585986602,
      "grad_norm": 0.0,
      "learning_rate": 6.173566889966678e-06,
      "loss": 0.0,
      "step": 55490
    },
    {
      "epoch": 0.38289857672114636,
      "grad_norm": 0.0,
      "learning_rate": 6.172876982621235e-06,
      "loss": 0.0,
      "step": 55500
    },
    {
      "epoch": 0.3829675674556907,
      "grad_norm": 0.0,
      "learning_rate": 6.17218707527579e-06,
      "loss": 0.0,
      "step": 55510
    },
    {
      "epoch": 0.3830365581902351,
      "grad_norm": 224.2919921875,
      "learning_rate": 6.171497167930347e-06,
      "loss": 0.3236,
      "step": 55520
    },
    {
      "epoch": 0.3831055489247794,
      "grad_norm": 0.2697772681713104,
      "learning_rate": 6.170807260584903e-06,
      "loss": 0.0001,
      "step": 55530
    },
    {
      "epoch": 0.38317453965932374,
      "grad_norm": 0.0,
      "learning_rate": 6.170117353239461e-06,
      "loss": 0.0004,
      "step": 55540
    },
    {
      "epoch": 0.3832435303938681,
      "grad_norm": 3.219560312572867e-05,
      "learning_rate": 6.169427445894017e-06,
      "loss": 0.0,
      "step": 55550
    },
    {
      "epoch": 0.38331252112841246,
      "grad_norm": 0.0,
      "learning_rate": 6.168737538548574e-06,
      "loss": 0.0,
      "step": 55560
    },
    {
      "epoch": 0.3833815118629568,
      "grad_norm": 4.1865173550093004e-09,
      "learning_rate": 6.16804763120313e-06,
      "loss": 0.0,
      "step": 55570
    },
    {
      "epoch": 0.38345050259750113,
      "grad_norm": 0.0,
      "learning_rate": 6.167357723857687e-06,
      "loss": 0.0,
      "step": 55580
    },
    {
      "epoch": 0.3835194933320455,
      "grad_norm": 0.0,
      "learning_rate": 6.166667816512243e-06,
      "loss": 0.0,
      "step": 55590
    },
    {
      "epoch": 0.38358848406658985,
      "grad_norm": 0.0,
      "learning_rate": 6.1659779091667996e-06,
      "loss": 0.1245,
      "step": 55600
    },
    {
      "epoch": 0.3836574748011342,
      "grad_norm": 0.0,
      "learning_rate": 6.165288001821356e-06,
      "loss": 0.0,
      "step": 55610
    },
    {
      "epoch": 0.3837264655356786,
      "grad_norm": 0.006121823564171791,
      "learning_rate": 6.164598094475912e-06,
      "loss": 0.0,
      "step": 55620
    },
    {
      "epoch": 0.38379545627022293,
      "grad_norm": 2.5452036922501975e-08,
      "learning_rate": 6.163908187130468e-06,
      "loss": 0.0,
      "step": 55630
    },
    {
      "epoch": 0.38386444700476724,
      "grad_norm": 0.0,
      "learning_rate": 6.163218279785025e-06,
      "loss": 0.0,
      "step": 55640
    },
    {
      "epoch": 0.3839334377393116,
      "grad_norm": 0.0,
      "learning_rate": 6.162528372439582e-06,
      "loss": 0.0,
      "step": 55650
    },
    {
      "epoch": 0.38400242847385596,
      "grad_norm": 0.0,
      "learning_rate": 6.1618384650941385e-06,
      "loss": 0.0,
      "step": 55660
    },
    {
      "epoch": 0.3840714192084003,
      "grad_norm": 1.4503708598567755e-06,
      "learning_rate": 6.161148557748695e-06,
      "loss": 0.0002,
      "step": 55670
    },
    {
      "epoch": 0.3841404099429447,
      "grad_norm": 0.0,
      "learning_rate": 6.1604586504032515e-06,
      "loss": 0.1825,
      "step": 55680
    },
    {
      "epoch": 0.384209400677489,
      "grad_norm": 0.0,
      "learning_rate": 6.159768743057808e-06,
      "loss": 0.0,
      "step": 55690
    },
    {
      "epoch": 0.38427839141203335,
      "grad_norm": 9.240924692122121e-10,
      "learning_rate": 6.1590788357123645e-06,
      "loss": 0.0,
      "step": 55700
    },
    {
      "epoch": 0.3843473821465777,
      "grad_norm": 0.0,
      "learning_rate": 6.158388928366921e-06,
      "loss": 0.0043,
      "step": 55710
    },
    {
      "epoch": 0.38441637288112207,
      "grad_norm": 5.683835411218752e-07,
      "learning_rate": 6.1576990210214775e-06,
      "loss": 0.0,
      "step": 55720
    },
    {
      "epoch": 0.38448536361566643,
      "grad_norm": 0.0,
      "learning_rate": 6.157009113676033e-06,
      "loss": 0.0,
      "step": 55730
    },
    {
      "epoch": 0.3845543543502108,
      "grad_norm": 5.987767053738935e-06,
      "learning_rate": 6.15631920633059e-06,
      "loss": 0.0,
      "step": 55740
    },
    {
      "epoch": 0.3846233450847551,
      "grad_norm": 2.0352820229163626e-06,
      "learning_rate": 6.155629298985146e-06,
      "loss": 0.0,
      "step": 55750
    },
    {
      "epoch": 0.38469233581929946,
      "grad_norm": 0.0,
      "learning_rate": 6.154939391639703e-06,
      "loss": 0.0,
      "step": 55760
    },
    {
      "epoch": 0.3847613265538438,
      "grad_norm": 0.0,
      "learning_rate": 6.15424948429426e-06,
      "loss": 0.0,
      "step": 55770
    },
    {
      "epoch": 0.3848303172883882,
      "grad_norm": 0.0,
      "learning_rate": 6.153559576948816e-06,
      "loss": 0.0,
      "step": 55780
    },
    {
      "epoch": 0.38489930802293254,
      "grad_norm": 0.0026983499992638826,
      "learning_rate": 6.152869669603373e-06,
      "loss": 0.1418,
      "step": 55790
    },
    {
      "epoch": 0.38496829875747685,
      "grad_norm": 0.0,
      "learning_rate": 6.152179762257929e-06,
      "loss": 0.0,
      "step": 55800
    },
    {
      "epoch": 0.3850372894920212,
      "grad_norm": 0.0,
      "learning_rate": 6.151489854912486e-06,
      "loss": 0.0,
      "step": 55810
    },
    {
      "epoch": 0.38510628022656557,
      "grad_norm": 168.53286743164062,
      "learning_rate": 6.150799947567042e-06,
      "loss": 0.0515,
      "step": 55820
    },
    {
      "epoch": 0.38517527096110993,
      "grad_norm": 0.13199575245380402,
      "learning_rate": 6.150110040221599e-06,
      "loss": 0.0,
      "step": 55830
    },
    {
      "epoch": 0.3852442616956543,
      "grad_norm": 0.0,
      "learning_rate": 6.149420132876155e-06,
      "loss": 0.0,
      "step": 55840
    },
    {
      "epoch": 0.38531325243019865,
      "grad_norm": 0.0,
      "learning_rate": 6.148730225530711e-06,
      "loss": 0.0,
      "step": 55850
    },
    {
      "epoch": 0.38538224316474295,
      "grad_norm": 0.0,
      "learning_rate": 6.1480403181852675e-06,
      "loss": 0.0,
      "step": 55860
    },
    {
      "epoch": 0.3854512338992873,
      "grad_norm": 0.0,
      "learning_rate": 6.147350410839825e-06,
      "loss": 0.0,
      "step": 55870
    },
    {
      "epoch": 0.3855202246338317,
      "grad_norm": 0.0,
      "learning_rate": 6.146660503494381e-06,
      "loss": 0.0086,
      "step": 55880
    },
    {
      "epoch": 0.38558921536837604,
      "grad_norm": 0.0,
      "learning_rate": 6.145970596148938e-06,
      "loss": 0.0,
      "step": 55890
    },
    {
      "epoch": 0.3856582061029204,
      "grad_norm": 0.0,
      "learning_rate": 6.145280688803494e-06,
      "loss": 0.0,
      "step": 55900
    },
    {
      "epoch": 0.3857271968374647,
      "grad_norm": 0.0,
      "learning_rate": 6.144590781458051e-06,
      "loss": 0.0235,
      "step": 55910
    },
    {
      "epoch": 0.38579618757200906,
      "grad_norm": 0.0,
      "learning_rate": 6.143900874112607e-06,
      "loss": 0.0011,
      "step": 55920
    },
    {
      "epoch": 0.3858651783065534,
      "grad_norm": 0.0,
      "learning_rate": 6.143210966767164e-06,
      "loss": 0.0002,
      "step": 55930
    },
    {
      "epoch": 0.3859341690410978,
      "grad_norm": 8.098172838799655e-05,
      "learning_rate": 6.14252105942172e-06,
      "loss": 0.0,
      "step": 55940
    },
    {
      "epoch": 0.38600315977564215,
      "grad_norm": 0.0,
      "learning_rate": 6.141831152076277e-06,
      "loss": 0.0,
      "step": 55950
    },
    {
      "epoch": 0.3860721505101865,
      "grad_norm": 0.0,
      "learning_rate": 6.141141244730832e-06,
      "loss": 0.0013,
      "step": 55960
    },
    {
      "epoch": 0.3861411412447308,
      "grad_norm": 0.0028657943475991488,
      "learning_rate": 6.140451337385389e-06,
      "loss": 0.0,
      "step": 55970
    },
    {
      "epoch": 0.3862101319792752,
      "grad_norm": 0.0,
      "learning_rate": 6.139761430039946e-06,
      "loss": 0.0024,
      "step": 55980
    },
    {
      "epoch": 0.38627912271381953,
      "grad_norm": 0.0,
      "learning_rate": 6.139071522694503e-06,
      "loss": 0.0001,
      "step": 55990
    },
    {
      "epoch": 0.3863481134483639,
      "grad_norm": 0.0,
      "learning_rate": 6.138381615349059e-06,
      "loss": 0.0,
      "step": 56000
    },
    {
      "epoch": 0.38641710418290826,
      "grad_norm": 0.0,
      "learning_rate": 6.137691708003616e-06,
      "loss": 0.0,
      "step": 56010
    },
    {
      "epoch": 0.38648609491745256,
      "grad_norm": 0.0,
      "learning_rate": 6.137001800658172e-06,
      "loss": 0.0,
      "step": 56020
    },
    {
      "epoch": 0.3865550856519969,
      "grad_norm": 0.0,
      "learning_rate": 6.136311893312729e-06,
      "loss": 0.0,
      "step": 56030
    },
    {
      "epoch": 0.3866240763865413,
      "grad_norm": 0.0,
      "learning_rate": 6.135621985967285e-06,
      "loss": 0.0,
      "step": 56040
    },
    {
      "epoch": 0.38669306712108564,
      "grad_norm": 0.0,
      "learning_rate": 6.134932078621842e-06,
      "loss": 0.0002,
      "step": 56050
    },
    {
      "epoch": 0.38676205785563,
      "grad_norm": 2.4157950306857856e-08,
      "learning_rate": 6.134242171276398e-06,
      "loss": 0.0105,
      "step": 56060
    },
    {
      "epoch": 0.38683104859017436,
      "grad_norm": 0.0,
      "learning_rate": 6.133552263930954e-06,
      "loss": 0.0,
      "step": 56070
    },
    {
      "epoch": 0.38690003932471867,
      "grad_norm": 0.0,
      "learning_rate": 6.13286235658551e-06,
      "loss": 0.0,
      "step": 56080
    },
    {
      "epoch": 0.38696903005926303,
      "grad_norm": 0.0,
      "learning_rate": 6.1321724492400684e-06,
      "loss": 0.0004,
      "step": 56090
    },
    {
      "epoch": 0.3870380207938074,
      "grad_norm": 0.0,
      "learning_rate": 6.131482541894624e-06,
      "loss": 0.0,
      "step": 56100
    },
    {
      "epoch": 0.38710701152835175,
      "grad_norm": 8.23408699943684e-05,
      "learning_rate": 6.1307926345491806e-06,
      "loss": 0.0,
      "step": 56110
    },
    {
      "epoch": 0.3871760022628961,
      "grad_norm": 0.0,
      "learning_rate": 6.130102727203737e-06,
      "loss": 0.0008,
      "step": 56120
    },
    {
      "epoch": 0.3872449929974404,
      "grad_norm": 0.0,
      "learning_rate": 6.1294128198582936e-06,
      "loss": 0.0,
      "step": 56130
    },
    {
      "epoch": 0.3873139837319848,
      "grad_norm": 0.0,
      "learning_rate": 6.12872291251285e-06,
      "loss": 0.0,
      "step": 56140
    },
    {
      "epoch": 0.38738297446652914,
      "grad_norm": 0.0,
      "learning_rate": 6.1280330051674065e-06,
      "loss": 0.0,
      "step": 56150
    },
    {
      "epoch": 0.3874519652010735,
      "grad_norm": 0.0,
      "learning_rate": 6.127343097821963e-06,
      "loss": 0.0,
      "step": 56160
    },
    {
      "epoch": 0.38752095593561786,
      "grad_norm": 1.6700771084288135e-05,
      "learning_rate": 6.1266531904765195e-06,
      "loss": 0.0,
      "step": 56170
    },
    {
      "epoch": 0.3875899466701622,
      "grad_norm": 0.0,
      "learning_rate": 6.125963283131075e-06,
      "loss": 0.0,
      "step": 56180
    },
    {
      "epoch": 0.3876589374047065,
      "grad_norm": 0.0,
      "learning_rate": 6.125273375785632e-06,
      "loss": 0.0,
      "step": 56190
    },
    {
      "epoch": 0.3877279281392509,
      "grad_norm": 0.0,
      "learning_rate": 6.12458346844019e-06,
      "loss": 0.0022,
      "step": 56200
    },
    {
      "epoch": 0.38779691887379525,
      "grad_norm": 0.0,
      "learning_rate": 6.1238935610947455e-06,
      "loss": 0.0,
      "step": 56210
    },
    {
      "epoch": 0.3878659096083396,
      "grad_norm": 0.0,
      "learning_rate": 6.123203653749302e-06,
      "loss": 0.0007,
      "step": 56220
    },
    {
      "epoch": 0.38793490034288397,
      "grad_norm": 1.635532259941101,
      "learning_rate": 6.1225137464038585e-06,
      "loss": 0.0003,
      "step": 56230
    },
    {
      "epoch": 0.3880038910774283,
      "grad_norm": 0.0,
      "learning_rate": 6.121823839058415e-06,
      "loss": 0.0,
      "step": 56240
    },
    {
      "epoch": 0.38807288181197264,
      "grad_norm": 1.7715757749670047e-09,
      "learning_rate": 6.1211339317129714e-06,
      "loss": 0.0,
      "step": 56250
    },
    {
      "epoch": 0.388141872546517,
      "grad_norm": 0.0,
      "learning_rate": 6.120444024367528e-06,
      "loss": 0.0,
      "step": 56260
    },
    {
      "epoch": 0.38821086328106136,
      "grad_norm": 0.0,
      "learning_rate": 6.119754117022084e-06,
      "loss": 0.0,
      "step": 56270
    },
    {
      "epoch": 0.3882798540156057,
      "grad_norm": 0.0,
      "learning_rate": 6.119064209676641e-06,
      "loss": 0.0,
      "step": 56280
    },
    {
      "epoch": 0.3883488447501501,
      "grad_norm": 0.06376428157091141,
      "learning_rate": 6.118374302331197e-06,
      "loss": 0.0,
      "step": 56290
    },
    {
      "epoch": 0.3884178354846944,
      "grad_norm": 6.970270987949334e-06,
      "learning_rate": 6.117684394985753e-06,
      "loss": 0.0001,
      "step": 56300
    },
    {
      "epoch": 0.38848682621923875,
      "grad_norm": 0.0,
      "learning_rate": 6.116994487640311e-06,
      "loss": 0.0,
      "step": 56310
    },
    {
      "epoch": 0.3885558169537831,
      "grad_norm": 1.4182120561599731,
      "learning_rate": 6.116304580294867e-06,
      "loss": 0.0002,
      "step": 56320
    },
    {
      "epoch": 0.38862480768832747,
      "grad_norm": 0.0004731070948764682,
      "learning_rate": 6.115614672949423e-06,
      "loss": 0.3451,
      "step": 56330
    },
    {
      "epoch": 0.38869379842287183,
      "grad_norm": 9.716685234195666e-10,
      "learning_rate": 6.11492476560398e-06,
      "loss": 0.0,
      "step": 56340
    },
    {
      "epoch": 0.38876278915741613,
      "grad_norm": 0.0,
      "learning_rate": 6.114234858258536e-06,
      "loss": 0.0,
      "step": 56350
    },
    {
      "epoch": 0.3888317798919605,
      "grad_norm": 0.0,
      "learning_rate": 6.113544950913093e-06,
      "loss": 0.0,
      "step": 56360
    },
    {
      "epoch": 0.38890077062650485,
      "grad_norm": 0.0,
      "learning_rate": 6.112855043567649e-06,
      "loss": 0.0001,
      "step": 56370
    },
    {
      "epoch": 0.3889697613610492,
      "grad_norm": 4.92419758302276e-08,
      "learning_rate": 6.112165136222206e-06,
      "loss": 0.2758,
      "step": 56380
    },
    {
      "epoch": 0.3890387520955936,
      "grad_norm": 0.0,
      "learning_rate": 6.111475228876762e-06,
      "loss": 0.0013,
      "step": 56390
    },
    {
      "epoch": 0.38910774283013794,
      "grad_norm": 0.00025528090191073716,
      "learning_rate": 6.110785321531319e-06,
      "loss": 0.0,
      "step": 56400
    },
    {
      "epoch": 0.38917673356468224,
      "grad_norm": 0.0,
      "learning_rate": 6.1100954141858744e-06,
      "loss": 0.0,
      "step": 56410
    },
    {
      "epoch": 0.3892457242992266,
      "grad_norm": 0.0,
      "learning_rate": 6.109405506840433e-06,
      "loss": 0.0,
      "step": 56420
    },
    {
      "epoch": 0.38931471503377096,
      "grad_norm": 0.0,
      "learning_rate": 6.108715599494989e-06,
      "loss": 0.0,
      "step": 56430
    },
    {
      "epoch": 0.3893837057683153,
      "grad_norm": 0.002434544963762164,
      "learning_rate": 6.108025692149545e-06,
      "loss": 0.0,
      "step": 56440
    },
    {
      "epoch": 0.3894526965028597,
      "grad_norm": 1.8608644625217607e-09,
      "learning_rate": 6.107335784804101e-06,
      "loss": 0.0003,
      "step": 56450
    },
    {
      "epoch": 0.389521687237404,
      "grad_norm": 0.0,
      "learning_rate": 6.106645877458658e-06,
      "loss": 0.0,
      "step": 56460
    },
    {
      "epoch": 0.38959067797194835,
      "grad_norm": 4.590802049619924e-08,
      "learning_rate": 6.105955970113214e-06,
      "loss": 0.0,
      "step": 56470
    },
    {
      "epoch": 0.3896596687064927,
      "grad_norm": 0.0,
      "learning_rate": 6.105266062767771e-06,
      "loss": 0.0,
      "step": 56480
    },
    {
      "epoch": 0.3897286594410371,
      "grad_norm": 0.0,
      "learning_rate": 6.104576155422327e-06,
      "loss": 0.0,
      "step": 56490
    },
    {
      "epoch": 0.38979765017558143,
      "grad_norm": 0.624225914478302,
      "learning_rate": 6.103886248076884e-06,
      "loss": 0.0048,
      "step": 56500
    },
    {
      "epoch": 0.3898666409101258,
      "grad_norm": 0.0,
      "learning_rate": 6.10319634073144e-06,
      "loss": 0.0056,
      "step": 56510
    },
    {
      "epoch": 0.3899356316446701,
      "grad_norm": 0.0,
      "learning_rate": 6.102506433385996e-06,
      "loss": 0.0,
      "step": 56520
    },
    {
      "epoch": 0.39000462237921446,
      "grad_norm": 0.0,
      "learning_rate": 6.101816526040554e-06,
      "loss": 0.0015,
      "step": 56530
    },
    {
      "epoch": 0.3900736131137588,
      "grad_norm": 0.0,
      "learning_rate": 6.1011266186951105e-06,
      "loss": 0.0,
      "step": 56540
    },
    {
      "epoch": 0.3901426038483032,
      "grad_norm": 4.329888392273773e-10,
      "learning_rate": 6.100436711349666e-06,
      "loss": 0.0,
      "step": 56550
    },
    {
      "epoch": 0.39021159458284754,
      "grad_norm": 0.11823488026857376,
      "learning_rate": 6.099746804004223e-06,
      "loss": 0.0,
      "step": 56560
    },
    {
      "epoch": 0.39028058531739185,
      "grad_norm": 0.0,
      "learning_rate": 6.099056896658779e-06,
      "loss": 0.0,
      "step": 56570
    },
    {
      "epoch": 0.3903495760519362,
      "grad_norm": 0.0019926249515265226,
      "learning_rate": 6.098366989313336e-06,
      "loss": 0.0,
      "step": 56580
    },
    {
      "epoch": 0.39041856678648057,
      "grad_norm": 0.0,
      "learning_rate": 6.097677081967892e-06,
      "loss": 0.0,
      "step": 56590
    },
    {
      "epoch": 0.39048755752102493,
      "grad_norm": 0.0,
      "learning_rate": 6.096987174622449e-06,
      "loss": 0.0,
      "step": 56600
    },
    {
      "epoch": 0.3905565482555693,
      "grad_norm": 0.0,
      "learning_rate": 6.096297267277005e-06,
      "loss": 0.0,
      "step": 56610
    },
    {
      "epoch": 0.39062553899011365,
      "grad_norm": 0.0,
      "learning_rate": 6.095607359931562e-06,
      "loss": 0.0,
      "step": 56620
    },
    {
      "epoch": 0.39069452972465796,
      "grad_norm": 0.0,
      "learning_rate": 6.094917452586118e-06,
      "loss": 0.0004,
      "step": 56630
    },
    {
      "epoch": 0.3907635204592023,
      "grad_norm": 0.0,
      "learning_rate": 6.094227545240675e-06,
      "loss": 0.0008,
      "step": 56640
    },
    {
      "epoch": 0.3908325111937467,
      "grad_norm": 0.0,
      "learning_rate": 6.093537637895232e-06,
      "loss": 0.0,
      "step": 56650
    },
    {
      "epoch": 0.39090150192829104,
      "grad_norm": 0.0,
      "learning_rate": 6.0928477305497875e-06,
      "loss": 0.0,
      "step": 56660
    },
    {
      "epoch": 0.3909704926628354,
      "grad_norm": 0.0,
      "learning_rate": 6.092157823204344e-06,
      "loss": 0.0,
      "step": 56670
    },
    {
      "epoch": 0.3910394833973797,
      "grad_norm": 0.07848416268825531,
      "learning_rate": 6.0914679158589005e-06,
      "loss": 0.0,
      "step": 56680
    },
    {
      "epoch": 0.39110847413192407,
      "grad_norm": 0.0,
      "learning_rate": 6.090778008513457e-06,
      "loss": 0.0,
      "step": 56690
    },
    {
      "epoch": 0.3911774648664684,
      "grad_norm": 0.0,
      "learning_rate": 6.0900881011680135e-06,
      "loss": 0.0,
      "step": 56700
    },
    {
      "epoch": 0.3912464556010128,
      "grad_norm": 4.518360963157164e-10,
      "learning_rate": 6.08939819382257e-06,
      "loss": 0.0,
      "step": 56710
    },
    {
      "epoch": 0.39131544633555715,
      "grad_norm": 4.793644237111039e-08,
      "learning_rate": 6.0887082864771265e-06,
      "loss": 0.0372,
      "step": 56720
    },
    {
      "epoch": 0.3913844370701015,
      "grad_norm": 0.0,
      "learning_rate": 6.088018379131683e-06,
      "loss": 0.0,
      "step": 56730
    },
    {
      "epoch": 0.3914534278046458,
      "grad_norm": 0.0,
      "learning_rate": 6.0873284717862395e-06,
      "loss": 0.0,
      "step": 56740
    },
    {
      "epoch": 0.3915224185391902,
      "grad_norm": 0.0,
      "learning_rate": 6.086638564440797e-06,
      "loss": 0.0,
      "step": 56750
    },
    {
      "epoch": 0.39159140927373454,
      "grad_norm": 0.0,
      "learning_rate": 6.085948657095353e-06,
      "loss": 0.0,
      "step": 56760
    },
    {
      "epoch": 0.3916604000082789,
      "grad_norm": 0.0,
      "learning_rate": 6.08525874974991e-06,
      "loss": 0.0001,
      "step": 56770
    },
    {
      "epoch": 0.39172939074282326,
      "grad_norm": 0.0,
      "learning_rate": 6.0845688424044654e-06,
      "loss": 0.0,
      "step": 56780
    },
    {
      "epoch": 0.39179838147736756,
      "grad_norm": 3.010267164427205e-07,
      "learning_rate": 6.083878935059022e-06,
      "loss": 0.0,
      "step": 56790
    },
    {
      "epoch": 0.3918673722119119,
      "grad_norm": 0.0,
      "learning_rate": 6.083189027713578e-06,
      "loss": 0.0,
      "step": 56800
    },
    {
      "epoch": 0.3919363629464563,
      "grad_norm": 0.0,
      "learning_rate": 6.082499120368135e-06,
      "loss": 0.0,
      "step": 56810
    },
    {
      "epoch": 0.39200535368100065,
      "grad_norm": 0.0,
      "learning_rate": 6.081809213022691e-06,
      "loss": 0.0,
      "step": 56820
    },
    {
      "epoch": 0.392074344415545,
      "grad_norm": 0.0,
      "learning_rate": 6.081119305677248e-06,
      "loss": 0.0,
      "step": 56830
    },
    {
      "epoch": 0.39214333515008937,
      "grad_norm": 9.613947622710839e-05,
      "learning_rate": 6.080429398331804e-06,
      "loss": 0.0,
      "step": 56840
    },
    {
      "epoch": 0.3922123258846337,
      "grad_norm": 0.0,
      "learning_rate": 6.079739490986361e-06,
      "loss": 0.0,
      "step": 56850
    },
    {
      "epoch": 0.39228131661917803,
      "grad_norm": 0.0,
      "learning_rate": 6.079049583640918e-06,
      "loss": 0.0,
      "step": 56860
    },
    {
      "epoch": 0.3923503073537224,
      "grad_norm": 0.0,
      "learning_rate": 6.078359676295475e-06,
      "loss": 0.0,
      "step": 56870
    },
    {
      "epoch": 0.39241929808826675,
      "grad_norm": 0.0,
      "learning_rate": 6.077669768950031e-06,
      "loss": 0.0,
      "step": 56880
    },
    {
      "epoch": 0.3924882888228111,
      "grad_norm": 0.0,
      "learning_rate": 6.076979861604587e-06,
      "loss": 0.0,
      "step": 56890
    },
    {
      "epoch": 0.3925572795573554,
      "grad_norm": 0.0,
      "learning_rate": 6.076289954259143e-06,
      "loss": 0.1911,
      "step": 56900
    },
    {
      "epoch": 0.3926262702918998,
      "grad_norm": 2.7228872776031494,
      "learning_rate": 6.0756000469137e-06,
      "loss": 0.0003,
      "step": 56910
    },
    {
      "epoch": 0.39269526102644414,
      "grad_norm": 0.0,
      "learning_rate": 6.074910139568256e-06,
      "loss": 0.0,
      "step": 56920
    },
    {
      "epoch": 0.3927642517609885,
      "grad_norm": 0.0,
      "learning_rate": 6.074220232222813e-06,
      "loss": 0.0,
      "step": 56930
    },
    {
      "epoch": 0.39283324249553286,
      "grad_norm": 0.0,
      "learning_rate": 6.073530324877369e-06,
      "loss": 0.0002,
      "step": 56940
    },
    {
      "epoch": 0.3929022332300772,
      "grad_norm": 4.683251009218736e-10,
      "learning_rate": 6.072840417531926e-06,
      "loss": 0.0,
      "step": 56950
    },
    {
      "epoch": 0.39297122396462153,
      "grad_norm": 0.0,
      "learning_rate": 6.072150510186482e-06,
      "loss": 0.0001,
      "step": 56960
    },
    {
      "epoch": 0.3930402146991659,
      "grad_norm": 0.0,
      "learning_rate": 6.07146060284104e-06,
      "loss": 0.0,
      "step": 56970
    },
    {
      "epoch": 0.39310920543371025,
      "grad_norm": 0.0,
      "learning_rate": 6.070770695495596e-06,
      "loss": 0.0,
      "step": 56980
    },
    {
      "epoch": 0.3931781961682546,
      "grad_norm": 235.7193603515625,
      "learning_rate": 6.0700807881501526e-06,
      "loss": 0.0758,
      "step": 56990
    },
    {
      "epoch": 0.393247186902799,
      "grad_norm": 0.0004806480719707906,
      "learning_rate": 6.069390880804708e-06,
      "loss": 0.0,
      "step": 57000
    },
    {
      "epoch": 0.3933161776373433,
      "grad_norm": 0.00013156459317542613,
      "learning_rate": 6.068700973459265e-06,
      "loss": 0.0009,
      "step": 57010
    },
    {
      "epoch": 0.39338516837188764,
      "grad_norm": 0.0,
      "learning_rate": 6.068011066113821e-06,
      "loss": 0.0,
      "step": 57020
    },
    {
      "epoch": 0.393454159106432,
      "grad_norm": 0.0,
      "learning_rate": 6.067321158768378e-06,
      "loss": 0.0,
      "step": 57030
    },
    {
      "epoch": 0.39352314984097636,
      "grad_norm": 0.0,
      "learning_rate": 6.066631251422934e-06,
      "loss": 0.0,
      "step": 57040
    },
    {
      "epoch": 0.3935921405755207,
      "grad_norm": 0.0,
      "learning_rate": 6.065941344077491e-06,
      "loss": 0.0,
      "step": 57050
    },
    {
      "epoch": 0.3936611313100651,
      "grad_norm": 1.2601099014282227,
      "learning_rate": 6.065251436732047e-06,
      "loss": 0.0002,
      "step": 57060
    },
    {
      "epoch": 0.3937301220446094,
      "grad_norm": 0.0,
      "learning_rate": 6.064561529386604e-06,
      "loss": 0.0,
      "step": 57070
    },
    {
      "epoch": 0.39379911277915375,
      "grad_norm": 0.0,
      "learning_rate": 6.063871622041161e-06,
      "loss": 0.0,
      "step": 57080
    },
    {
      "epoch": 0.3938681035136981,
      "grad_norm": 0.0006193273002281785,
      "learning_rate": 6.0631817146957175e-06,
      "loss": 0.0052,
      "step": 57090
    },
    {
      "epoch": 0.39393709424824247,
      "grad_norm": 4.476706783052009e-10,
      "learning_rate": 6.062491807350274e-06,
      "loss": 0.0,
      "step": 57100
    },
    {
      "epoch": 0.39400608498278683,
      "grad_norm": 0.0,
      "learning_rate": 6.0618019000048305e-06,
      "loss": 0.0013,
      "step": 57110
    },
    {
      "epoch": 0.39407507571733114,
      "grad_norm": 0.0,
      "learning_rate": 6.061111992659386e-06,
      "loss": 0.0,
      "step": 57120
    },
    {
      "epoch": 0.3941440664518755,
      "grad_norm": 0.0,
      "learning_rate": 6.060422085313943e-06,
      "loss": 0.0,
      "step": 57130
    },
    {
      "epoch": 0.39421305718641986,
      "grad_norm": 0.0,
      "learning_rate": 6.059732177968499e-06,
      "loss": 0.0001,
      "step": 57140
    },
    {
      "epoch": 0.3942820479209642,
      "grad_norm": 0.0,
      "learning_rate": 6.0590422706230556e-06,
      "loss": 0.0005,
      "step": 57150
    },
    {
      "epoch": 0.3943510386555086,
      "grad_norm": 4.48142495335091e-10,
      "learning_rate": 6.058352363277612e-06,
      "loss": 0.0,
      "step": 57160
    },
    {
      "epoch": 0.39442002939005294,
      "grad_norm": 0.0,
      "learning_rate": 6.0576624559321686e-06,
      "loss": 0.0,
      "step": 57170
    },
    {
      "epoch": 0.39448902012459724,
      "grad_norm": 0.0,
      "learning_rate": 6.056972548586725e-06,
      "loss": 0.0,
      "step": 57180
    },
    {
      "epoch": 0.3945580108591416,
      "grad_norm": 0.0,
      "learning_rate": 6.056282641241282e-06,
      "loss": 0.0043,
      "step": 57190
    },
    {
      "epoch": 0.39462700159368597,
      "grad_norm": 0.0001748184149619192,
      "learning_rate": 6.055592733895839e-06,
      "loss": 0.0726,
      "step": 57200
    },
    {
      "epoch": 0.3946959923282303,
      "grad_norm": 1.9876635633409023e-05,
      "learning_rate": 6.054902826550395e-06,
      "loss": 0.0,
      "step": 57210
    },
    {
      "epoch": 0.3947649830627747,
      "grad_norm": 1.5168774325502454e-07,
      "learning_rate": 6.054212919204952e-06,
      "loss": 0.0,
      "step": 57220
    },
    {
      "epoch": 0.394833973797319,
      "grad_norm": 2.2972858459979761e-07,
      "learning_rate": 6.0535230118595075e-06,
      "loss": 0.0,
      "step": 57230
    },
    {
      "epoch": 0.39490296453186335,
      "grad_norm": 0.0,
      "learning_rate": 6.052833104514064e-06,
      "loss": 0.0,
      "step": 57240
    },
    {
      "epoch": 0.3949719552664077,
      "grad_norm": 0.0,
      "learning_rate": 6.0521431971686205e-06,
      "loss": 0.1383,
      "step": 57250
    },
    {
      "epoch": 0.3950409460009521,
      "grad_norm": 0.0010140405502170324,
      "learning_rate": 6.051453289823177e-06,
      "loss": 0.0,
      "step": 57260
    },
    {
      "epoch": 0.39510993673549644,
      "grad_norm": 0.0,
      "learning_rate": 6.0507633824777335e-06,
      "loss": 0.0,
      "step": 57270
    },
    {
      "epoch": 0.3951789274700408,
      "grad_norm": 0.0,
      "learning_rate": 6.05007347513229e-06,
      "loss": 0.0,
      "step": 57280
    },
    {
      "epoch": 0.3952479182045851,
      "grad_norm": 2.835417012292396e-09,
      "learning_rate": 6.0493835677868464e-06,
      "loss": 0.0,
      "step": 57290
    },
    {
      "epoch": 0.39531690893912946,
      "grad_norm": 0.011000720784068108,
      "learning_rate": 6.048693660441404e-06,
      "loss": 0.0,
      "step": 57300
    },
    {
      "epoch": 0.3953858996736738,
      "grad_norm": 0.0049164132215082645,
      "learning_rate": 6.04800375309596e-06,
      "loss": 0.0,
      "step": 57310
    },
    {
      "epoch": 0.3954548904082182,
      "grad_norm": 0.0,
      "learning_rate": 6.047313845750517e-06,
      "loss": 0.0,
      "step": 57320
    },
    {
      "epoch": 0.39552388114276255,
      "grad_norm": 0.0,
      "learning_rate": 6.046623938405073e-06,
      "loss": 0.0839,
      "step": 57330
    },
    {
      "epoch": 0.39559287187730685,
      "grad_norm": 0.0,
      "learning_rate": 6.045934031059629e-06,
      "loss": 0.0001,
      "step": 57340
    },
    {
      "epoch": 0.3956618626118512,
      "grad_norm": 0.0,
      "learning_rate": 6.045244123714185e-06,
      "loss": 0.0,
      "step": 57350
    },
    {
      "epoch": 0.3957308533463956,
      "grad_norm": 0.0,
      "learning_rate": 6.044554216368742e-06,
      "loss": 0.0,
      "step": 57360
    },
    {
      "epoch": 0.39579984408093993,
      "grad_norm": 0.0,
      "learning_rate": 6.043864309023298e-06,
      "loss": 0.0,
      "step": 57370
    },
    {
      "epoch": 0.3958688348154843,
      "grad_norm": 0.0,
      "learning_rate": 6.043174401677855e-06,
      "loss": 0.0,
      "step": 57380
    },
    {
      "epoch": 0.39593782555002865,
      "grad_norm": 0.0,
      "learning_rate": 6.042484494332411e-06,
      "loss": 0.0614,
      "step": 57390
    },
    {
      "epoch": 0.39600681628457296,
      "grad_norm": 0.0,
      "learning_rate": 6.041794586986968e-06,
      "loss": 0.0,
      "step": 57400
    },
    {
      "epoch": 0.3960758070191173,
      "grad_norm": 0.0,
      "learning_rate": 6.041104679641525e-06,
      "loss": 0.0,
      "step": 57410
    },
    {
      "epoch": 0.3961447977536617,
      "grad_norm": 0.0,
      "learning_rate": 6.040414772296082e-06,
      "loss": 0.0,
      "step": 57420
    },
    {
      "epoch": 0.39621378848820604,
      "grad_norm": 0.0,
      "learning_rate": 6.039724864950638e-06,
      "loss": 0.0,
      "step": 57430
    },
    {
      "epoch": 0.3962827792227504,
      "grad_norm": 0.0,
      "learning_rate": 6.039034957605195e-06,
      "loss": 0.0,
      "step": 57440
    },
    {
      "epoch": 0.3963517699572947,
      "grad_norm": 0.0,
      "learning_rate": 6.038345050259751e-06,
      "loss": 0.0285,
      "step": 57450
    },
    {
      "epoch": 0.39642076069183907,
      "grad_norm": 0.0,
      "learning_rate": 6.037655142914307e-06,
      "loss": 0.0,
      "step": 57460
    },
    {
      "epoch": 0.39648975142638343,
      "grad_norm": 0.0,
      "learning_rate": 6.036965235568863e-06,
      "loss": 0.0,
      "step": 57470
    },
    {
      "epoch": 0.3965587421609278,
      "grad_norm": 0.0,
      "learning_rate": 6.03627532822342e-06,
      "loss": 0.0,
      "step": 57480
    },
    {
      "epoch": 0.39662773289547215,
      "grad_norm": 0.0,
      "learning_rate": 6.035585420877976e-06,
      "loss": 0.0,
      "step": 57490
    },
    {
      "epoch": 0.3966967236300165,
      "grad_norm": 0.0,
      "learning_rate": 6.034895513532533e-06,
      "loss": 0.1054,
      "step": 57500
    },
    {
      "epoch": 0.3967657143645608,
      "grad_norm": 0.0,
      "learning_rate": 6.034205606187089e-06,
      "loss": 0.0018,
      "step": 57510
    },
    {
      "epoch": 0.3968347050991052,
      "grad_norm": 0.0,
      "learning_rate": 6.0335156988416466e-06,
      "loss": 0.0018,
      "step": 57520
    },
    {
      "epoch": 0.39690369583364954,
      "grad_norm": 0.0,
      "learning_rate": 6.032825791496203e-06,
      "loss": 0.0004,
      "step": 57530
    },
    {
      "epoch": 0.3969726865681939,
      "grad_norm": 0.0,
      "learning_rate": 6.0321358841507595e-06,
      "loss": 0.0,
      "step": 57540
    },
    {
      "epoch": 0.39704167730273826,
      "grad_norm": 0.001273843110539019,
      "learning_rate": 6.031445976805316e-06,
      "loss": 0.0,
      "step": 57550
    },
    {
      "epoch": 0.39711066803728257,
      "grad_norm": 3.107517443368124e-08,
      "learning_rate": 6.0307560694598725e-06,
      "loss": 0.0,
      "step": 57560
    },
    {
      "epoch": 0.3971796587718269,
      "grad_norm": 0.0,
      "learning_rate": 6.030066162114428e-06,
      "loss": 0.0,
      "step": 57570
    },
    {
      "epoch": 0.3972486495063713,
      "grad_norm": 0.0,
      "learning_rate": 6.029376254768985e-06,
      "loss": 0.0057,
      "step": 57580
    },
    {
      "epoch": 0.39731764024091565,
      "grad_norm": 0.0,
      "learning_rate": 6.028686347423541e-06,
      "loss": 0.0001,
      "step": 57590
    },
    {
      "epoch": 0.39738663097546,
      "grad_norm": 0.0,
      "learning_rate": 6.027996440078098e-06,
      "loss": 0.0001,
      "step": 57600
    },
    {
      "epoch": 0.39745562171000437,
      "grad_norm": 29.133642196655273,
      "learning_rate": 6.027306532732654e-06,
      "loss": 0.0071,
      "step": 57610
    },
    {
      "epoch": 0.3975246124445487,
      "grad_norm": 0.0,
      "learning_rate": 6.026616625387211e-06,
      "loss": 0.0,
      "step": 57620
    },
    {
      "epoch": 0.39759360317909304,
      "grad_norm": 0.0,
      "learning_rate": 6.025926718041768e-06,
      "loss": 0.0002,
      "step": 57630
    },
    {
      "epoch": 0.3976625939136374,
      "grad_norm": 0.0,
      "learning_rate": 6.0252368106963244e-06,
      "loss": 0.0001,
      "step": 57640
    },
    {
      "epoch": 0.39773158464818176,
      "grad_norm": 1.2381713531794958e-05,
      "learning_rate": 6.024546903350881e-06,
      "loss": 0.0,
      "step": 57650
    },
    {
      "epoch": 0.3978005753827261,
      "grad_norm": 0.0,
      "learning_rate": 6.0238569960054374e-06,
      "loss": 0.0,
      "step": 57660
    },
    {
      "epoch": 0.3978695661172704,
      "grad_norm": 0.0,
      "learning_rate": 6.023167088659994e-06,
      "loss": 0.0,
      "step": 57670
    },
    {
      "epoch": 0.3979385568518148,
      "grad_norm": 8.026274755934537e-09,
      "learning_rate": 6.0224771813145496e-06,
      "loss": 0.0002,
      "step": 57680
    },
    {
      "epoch": 0.39800754758635914,
      "grad_norm": 0.0,
      "learning_rate": 6.021787273969106e-06,
      "loss": 0.0,
      "step": 57690
    },
    {
      "epoch": 0.3980765383209035,
      "grad_norm": 0.0,
      "learning_rate": 6.0210973666236625e-06,
      "loss": 0.0,
      "step": 57700
    },
    {
      "epoch": 0.39814552905544787,
      "grad_norm": 0.009298167191445827,
      "learning_rate": 6.020407459278219e-06,
      "loss": 0.0,
      "step": 57710
    },
    {
      "epoch": 0.3982145197899922,
      "grad_norm": 1.642198913032189e-05,
      "learning_rate": 6.0197175519327755e-06,
      "loss": 0.0,
      "step": 57720
    },
    {
      "epoch": 0.39828351052453653,
      "grad_norm": 0.0,
      "learning_rate": 6.019027644587332e-06,
      "loss": 0.0,
      "step": 57730
    },
    {
      "epoch": 0.3983525012590809,
      "grad_norm": 0.0,
      "learning_rate": 6.018337737241889e-06,
      "loss": 0.0,
      "step": 57740
    },
    {
      "epoch": 0.39842149199362525,
      "grad_norm": 0.0,
      "learning_rate": 6.017647829896446e-06,
      "loss": 0.0,
      "step": 57750
    },
    {
      "epoch": 0.3984904827281696,
      "grad_norm": 0.0,
      "learning_rate": 6.016957922551002e-06,
      "loss": 0.0,
      "step": 57760
    },
    {
      "epoch": 0.398559473462714,
      "grad_norm": 0.02610335871577263,
      "learning_rate": 6.016268015205559e-06,
      "loss": 0.0,
      "step": 57770
    },
    {
      "epoch": 0.3986284641972583,
      "grad_norm": 0.0,
      "learning_rate": 6.015578107860115e-06,
      "loss": 0.0,
      "step": 57780
    },
    {
      "epoch": 0.39869745493180264,
      "grad_norm": 0.0,
      "learning_rate": 6.014888200514671e-06,
      "loss": 0.0,
      "step": 57790
    },
    {
      "epoch": 0.398766445666347,
      "grad_norm": 2.4291327918035677e-06,
      "learning_rate": 6.0141982931692274e-06,
      "loss": 0.0,
      "step": 57800
    },
    {
      "epoch": 0.39883543640089136,
      "grad_norm": 1.2050320741252563e-08,
      "learning_rate": 6.013508385823784e-06,
      "loss": 0.0,
      "step": 57810
    },
    {
      "epoch": 0.3989044271354357,
      "grad_norm": 0.0,
      "learning_rate": 6.0128184784783404e-06,
      "loss": 0.0,
      "step": 57820
    },
    {
      "epoch": 0.3989734178699801,
      "grad_norm": 0.0,
      "learning_rate": 6.012128571132897e-06,
      "loss": 0.0,
      "step": 57830
    },
    {
      "epoch": 0.3990424086045244,
      "grad_norm": 280.42071533203125,
      "learning_rate": 6.011438663787453e-06,
      "loss": 0.0536,
      "step": 57840
    },
    {
      "epoch": 0.39911139933906875,
      "grad_norm": 0.0,
      "learning_rate": 6.010748756442011e-06,
      "loss": 0.0213,
      "step": 57850
    },
    {
      "epoch": 0.3991803900736131,
      "grad_norm": 0.0,
      "learning_rate": 6.010058849096567e-06,
      "loss": 0.0,
      "step": 57860
    },
    {
      "epoch": 0.3992493808081575,
      "grad_norm": 0.0,
      "learning_rate": 6.009368941751124e-06,
      "loss": 0.0,
      "step": 57870
    },
    {
      "epoch": 0.39931837154270183,
      "grad_norm": 0.0,
      "learning_rate": 6.00867903440568e-06,
      "loss": 0.0,
      "step": 57880
    },
    {
      "epoch": 0.39938736227724614,
      "grad_norm": 0.0,
      "learning_rate": 6.007989127060237e-06,
      "loss": 0.0,
      "step": 57890
    },
    {
      "epoch": 0.3994563530117905,
      "grad_norm": 0.0,
      "learning_rate": 6.007299219714793e-06,
      "loss": 0.0,
      "step": 57900
    },
    {
      "epoch": 0.39952534374633486,
      "grad_norm": 0.0,
      "learning_rate": 6.006609312369349e-06,
      "loss": 0.0,
      "step": 57910
    },
    {
      "epoch": 0.3995943344808792,
      "grad_norm": 0.13083402812480927,
      "learning_rate": 6.005919405023905e-06,
      "loss": 0.0,
      "step": 57920
    },
    {
      "epoch": 0.3996633252154236,
      "grad_norm": 0.0,
      "learning_rate": 6.005229497678462e-06,
      "loss": 0.0,
      "step": 57930
    },
    {
      "epoch": 0.39973231594996794,
      "grad_norm": 9.836826211540028e-05,
      "learning_rate": 6.004539590333018e-06,
      "loss": 0.0,
      "step": 57940
    },
    {
      "epoch": 0.39980130668451225,
      "grad_norm": 0.0,
      "learning_rate": 6.003849682987575e-06,
      "loss": 0.0,
      "step": 57950
    },
    {
      "epoch": 0.3998702974190566,
      "grad_norm": 0.0,
      "learning_rate": 6.003159775642132e-06,
      "loss": 0.0,
      "step": 57960
    },
    {
      "epoch": 0.39993928815360097,
      "grad_norm": 0.0,
      "learning_rate": 6.002469868296689e-06,
      "loss": 0.0,
      "step": 57970
    },
    {
      "epoch": 0.40000827888814533,
      "grad_norm": 0.0,
      "learning_rate": 6.001779960951245e-06,
      "loss": 0.0,
      "step": 57980
    },
    {
      "epoch": 0.4000772696226897,
      "grad_norm": 0.0,
      "learning_rate": 6.001090053605802e-06,
      "loss": 0.0,
      "step": 57990
    },
    {
      "epoch": 0.400146260357234,
      "grad_norm": 1.2928440810355823e-05,
      "learning_rate": 6.000400146260358e-06,
      "loss": 0.0,
      "step": 58000
    },
    {
      "epoch": 0.40021525109177836,
      "grad_norm": 0.0,
      "learning_rate": 5.999710238914915e-06,
      "loss": 0.0,
      "step": 58010
    },
    {
      "epoch": 0.4002842418263227,
      "grad_norm": 0.046858225017786026,
      "learning_rate": 5.99902033156947e-06,
      "loss": 0.0,
      "step": 58020
    },
    {
      "epoch": 0.4003532325608671,
      "grad_norm": 2.005464693866088e-06,
      "learning_rate": 5.998330424224027e-06,
      "loss": 0.0,
      "step": 58030
    },
    {
      "epoch": 0.40042222329541144,
      "grad_norm": 1.0044138321063656e-07,
      "learning_rate": 5.997640516878583e-06,
      "loss": 0.0,
      "step": 58040
    },
    {
      "epoch": 0.4004912140299558,
      "grad_norm": 0.00011691428517224267,
      "learning_rate": 5.99695060953314e-06,
      "loss": 0.0,
      "step": 58050
    },
    {
      "epoch": 0.4005602047645001,
      "grad_norm": 0.0,
      "learning_rate": 5.996260702187696e-06,
      "loss": 0.0,
      "step": 58060
    },
    {
      "epoch": 0.40062919549904447,
      "grad_norm": 0.0,
      "learning_rate": 5.9955707948422535e-06,
      "loss": 0.0,
      "step": 58070
    },
    {
      "epoch": 0.4006981862335888,
      "grad_norm": 4.244372121320339e-06,
      "learning_rate": 5.99488088749681e-06,
      "loss": 0.0,
      "step": 58080
    },
    {
      "epoch": 0.4007671769681332,
      "grad_norm": 0.0,
      "learning_rate": 5.9941909801513665e-06,
      "loss": 0.0,
      "step": 58090
    },
    {
      "epoch": 0.40083616770267755,
      "grad_norm": 0.0003876199189107865,
      "learning_rate": 5.993501072805923e-06,
      "loss": 0.0002,
      "step": 58100
    },
    {
      "epoch": 0.40090515843722185,
      "grad_norm": 0.003690573154017329,
      "learning_rate": 5.9928111654604795e-06,
      "loss": 0.0,
      "step": 58110
    },
    {
      "epoch": 0.4009741491717662,
      "grad_norm": 0.0,
      "learning_rate": 5.992121258115036e-06,
      "loss": 0.0,
      "step": 58120
    },
    {
      "epoch": 0.4010431399063106,
      "grad_norm": 0.008085214532911777,
      "learning_rate": 5.991431350769592e-06,
      "loss": 0.0,
      "step": 58130
    },
    {
      "epoch": 0.40111213064085494,
      "grad_norm": 0.0,
      "learning_rate": 5.990741443424148e-06,
      "loss": 0.002,
      "step": 58140
    },
    {
      "epoch": 0.4011811213753993,
      "grad_norm": 4.208536211081082e-06,
      "learning_rate": 5.990051536078705e-06,
      "loss": 0.0,
      "step": 58150
    },
    {
      "epoch": 0.40125011210994366,
      "grad_norm": 0.19480331242084503,
      "learning_rate": 5.989361628733261e-06,
      "loss": 0.0165,
      "step": 58160
    },
    {
      "epoch": 0.40131910284448796,
      "grad_norm": 4.850200525652326e-07,
      "learning_rate": 5.988671721387818e-06,
      "loss": 0.0,
      "step": 58170
    },
    {
      "epoch": 0.4013880935790323,
      "grad_norm": 0.0,
      "learning_rate": 5.987981814042375e-06,
      "loss": 0.0,
      "step": 58180
    },
    {
      "epoch": 0.4014570843135767,
      "grad_norm": 0.08359970152378082,
      "learning_rate": 5.987291906696931e-06,
      "loss": 0.0001,
      "step": 58190
    },
    {
      "epoch": 0.40152607504812104,
      "grad_norm": 6.887652830300794e-07,
      "learning_rate": 5.986601999351488e-06,
      "loss": 0.0,
      "step": 58200
    },
    {
      "epoch": 0.4015950657826654,
      "grad_norm": 0.0013147451682016253,
      "learning_rate": 5.985912092006044e-06,
      "loss": 0.0,
      "step": 58210
    },
    {
      "epoch": 0.4016640565172097,
      "grad_norm": 0.0,
      "learning_rate": 5.985222184660601e-06,
      "loss": 0.0,
      "step": 58220
    },
    {
      "epoch": 0.40173304725175407,
      "grad_norm": 0.0,
      "learning_rate": 5.984532277315157e-06,
      "loss": 0.0,
      "step": 58230
    },
    {
      "epoch": 0.40180203798629843,
      "grad_norm": 0.0,
      "learning_rate": 5.983842369969714e-06,
      "loss": 0.0,
      "step": 58240
    },
    {
      "epoch": 0.4018710287208428,
      "grad_norm": 0.0,
      "learning_rate": 5.9831524626242695e-06,
      "loss": 0.0,
      "step": 58250
    },
    {
      "epoch": 0.40194001945538715,
      "grad_norm": 0.0,
      "learning_rate": 5.982462555278826e-06,
      "loss": 0.0,
      "step": 58260
    },
    {
      "epoch": 0.4020090101899315,
      "grad_norm": 0.0,
      "learning_rate": 5.9817726479333825e-06,
      "loss": 0.0,
      "step": 58270
    },
    {
      "epoch": 0.4020780009244758,
      "grad_norm": 0.0,
      "learning_rate": 5.981082740587939e-06,
      "loss": 0.0,
      "step": 58280
    },
    {
      "epoch": 0.4021469916590202,
      "grad_norm": 0.0,
      "learning_rate": 5.980392833242496e-06,
      "loss": 0.0002,
      "step": 58290
    },
    {
      "epoch": 0.40221598239356454,
      "grad_norm": 0.0,
      "learning_rate": 5.979702925897053e-06,
      "loss": 0.0,
      "step": 58300
    },
    {
      "epoch": 0.4022849731281089,
      "grad_norm": 9.228948329109699e-06,
      "learning_rate": 5.979013018551609e-06,
      "loss": 0.0001,
      "step": 58310
    },
    {
      "epoch": 0.40235396386265326,
      "grad_norm": 0.0,
      "learning_rate": 5.978323111206166e-06,
      "loss": 0.0007,
      "step": 58320
    },
    {
      "epoch": 0.4024229545971976,
      "grad_norm": 0.5624728798866272,
      "learning_rate": 5.977633203860722e-06,
      "loss": 0.0005,
      "step": 58330
    },
    {
      "epoch": 0.40249194533174193,
      "grad_norm": 0.0,
      "learning_rate": 5.976943296515279e-06,
      "loss": 0.0,
      "step": 58340
    },
    {
      "epoch": 0.4025609360662863,
      "grad_norm": 0.0,
      "learning_rate": 5.976253389169835e-06,
      "loss": 0.0,
      "step": 58350
    },
    {
      "epoch": 0.40262992680083065,
      "grad_norm": 7.086819096002728e-05,
      "learning_rate": 5.975563481824391e-06,
      "loss": 0.0,
      "step": 58360
    },
    {
      "epoch": 0.402698917535375,
      "grad_norm": 0.0,
      "learning_rate": 5.974873574478947e-06,
      "loss": 0.0,
      "step": 58370
    },
    {
      "epoch": 0.4027679082699194,
      "grad_norm": 0.0,
      "learning_rate": 5.974183667133504e-06,
      "loss": 0.0,
      "step": 58380
    },
    {
      "epoch": 0.4028368990044637,
      "grad_norm": 0.0,
      "learning_rate": 5.97349375978806e-06,
      "loss": 0.0,
      "step": 58390
    },
    {
      "epoch": 0.40290588973900804,
      "grad_norm": 0.0003629494458436966,
      "learning_rate": 5.972803852442618e-06,
      "loss": 0.0001,
      "step": 58400
    },
    {
      "epoch": 0.4029748804735524,
      "grad_norm": 0.0,
      "learning_rate": 5.972113945097174e-06,
      "loss": 0.0,
      "step": 58410
    },
    {
      "epoch": 0.40304387120809676,
      "grad_norm": 0.0,
      "learning_rate": 5.971424037751731e-06,
      "loss": 0.0,
      "step": 58420
    },
    {
      "epoch": 0.4031128619426411,
      "grad_norm": 4.6831974032102153e-07,
      "learning_rate": 5.970734130406287e-06,
      "loss": 0.012,
      "step": 58430
    },
    {
      "epoch": 0.4031818526771855,
      "grad_norm": 0.3016344904899597,
      "learning_rate": 5.970044223060844e-06,
      "loss": 0.0001,
      "step": 58440
    },
    {
      "epoch": 0.4032508434117298,
      "grad_norm": 0.0,
      "learning_rate": 5.9693543157154e-06,
      "loss": 0.0,
      "step": 58450
    },
    {
      "epoch": 0.40331983414627415,
      "grad_norm": 0.0,
      "learning_rate": 5.968664408369957e-06,
      "loss": 0.0,
      "step": 58460
    },
    {
      "epoch": 0.4033888248808185,
      "grad_norm": 0.0,
      "learning_rate": 5.967974501024512e-06,
      "loss": 0.0,
      "step": 58470
    },
    {
      "epoch": 0.40345781561536287,
      "grad_norm": 0.0,
      "learning_rate": 5.967284593679069e-06,
      "loss": 0.0,
      "step": 58480
    },
    {
      "epoch": 0.40352680634990723,
      "grad_norm": 0.0,
      "learning_rate": 5.966594686333625e-06,
      "loss": 0.0,
      "step": 58490
    },
    {
      "epoch": 0.40359579708445154,
      "grad_norm": 0.0,
      "learning_rate": 5.965904778988182e-06,
      "loss": 0.0009,
      "step": 58500
    },
    {
      "epoch": 0.4036647878189959,
      "grad_norm": 0.0,
      "learning_rate": 5.965214871642739e-06,
      "loss": 0.0002,
      "step": 58510
    },
    {
      "epoch": 0.40373377855354026,
      "grad_norm": 0.0,
      "learning_rate": 5.964524964297296e-06,
      "loss": 0.0,
      "step": 58520
    },
    {
      "epoch": 0.4038027692880846,
      "grad_norm": 7.296505941667419e-07,
      "learning_rate": 5.963835056951852e-06,
      "loss": 0.0,
      "step": 58530
    },
    {
      "epoch": 0.403871760022629,
      "grad_norm": 0.0,
      "learning_rate": 5.963145149606409e-06,
      "loss": 0.0,
      "step": 58540
    },
    {
      "epoch": 0.40394075075717334,
      "grad_norm": 0.0,
      "learning_rate": 5.962455242260965e-06,
      "loss": 0.0,
      "step": 58550
    },
    {
      "epoch": 0.40400974149171764,
      "grad_norm": 0.0,
      "learning_rate": 5.9617653349155216e-06,
      "loss": 0.0002,
      "step": 58560
    },
    {
      "epoch": 0.404078732226262,
      "grad_norm": 0.0,
      "learning_rate": 5.961075427570078e-06,
      "loss": 0.0,
      "step": 58570
    },
    {
      "epoch": 0.40414772296080637,
      "grad_norm": 2.373796269239392e-05,
      "learning_rate": 5.9603855202246345e-06,
      "loss": 0.0111,
      "step": 58580
    },
    {
      "epoch": 0.4042167136953507,
      "grad_norm": 0.0,
      "learning_rate": 5.95969561287919e-06,
      "loss": 0.0,
      "step": 58590
    },
    {
      "epoch": 0.4042857044298951,
      "grad_norm": 0.0,
      "learning_rate": 5.959005705533747e-06,
      "loss": 0.0,
      "step": 58600
    },
    {
      "epoch": 0.4043546951644394,
      "grad_norm": 6.994843261054484e-06,
      "learning_rate": 5.958315798188303e-06,
      "loss": 0.0,
      "step": 58610
    },
    {
      "epoch": 0.40442368589898375,
      "grad_norm": 0.0,
      "learning_rate": 5.9576258908428605e-06,
      "loss": 0.0,
      "step": 58620
    },
    {
      "epoch": 0.4044926766335281,
      "grad_norm": 0.0,
      "learning_rate": 5.956935983497417e-06,
      "loss": 0.0,
      "step": 58630
    },
    {
      "epoch": 0.4045616673680725,
      "grad_norm": 0.0,
      "learning_rate": 5.9562460761519735e-06,
      "loss": 0.0,
      "step": 58640
    },
    {
      "epoch": 0.40463065810261684,
      "grad_norm": 0.0,
      "learning_rate": 5.95555616880653e-06,
      "loss": 0.0,
      "step": 58650
    },
    {
      "epoch": 0.4046996488371612,
      "grad_norm": 0.0,
      "learning_rate": 5.9548662614610865e-06,
      "loss": 0.0,
      "step": 58660
    },
    {
      "epoch": 0.4047686395717055,
      "grad_norm": 2.2283454248483991e-10,
      "learning_rate": 5.954176354115643e-06,
      "loss": 0.0,
      "step": 58670
    },
    {
      "epoch": 0.40483763030624986,
      "grad_norm": 0.0,
      "learning_rate": 5.9534864467701994e-06,
      "loss": 0.0,
      "step": 58680
    },
    {
      "epoch": 0.4049066210407942,
      "grad_norm": 0.0,
      "learning_rate": 5.952796539424756e-06,
      "loss": 0.0003,
      "step": 58690
    },
    {
      "epoch": 0.4049756117753386,
      "grad_norm": 0.0,
      "learning_rate": 5.952106632079312e-06,
      "loss": 0.0,
      "step": 58700
    },
    {
      "epoch": 0.40504460250988295,
      "grad_norm": 0.0,
      "learning_rate": 5.951416724733868e-06,
      "loss": 0.0,
      "step": 58710
    },
    {
      "epoch": 0.40511359324442725,
      "grad_norm": 0.0,
      "learning_rate": 5.9507268173884246e-06,
      "loss": 0.0025,
      "step": 58720
    },
    {
      "epoch": 0.4051825839789716,
      "grad_norm": 0.0,
      "learning_rate": 5.950036910042982e-06,
      "loss": 0.0003,
      "step": 58730
    },
    {
      "epoch": 0.40525157471351597,
      "grad_norm": 0.0,
      "learning_rate": 5.949347002697538e-06,
      "loss": 0.0,
      "step": 58740
    },
    {
      "epoch": 0.40532056544806033,
      "grad_norm": 0.0,
      "learning_rate": 5.948657095352095e-06,
      "loss": 0.0007,
      "step": 58750
    },
    {
      "epoch": 0.4053895561826047,
      "grad_norm": 4.398719610776425e-08,
      "learning_rate": 5.947967188006651e-06,
      "loss": 0.0259,
      "step": 58760
    },
    {
      "epoch": 0.40545854691714905,
      "grad_norm": 0.0,
      "learning_rate": 5.947277280661208e-06,
      "loss": 0.0129,
      "step": 58770
    },
    {
      "epoch": 0.40552753765169336,
      "grad_norm": 0.0,
      "learning_rate": 5.946587373315764e-06,
      "loss": 0.0,
      "step": 58780
    },
    {
      "epoch": 0.4055965283862377,
      "grad_norm": 0.0,
      "learning_rate": 5.945897465970321e-06,
      "loss": 0.0001,
      "step": 58790
    },
    {
      "epoch": 0.4056655191207821,
      "grad_norm": 0.577299952507019,
      "learning_rate": 5.945207558624877e-06,
      "loss": 0.0001,
      "step": 58800
    },
    {
      "epoch": 0.40573450985532644,
      "grad_norm": 0.0,
      "learning_rate": 5.944586642013977e-06,
      "loss": 0.1159,
      "step": 58810
    },
    {
      "epoch": 0.4058035005898708,
      "grad_norm": 0.0,
      "learning_rate": 5.943896734668535e-06,
      "loss": 0.0002,
      "step": 58820
    },
    {
      "epoch": 0.4058724913244151,
      "grad_norm": 6.334661483764648,
      "learning_rate": 5.943206827323091e-06,
      "loss": 0.0011,
      "step": 58830
    },
    {
      "epoch": 0.40594148205895947,
      "grad_norm": 0.0,
      "learning_rate": 5.9425169199776475e-06,
      "loss": 0.0,
      "step": 58840
    },
    {
      "epoch": 0.40601047279350383,
      "grad_norm": 0.0,
      "learning_rate": 5.941827012632204e-06,
      "loss": 0.0,
      "step": 58850
    },
    {
      "epoch": 0.4060794635280482,
      "grad_norm": 0.0,
      "learning_rate": 5.9411371052867605e-06,
      "loss": 0.0021,
      "step": 58860
    },
    {
      "epoch": 0.40614845426259255,
      "grad_norm": 4.860330093947596e-08,
      "learning_rate": 5.940447197941317e-06,
      "loss": 0.0,
      "step": 58870
    },
    {
      "epoch": 0.4062174449971369,
      "grad_norm": 229.7114715576172,
      "learning_rate": 5.9397572905958734e-06,
      "loss": 0.048,
      "step": 58880
    },
    {
      "epoch": 0.4062864357316812,
      "grad_norm": 5.710251116397558e-08,
      "learning_rate": 5.93906738325043e-06,
      "loss": 0.0015,
      "step": 58890
    },
    {
      "epoch": 0.4063554264662256,
      "grad_norm": 0.0,
      "learning_rate": 5.938377475904986e-06,
      "loss": 0.0,
      "step": 58900
    },
    {
      "epoch": 0.40642441720076994,
      "grad_norm": 0.0,
      "learning_rate": 5.937687568559543e-06,
      "loss": 0.0,
      "step": 58910
    },
    {
      "epoch": 0.4064934079353143,
      "grad_norm": 0.0,
      "learning_rate": 5.9369976612140985e-06,
      "loss": 0.0693,
      "step": 58920
    },
    {
      "epoch": 0.40656239866985866,
      "grad_norm": 4.417508581155971e-10,
      "learning_rate": 5.936307753868657e-06,
      "loss": 0.0,
      "step": 58930
    },
    {
      "epoch": 0.40663138940440297,
      "grad_norm": 0.0,
      "learning_rate": 5.935617846523213e-06,
      "loss": 0.0,
      "step": 58940
    },
    {
      "epoch": 0.4067003801389473,
      "grad_norm": 0.0,
      "learning_rate": 5.934927939177769e-06,
      "loss": 0.0003,
      "step": 58950
    },
    {
      "epoch": 0.4067693708734917,
      "grad_norm": 0.0,
      "learning_rate": 5.934238031832325e-06,
      "loss": 0.0875,
      "step": 58960
    },
    {
      "epoch": 0.40683836160803605,
      "grad_norm": 0.0,
      "learning_rate": 5.933548124486882e-06,
      "loss": 0.0,
      "step": 58970
    },
    {
      "epoch": 0.4069073523425804,
      "grad_norm": 0.0,
      "learning_rate": 5.932858217141438e-06,
      "loss": 0.0009,
      "step": 58980
    },
    {
      "epoch": 0.40697634307712477,
      "grad_norm": 0.0,
      "learning_rate": 5.932168309795995e-06,
      "loss": 0.0,
      "step": 58990
    },
    {
      "epoch": 0.4070453338116691,
      "grad_norm": 0.0,
      "learning_rate": 5.931478402450551e-06,
      "loss": 0.0,
      "step": 59000
    },
    {
      "epoch": 0.40711432454621344,
      "grad_norm": 0.0,
      "learning_rate": 5.930788495105108e-06,
      "loss": 0.0,
      "step": 59010
    },
    {
      "epoch": 0.4071833152807578,
      "grad_norm": 0.0,
      "learning_rate": 5.930098587759664e-06,
      "loss": 0.1049,
      "step": 59020
    },
    {
      "epoch": 0.40725230601530216,
      "grad_norm": 8.108017921447754,
      "learning_rate": 5.92940868041422e-06,
      "loss": 0.0011,
      "step": 59030
    },
    {
      "epoch": 0.4073212967498465,
      "grad_norm": 0.0,
      "learning_rate": 5.928718773068778e-06,
      "loss": 0.0,
      "step": 59040
    },
    {
      "epoch": 0.4073902874843908,
      "grad_norm": 2.02359977663491e-08,
      "learning_rate": 5.928028865723335e-06,
      "loss": 0.2521,
      "step": 59050
    },
    {
      "epoch": 0.4074592782189352,
      "grad_norm": 2.163207000194234e-06,
      "learning_rate": 5.92733895837789e-06,
      "loss": 0.0387,
      "step": 59060
    },
    {
      "epoch": 0.40752826895347954,
      "grad_norm": 0.0,
      "learning_rate": 5.926649051032447e-06,
      "loss": 0.0147,
      "step": 59070
    },
    {
      "epoch": 0.4075972596880239,
      "grad_norm": 0.007901228964328766,
      "learning_rate": 5.925959143687003e-06,
      "loss": 0.0,
      "step": 59080
    },
    {
      "epoch": 0.40766625042256827,
      "grad_norm": 0.0,
      "learning_rate": 5.92526923634156e-06,
      "loss": 0.0,
      "step": 59090
    },
    {
      "epoch": 0.4077352411571126,
      "grad_norm": 0.0,
      "learning_rate": 5.924579328996116e-06,
      "loss": 0.0048,
      "step": 59100
    },
    {
      "epoch": 0.40780423189165693,
      "grad_norm": 0.0,
      "learning_rate": 5.923889421650673e-06,
      "loss": 0.0,
      "step": 59110
    },
    {
      "epoch": 0.4078732226262013,
      "grad_norm": 0.0,
      "learning_rate": 5.923199514305229e-06,
      "loss": 0.007,
      "step": 59120
    },
    {
      "epoch": 0.40794221336074565,
      "grad_norm": 0.0,
      "learning_rate": 5.922509606959786e-06,
      "loss": 0.0,
      "step": 59130
    },
    {
      "epoch": 0.40801120409529,
      "grad_norm": 0.0,
      "learning_rate": 5.921819699614342e-06,
      "loss": 0.0005,
      "step": 59140
    },
    {
      "epoch": 0.4080801948298344,
      "grad_norm": 0.0,
      "learning_rate": 5.9211297922688995e-06,
      "loss": 0.0,
      "step": 59150
    },
    {
      "epoch": 0.4081491855643787,
      "grad_norm": 0.0,
      "learning_rate": 5.920439884923456e-06,
      "loss": 0.0,
      "step": 59160
    },
    {
      "epoch": 0.40821817629892304,
      "grad_norm": 4.628163408071373e-10,
      "learning_rate": 5.919749977578012e-06,
      "loss": 0.0,
      "step": 59170
    },
    {
      "epoch": 0.4082871670334674,
      "grad_norm": 0.0007107885321602225,
      "learning_rate": 5.919060070232568e-06,
      "loss": 0.0,
      "step": 59180
    },
    {
      "epoch": 0.40835615776801176,
      "grad_norm": 0.0,
      "learning_rate": 5.918370162887125e-06,
      "loss": 0.0,
      "step": 59190
    },
    {
      "epoch": 0.4084251485025561,
      "grad_norm": 0.0,
      "learning_rate": 5.917680255541681e-06,
      "loss": 0.0001,
      "step": 59200
    },
    {
      "epoch": 0.4084941392371005,
      "grad_norm": 0.0,
      "learning_rate": 5.916990348196238e-06,
      "loss": 0.0,
      "step": 59210
    },
    {
      "epoch": 0.4085631299716448,
      "grad_norm": 0.0,
      "learning_rate": 5.916300440850794e-06,
      "loss": 0.0002,
      "step": 59220
    },
    {
      "epoch": 0.40863212070618915,
      "grad_norm": 8.786363303592282e-10,
      "learning_rate": 5.915610533505351e-06,
      "loss": 0.0,
      "step": 59230
    },
    {
      "epoch": 0.4087011114407335,
      "grad_norm": 0.0,
      "learning_rate": 5.914920626159907e-06,
      "loss": 0.0,
      "step": 59240
    },
    {
      "epoch": 0.40877010217527787,
      "grad_norm": 0.00012982876796741039,
      "learning_rate": 5.9142307188144636e-06,
      "loss": 0.0001,
      "step": 59250
    },
    {
      "epoch": 0.40883909290982223,
      "grad_norm": 0.0,
      "learning_rate": 5.913540811469021e-06,
      "loss": 0.0,
      "step": 59260
    },
    {
      "epoch": 0.40890808364436654,
      "grad_norm": 0.0,
      "learning_rate": 5.912850904123577e-06,
      "loss": 0.0,
      "step": 59270
    },
    {
      "epoch": 0.4089770743789109,
      "grad_norm": 0.0,
      "learning_rate": 5.912160996778134e-06,
      "loss": 0.0,
      "step": 59280
    },
    {
      "epoch": 0.40904606511345526,
      "grad_norm": 0.0,
      "learning_rate": 5.9114710894326895e-06,
      "loss": 0.0002,
      "step": 59290
    },
    {
      "epoch": 0.4091150558479996,
      "grad_norm": 0.00011850119335576892,
      "learning_rate": 5.910781182087246e-06,
      "loss": 0.0032,
      "step": 59300
    },
    {
      "epoch": 0.409184046582544,
      "grad_norm": 0.0,
      "learning_rate": 5.9100912747418025e-06,
      "loss": 0.0,
      "step": 59310
    },
    {
      "epoch": 0.40925303731708834,
      "grad_norm": 0.0,
      "learning_rate": 5.909401367396359e-06,
      "loss": 0.0,
      "step": 59320
    },
    {
      "epoch": 0.40932202805163265,
      "grad_norm": 0.0003999821492470801,
      "learning_rate": 5.9087114600509155e-06,
      "loss": 0.0,
      "step": 59330
    },
    {
      "epoch": 0.409391018786177,
      "grad_norm": 0.0,
      "learning_rate": 5.908021552705472e-06,
      "loss": 0.0088,
      "step": 59340
    },
    {
      "epoch": 0.40946000952072137,
      "grad_norm": 0.0,
      "learning_rate": 5.9073316453600285e-06,
      "loss": 0.0,
      "step": 59350
    },
    {
      "epoch": 0.40952900025526573,
      "grad_norm": 9.623555342841428e-06,
      "learning_rate": 5.906641738014585e-06,
      "loss": 0.0,
      "step": 59360
    },
    {
      "epoch": 0.4095979909898101,
      "grad_norm": 0.0,
      "learning_rate": 5.905951830669142e-06,
      "loss": 0.0,
      "step": 59370
    },
    {
      "epoch": 0.4096669817243544,
      "grad_norm": 0.0,
      "learning_rate": 5.905261923323699e-06,
      "loss": 0.0,
      "step": 59380
    },
    {
      "epoch": 0.40973597245889876,
      "grad_norm": 0.0,
      "learning_rate": 5.904572015978255e-06,
      "loss": 0.0,
      "step": 59390
    },
    {
      "epoch": 0.4098049631934431,
      "grad_norm": 0.0,
      "learning_rate": 5.903882108632811e-06,
      "loss": 0.0051,
      "step": 59400
    },
    {
      "epoch": 0.4098739539279875,
      "grad_norm": 2.372068905742708e-07,
      "learning_rate": 5.903192201287367e-06,
      "loss": 0.0,
      "step": 59410
    },
    {
      "epoch": 0.40994294466253184,
      "grad_norm": 0.4272094964981079,
      "learning_rate": 5.902502293941924e-06,
      "loss": 0.0001,
      "step": 59420
    },
    {
      "epoch": 0.4100119353970762,
      "grad_norm": 0.0,
      "learning_rate": 5.90181238659648e-06,
      "loss": 0.0,
      "step": 59430
    },
    {
      "epoch": 0.4100809261316205,
      "grad_norm": 0.0,
      "learning_rate": 5.901122479251037e-06,
      "loss": 0.0002,
      "step": 59440
    },
    {
      "epoch": 0.41014991686616487,
      "grad_norm": 0.0,
      "learning_rate": 5.900432571905593e-06,
      "loss": 0.0,
      "step": 59450
    },
    {
      "epoch": 0.4102189076007092,
      "grad_norm": 0.0,
      "learning_rate": 5.89974266456015e-06,
      "loss": 0.0006,
      "step": 59460
    },
    {
      "epoch": 0.4102878983352536,
      "grad_norm": 4.763747174507671e-10,
      "learning_rate": 5.899052757214706e-06,
      "loss": 0.482,
      "step": 59470
    },
    {
      "epoch": 0.41035688906979795,
      "grad_norm": 0.0,
      "learning_rate": 5.898362849869264e-06,
      "loss": 0.1172,
      "step": 59480
    },
    {
      "epoch": 0.41042587980434225,
      "grad_norm": 0.002104072831571102,
      "learning_rate": 5.89767294252382e-06,
      "loss": 0.001,
      "step": 59490
    },
    {
      "epoch": 0.4104948705388866,
      "grad_norm": 0.0,
      "learning_rate": 5.896983035178377e-06,
      "loss": 0.0,
      "step": 59500
    },
    {
      "epoch": 0.410563861273431,
      "grad_norm": 0.0,
      "learning_rate": 5.896293127832932e-06,
      "loss": 0.0,
      "step": 59510
    },
    {
      "epoch": 0.41063285200797534,
      "grad_norm": 0.0,
      "learning_rate": 5.895603220487489e-06,
      "loss": 0.0,
      "step": 59520
    },
    {
      "epoch": 0.4107018427425197,
      "grad_norm": 0.0,
      "learning_rate": 5.894913313142045e-06,
      "loss": 0.0,
      "step": 59530
    },
    {
      "epoch": 0.41077083347706406,
      "grad_norm": 0.0,
      "learning_rate": 5.894223405796602e-06,
      "loss": 0.0,
      "step": 59540
    },
    {
      "epoch": 0.41083982421160836,
      "grad_norm": 0.0,
      "learning_rate": 5.893533498451158e-06,
      "loss": 0.0,
      "step": 59550
    },
    {
      "epoch": 0.4109088149461527,
      "grad_norm": 0.0,
      "learning_rate": 5.892843591105715e-06,
      "loss": 0.0,
      "step": 59560
    },
    {
      "epoch": 0.4109778056806971,
      "grad_norm": 0.0,
      "learning_rate": 5.892153683760271e-06,
      "loss": 0.0003,
      "step": 59570
    },
    {
      "epoch": 0.41104679641524144,
      "grad_norm": 0.0,
      "learning_rate": 5.891463776414828e-06,
      "loss": 0.0,
      "step": 59580
    },
    {
      "epoch": 0.4111157871497858,
      "grad_norm": 0.0,
      "learning_rate": 5.890773869069385e-06,
      "loss": 0.0,
      "step": 59590
    },
    {
      "epoch": 0.4111847778843301,
      "grad_norm": 0.00011904434359166771,
      "learning_rate": 5.890083961723942e-06,
      "loss": 0.0,
      "step": 59600
    },
    {
      "epoch": 0.41125376861887447,
      "grad_norm": 0.0,
      "learning_rate": 5.889394054378498e-06,
      "loss": 0.0,
      "step": 59610
    },
    {
      "epoch": 0.41132275935341883,
      "grad_norm": 0.0,
      "learning_rate": 5.8887041470330546e-06,
      "loss": 0.0,
      "step": 59620
    },
    {
      "epoch": 0.4113917500879632,
      "grad_norm": 0.0,
      "learning_rate": 5.88801423968761e-06,
      "loss": 0.0,
      "step": 59630
    },
    {
      "epoch": 0.41146074082250755,
      "grad_norm": 9.936236165231094e-05,
      "learning_rate": 5.887324332342167e-06,
      "loss": 0.0,
      "step": 59640
    },
    {
      "epoch": 0.4115297315570519,
      "grad_norm": 2.134287058197515e-08,
      "learning_rate": 5.886634424996723e-06,
      "loss": 0.0,
      "step": 59650
    },
    {
      "epoch": 0.4115987222915962,
      "grad_norm": 0.0,
      "learning_rate": 5.88594451765128e-06,
      "loss": 0.0,
      "step": 59660
    },
    {
      "epoch": 0.4116677130261406,
      "grad_norm": 0.0,
      "learning_rate": 5.885254610305836e-06,
      "loss": 0.0,
      "step": 59670
    },
    {
      "epoch": 0.41173670376068494,
      "grad_norm": 0.0,
      "learning_rate": 5.884564702960393e-06,
      "loss": 0.0001,
      "step": 59680
    },
    {
      "epoch": 0.4118056944952293,
      "grad_norm": 0.0,
      "learning_rate": 5.883874795614949e-06,
      "loss": 0.0,
      "step": 59690
    },
    {
      "epoch": 0.41187468522977366,
      "grad_norm": 0.0,
      "learning_rate": 5.8831848882695065e-06,
      "loss": 0.0,
      "step": 59700
    },
    {
      "epoch": 0.41194367596431797,
      "grad_norm": 0.0,
      "learning_rate": 5.882494980924063e-06,
      "loss": 0.0003,
      "step": 59710
    },
    {
      "epoch": 0.41201266669886233,
      "grad_norm": 0.0,
      "learning_rate": 5.8818050735786195e-06,
      "loss": 0.0,
      "step": 59720
    },
    {
      "epoch": 0.4120816574334067,
      "grad_norm": 0.0,
      "learning_rate": 5.881115166233176e-06,
      "loss": 0.0,
      "step": 59730
    },
    {
      "epoch": 0.41215064816795105,
      "grad_norm": 0.0,
      "learning_rate": 5.880425258887732e-06,
      "loss": 0.0,
      "step": 59740
    },
    {
      "epoch": 0.4122196389024954,
      "grad_norm": 0.0,
      "learning_rate": 5.879735351542288e-06,
      "loss": 0.0,
      "step": 59750
    },
    {
      "epoch": 0.41228862963703977,
      "grad_norm": 675.55224609375,
      "learning_rate": 5.879045444196845e-06,
      "loss": 0.5651,
      "step": 59760
    },
    {
      "epoch": 0.4123576203715841,
      "grad_norm": 0.0,
      "learning_rate": 5.878355536851401e-06,
      "loss": 0.0313,
      "step": 59770
    },
    {
      "epoch": 0.41242661110612844,
      "grad_norm": 0.0,
      "learning_rate": 5.8776656295059576e-06,
      "loss": 0.0,
      "step": 59780
    },
    {
      "epoch": 0.4124956018406728,
      "grad_norm": 0.0,
      "learning_rate": 5.876975722160514e-06,
      "loss": 0.0018,
      "step": 59790
    },
    {
      "epoch": 0.41256459257521716,
      "grad_norm": 0.0,
      "learning_rate": 5.8762858148150705e-06,
      "loss": 0.0,
      "step": 59800
    },
    {
      "epoch": 0.4126335833097615,
      "grad_norm": 0.0,
      "learning_rate": 5.875595907469628e-06,
      "loss": 0.0,
      "step": 59810
    },
    {
      "epoch": 0.4127025740443058,
      "grad_norm": 3.3830306165327784e-06,
      "learning_rate": 5.874906000124184e-06,
      "loss": 0.0,
      "step": 59820
    },
    {
      "epoch": 0.4127715647788502,
      "grad_norm": 0.0,
      "learning_rate": 5.874216092778741e-06,
      "loss": 0.0,
      "step": 59830
    },
    {
      "epoch": 0.41284055551339455,
      "grad_norm": 0.0,
      "learning_rate": 5.873526185433297e-06,
      "loss": 0.0,
      "step": 59840
    },
    {
      "epoch": 0.4129095462479389,
      "grad_norm": 0.0,
      "learning_rate": 5.872836278087853e-06,
      "loss": 0.0153,
      "step": 59850
    },
    {
      "epoch": 0.41297853698248327,
      "grad_norm": 0.0,
      "learning_rate": 5.8721463707424095e-06,
      "loss": 0.0035,
      "step": 59860
    },
    {
      "epoch": 0.41304752771702763,
      "grad_norm": 8.570363263515901e-08,
      "learning_rate": 5.871456463396966e-06,
      "loss": 0.0,
      "step": 59870
    },
    {
      "epoch": 0.41311651845157193,
      "grad_norm": 0.0,
      "learning_rate": 5.8707665560515225e-06,
      "loss": 0.2245,
      "step": 59880
    },
    {
      "epoch": 0.4131855091861163,
      "grad_norm": 0.0,
      "learning_rate": 5.870076648706079e-06,
      "loss": 0.0,
      "step": 59890
    },
    {
      "epoch": 0.41325449992066066,
      "grad_norm": 0.0,
      "learning_rate": 5.8693867413606354e-06,
      "loss": 0.0,
      "step": 59900
    },
    {
      "epoch": 0.413323490655205,
      "grad_norm": 0.0,
      "learning_rate": 5.868696834015192e-06,
      "loss": 0.0001,
      "step": 59910
    },
    {
      "epoch": 0.4133924813897494,
      "grad_norm": 0.0014781218487769365,
      "learning_rate": 5.868006926669749e-06,
      "loss": 0.0,
      "step": 59920
    },
    {
      "epoch": 0.4134614721242937,
      "grad_norm": 0.0,
      "learning_rate": 5.867317019324306e-06,
      "loss": 0.0,
      "step": 59930
    },
    {
      "epoch": 0.41353046285883804,
      "grad_norm": 0.0,
      "learning_rate": 5.866627111978862e-06,
      "loss": 0.0002,
      "step": 59940
    },
    {
      "epoch": 0.4135994535933824,
      "grad_norm": 0.0,
      "learning_rate": 5.865937204633419e-06,
      "loss": 0.0,
      "step": 59950
    },
    {
      "epoch": 0.41366844432792677,
      "grad_norm": 0.0,
      "learning_rate": 5.865247297287975e-06,
      "loss": 0.0001,
      "step": 59960
    },
    {
      "epoch": 0.4137374350624711,
      "grad_norm": 0.0,
      "learning_rate": 5.864557389942531e-06,
      "loss": 0.0006,
      "step": 59970
    },
    {
      "epoch": 0.4138064257970155,
      "grad_norm": 45.626197814941406,
      "learning_rate": 5.863867482597087e-06,
      "loss": 0.0077,
      "step": 59980
    },
    {
      "epoch": 0.4138754165315598,
      "grad_norm": 102.7043228149414,
      "learning_rate": 5.863177575251644e-06,
      "loss": 0.0149,
      "step": 59990
    },
    {
      "epoch": 0.41394440726610415,
      "grad_norm": 0.0,
      "learning_rate": 5.8624876679062e-06,
      "loss": 0.0001,
      "step": 60000
    }
  ],
  "logging_steps": 10,
  "max_steps": 144947,
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}